@stabrise/scaledp 0.1.0 → 0.1.1

package/dist/ocr-Dkkbk0Wl.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ocr-Dkkbk0Wl.js","names":["boxesOf","boxesOf"],"sources":["../src/ocr/db-postprocess.ts","../src/ocr/dbnet-onnx.ts","../src/ocr/detector-registry.ts","../src/ocr/line-orientation.ts","../src/ocr/line-orientation-stage.ts","../src/ocr/presets.ts","../src/ocr/paddle-service.ts","../src/ocr/paddle.ts","../src/ocr/paddle-recognizer.ts","../src/ocr/script-detect.ts","../src/ocr/tesseract.ts","../src/ocr/tesseract-recognizer.ts"],"sourcesContent":["/**\n * Differentiable Binarization post-processing.\n *\n * Port of `scaledp/models/detectors/paddle_onnx/db_postprocess.py`. Takes the\n * model's probability map and turns it into text quads, with the same\n * thresholds, ordering and filtering as Python so box output matches.\n */\n\nimport { boxPoints, minAreaRect, type Point, polygonArea, polygonPerimeter } from '../core/geometry.js'\n\nexport interface DbPostProcessOptions {\n    /** Probability above which a pixel counts as text. */\n    thresh: number\n    /** Mean in-box probability a candidate must reach to survive. */\n    boxThresh: number\n    /** Cap on contours considered, guarding against pathological maps. */\n    maxCandidates: number\n    /** How far to grow each box; DB shrinks regions during training. */\n    unclipRatio: number\n    /** Minimum short side, in model pixels. */\n    minSize: number\n}\n\n/**\n * ScaleDP's own values, from `predict_det.py`. Note `boxThresh` is 0.3 there\n * and the stage-level `scoreThreshold` never reaches this code -- a Python bug\n * that is not reproduced: here `boxThresh` is what the stage actually sets.\n */\nexport const DB_POSTPROCESS_DEFAULTS: DbPostProcessOptions = Object.freeze({\n    thresh: 0.5,\n    boxThresh: 0.3,\n    maxCandidates: 1000,\n    unclipRatio: 2.5,\n    minSize: 3,\n})\n\nexport interface ProbabilityMap {\n    data: Float32Array\n    width: number\n    height: number\n}\n\nexport interface DetectedQuad {\n    /** Four corners, ordered top-left, top-right, bottom-right, bottom-left. */\n    points: [Point, Point, Point, Point]\n    score: number\n}\n\n/**\n * Boundary pixels of each 8-connected foreground component.\n *\n * cv2.findContours traces outlines; we collect boundary pixels instead. For the\n * only consumers here -- minAreaRect and its convex hull -- the two are\n * equivalent, and this avoids porting Suzuki-Abe border following.\n */\nexport function findComponentBoundaries(\n    map: ProbabilityMap,\n    thresh: number,\n    maxCandidates: number\n): Point[][] {\n    const { data, width, height } = map\n    const visited = new Uint8Array(width * height)\n    const components: Point[][] = []\n\n    const isForeground = (x: number, y: number): boolean =>\n        x >= 0 && y >= 0 && x < width && y < height && (data[y * width + x] as number) > thresh\n\n    const stack: number[] = []\n    for (let start = 0; start < visited.length && components.length < maxCandidates; start++) {\n        if (visited[start] || !((data[start] as number) > thresh)) continue\n\n        const boundary: Point[] = []\n        stack.length = 0\n        stack.push(start)\n        visited[start] = 1\n\n        while (stack.length > 0) {\n            const index = stack.pop() as number\n            const x = index % width\n            const y = (index - x) / width\n\n            let onBoundary = false\n            for (let dy = -1; dy <= 1; dy++) {\n                for (let dx = -1; dx <= 1; dx++) {\n                    if (dx === 0 && dy === 0) continue\n                    const nx = x + dx\n                    const ny = y + dy\n                    if (!isForeground(nx, ny)) {\n                        onBoundary = true\n                        continue\n                    }\n                    const neighbour = ny * width + nx\n                    if (!visited[neighbour]) {\n                        visited[neighbour] = 1\n                        stack.push(neighbour)\n                    }\n                }\n            }\n            if (onBoundary) boundary.push([x, y])\n        }\n        if (boundary.length >= 3) components.push(boundary)\n    }\n    return components\n}\n\n/**\n * Corners of the minimum-area rect, ordered top-left, top-right, bottom-right,\n * bottom-left, together with the rect's shorter side.\n *\n * Port of `get_mini_boxes`: sort the four corners by x, then decide within each\n * pair which is the upper one.\n */\nexport function miniBox(points: readonly Point[]): { points: [Point, Point, Point, Point]; sside: number } {\n    const rect = minAreaRect(points)\n    const corners = [...boxPoints(rect)].sort((a, b) => a[0] - b[0])\n\n    const [p0, p1, p2, p3] = corners as [Point, Point, Point, Point]\n    const [topLeft, bottomLeft] = p0[1] <= p1[1] ? [p0, p1] : [p1, p0]\n    const [topRight, bottomRight] = p2[1] <= p3[1] ? [p2, p3] : [p3, p2]\n\n    return {\n        points: [topLeft, topRight, bottomRight, bottomLeft],\n        sside: Math.min(rect.size[0], rect.size[1]),\n    }\n}\n\n/**\n * Mean probability inside a quad -- port of `box_score_fast`.\n *\n * Python builds an integer mask with `cv2.fillPoly` and averages the\n * probability map under it, so this reproduces OpenCV's fill convention:\n * vertices are truncated to integers and treated as pixel *centres*, scanlines\n * run at integer y, and both ends of each span are inclusive. Sampling at\n * pixel centres instead (the more usual rasterisation rule) drops the boundary\n * row and column, which shifts the score by around 1% and changes which\n * candidates clear `boxThresh`.\n */\nexport function boxScore(map: ProbabilityMap, points: readonly Point[]): number {\n    const { data, width, height } = map\n    const xs = points.map((p) => p[0])\n    const ys = points.map((p) => p[1])\n\n    const clamp = (value: number, max: number) => Math.min(Math.max(value, 0), max)\n    const xmin = clamp(Math.floor(Math.min(...xs)), width - 1)\n    const xmax = clamp(Math.ceil(Math.max(...xs)), width - 1)\n    const ymin = clamp(Math.floor(Math.min(...ys)), height - 1)\n    const ymax = clamp(Math.ceil(Math.max(...ys)), height - 1)\n    if (xmax < xmin || ymax < ymin) return 0\n\n    // Translate into mask space and truncate, matching numpy's astype(int32).\n    const local = points.map(([x, y]) => [Math.trunc(x - xmin), Math.trunc(y - ymin)] as Point)\n    const maskWidth = xmax - xmin\n    const maskHeight = ymax - ymin\n\n    let sum = 0\n    let count = 0\n    for (let y = 0; y <= maskHeight; y++) {\n        let left = Number.POSITIVE_INFINITY\n        let right = Number.NEGATIVE_INFINITY\n\n        for (let i = 0; i < local.length; i++) {\n            const a = local[i] as Point\n            const b = local[(i + 1) % local.length] as Point\n            if (a[1] === b[1]) {\n                // A horizontal edge contributes both endpoints on its own row.\n                if (a[1] !== y) continue\n                left = Math.min(left, a[0], b[0])\n                right = Math.max(right, a[0], b[0])\n                continue\n            }\n            if (y < Math.min(a[1], b[1]) || y > Math.max(a[1], b[1])) continue\n            const x = a[0] + ((y - a[1]) / (b[1] - a[1])) * (b[0] - a[0])\n            left = Math.min(left, x)\n            right = Math.max(right, x)\n        }\n        if (left > right) continue\n\n        // The quads here are always convex, so the span between the extreme\n        // crossings is exactly the covered run.\n        const from = Math.max(0, Math.ceil(left))\n        const to = Math.min(maskWidth, Math.floor(right))\n        const rowOffset = (y + ymin) * width + xmin\n        for (let x = from; x <= to; x++) {\n            sum += data[rowOffset + x] as number\n            count++\n        }\n    }\n    return count === 0 ? 0 : sum / count\n}\n\n/**\n * Grow a detected rect outward -- port of `unclip`.\n *\n * Python offsets the polygon with a Clipper round join and then takes the\n * minimum-area rect of the result. For a rectangle those two steps compose\n * exactly: offsetting outward by `d` produces a rounded rectangle whose\n * min-area rect is the original grown by `d` on each of the four sides. That\n * identity is what lets this avoid a Clipper dependency entirely.\n */\nexport function unclipRect(points: readonly Point[], unclipRatio: number): [Point, Point, Point, Point] {\n    const rect = minAreaRect(points)\n    const perimeter = polygonPerimeter(points)\n    const distance = perimeter === 0 ? 0 : (polygonArea(points) * unclipRatio) / perimeter\n\n    return boxPoints({\n        center: rect.center,\n        size: [rect.size[0] + distance * 2, rect.size[1] + distance * 2],\n        angle: rect.angle,\n    })\n}\n\n/**\n * Reorder four points clockwise from the top-left -- port of\n * `order_points_clockwise`. Top-left has the smallest x+y, bottom-right the\n * largest; the remaining two are separated by y-x.\n */\nexport function orderPointsClockwise(points: readonly Point[]): [Point, Point, Point, Point] {\n    const bySum = [...points].sort((a, b) => a[0] + a[1] - (b[0] + b[1]))\n    const topLeft = bySum[0] as Point\n    const bottomRight = bySum[bySum.length - 1] as Point\n\n    const rest = bySum.slice(1, -1).sort((a, b) => a[1] - a[0] - (b[1] - b[0]))\n    return [topLeft, rest[0] as Point, bottomRight, rest[1] as Point]\n}\n\n/**\n * Probability map -> text quads in source-image coordinates.\n *\n * `scale` is the uniform factor the source was resized by. Coordinates restore\n * by *dividing* by it, with no offset to subtract, because the letterbox pads\n * bottom and right only.\n */\nexport function quadsFromProbabilityMap(\n    map: ProbabilityMap,\n    source: { width: number; height: number },\n    scale: number,\n    options: Partial<DbPostProcessOptions> = {}\n): DetectedQuad[] {\n    const opts = { ...DB_POSTPROCESS_DEFAULTS, ...options }\n    const quads: DetectedQuad[] = []\n\n    for (const boundary of findComponentBoundaries(map, opts.thresh, opts.maxCandidates)) {\n        const candidate = miniBox(boundary)\n        if (candidate.sside < opts.minSize) continue\n\n        const score = boxScore(map, candidate.points)\n        if (score < opts.boxThresh) continue\n\n        const expanded = miniBox(unclipRect(candidate.points, opts.unclipRatio))\n        // Python's second gate is `min_size + 2`, i.e. tighter than the first --\n        // unclipping can only grow a box, so anything still tiny is noise.\n        if (expanded.sside < opts.minSize + 2) continue\n\n        // Python clips to [0, dest], inclusive of the far edge -- not [0, dest-1].\n        const restored = expanded.points.map(\n            ([x, y]) =>\n                [\n                    Math.min(Math.max(Math.round(x / scale), 0), source.width),\n                    Math.min(Math.max(Math.round(y / scale), 0), source.height),\n                ] as Point\n        )\n\n        const ordered = orderPointsClockwise(restored)\n        // Drop slivers: Python filters boxes whose sides are 3px or less.\n        const width = Math.hypot(ordered[0][0] - ordered[1][0], ordered[0][1] - ordered[1][1])\n        const height = Math.hypot(ordered[0][0] - ordered[3][0], ordered[0][1] - ordered[3][1])\n        if (width <= 3 || height <= 3) continue\n\n        quads.push({ points: ordered, score })\n    }\n    return quads\n}\n","/**\n * DBNet ONNX text detection -- the direct mirror of\n * `scaledp/models/detectors/DBNetOnnxDetector.py`, so ScaleDP's own detection\n * model runs unchanged in the browser.\n *\n * Preprocessing reproduces `paddle_onnx/operators.py` exactly, including one\n * quirk that matters: the Python path converts RGB to BGR and never converts\n * back, so the model is fed BGR channels normalised against *RGB* ImageNet\n * statistics. Feeding true RGB instead shifts the boxes.\n *\n * The model detects text *lines*, not words -- one region per line, in Python\n * as here. Nothing downstream subdivides them either: `TesseractRecognizer`\n * reads each region with PSM.SINGLE_WORD and returns one box per region, which\n * is what ScaleDP does too. `TesseractOcr`, which runs tesseract's own layout\n * analysis over the whole page, is the stage that yields word boxes.\n */\n\nimport { getConfig } from '../core/config.js'\nimport { DetectionError } from '../core/errors.js'\nimport {\n    decodeImage,\n    IMAGENET_MEAN,\n    IMAGENET_STD,\n    letterbox,\n    toImageData,\n    toNchwFloat32,\n} from '../core/image.js'\nimport { ensureModelFiles, type ModelSpec } from '../core/model-cache.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage } from '../core/pipeline.js'\nimport { type Box, boxFromPolygon, mergeOverlappingBoxes } from '../schemas/box.js'\nimport { createDetectorOutput, type DetectorOutput } from '../schemas/detector-output.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\nimport { DB_POSTPROCESS_DEFAULTS, type ProbabilityMap, quadsFromProbabilityMap } from './db-postprocess.js'\nimport { createSession } from './ort.js'\n\n/** Fixed input size from ScaleDP's `DetResizeForTest` config. */\nexport const DBNET_INPUT_SIZE = 1280\n\n/** Model ScaleDP's DBNetOnnxDetector documents and its tests use. */\nexport const DEFAULT_DBNET_MODEL = 'StabRise/text_detection_dbnet_ml_v0.2'\n\nexport interface DbnetOnnxDetectorParams extends BaseStageParams {\n    /** Hugging Face repo id, or a URL when self-hosting. */\n    model: string\n    /** Mean in-box probability a candidate must reach. */\n    scoreThreshold: number\n    /** Probability above which a pixel counts as text. */\n    binaryThreshold: number\n    /** How far to grow each box; DB shrinks text regions during training. */\n    unclipRatio: number\n    /**\n     * Merge boxes that overlap and share a line. ScaleDP uses an unusually low\n     * IoU of 0.02 here, because adjacent words in a line barely overlap.\n     *\n     * In practice it rarely changes anything for this model: the regions are\n     * already whole lines and do not overlap. ScaleDP merges unconditionally\n     * and its box count is likewise unchanged. Kept as a parameter because it\n     * does matter for detectors that emit overlapping candidates.\n     */\n    mergeBoxes: boolean\n}\n\nexport const DBNET_DETECTOR_DEFAULTS: DbnetOnnxDetectorParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    outputCol: 'boxes',\n    keepInputData: true,\n    model: DEFAULT_DBNET_MODEL,\n    scoreThreshold: DB_POSTPROCESS_DEFAULTS.boxThresh,\n    binaryThreshold: DB_POSTPROCESS_DEFAULTS.thresh,\n    unclipRatio: DB_POSTPROCESS_DEFAULTS.unclipRatio,\n    mergeBoxes: true,\n})\n\nfunction modelSpec(model: string): ModelSpec {\n    return /^https?:\\/\\//.test(model)\n        ? { repo: 'dbnet', files: [{ path: model }] }\n        : { repo: model, files: [{ path: 'model.onnx' }] }\n}\n\nexport class DbnetOnnxDetector extends Stage<DbnetOnnxDetectorParams> {\n    readonly name = 'DbnetOnnxDetector'\n\n    private session: import('onnxruntime-web').InferenceSession | null = null\n    private loading: Promise<import('onnxruntime-web').InferenceSession> | null = null\n\n    constructor(options: Partial<DbnetOnnxDetectorParams> = {}) {\n        super(resolveParams(DBNET_DETECTOR_DEFAULTS, options))\n    }\n\n    override async init(): Promise<void> {\n        await this.getSession()\n    }\n\n    private getSession(): Promise<import('onnxruntime-web').InferenceSession> {\n        if (this.session) return Promise.resolve(this.session)\n        if (this.loading) return this.loading\n\n        this.loading = (async () => {\n            const spec = modelSpec(this.params.model)\n            const files = await ensureModelFiles(spec)\n            const bytes = files[spec.files[0]?.path ?? '']\n            if (!bytes) throw new DetectionError(`Model ${this.params.model} not found`, this.name)\n\n            const session = await createSession(bytes, {\n                executionProviders: getConfig().executionProviders,\n            })\n            this.session = session\n            return session\n        })()\n\n        // Clear on failure so a transient network error can be retried rather\n        // than cached as a permanently rejected promise.\n        this.loading.catch(() => {\n            this.loading = null\n        })\n        return this.loading\n    }\n\n    override async dispose(): Promise<void> {\n        await this.session?.release()\n        this.session = null\n        this.loading = null\n    }\n\n    protected async apply(input: unknown, row: Row): Promise<DetectorOutput> {\n        const image = input as ScaleDpImage | undefined\n        // Check `exception` first. A failed upstream stage returns a well-formed but\n        // empty Image, so testing the bytes first would report \"no decoded bytes\" and\n        // bury the real cause.\n        if (image?.exception) {\n            throw new DetectionError(`Upstream stage failed: ${image.exception}`, this.name)\n        }\n        if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new DetectionError('Expected an Image with decoded bytes', this.name)\n        }\n\n        const bitmap = await decodeImage(image.data)\n        let boxes: Box[]\n        try {\n            boxes = await this.detect(bitmap)\n        } finally {\n            bitmap.close()\n        }\n\n        return createDetectorOutput({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'dbnet-onnx',\n            bboxes: boxes,\n        })\n    }\n\n    /** Run the model over one decoded image and return boxes in its coordinates. */\n    async detect(source: ImageBitmap | OffscreenCanvas): Promise<Box[]> {\n        const session = await this.getSession()\n\n        // Letterbox to a fixed 1280x1280, padding bottom and right only with\n        // white. Because nothing is padded at the top or left, coordinates\n        // restore by dividing by the scale with no offset to subtract.\n        const fitted = letterbox(\n            source,\n            { width: DBNET_INPUT_SIZE, height: DBNET_INPUT_SIZE },\n            {\n                padding: 'end',\n                fill: '#ffffff',\n            }\n        )\n        const tensorData = toNchwFloat32(toImageData(fitted.canvas), {\n            mean: IMAGENET_MEAN,\n            std: IMAGENET_STD,\n            bgr: true,\n        })\n\n        const { Tensor } = await import('onnxruntime-web')\n        const inputName = session.inputNames[0]\n        const outputName = session.outputNames[0]\n        if (!inputName || !outputName) {\n            throw new DetectionError('Model exposes no input or output', this.name)\n        }\n\n        const outputs = await session.run({\n            [inputName]: new Tensor('float32', tensorData, [1, 3, DBNET_INPUT_SIZE, DBNET_INPUT_SIZE]),\n        })\n        const output = outputs[outputName]\n        if (!output) throw new DetectionError(`Model produced no \"${outputName}\" output`, this.name)\n\n        // Output is [N, 1, H, W]; the first channel is the probability map.\n        const [, , height = DBNET_INPUT_SIZE, width = DBNET_INPUT_SIZE] = output.dims\n        const map: ProbabilityMap = {\n            data: output.data as Float32Array,\n            width,\n            height,\n        }\n\n        const quads = quadsFromProbabilityMap(\n            map,\n            { width: source.width, height: source.height },\n            fitted.scale,\n            {\n                thresh: this.params.binaryThreshold,\n                boxThresh: this.params.scoreThreshold,\n                unclipRatio: this.params.unclipRatio,\n            }\n        )\n\n        const boxes = quads.map((quad) => boxFromPolygon(quad.points, { score: quad.score }))\n        return this.params.mergeBoxes ? mergeOverlappingBoxes(boxes, 0.02, 10, 0.3) : boxes\n    }\n\n    protected onError(message: string, row: Row): DetectorOutput {\n        return createDetectorOutput({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'dbnet-onnx',\n            exception: message,\n        })\n    }\n}\n","/**\n * Text detectors that can run in the browser.\n *\n * Detection and recognition are separable in ScaleDP -- a detector stage feeds\n * boxes to a recognizer -- so which detector produced a page's boxes is a real\n * choice worth making addressable, the same way OCR presets and NER models are.\n */\n\nexport type DetectorKind = 'paddle' | 'dbnet-onnx'\n\nexport interface DetectorModel {\n    /** Short id callers pass to the picker. */\n    id: string\n    /** Human-readable name including the download size. */\n    name: string\n    kind: DetectorKind\n    /**\n     * Hugging Face repo, for ONNX detectors. Paddle detectors come from the\n     * OCR preset instead, so this is unset for them.\n     */\n    repo?: string\n    approxBytes?: number\n    notes: string\n}\n\nexport const DETECTOR_MODELS: readonly DetectorModel[] = Object.freeze([\n    {\n        id: 'paddle',\n        name: 'PaddleOCR DB (follows the OCR preset)',\n        kind: 'paddle',\n        notes: 'Shares the detection model already downloaded for the selected OCR preset.',\n    },\n    {\n        id: 'dbnet-v0.2',\n        name: 'StabRise DBNet ONNX v0.2 (~5 MB)',\n        kind: 'dbnet-onnx',\n        repo: 'StabRise/text_detection_dbnet_ml_v0.2',\n        approxBytes: 4_800_000,\n        notes: 'The same detection model ScaleDP uses server-side.',\n    },\n    {\n        id: 'dbnet-v0.1',\n        name: 'StabRise DBNet ONNX v0.1 (~5 MB)',\n        kind: 'dbnet-onnx',\n        repo: 'StabRise/text_detection_dbnet_ml_v0.1',\n        approxBytes: 4_800_000,\n        notes: 'The earlier revision, kept for comparison.',\n    },\n])\n\nexport const DEFAULT_DETECTOR_ID = 'paddle'\n\nexport function getDetectorModel(id: string): DetectorModel | undefined {\n    return DETECTOR_MODELS.find((m) => m.id === id)\n}\n","/**\n * 0 / 180 degree line-orientation classifier.\n *\n * Port of `scaledp/models/detectors/HasDetectLineOrientation.py`. Tiny, and it\n * meaningfully improves recognition on rotated crops -- an upside-down line\n * otherwise recognises as noise.\n *\n * Reproduces the same BGR-with-RGB-statistics quirk as the DBNet path: the\n * Python code converts to BGR and normalises with RGB ImageNet constants.\n */\n\nimport { DetectionError } from '../core/errors.js'\nimport {\n    context2d,\n    createCanvas,\n    IMAGENET_MEAN,\n    IMAGENET_STD,\n    toImageData,\n    toNchwFloat32,\n} from '../core/image.js'\nimport { ensureModelFiles } from '../core/model-cache.js'\nimport { createSession } from './ort.js'\n\nexport const DEFAULT_ORIENTATION_MODEL = 'StabRise/line_orientation_detection_v0.1'\n\n/** Model input, width x height. */\nexport const ORIENTATION_INPUT = { width: 160, height: 80 } as const\n\nexport type LineOrientation = '0_degree' | '180_degree'\n\nconst LABELS: readonly LineOrientation[] = ['0_degree', '180_degree']\n\nexport class LineOrientationClassifier {\n    private session: import('onnxruntime-web').InferenceSession | null = null\n    private loading: Promise<import('onnxruntime-web').InferenceSession> | null = null\n\n    constructor(readonly model: string = DEFAULT_ORIENTATION_MODEL) {}\n\n    private getSession(): Promise<import('onnxruntime-web').InferenceSession> {\n        if (this.session) return Promise.resolve(this.session)\n        if (this.loading) return this.loading\n\n        this.loading = (async () => {\n            const files = await ensureModelFiles({\n                repo: this.model,\n                files: [{ path: 'model.onnx' }],\n            })\n            const bytes = files['model.onnx']\n            if (!bytes) {\n                throw new DetectionError(`Model ${this.model} has no model.onnx`, 'LineOrientation')\n            }\n            const session = await createSession(bytes)\n            this.session = session\n            return session\n        })()\n\n        this.loading.catch(() => {\n            this.loading = null\n        })\n        return this.loading\n    }\n\n    async classify(source: ImageBitmap | OffscreenCanvas): Promise<LineOrientation> {\n        const session = await this.getSession()\n\n        // Plain resize, not a letterbox: the Python path calls cv2.resize\n        // directly and lets the aspect ratio distort.\n        const canvas = createCanvas(ORIENTATION_INPUT.width, ORIENTATION_INPUT.height)\n        context2d(canvas).drawImage(source, 0, 0, canvas.width, canvas.height)\n\n        const data = toNchwFloat32(toImageData(canvas), {\n            mean: IMAGENET_MEAN,\n            std: IMAGENET_STD,\n            bgr: true,\n        })\n\n        const { Tensor } = await import('onnxruntime-web')\n        const inputName = session.inputNames[0]\n        const outputName = session.outputNames[0]\n        if (!inputName || !outputName) {\n            throw new DetectionError('Model exposes no input or output', 'LineOrientation')\n        }\n\n        const outputs = await session.run({\n            [inputName]: new Tensor('float32', data, [\n                1,\n                3,\n                ORIENTATION_INPUT.height,\n                ORIENTATION_INPUT.width,\n            ]),\n        })\n        const logits = outputs[outputName]?.data as Float32Array | undefined\n        if (!logits || logits.length < 2) {\n            throw new DetectionError('Classifier produced no logits', 'LineOrientation')\n        }\n\n        return (logits[1] as number) > (logits[0] as number)\n            ? (LABELS[1] as LineOrientation)\n            : (LABELS[0] as LineOrientation)\n    }\n\n    async dispose(): Promise<void> {\n        await this.session?.release()\n        this.session = null\n        this.loading = null\n    }\n}\n","/**\n * Line-orientation detection and correction.\n *\n * Port of ScaleDP's `HasDetectLineOrientation`, which `TesseractRecognizer`\n * mixes in: each detected box is cropped, classified 0 or 180 degrees, and the\n * upside-down ones are turned before recognition. An inverted line otherwise\n * recognises as noise.\n *\n * Python does this per crop inside the recognizer. Here it is a stage of its\n * own, because `PaddleTextRecognizer` detects and recognises in a single pass\n * and has no seam to hook into. Flipping each inverted region in place on a\n * copy of the page gets the same result and costs one recognition pass rather\n * than one per box; the regions are rectangles, so a 180-degree turn leaves\n * every box's coordinates untouched and downstream stages need no adjustment.\n */\n\nimport { DetectionError } from '../core/errors.js'\nimport { context2d, createCanvas, cropBox, decodeImage, encodeImage } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { type Box, isRotated } from '../schemas/box.js'\nimport type { DetectorOutput } from '../schemas/detector-output.js'\nimport type { Document } from '../schemas/document.js'\nimport { createImage, type ImageFormat, type ScaleDpImage } from '../schemas/image.js'\nimport {\n    DEFAULT_ORIENTATION_MODEL,\n    type LineOrientation,\n    LineOrientationClassifier,\n} from './line-orientation.js'\n\nexport interface LineOrientationDetectorParams extends BaseStageParams {\n    /** [imageColumn, boxColumn]. */\n    inputCols: string[]\n    /** Column the corrected image is written to. */\n    outputCol: string\n    /** Column the per-box orientation labels are written to. */\n    orientationCol: string\n    model: string\n    /** Turn the inverted regions. Off classifies only, leaving the page as-is. */\n    correct: boolean\n    /**\n     * Classify only boxes that are already rotated, as ScaleDP's `onlyRotated`\n     * does, and defaults to.\n     *\n     * This is not just a saved inference per box. The classifier has a real\n     * false-positive rate: on an upright invoice it called 1 of 81 upright\n     * regions inverted, and turning that region cost about 40 characters of\n     * recognition. Restricting it to boxes that are already rotated is where\n     * the signal actually is.\n     *\n     * Set false for pages that may contain upside-down *horizontal* text, which\n     * is the case this misses.\n     */\n    onlyRotated: boolean\n    /** Grow each box before cropping, so glyph edges are not clipped. */\n    padding: number\n    imageType: ImageFormat\n}\n\nexport const LINE_ORIENTATION_DEFAULTS: LineOrientationDetectorParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    inputCols: ['image', 'boxes'],\n    outputCol: 'oriented',\n    orientationCol: 'orientations',\n    keepInputData: true,\n    model: DEFAULT_ORIENTATION_MODEL,\n    correct: true,\n    onlyRotated: true,\n    padding: 2,\n    imageType: 'png' as ImageFormat,\n})\n\nfunction boxesOf(source: unknown): Box[] {\n    if (typeof source !== 'object' || source === null) return []\n    return (source as DetectorOutput | Document).bboxes ?? []\n}\n\nexport class LineOrientationDetector extends Stage<LineOrientationDetectorParams> {\n    readonly name = 'LineOrientationDetector'\n\n    private classifier: LineOrientationClassifier | null = null\n\n    constructor(options: Partial<LineOrientationDetectorParams> = {}) {\n        super(\n            resolveParams(LINE_ORIENTATION_DEFAULTS, options, {\n                inputCols: (value) => {\n                    if (value.length !== 2) {\n                        throw new RangeError('inputCols must be [imageColumn, boxColumn]')\n                    }\n                },\n            })\n        )\n    }\n\n    override async init(): Promise<void> {\n        this.classifier ??= new LineOrientationClassifier(this.params.model)\n    }\n\n    /** One row in, one row out -- but two columns written, so expand not apply. */\n    protected override async expand(_input: unknown, row: Row, ctx: StageContext): Promise<Row[]> {\n        const { inputCols, outputCol, orientationCol, correct, onlyRotated, padding } = this.params\n        const [imageCol, boxCol] = inputCols as [string, string]\n        const image = row[imageCol] as ScaleDpImage | undefined\n\n        if (image?.exception) {\n            throw new DetectionError(`Upstream stage failed: ${image.exception}`, this.name)\n        }\n        if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new DetectionError('Expected an Image with decoded bytes', this.name)\n        }\n\n        await this.init()\n        const classifier = this.classifier as LineOrientationClassifier\n        const boxes = boxesOf(row[boxCol])\n\n        const bitmap = await decodeImage(image.data)\n        try {\n            const canvas = createCanvas(bitmap.width, bitmap.height)\n            const ctx2d = context2d(canvas)\n            ctx2d.drawImage(bitmap, 0, 0)\n\n            const orientations: LineOrientation[] = []\n            let flipped = 0\n\n            for (const box of boxes) {\n                ctx.signal?.throwIfAborted()\n\n                if (onlyRotated && !isRotated(box)) {\n                    orientations.push('0_degree')\n                    continue\n                }\n\n                const crop = cropBox(bitmap, box, { padding })\n                const orientation = await classifier.classify(crop)\n                orientations.push(orientation)\n\n                if (orientation === '180_degree' && correct) {\n                    flipRegion(ctx2d, canvas, box)\n                    flipped++\n                }\n            }\n\n            const corrected =\n                flipped > 0\n                    ? createImage({\n                          path: image.path,\n                          resolution: image.resolution,\n                          data: await encodeImage(canvas, `image/${this.params.imageType}` as never),\n                          imageType: this.params.imageType,\n                          width: canvas.width,\n                          height: canvas.height,\n                      })\n                    : // Nothing was inverted, so hand the original through rather\n                      // than paying a re-encode for an identical image.\n                      image\n\n            return [{ ...row, [outputCol]: corrected, [orientationCol]: orientations }]\n        } finally {\n            bitmap.close()\n        }\n    }\n\n    protected async apply(): Promise<never> {\n        throw new DetectionError('unreachable: expand handles every row', this.name)\n    }\n\n    protected onError(message: string, row: Row): ScaleDpImage {\n        return createImage({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            exception: message,\n        })\n    }\n\n    override async dispose(): Promise<void> {\n        await this.classifier?.dispose()\n        this.classifier = null\n    }\n}\n\n/**\n * Turn one box's region 180 degrees in place.\n *\n * A rectangle maps onto itself under a 180-degree rotation about its own\n * centre, whatever its angle, so the box's coordinates stay valid and\n * downstream stages need no adjustment.\n *\n * The draw is clipped to the box's own rotated outline. Without that a rotated\n * box turns its whole axis-aligned envelope, dragging neighbouring text through\n * the rotation with it -- and the envelope of a skewed box is meaningfully\n * larger than the box.\n */\nfunction flipRegion(ctx: OffscreenCanvasRenderingContext2D, canvas: OffscreenCanvas, box: Box): void {\n    const pad = 1\n    const x = Math.max(0, Math.min(Math.round(box.x) - pad, canvas.width - 1))\n    const y = Math.max(0, Math.min(Math.round(box.y) - pad, canvas.height - 1))\n    const width = Math.min(Math.round(box.width) + pad * 2, canvas.width - x)\n    const height = Math.min(Math.round(box.height) + pad * 2, canvas.height - y)\n    if (width < 1 || height < 1) return\n\n    // Drawing a canvas onto itself through a transform reads and writes the same\n    // backing store, so take a copy first.\n    const region = createCanvas(width, height)\n    context2d(region).drawImage(canvas, x, y, width, height, 0, 0, width, height)\n\n    const cx = x + width / 2\n    const cy = y + height / 2\n\n    ctx.save()\n    ctx.beginPath()\n    if (Math.abs(box.angle) < 3) {\n        ctx.rect(x, y, width, height)\n    } else {\n        const rad = (box.angle * Math.PI) / 180\n        const cos = Math.cos(rad)\n        const sin = Math.sin(rad)\n        const corners: [number, number][] = [\n            [-box.width / 2, -box.height / 2],\n            [box.width / 2, -box.height / 2],\n            [box.width / 2, box.height / 2],\n            [-box.width / 2, box.height / 2],\n        ]\n        corners.forEach(([px, py], index) => {\n            const rx = px * cos - py * sin + cx\n            const ry = px * sin + py * cos + cy\n            if (index === 0) ctx.moveTo(rx, ry)\n            else ctx.lineTo(rx, ry)\n        })\n        ctx.closePath()\n    }\n    ctx.clip()\n\n    ctx.translate(cx, cy)\n    ctx.rotate(Math.PI)\n    ctx.drawImage(region, -width / 2, -height / 2)\n    ctx.restore()\n}\n","/**\n * PaddleOCR language presets.\n *\n * Each preset pairs a detection model, a recognition model and a character\n * dictionary. There is no single model covering every script, so the choice is\n * a real one: pick the pairing that matches the documents being processed.\n *\n * Curated from ppu-paddle-ocr's catalogue -- the full list also carries v3/v4\n * models that v5/v6 supersede.\n */\n\nexport interface OcrPreset {\n    /** Key accepted by `PaddleTextDetector`/`PaddleTextRecognizer`. */\n    value: string\n    /** Human-readable name, suitable for a language picker. */\n    label: string\n    /** Scripts this preset can read. */\n    scripts: readonly string[]\n}\n\nexport const PADDLE_OCR_PRESETS: readonly OcrPreset[] = Object.freeze([\n    { value: 'v6-small', label: 'Latin / CJK (default)', scripts: ['Latin', 'Han', 'Hiragana', 'Katakana'] },\n    {\n        value: 'v6-medium',\n        label: 'Latin / CJK (medium, more accurate)',\n        scripts: ['Latin', 'Han', 'Hiragana', 'Katakana'],\n    },\n    { value: 'v6-tiny', label: 'Latin / CJK (tiny, fastest)', scripts: ['Latin', 'Han'] },\n    { value: 'v5-latin-mobile', label: 'Latin (French, German, Spanish, ...)', scripts: ['Latin'] },\n    {\n        value: 'v5-eslav-mobile',\n        label: 'Latin + Cyrillic (Russian, Ukrainian, ...)',\n        scripts: ['Latin', 'Cyrillic'],\n    },\n    { value: 'v5-cyrillic-mobile', label: 'Cyrillic only', scripts: ['Cyrillic'] },\n    { value: 'v5-devanagari-mobile', label: 'Latin + Hindi (Devanagari)', scripts: ['Latin', 'Devanagari'] },\n    { value: 'v5-arabic-mobile', label: 'Latin + Arabic', scripts: ['Latin', 'Arabic'] },\n    { value: 'v5-greek-mobile', label: 'Latin + Greek', scripts: ['Latin', 'Greek'] },\n    { value: 'v5-korean-mobile', label: 'Latin + Korean', scripts: ['Latin', 'Hangul'] },\n    { value: 'v5-thai-mobile', label: 'Latin + Thai', scripts: ['Latin', 'Thai'] },\n    { value: 'v5-tamil-mobile', label: 'Latin + Tamil', scripts: ['Latin', 'Tamil'] },\n    { value: 'v5-telugu-mobile', label: 'Latin + Telugu', scripts: ['Latin', 'Telugu'] },\n    { value: 'v5-en-mobile', label: 'English only (fastest)', scripts: ['Latin'] },\n])\n\nexport const DEFAULT_OCR_PRESET = 'v6-small'\n\nexport function isKnownPreset(value: string): boolean {\n    return PADDLE_OCR_PRESETS.some((p) => p.value === value)\n}\n\n/** Shared by every Paddle stage, so an unknown preset fails at construction. */\nexport function validatePreset(value: string): void {\n    if (!isKnownPreset(value)) {\n        throw new RangeError(`Unknown OCR preset \"${value}\". See PADDLE_OCR_PRESETS for valid values.`)\n    }\n}\n\n/** Presets able to read a script name as reported by OSD script detection. */\nexport function presetsForScript(script: string): OcrPreset[] {\n    return PADDLE_OCR_PRESETS.filter((p) => p.scripts.includes(script))\n}\n","/**\n * PaddleOcrService lifecycle: model fetching, caching and per-preset reuse.\n *\n * Models are pulled through our own ModelCache and handed to the service as\n * ArrayBuffers, rather than letting ppu-paddle-ocr fetch them itself. Its\n * browser build re-downloads ~6 MB on every page load, relying only on the HTTP\n * cache; routing through IndexedDB makes a repeat visit instant and offline-safe.\n *\n * Two entry points, because the two halves of a preset are separable:\n * `getPaddleService` builds the full detect-then-read service, while\n * `getPaddleRecognizer` loads only the recognition model and its dictionary --\n * what `PaddleRecognizer` needs when some other detector already found the\n * boxes. Both key their cache on the same repo, so a pipeline using both pays\n * for each file once.\n */\n\nimport { getConfig } from '../core/config.js'\nimport type { ModelSpec } from '../core/model-cache.js'\nimport { ensureModelFiles, evict, isCached } from '../core/model-cache.js'\nimport { createSession, loadOrt } from './ort.js'\nimport { DEFAULT_OCR_PRESET } from './presets.js'\n\ntype OrtSession = Awaited<ReturnType<typeof createSession>>\n\ntype PpuWeb = typeof import('ppu-paddle-ocr/web')\ntype PaddleOcrService = InstanceType<PpuWeb['PaddleOcrService']>\n\n/**\n * ppu's recognition half, bound to a session: `run(canvas, boxes)` reads the\n * regions it is handed and does no detection of its own.\n */\nexport type PaddleRecognitionService = InstanceType<PpuWeb['RecognitionService']>\n\n/** Recognition tuning a stage may vary without paying for a second session. */\nexport interface PaddleRecognizerOptions {\n    /** Crops per batched inference. 1 disables batching. */\n    recBatchSize?: number\n    /** Recover inter-word spaces the greedy CTC decode drops. */\n    spaceRecovery?: boolean\n}\n\nlet modulePromise: Promise<PpuWeb> | null = null\n\nasync function loadPpu(): Promise<PpuWeb> {\n    if (modulePromise) return modulePromise\n    modulePromise = (async () => {\n        // ORT must be configured before ppu-paddle-ocr creates any session,\n        // otherwise its own wasmPaths default wins the \"if unset\" check.\n        await loadOrt()\n        try {\n            return await import('ppu-paddle-ocr/web')\n        } catch (cause) {\n            throw new Error(\n                'ppu-paddle-ocr is required for PaddleOCR stages. Install it: npm i ppu-paddle-ocr',\n                { cause }\n            )\n        }\n    })()\n    return modulePromise\n}\n\n/** The three files a preset needs, as a ModelSpec our cache understands. */\nasync function specForPreset(preset: string): Promise<{ spec: ModelSpec; roles: string[] }> {\n    const ppu = await loadPpu()\n    const urls =\n        preset in ppu.MODEL_PRESETS\n            ? ppu.MODEL_PRESETS[preset as keyof typeof ppu.MODEL_PRESETS]\n            : ppu.DEFAULT_MODEL\n\n    const roles = ['detection', 'recognition', 'charactersDictionary'] as const\n    // ppu gives absolute URLs; the cache keys on repo-relative paths, so the\n    // full URL doubles as the key and `modelHost` is bypassed for these.\n    return {\n        spec: { repo: `ppu-paddle-ocr/${preset}`, files: roles.map((r) => ({ path: urls[r] })) },\n        roles: [...roles],\n    }\n}\n\n/**\n * Absolute model URLs bypass `modelHost`: ppu-paddle-ocr publishes its own\n * catalogue and the paths are meaningful only against that host.\n */\nasync function fetchPresetFiles(preset: string): Promise<Record<string, ArrayBuffer>> {\n    const { spec, roles } = await specForPreset(preset)\n    const files = await ensureModelFiles({\n        ...spec,\n        files: spec.files.map((f) => ({ ...f })),\n    })\n    const out: Record<string, ArrayBuffer> = {}\n    for (const [i, role] of roles.entries()) {\n        const path = spec.files[i]?.path\n        if (path && files[path]) out[role] = files[path] as ArrayBuffer\n    }\n    return out\n}\n\n// Keyed by preset: switching language keeps the previous service alive, so\n// toggling back is instant rather than a re-download plus re-init.\nconst services = new Map<string, Promise<PaddleOcrService>>()\n\nexport async function getPaddleService(preset = DEFAULT_OCR_PRESET): Promise<PaddleOcrService> {\n    const existing = services.get(preset)\n    if (existing) return existing\n\n    const promise = (async () => {\n        const ppu = await loadPpu()\n        const model = await fetchPresetFiles(preset)\n        const service = new ppu.PaddleOcrService({\n            model,\n            session: {\n                executionProviders: [...getConfig().executionProviders] as never,\n                graphOptimizationLevel: 'all',\n            },\n        })\n        await service.initialize()\n        return service\n    })()\n\n    // Evict on failure so a transient network error is retried rather than\n    // cached as a permanently rejected promise.\n    promise.catch(() => services.delete(preset))\n    services.set(preset, promise)\n    return promise\n}\n\n/**\n * The recognition ONNX session and its character dictionary, per preset.\n *\n * Deliberately separate from `services`: a recognizer running behind someone\n * else's detector never needs the detection model, and skipping it saves a\n * download of a few MB. The cache `repo` is the same either way, so the two\n * files are shared with the full service rather than stored twice.\n */\nconst recognitionAssets = new Map<string, Promise<{ session: OrtSession; dictionary: string[] }>>()\n\n/**\n * Split a PaddleOCR dictionary into an ordered array, one entry per line.\n *\n * ppu exports `parseDictionary` from a module its `exports` map does not\n * publish, so this reproduces it -- and it is the whole of it. Blank entries\n * are preserved: the index is the class id the model emits, so dropping one\n * would shift every character after it.\n */\nfunction parseDictionary(source: ArrayBuffer): string[] {\n    return new TextDecoder('utf-8').decode(source).split(/\\r?\\n/)\n}\n\nasync function loadRecognitionAssets(preset: string): Promise<{ session: OrtSession; dictionary: string[] }> {\n    const existing = recognitionAssets.get(preset)\n    if (existing) return existing\n\n    const promise = (async () => {\n        // Nothing below imports ppu, but loading it first keeps the \"ORT\n        // configured before any session exists\" ordering `loadPpu` enforces --\n        // and importing `ppu-paddle-ocr/web` is also what registers ppu-ocv's\n        // web canvas platform, which the crop path needs. Skipping\n        // `PaddleOcrService.initialize()` skips nothing else: the web recognizer\n        // runs the canvas-native engine and never touches OpenCV.\n        await loadPpu()\n        const { spec, roles } = await specForPreset(preset)\n        // Same `spec.repo`, so these two land on the cache keys the full\n        // service already uses -- the detection file is simply never asked for.\n        const wanted = ['recognition', 'charactersDictionary'].map(\n            (role) => spec.files[roles.indexOf(role)]?.path ?? ''\n        )\n        const files = await ensureModelFiles({ ...spec, files: wanted.map((path) => ({ path })) })\n        const model = files[wanted[0] as string]\n        const dict = files[wanted[1] as string]\n        if (!model || !dict) {\n            throw new Error(`PaddleOCR preset \"${preset}\" is missing its recognition model or dictionary.`)\n        }\n\n        const dictionary = parseDictionary(dict)\n        if (dictionary.length === 0) {\n            throw new Error(`PaddleOCR preset \"${preset}\" has an empty character dictionary.`)\n        }\n        // WebGPU cannot run PP-OCR's recognition graph -- it rewrites the\n        // convolutions into `com.ms.internal.nhwc` and has no kernel for them,\n        // which fails at session creation. `PaddleOcrService` retries on WASM\n        // for exactly this, so a session built here has to as well.\n        return { session: await createSession(model, { fallbackToWasm: true }), dictionary }\n    })()\n\n    promise.catch(() => recognitionAssets.delete(preset))\n    recognitionAssets.set(preset, promise)\n    return promise\n}\n\n/**\n * Recognition only: read the regions you hand it, no detection.\n *\n * `minimumConfidence` is pinned to 0 and `maxCropSourceSideLength` is set past\n * any canvas we pass. ppu would otherwise silently drop low-scoring results --\n * breaking the caller's box-to-result mapping, since `run` also sorts what it\n * returns into reading order -- and downscale a tall batch of stacked crops.\n * Filtering is the calling stage's job, where the threshold is a parameter.\n */\nexport async function getPaddleRecognizer(\n    preset = DEFAULT_OCR_PRESET,\n    options: PaddleRecognizerOptions = {}\n): Promise<PaddleRecognitionService> {\n    const ppu = await loadPpu()\n    const { session, dictionary } = await loadRecognitionAssets(preset)\n    return new ppu.RecognitionService(session, {\n        charactersDictionary: dictionary,\n        minimumConfidence: 0,\n        maxCropSourceSideLength: Number.MAX_SAFE_INTEGER,\n        ...(options.recBatchSize === undefined ? {} : { recBatchSize: options.recBatchSize }),\n        ...(options.spaceRecovery === undefined ? {} : { spaceRecovery: options.spaceRecovery }),\n    })\n}\n\nasync function releaseRecognition(preset: string): Promise<void> {\n    const pending = recognitionAssets.get(preset)\n    if (!pending) return\n    recognitionAssets.delete(preset)\n    await pending.then(({ session }) => session.release()).catch(() => undefined)\n}\n\nexport async function isPresetCached(preset: string): Promise<boolean> {\n    const { spec } = await specForPreset(preset)\n    return isCached(spec)\n}\n\n/** Pre-warm a preset so the first OCR call does not pay the download. */\nexport async function loadPreset(preset: string): Promise<void> {\n    await getPaddleService(preset)\n}\n\nexport async function removePreset(preset: string): Promise<void> {\n    const { spec } = await specForPreset(preset)\n    const service = services.get(preset)\n    services.delete(preset)\n    await service?.then((s) => s.destroy()).catch(() => undefined)\n    await releaseRecognition(preset)\n    await evict(spec)\n}\n\n/** Tear down every cached service, recognition-only sessions included. */\nexport async function disposePaddleServices(): Promise<void> {\n    const pending = [...services.values()]\n    services.clear()\n    const recognition = [...recognitionAssets.keys()]\n    await Promise.all([\n        ...pending.map((p) => p.then((s) => s.destroy()).catch(() => undefined)),\n        ...recognition.map((preset) => releaseRecognition(preset)),\n    ])\n}\n","/**\n * PaddleOCR text detection and recognition -- the default OCR engines.\n *\n * `PaddleTextDetector` mirrors ScaleDP's detector stages (image -> boxes) and\n * `PaddleTextRecognizer` mirrors its OCR stages (image -> Document with text\n * and boxes). Both run PP-OCR models through ppu-paddle-ocr on onnxruntime-web.\n */\n\nimport { OcrError } from '../core/errors.js'\nimport { decodeImage, imageDataToCanvas } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage } from '../core/pipeline.js'\nimport { boxesToFormattedText, boxesToText } from '../core/text.js'\nimport { type Box, boxFromBBox } from '../schemas/box.js'\nimport { createDetectorOutput, type DetectorOutput } from '../schemas/detector-output.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\nimport { getPaddleService } from './paddle-service.js'\nimport { DEFAULT_OCR_PRESET, validatePreset } from './presets.js'\n\n/** Recognizing per box gives word-level output; per line merges them first. */\nexport type RecognitionStrategy = 'per-box' | 'per-line' | 'cross-line'\n\nexport interface PaddleOcrParams extends BaseStageParams {\n    /** Language/script preset. See PADDLE_OCR_PRESETS. */\n    preset: string\n    /** Drop results below this confidence (0-1). */\n    scoreThreshold: number\n}\n\nexport interface PaddleTextDetectorParams extends PaddleOcrParams {}\n\nexport const PADDLE_TEXT_DETECTOR_DEFAULTS: PaddleTextDetectorParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    outputCol: 'boxes',\n    keepInputData: true,\n    preset: DEFAULT_OCR_PRESET,\n    scoreThreshold: 0,\n})\n\n/** Decode a stage input into something ppu-paddle-ocr accepts. */\nasync function toCanvas(input: unknown): Promise<OffscreenCanvas> {\n    if (typeof OffscreenCanvas !== 'undefined' && input instanceof OffscreenCanvas) return input\n    if (typeof ImageData !== 'undefined' && input instanceof ImageData) {\n        return imageDataToCanvas(input)\n    }\n\n    const image = input as ScaleDpImage | undefined\n    // Check `exception` first. A failed upstream stage returns a well-formed but\n    // empty Image, so testing the bytes first would report \"no decoded bytes\" and\n    // bury the real cause.\n    if (image?.exception) {\n        throw new OcrError(`Upstream stage failed: ${image.exception}`, 'toCanvas')\n    }\n    if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n        throw new OcrError('Expected an Image with decoded bytes', 'toCanvas')\n    }\n\n    const bitmap = await decodeImage(image.data)\n    try {\n        const canvas = new OffscreenCanvas(bitmap.width, bitmap.height)\n        const ctx = canvas.getContext('2d')\n        if (!ctx) throw new OcrError('Failed to acquire a 2D context', 'toCanvas')\n        ctx.drawImage(bitmap, 0, 0)\n        return canvas\n    } finally {\n        bitmap.close()\n    }\n}\n\n/** Text detection only: image -> word boxes, no recognition. */\nexport class PaddleTextDetector extends Stage<PaddleTextDetectorParams> {\n    readonly name = 'PaddleTextDetector'\n\n    constructor(options: Partial<PaddleTextDetectorParams> = {}) {\n        super(resolveParams(PADDLE_TEXT_DETECTOR_DEFAULTS, options, { preset: validatePreset }))\n    }\n\n    override async init(): Promise<void> {\n        await getPaddleService(this.params.preset)\n    }\n\n    protected async apply(input: unknown, row: Row): Promise<DetectorOutput> {\n        const service = await getPaddleService(this.params.preset)\n        const canvas = await toCanvas(input)\n        const { boxes } = await service.detect(canvas as never)\n\n        return createDetectorOutput({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'paddle',\n            // ppu returns axis-aligned boxes in original image coordinates.\n            bboxes: boxes.map((b) => boxFromBBox([b.x, b.y, b.x + b.width, b.y + b.height], { score: 1 })),\n        })\n    }\n\n    protected onError(message: string, row: Row): DetectorOutput {\n        return createDetectorOutput({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'paddle',\n            exception: message,\n        })\n    }\n}\n\nexport interface PaddleTextRecognizerParams extends PaddleOcrParams {\n    /**\n     * How recognized regions are grouped. 'per-box' keeps them as detected,\n     * 'per-line' merges them into lines.\n     *\n     * ScaleDP's docstring calls 'per-box' word-level. That describes the\n     * grouping, not the geometry: the boxes are whatever the preset's detector\n     * produced, and PaddleOCR's detector is line-level. No strategy can\n     * subdivide a region. `TesseractOcr` is the stage that returns words.\n     */\n    strategy: RecognitionStrategy\n    /** Rebuild the original layout with spaces and blank lines. */\n    keepFormatting: boolean\n    /** Line-grouping tolerance in pixels; 0 derives it from character height. */\n    lineTolerance: number\n}\n\nexport const PADDLE_TEXT_RECOGNIZER_DEFAULTS: PaddleTextRecognizerParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    outputCol: 'text',\n    keepInputData: true,\n    preset: DEFAULT_OCR_PRESET,\n    scoreThreshold: 0.5,\n    strategy: 'per-box' as RecognitionStrategy,\n    keepFormatting: false,\n    lineTolerance: 0,\n})\n\n/** Full OCR: image -> Document with text and word-level boxes. */\nexport class PaddleTextRecognizer extends Stage<PaddleTextRecognizerParams> {\n    readonly name = 'PaddleTextRecognizer'\n\n    constructor(options: Partial<PaddleTextRecognizerParams> = {}) {\n        super(resolveParams(PADDLE_TEXT_RECOGNIZER_DEFAULTS, options, { preset: validatePreset }))\n    }\n\n    override async init(): Promise<void> {\n        await getPaddleService(this.params.preset)\n    }\n\n    protected async apply(input: unknown, row: Row): Promise<Document> {\n        const { preset, strategy, scoreThreshold, keepFormatting, lineTolerance } = this.params\n        const service = await getPaddleService(preset)\n        const canvas = await toCanvas(input)\n\n        const result = await service.recognize(canvas as never, {\n            flatten: true,\n            strategy,\n            // ppu caches globally, keyed on pixels. Without this, switching\n            // preset returns the *previous* model's result for the same image.\n            noCache: true,\n        })\n\n        const items = 'results' in result ? result.results : []\n        const bboxes: Box[] = items\n            .filter((item) => item.confidence >= scoreThreshold)\n            .map((item) =>\n                boxFromBBox(\n                    [item.box.x, item.box.y, item.box.x + item.box.width, item.box.y + item.box.height],\n                    { text: item.text, score: item.confidence }\n                )\n            )\n\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'ocr',\n            text: keepFormatting ? boxesToFormattedText(bboxes, lineTolerance) : boxesToText(bboxes),\n            bboxes,\n        })\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'ocr',\n            exception: message,\n        })\n    }\n}\n","/**\n * Read the boxes a detector already found, with a PaddleOCR preset.\n *\n * The Paddle counterpart to `TesseractRecognizer`, and for the same reason:\n * `PaddleTextRecognizer` detects and recognises in a single pass over the page,\n * so boxes produced by a *separate* detector never reach it. This stage takes\n * those boxes, straightens each one, and reads it -- which is what lets DBNet or\n * YOLO feed PaddleOCR recognition, rotated regions included.\n *\n * There is no `strategy` parameter. ppu's 'per-line' and 'cross-line' merge\n * boxes before reading them; the contract here is one result per box handed in,\n * which only 'per-box' can honour.\n */\n\nimport { OcrError } from '../core/errors.js'\nimport { context2d, createCanvas, cropBox, decodeImage, resize, rotate180 } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { boxesToFormattedText, boxesToText } from '../core/text.js'\nimport { type Box, isRotated } from '../schemas/box.js'\nimport type { DetectorOutput } from '../schemas/detector-output.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\nimport { DEFAULT_ORIENTATION_MODEL, LineOrientationClassifier } from './line-orientation.js'\nimport { getPaddleRecognizer, type PaddleRecognitionService } from './paddle-service.js'\nimport { DEFAULT_OCR_PRESET, validatePreset } from './presets.js'\n\nexport interface PaddleRecognizerParams extends BaseStageParams {\n    /** [imageColumn, boxColumn]. */\n    inputCols: string[]\n    /** Language/script preset. See PADDLE_OCR_PRESETS. */\n    preset: string\n    /** Resize the page by this factor before cropping. */\n    scaleFactor: number\n    /** Grow each box before cropping. ScaleDP hardcodes 5. */\n    padding: number\n    /** Drop regions below this confidence (0-1). */\n    scoreThreshold: number\n    /** Rebuild the original layout with spaces and blank lines. */\n    keepFormatting: boolean\n    /** Line-grouping tolerance in pixels; 0 derives it from character height. */\n    lineTolerance: number\n    /**\n     * Classify each crop 0/180 degrees and turn the inverted ones.\n     *\n     * Off by default, unlike `TesseractRecognizer`. Paddle already turns a crop\n     * that is markedly taller than wide, so only the 180 degree flip is missing\n     * -- and catching it costs a separate ~9 MB model an ordinary page never\n     * needs. Turn it on for scans that come in upside down.\n     */\n    detectLineOrientation: boolean\n    /** Recognize only boxes that are rotated or came back inverted. */\n    onlyRotated: boolean\n    oriModel: string\n    /**\n     * Recover inter-word spaces the greedy CTC decode drops. Helps Latin text\n     * where the model collapses word gaps; can add spurious ones in dense\n     * symbol runs.\n     */\n    spaceRecovery: boolean\n    /** Crops per batched inference. 1 disables batching. */\n    recBatchSize: number\n}\n\nexport const PADDLE_RECOGNIZER_DEFAULTS: PaddleRecognizerParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    inputCols: ['image', 'boxes'],\n    outputCol: 'text',\n    keepInputData: true,\n    preset: DEFAULT_OCR_PRESET,\n    scaleFactor: 1,\n    padding: 5,\n    scoreThreshold: 0.5,\n    keepFormatting: false,\n    lineTolerance: 0,\n    detectLineOrientation: false,\n    onlyRotated: false,\n    oriModel: DEFAULT_ORIENTATION_MODEL,\n    spaceRecovery: false,\n    recBatchSize: 6,\n})\n\n/**\n * Tallest sheet crops are stacked onto.\n *\n * Well under every browser's canvas ceiling, and low enough that a sheet stays\n * cheap to allocate. A single crop taller than this gets a sheet to itself.\n */\nconst MAX_SHEET_HEIGHT = 8192\n\nfunction boxesOf(source: unknown): Box[] {\n    if (typeof source !== 'object' || source === null) return []\n    return (source as DetectorOutput).bboxes ?? []\n}\n\nexport class PaddleRecognizer extends Stage<PaddleRecognizerParams> {\n    readonly name = 'PaddleRecognizer'\n\n    private orientation: LineOrientationClassifier | null = null\n    private recognition: PaddleRecognitionService | null = null\n\n    constructor(options: Partial<PaddleRecognizerParams> = {}) {\n        super(\n            resolveParams(PADDLE_RECOGNIZER_DEFAULTS, options, {\n                inputCols: (value) => {\n                    if (value.length !== 2) {\n                        throw new RangeError('inputCols must be [imageColumn, boxColumn]')\n                    }\n                },\n                preset: validatePreset,\n            })\n        )\n    }\n\n    override async init(): Promise<void> {\n        const { preset, recBatchSize, spaceRecovery, detectLineOrientation, oriModel } = this.params\n        this.recognition ??= await getPaddleRecognizer(preset, { recBatchSize, spaceRecovery })\n        if (detectLineOrientation) {\n            this.orientation ??= new LineOrientationClassifier(oriModel)\n        }\n    }\n\n    protected async apply(_input: unknown, row: Row, ctx: StageContext): Promise<Document> {\n        const {\n            inputCols,\n            scaleFactor,\n            padding,\n            scoreThreshold,\n            keepFormatting,\n            lineTolerance,\n            detectLineOrientation,\n            onlyRotated,\n        } = this.params\n        const [imageCol, boxCol] = inputCols as [string, string]\n        const image = row[imageCol] as ScaleDpImage | undefined\n\n        // Check `exception` first. A failed upstream stage returns a well-formed\n        // but empty Image, so testing the bytes first would report \"no decoded\n        // bytes\" and bury the real cause.\n        if (image?.exception) {\n            throw new OcrError(`Upstream stage failed: ${image.exception}`, this.name)\n        }\n        if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new OcrError('Expected an Image with decoded bytes', this.name)\n        }\n\n        const source = row[boxCol]\n        if (source === undefined) {\n            throw new OcrError(\n                `No boxes in column \"${boxCol}\". This stage reads a detector's output; ` +\n                    'run a text detector before it.',\n                this.name\n            )\n        }\n\n        await this.init()\n        const recognition = this.recognition as PaddleRecognitionService\n        const bitmap = await decodeImage(image.data)\n\n        // ScaleDP resizes the *page* and scales each box to index into it, which\n        // is how a small line is handed to the model at a readable size. The\n        // boxes it reports back are the originals, untouched -- so nothing here\n        // has to map coordinates out again.\n        const canvas = scaleFactor === 1 ? bitmap : resize(bitmap, scaleFactor)\n\n        // Kept index-aligned: `crops[i]` is what `kept[i]` became.\n        const kept: Box[] = []\n        const crops: OffscreenCanvas[] = []\n\n        try {\n            for (const box of boxesOf(source)) {\n                ctx.signal?.throwIfAborted()\n\n                // Straightens rotated boxes rather than taking their envelope,\n                // which is what makes a skewed line readable at all -- and what\n                // ppu's own cropping, being axis-aligned, cannot do.\n                let crop = cropBox(canvas, box, { scaleFactor, padding })\n\n                let inverted = false\n                if (detectLineOrientation && this.orientation) {\n                    inverted = (await this.orientation.classify(crop)) === '180_degree'\n                    if (inverted) crop = rotate180(crop)\n                }\n\n                // ScaleDP's onlyRotated: an upright, right-way-up box was\n                // already handled by whatever pass produced it.\n                if (onlyRotated && !isRotated(box) && !inverted) continue\n\n                kept.push(box)\n                crops.push(crop)\n            }\n        } finally {\n            bitmap.close()\n        }\n\n        const read = await recognizeCrops(recognition, crops, ctx)\n\n        const bboxes: Box[] = []\n        for (const [i, box] of kept.entries()) {\n            const result = read[i]\n            if (!result?.text) continue\n            if (result.score < scoreThreshold) continue\n            bboxes.push({ ...box, text: result.text, score: result.score })\n        }\n\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? image.path),\n            type: 'paddle-recognizer',\n            text: keepFormatting ? boxesToFormattedText(bboxes, lineTolerance) : boxesToText(bboxes),\n            bboxes,\n        })\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'paddle-recognizer',\n            exception: message,\n        })\n    }\n\n    override async dispose(): Promise<void> {\n        await this.orientation?.dispose()\n        this.orientation = null\n        // The recognition session is shared per preset and owned by\n        // paddle-service; `disposePaddleServices()` is what releases it.\n        this.recognition = null\n    }\n}\n\ninterface ReadResult {\n    text: string\n    score: number\n}\n\n/**\n * Read every crop, batching across them.\n *\n * ppu batches `recBatchSize` crops into a single inference, but only within one\n * `run()` call, and `run()` cuts its crops out of the one canvas it is given. So\n * the crops are stacked onto sheet canvases -- each at x 0, one below the last --\n * and handed back as the boxes to read. ppu re-crops exactly those rects, so the\n * unused width beside a narrow crop is never sampled. Calling `run()` once per\n * box would instead pay one inference, and one main-thread yield, per line.\n *\n * Results come back index-aligned to `crops`; a crop ppu rejected or did not\n * return is left `undefined`. `run()` sorts what it returns into reading order,\n * so results are matched on the slot's y offset, never on array position.\n */\nasync function recognizeCrops(\n    recognition: PaddleRecognitionService,\n    crops: readonly OffscreenCanvas[],\n    ctx: StageContext\n): Promise<(ReadResult | undefined)[]> {\n    const out: (ReadResult | undefined)[] = new Array(crops.length)\n\n    for (let start = 0; start < crops.length; ) {\n        ctx.signal?.throwIfAborted()\n\n        // Take as many crops as fit on one sheet, but always at least one, so a\n        // crop taller than the ceiling is still read.\n        let end = start\n        let height = 0\n        let width = 0\n        while (end < crops.length) {\n            const crop = crops[end] as OffscreenCanvas\n            if (end > start && height + crop.height > MAX_SHEET_HEIGHT) break\n            height += crop.height\n            width = Math.max(width, crop.width)\n            end++\n        }\n\n        const sheet = createCanvas(width, height)\n        const sheetCtx = context2d(sheet)\n        const slots: { x: number; y: number; width: number; height: number }[] = []\n        const atOffset = new Map<number, number>()\n        let offset = 0\n        for (let i = start; i < end; i++) {\n            const crop = crops[i] as OffscreenCanvas\n            sheetCtx.drawImage(crop, 0, offset)\n            slots.push({ x: 0, y: offset, width: crop.width, height: crop.height })\n            atOffset.set(offset, i)\n            offset += crop.height\n        }\n\n        const results = await recognition.run(sheet as never, slots, undefined, 'per-box')\n        for (const result of results) {\n            const index = atOffset.get(result.box.y)\n            if (index !== undefined) out[index] = { text: result.text.trim(), score: result.confidence }\n        }\n\n        start = end\n    }\n\n    return out\n}\n","/**\n * Writing-script detection via Tesseract's OSD model.\n *\n * Kept separate from recognition on purpose: when the recognition model is\n * wrong for the page the text comes back garbled, and that is precisely when\n * knowing the script is most useful. Running OSD independently means the\n * answer stays trustworthy.\n *\n * Feeds the preset picker -- see `presetsForScript`.\n */\n\nimport { OcrError } from '../core/errors.js'\nimport { decodeImage, imageDataToCanvas, toImageData } from '../core/image.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\nimport { type OcrPreset, presetsForScript } from './presets.js'\n\nexport interface DetectedScript {\n    /** Script name as Tesseract reports it, e.g. 'Latin', 'Cyrillic', 'Han'. */\n    script: string\n    /** Tesseract's own confidence; not a probability. */\n    confidence: number\n}\n\ntype TesseractJs = typeof import('tesseract.js')\ntype OsdWorker = Awaited<ReturnType<TesseractJs['createWorker']>>\n\nlet workerPromise: Promise<OsdWorker> | null = null\n\nasync function getOsdWorker(): Promise<OsdWorker> {\n    if (workerPromise) return workerPromise\n\n    workerPromise = (async () => {\n        let mod: TesseractJs\n        try {\n            mod = await import('tesseract.js')\n        } catch (cause) {\n            throw new Error('tesseract.js is required for script detection. Install it: npm i tesseract.js', {\n                cause,\n            })\n        }\n        // OEM 0 (legacy engine) is the only one that carries OSD data.\n        return mod.createWorker('osd', mod.OEM.TESSERACT_ONLY)\n    })()\n\n    workerPromise.catch(() => {\n        workerPromise = null\n    })\n    return workerPromise\n}\n\n/** Detect the dominant writing script on a page. Returns null when unsure. */\nexport async function detectScript(\n    source: ImageBitmap | OffscreenCanvas | ImageData | ScaleDpImage\n): Promise<DetectedScript | null> {\n    const worker = await getOsdWorker()\n\n    let canvas: OffscreenCanvas\n    if (typeof OffscreenCanvas !== 'undefined' && source instanceof OffscreenCanvas) {\n        canvas = source\n    } else if (typeof ImageData !== 'undefined' && source instanceof ImageData) {\n        canvas = imageDataToCanvas(source)\n    } else if (typeof ImageBitmap !== 'undefined' && source instanceof ImageBitmap) {\n        canvas = imageDataToCanvas(toImageData(source))\n    } else {\n        const image = source as ScaleDpImage\n        if (!(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new OcrError('Expected an Image with decoded bytes', 'detectScript')\n        }\n        const bitmap = await decodeImage(image.data)\n        try {\n            canvas = imageDataToCanvas(toImageData(bitmap))\n        } finally {\n            bitmap.close()\n        }\n    }\n\n    const blob = await canvas.convertToBlob({ type: 'image/png' })\n    const { data } = await worker.detect(blob)\n    if (!data?.script) return null\n    return { script: data.script, confidence: data.script_confidence ?? 0 }\n}\n\n/** Presets able to read the detected script, best-first. */\nexport async function suggestPresets(\n    source: ImageBitmap | OffscreenCanvas | ImageData | ScaleDpImage\n): Promise<OcrPreset[]> {\n    const detected = await detectScript(source)\n    return detected ? presetsForScript(detected.script) : []\n}\n\n/** Tear down the shared OSD worker. */\nexport async function disposeScriptDetection(): Promise<void> {\n    const pending = workerPromise\n    workerPromise = null\n    await pending?.then((w) => w.terminate()).catch(() => undefined)\n}\n","/**\n * Tesseract OCR, mirroring ScaleDP's `TesseractOcr` stage.\n *\n * Recognition runs on tesseract-wasm; script detection uses tesseract.js's OSD\n * model. The two are deliberately independent -- script detection still works\n * when the recognition model is wrong for the page and garbles the text, which\n * is exactly when you most want to know the script.\n *\n * Both the worker URL and the traineddata location come from `configure()`.\n * The pdftools prototype hardcoded `/tesseract-worker.js` and a raw GitHub URL,\n * neither of which a library can assume.\n */\n\nimport { getConfig } from '../core/config.js'\nimport { OcrError } from '../core/errors.js'\nimport { decodeImage, toImageData } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage } from '../core/pipeline.js'\nimport { boxesToFormattedText, boxesToText } from '../core/text.js'\nimport { type Box, boxFromBBox } from '../schemas/box.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\n\n/** tessdata_fast is a good default: far smaller than tessdata, barely less accurate. */\nexport const DEFAULT_TESSDATA_URL = 'https://raw.githubusercontent.com/tesseract-ocr/tessdata_fast/main/'\n\nexport interface TesseractOcrParams extends BaseStageParams {\n    /** Language codes, e.g. ['eng'] or ['eng', 'deu']. */\n    lang: readonly string[]\n    /** Drop words below this confidence (0-1). */\n    scoreThreshold: number\n    keepFormatting: boolean\n    lineTolerance: number\n}\n\nexport const TESSERACT_OCR_DEFAULTS: TesseractOcrParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    outputCol: 'text',\n    keepInputData: true,\n    lang: ['eng'] as readonly string[],\n    scoreThreshold: 0.5,\n    keepFormatting: false,\n    lineTolerance: 0,\n})\n\ntype TesseractWasm = typeof import('tesseract-wasm')\ntype OcrClient = InstanceType<TesseractWasm['OCRClient']>\n\nlet clientPromise: Promise<OcrClient> | null = null\nlet loadedLanguage: string | null = null\n\nfunction trainedDataUrl(lang: string): string {\n    const base = (getConfig().tesseract.dataUrl ?? DEFAULT_TESSDATA_URL).replace(/\\/?$/, '/')\n    return `${base}${lang}.traineddata`\n}\n\n/**\n * The shared Tesseract client, loading the model for `lang` if needed.\n *\n * Exported so the crop-based recognizer reuses this one client rather than\n * standing up a second worker and loading the same traineddata twice.\n */\nexport async function getTesseractClient(lang: string): Promise<OcrClient> {\n    if (clientPromise && loadedLanguage === lang) return clientPromise\n\n    // A client holds one model; switching language means a fresh one.\n    if (clientPromise) {\n        const previous = clientPromise\n        clientPromise = null\n        await previous.then((c) => c.destroy()).catch(() => undefined)\n    }\n\n    loadedLanguage = lang\n    clientPromise = (async () => {\n        let mod: TesseractWasm\n        try {\n            mod = await import('tesseract-wasm')\n        } catch (cause) {\n            throw new Error(\n                'tesseract-wasm is required for the Tesseract engine. Install it: npm i tesseract-wasm',\n                { cause }\n            )\n        }\n\n        const { workerUrl } = getConfig().tesseract\n        const client = new mod.OCRClient(workerUrl ? { workerURL: workerUrl } : {})\n        const response = await fetch(trainedDataUrl(lang))\n        if (!response.ok) {\n            throw new OcrError(\n                `Failed to fetch ${lang}.traineddata: ${response.status} ${response.statusText}`,\n                'TesseractOcr'\n            )\n        }\n        await client.loadModel(new Uint8Array(await response.arrayBuffer()))\n        return client\n    })()\n\n    clientPromise.catch(() => {\n        clientPromise = null\n        loadedLanguage = null\n    })\n    return clientPromise\n}\n\n/** Tear down the shared Tesseract client. */\nexport async function disposeTesseract(): Promise<void> {\n    const pending = clientPromise\n    clientPromise = null\n    loadedLanguage = null\n    await pending?.then((c) => c.destroy()).catch(() => undefined)\n}\n\nexport class TesseractOcr extends Stage<TesseractOcrParams> {\n    readonly name = 'TesseractOcr'\n\n    constructor(options: Partial<TesseractOcrParams> = {}) {\n        super(\n            resolveParams(TESSERACT_OCR_DEFAULTS, options, {\n                lang: (value) => {\n                    if (value.length === 0) throw new RangeError('lang must not be empty')\n                },\n            })\n        )\n    }\n\n    /** tesseract-wasm loads one model, so multi-language means a joined code. */\n    private get language(): string {\n        return this.params.lang.join('+')\n    }\n\n    override async init(): Promise<void> {\n        await getTesseractClient(this.language)\n    }\n\n    protected async apply(input: unknown, row: Row): Promise<Document> {\n        const image = input as ScaleDpImage | undefined\n        // Check `exception` first. A failed upstream stage returns a well-formed but\n        // empty Image, so testing the bytes first would report \"no decoded bytes\" and\n        // bury the real cause.\n        if (image?.exception) {\n            throw new OcrError(`Upstream stage failed: ${image.exception}`, this.name)\n        }\n        if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new OcrError('Expected an Image with decoded bytes', this.name)\n        }\n\n        const client = await getTesseractClient(this.language)\n        const bitmap = await decodeImage(image.data)\n        let words: import('tesseract-wasm').TextItem[]\n        try {\n            await client.loadImage(toImageData(bitmap))\n            // getTextBoxes, not getBoundingBoxes: the latter is layout analysis\n            // only and returns geometry with no text and no confidence.\n            words = await client.getTextBoxes('word')\n        } finally {\n            bitmap.close()\n        }\n\n        const { scoreThreshold, keepFormatting, lineTolerance } = this.params\n        const bboxes: Box[] = words\n            // tesseract-wasm reports confidence on a 0-1 scale, not 0-100.\n            .filter((word) => word.confidence >= scoreThreshold && word.text.trim().length > 0)\n            .map((word) =>\n                boxFromBBox([word.rect.left, word.rect.top, word.rect.right, word.rect.bottom], {\n                    text: word.text.trim(),\n                    score: word.confidence,\n                })\n            )\n\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'tesseract',\n            text: keepFormatting ? boxesToFormattedText(bboxes, lineTolerance) : boxesToText(bboxes),\n            bboxes,\n        })\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'tesseract',\n            exception: message,\n        })\n    }\n\n    override async dispose(): Promise<void> {\n        await disposeTesseract()\n    }\n}\n","/**\n * Recognize text inside boxes a detector already found.\n *\n * Port of ScaleDP's `TesseractRecognizer` (a `BaseRecognizer`), which is the\n * half of the OCR story `PaddleTextRecognizer` cannot cover: Paddle detects and\n * recognises in a single pass over the page, so boxes produced by a *separate*\n * detector never reach it -- rotated boxes in particular. This stage takes those\n * boxes, straightens each one, and reads it.\n */\n\nimport { OcrError } from '../core/errors.js'\nimport type { Point } from '../core/geometry.js'\nimport { cropBox, cropGeometry, decodeImage, rotate180, toImageData } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { boxesToFormattedText, boxesToText } from '../core/text.js'\nimport { type Box, boxFromPolygon, isRotated } from '../schemas/box.js'\nimport type { DetectorOutput } from '../schemas/detector-output.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport type { ScaleDpImage } from '../schemas/image.js'\nimport { DEFAULT_ORIENTATION_MODEL, LineOrientationClassifier } from './line-orientation.js'\nimport { getTesseractClient } from './tesseract.js'\n\nexport interface TesseractRecognizerParams extends BaseStageParams {\n    /** [imageColumn, boxColumn]. */\n    inputCols: string[]\n    lang: readonly string[]\n    /** Resize the page by this factor before cropping. */\n    scaleFactor: number\n    /** Grow each box before cropping. ScaleDP hardcodes 5. */\n    padding: number\n    /** Drop words below this confidence (0-1). */\n    scoreThreshold: number\n    keepFormatting: boolean\n    lineTolerance: number\n    /**\n     * Granularity of the boxes returned.\n     *\n     * 'region' keeps ScaleDP's behaviour: one box per region the detector\n     * found, carrying everything read inside it. Since the detectors here are\n     * line-level, so are those boxes.\n     *\n     * 'word' returns instead the boxes tesseract reports for each word, mapped\n     * back through the crop -- rotation included -- into page coordinates. Not\n     * in Python ScaleDP, which always returns one box per region.\n     */\n    boxLevel: 'region' | 'word'\n    /** Classify each crop 0/180 degrees and turn the inverted ones. */\n    detectLineOrientation: boolean\n    /**\n     * Recognize only boxes that are rotated or came back inverted.\n     *\n     * ScaleDP defaults this to true because there the stage refines an OCR pass\n     * that already ran. Standalone it is the primary recognizer, and skipping\n     * every upright box would return an empty document for the ordinary case,\n     * so the default is flipped here. See docs/porting.md.\n     */\n    onlyRotated: boolean\n    oriModel: string\n}\n\nexport const TESSERACT_RECOGNIZER_DEFAULTS: TesseractRecognizerParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'image',\n    inputCols: ['image', 'boxes'],\n    outputCol: 'text',\n    keepInputData: true,\n    lang: ['eng'] as readonly string[],\n    scaleFactor: 1,\n    padding: 5,\n    scoreThreshold: 0.5,\n    keepFormatting: false,\n    lineTolerance: 0,\n    boxLevel: 'region' as 'region' | 'word',\n    detectLineOrientation: true,\n    onlyRotated: false,\n    oriModel: DEFAULT_ORIENTATION_MODEL,\n})\n\nfunction boxesOf(source: unknown): Box[] {\n    if (typeof source !== 'object' || source === null) return []\n    return (source as DetectorOutput).bboxes ?? []\n}\n\nexport class TesseractRecognizer extends Stage<TesseractRecognizerParams> {\n    readonly name = 'TesseractRecognizer'\n\n    private orientation: LineOrientationClassifier | null = null\n\n    constructor(options: Partial<TesseractRecognizerParams> = {}) {\n        super(\n            resolveParams(TESSERACT_RECOGNIZER_DEFAULTS, options, {\n                inputCols: (value) => {\n                    if (value.length !== 2) {\n                        throw new RangeError('inputCols must be [imageColumn, boxColumn]')\n                    }\n                },\n                lang: (value) => {\n                    if (value.length === 0) throw new RangeError('lang must not be empty')\n                },\n            })\n        )\n    }\n\n    private get language(): string {\n        return this.params.lang.join('+')\n    }\n\n    override async init(): Promise<void> {\n        await getTesseractClient(this.language)\n        if (this.params.detectLineOrientation) {\n            this.orientation ??= new LineOrientationClassifier(this.params.oriModel)\n        }\n    }\n\n    protected async apply(_input: unknown, row: Row, ctx: StageContext): Promise<Document> {\n        const {\n            inputCols,\n            scaleFactor,\n            padding,\n            scoreThreshold,\n            keepFormatting,\n            lineTolerance,\n            boxLevel,\n            detectLineOrientation,\n            onlyRotated,\n        } = this.params\n        const [imageCol, boxCol] = inputCols as [string, string]\n        const image = row[imageCol] as ScaleDpImage | undefined\n\n        if (image?.exception) {\n            throw new OcrError(`Upstream stage failed: ${image.exception}`, this.name)\n        }\n        if (!image || !(image.data instanceof Uint8Array) || image.data.byteLength === 0) {\n            throw new OcrError('Expected an Image with decoded bytes', this.name)\n        }\n\n        const source = row[boxCol]\n        if (source === undefined) {\n            throw new OcrError(\n                `No boxes in column \"${boxCol}\". This stage reads a detector's output; ` +\n                    'run a text detector before it.',\n                this.name\n            )\n        }\n\n        await this.init()\n        const client = await getTesseractClient(this.language)\n        const bitmap = await decodeImage(image.data)\n        const recognized: Box[] = []\n\n        try {\n            for (const box of boxesOf(source)) {\n                ctx.signal?.throwIfAborted()\n\n                // Straightens rotated boxes rather than taking their envelope,\n                // which is what makes a skewed line readable at all. The same\n                // geometry maps word boxes back out again.\n                const geometry = cropGeometry(box, { scaleFactor, padding })\n                let crop = cropBox(bitmap, box, { scaleFactor, padding })\n\n                let inverted = false\n                if (detectLineOrientation && this.orientation) {\n                    inverted = (await this.orientation.classify(crop)) === '180_degree'\n                    if (inverted) crop = rotate180(crop)\n                }\n\n                // ScaleDP's onlyRotated: an upright, right-way-up box was\n                // already handled by whatever pass produced it.\n                if (onlyRotated && !isRotated(box) && !inverted) continue\n\n                await client.loadImage(toImageData(crop))\n                const items = await client.getTextBoxes('word')\n\n                const words = items.filter((item) => item.text.trim())\n                if (words.length === 0) continue\n\n                // tesseract-wasm reports confidence on a 0-1 scale, not 0-100.\n                // ScaleDP writes its confidence to `conf` and then filters on\n                // `score`, so its threshold silently applies to the detector's\n                // score instead; the value goes where it is read here.\n                // Averaged over every item tesseract returned, empty ones\n                // included, which is what the region mode did before word boxes\n                // existed -- so a pipeline that was passing this gate still is.\n                const score = items.reduce((sum, item) => sum + item.confidence, 0) / (items.length || 1)\n\n                // The gate is the region's score in both modes, so `boxLevel`\n                // changes how finely the result is cut up and nothing else. Per\n                // word it would also change *what was read*: a word scoring 0.3\n                // between two at 0.9 rides out on the region's mean, and would\n                // vanish on its own -- so the same page would yield different\n                // text depending on the box size asked for. Each word still\n                // carries its own confidence for filtering further downstream.\n                if (score < scoreThreshold) continue\n\n                if (boxLevel === 'word') {\n                    for (const item of words) {\n                        recognized.push(\n                            wordBox(\n                                item.rect,\n                                geometry,\n                                inverted,\n                                scaleFactor,\n                                item.text.trim(),\n                                item.confidence\n                            )\n                        )\n                    }\n                    continue\n                }\n\n                recognized.push({ ...box, text: words.map((item) => item.text.trim()).join(' '), score })\n            }\n        } finally {\n            bitmap.close()\n        }\n\n        // Coordinates came from the scaled page, so bring them back. Word boxes\n        // are already in page coordinates -- `wordBox` divides as it maps, so it\n        // rounds once rather than twice.\n        const bboxes =\n            scaleFactor === 1 || boxLevel === 'word'\n                ? recognized\n                : recognized.map((box) => ({\n                      ...box,\n                      x: Math.round(box.x / scaleFactor),\n                      y: Math.round(box.y / scaleFactor),\n                      width: Math.round(box.width / scaleFactor),\n                      height: Math.round(box.height / scaleFactor),\n                  }))\n\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? image.path),\n            type: 'tesseract-recognizer',\n            text: keepFormatting ? boxesToFormattedText(bboxes, lineTolerance) : boxesToText(bboxes),\n            bboxes,\n        })\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'tesseract-recognizer',\n            exception: message,\n        })\n    }\n\n    override async dispose(): Promise<void> {\n        await this.orientation?.dispose()\n        this.orientation = null\n    }\n}\n\n/**\n * One word's box, in the page's coordinates.\n *\n * tesseract reports the rect in the crop's own space: straightened, padded,\n * scaled, and turned the right way up if the line was upside down. Undo the\n * turn, then push the four corners back through the crop's own transform, so a\n * word inside a skewed line comes back skewed the same way.\n */\nfunction wordBox(\n    rect: { left: number; top: number; right: number; bottom: number },\n    geometry: ReturnType<typeof cropGeometry>,\n    inverted: boolean,\n    scaleFactor: number,\n    text: string,\n    score: number\n): Box {\n    const { width, height, map } = geometry\n    // rotate180 maps (x, y) to (w - x, h - y), so the rect's corners swap.\n    const [left, top, right, bottom] = inverted\n        ? [width - rect.right, height - rect.bottom, width - rect.left, height - rect.top]\n        : [rect.left, rect.top, rect.right, rect.bottom]\n\n    const toPage = (x: number, y: number): Point => {\n        const [px, py] = map(x, y)\n        return [px / scaleFactor, py / scaleFactor]\n    }\n    const corners: Point[] = [\n        toPage(left, top),\n        toPage(right, top),\n        toPage(right, bottom),\n        toPage(left, bottom),\n    ]\n    return boxFromPolygon(corners, { text, score })\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AA4BA,MAAa,0BAAgD,OAAO,OAAO;CACvE,QAAQ;CACR,WAAW;CACX,eAAe;CACf,aAAa;CACb,SAAS;AACb,CAAC;;;;;;;;AAqBD,SAAgB,wBACZ,KACA,QACA,eACS;CACT,MAAM,EAAE,MAAM,OAAO,WAAW;CAChC,MAAM,UAAU,IAAI,WAAW,QAAQ,MAAM;CAC7C,MAAM,aAAwB,CAAC;CAE/B,MAAM,gBAAgB,GAAW,MAC7B,KAAK,KAAK,KAAK,KAAK,IAAI,SAAS,IAAI,UAAW,KAAK,IAAI,QAAQ,KAAgB;CAErF,MAAM,QAAkB,CAAC;CACzB,KAAK,IAAI,QAAQ,GAAG,QAAQ,QAAQ,UAAU,WAAW,SAAS,eAAe,SAAS;EACtF,IAAI,QAAQ,UAAU,EAAG,KAAK,SAAoB,SAAS;EAE3D,MAAM,WAAoB,CAAC;EAC3B,MAAM,SAAS;EACf,MAAM,KAAK,KAAK;EAChB,QAAQ,SAAS;EAEjB,OAAO,MAAM,SAAS,GAAG;GACrB,MAAM,QAAQ,MAAM,IAAI;GACxB,MAAM,IAAI,QAAQ;GAClB,MAAM,KAAK,QAAQ,KAAK;GAExB,IAAI,aAAa;GACjB,KAAK,IAAI,KAAK,IAAI,MAAM,GAAG,MACvB,KAAK,IAAI,KAAK,IAAI,MAAM,GAAG,MAAM;IAC7B,IAAI,OAAO,KAAK,OAAO,GAAG;IAC1B,MAAM,KAAK,IAAI;IACf,MAAM,KAAK,IAAI;IACf,IAAI,CAAC,aAAa,IAAI,EAAE,GAAG;KACvB,aAAa;KACb;IACJ;IACA,MAAM,YAAY,KAAK,QAAQ;IAC/B,IAAI,CAAC,QAAQ,YAAY;KACrB,QAAQ,aAAa;KACrB,MAAM,KAAK,SAAS;IACxB;GACJ;GAEJ,IAAI,YAAY,SAAS,KAAK,CAAC,GAAG,CAAC,CAAC;EACxC;EACA,IAAI,SAAS,UAAU,GAAG,WAAW,KAAK,QAAQ;CACtD;CACA,OAAO;AACX;;;;;;;;AASA,SAAgB,QAAQ,QAAmF;CACvG,MAAM,OAAO,YAAY,MAAM;CAG/B,MAAM,CAAC,IAAI,IAAI,IAAI,MAFH,CAAC,GAAG,UAAU,IAAI,CAAC,CAAC,CAAC,MAAM,GAAG,MAAM,EAAE,KAAK,EAAE,EAE9B;CAC/B,MAAM,CAAC,SAAS,cAAc,GAAG,MAAM,GAAG,KAAK,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE;CACjE,MAAM,CAAC,UAAU,eAAe,GAAG,MAAM,GAAG,KAAK,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE;CAEnE,OAAO;EACH,QAAQ;GAAC;GAAS;GAAU;GAAa;EAAU;EACnD,OAAO,KAAK,IAAI,KAAK,KAAK,IAAI,KAAK,KAAK,EAAE;CAC9C;AACJ;;;;;;;;;;;;AAaA,SAAgB,SAAS,KAAqB,QAAkC;CAC5E,MAAM,EAAE,MAAM,OAAO,WAAW;CAChC,MAAM,KAAK,OAAO,KAAK,MAAM,EAAE,EAAE;CACjC,MAAM,KAAK,OAAO,KAAK,MAAM,EAAE,EAAE;CAEjC,MAAM,SAAS,OAAe,QAAgB,KAAK,IAAI,KAAK,IAAI,OAAO,CAAC,GAAG,GAAG;CAC9E,MAAM,OAAO,MAAM,KAAK,MAAM,KAAK,IAAI,GAAG,EAAE,CAAC,GAAG,QAAQ,CAAC;CACzD,MAAM,OAAO,MAAM,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,CAAC,GAAG,QAAQ,CAAC;CACxD,MAAM,OAAO,MAAM,KAAK,MAAM,KAAK,IAAI,GAAG,EAAE,CAAC,GAAG,SAAS,CAAC;CAC1D,MAAM,OAAO,MAAM,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,CAAC,GAAG,SAAS,CAAC;CACzD,IAAI,OAAO,QAAQ,OAAO,MAAM,OAAO;CAGvC,MAAM,QAAQ,OAAO,KAAK,CAAC,GAAG,OAAO,CAAC,KAAK,MAAM,IAAI,IAAI,GAAG,KAAK,MAAM,IAAI,IAAI,CAAC,CAAU;CAC1F,MAAM,YAAY,OAAO;CACzB,MAAM,aAAa,OAAO;CAE1B,IAAI,MAAM;CACV,IAAI,QAAQ;CACZ,KAAK,IAAI,IAAI,GAAG,KAAK,YAAY,KAAK;EAClC,IAAI,OAAO,OAAO;EAClB,IAAI,QAAQ,OAAO;EAEnB,KAAK,IAAI,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;GACnC,MAAM,IAAI,MAAM;GAChB,MAAM,IAAI,OAAO,IAAI,KAAK,MAAM;GAChC,IAAI,EAAE,OAAO,EAAE,IAAI;IAEf,IAAI,EAAE,OAAO,GAAG;IAChB,OAAO,KAAK,IAAI,MAAM,EAAE,IAAI,EAAE,EAAE;IAChC,QAAQ,KAAK,IAAI,OAAO,EAAE,IAAI,EAAE,EAAE;IAClC;GACJ;GACA,IAAI,IAAI,KAAK,IAAI,EAAE,IAAI,EAAE,EAAE,KAAK,IAAI,KAAK,IAAI,EAAE,IAAI,EAAE,EAAE,GAAG;GAC1D,MAAM,IAAI,EAAE,MAAO,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,OAAQ,EAAE,KAAK,EAAE;GAC1D,OAAO,KAAK,IAAI,MAAM,CAAC;GACvB,QAAQ,KAAK,IAAI,OAAO,CAAC;EAC7B;EACA,IAAI,OAAO,OAAO;EAIlB,MAAM,OAAO,KAAK,IAAI,GAAG,KAAK,KAAK,IAAI,CAAC;EACxC,MAAM,KAAK,KAAK,IAAI,WAAW,KAAK,MAAM,KAAK,CAAC;EAChD,MAAM,aAAa,IAAI,QAAQ,QAAQ;EACvC,KAAK,IAAI,IAAI,MAAM,KAAK,IAAI,KAAK;GAC7B,OAAO,KAAK,YAAY;GACxB;EACJ;CACJ;CACA,OAAO,UAAU,IAAI,IAAI,MAAM;AACnC;;;;;;;;;;AAWA,SAAgB,WAAW,QAA0B,aAAmD;CACpG,MAAM,OAAO,YAAY,MAAM;CAC/B,MAAM,YAAY,iBAAiB,MAAM;CACzC,MAAM,WAAW,cAAc,IAAI,IAAK,YAAY,MAAM,IAAI,cAAe;CAE7E,OAAO,UAAU;EACb,QAAQ,KAAK;EACb,MAAM,CAAC,KAAK,KAAK,KAAK,WAAW,GAAG,KAAK,KAAK,KAAK,WAAW,CAAC;EAC/D,OAAO,KAAK;CAChB,CAAC;AACL;;;;;;AAOA,SAAgB,qBAAqB,QAAwD;CACzF,MAAM,QAAQ,CAAC,GAAG,MAAM,CAAC,CAAC,MAAM,GAAG,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG;CACpE,MAAM,UAAU,MAAM;CACtB,MAAM,cAAc,MAAM,MAAM,SAAS;CAEzC,MAAM,OAAO,MAAM,MAAM,GAAG,EAAE,CAAC,CAAC,MAAM,GAAG,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG;CAC1E,OAAO;EAAC;EAAS,KAAK;EAAa;EAAa,KAAK;CAAW;AACpE;;;;;;;;AASA,SAAgB,wBACZ,KACA,QACA,OACA,UAAyC,CAAC,GAC5B;CACd,MAAM,OAAO;EAAE,GAAG;EAAyB,GAAG;CAAQ;CACtD,MAAM,QAAwB,CAAC;CAE/B,KAAK,MAAM,YAAY,wBAAwB,KAAK,KAAK,QAAQ,KAAK,aAAa,GAAG;EAClF,MAAM,YAAY,QAAQ,QAAQ;EAClC,IAAI,UAAU,QAAQ,KAAK,SAAS;EAEpC,MAAM,QAAQ,SAAS,KAAK,UAAU,MAAM;EAC5C,IAAI,QAAQ,KAAK,WAAW;EAE5B,MAAM,WAAW,QAAQ,WAAW,UAAU,QAAQ,KAAK,WAAW,CAAC;EAGvE,IAAI,SAAS,QAAQ,KAAK,UAAU,GAAG;EAWvC,MAAM,UAAU,qBARC,SAAS,OAAO,KAC5B,CAAC,GAAG,OACD,CACI,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,IAAI,KAAK,GAAG,CAAC,GAAG,OAAO,KAAK,GACzD,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,IAAI,KAAK,GAAG,CAAC,GAAG,OAAO,MAAM,CAC9D,CAGoC,CAAC;EAE7C,MAAM,QAAQ,KAAK,MAAM,QAAQ,EAAE,CAAC,KAAK,QAAQ,EAAE,CAAC,IAAI,QAAQ,EAAE,CAAC,KAAK,QAAQ,EAAE,CAAC,EAAE;EACrF,MAAM,SAAS,KAAK,MAAM,QAAQ,EAAE,CAAC,KAAK,QAAQ,EAAE,CAAC,IAAI,QAAQ,EAAE,CAAC,KAAK,QAAQ,EAAE,CAAC,EAAE;EACtF,IAAI,SAAS,KAAK,UAAU,GAAG;EAE/B,MAAM,KAAK;GAAE,QAAQ;GAAS;EAAM,CAAC;CACzC;CACA,OAAO;AACX;;;;;;;;;;;;;;;;;;;;AC1OA,MAAa,mBAAmB;;AAGhC,MAAa,sBAAsB;AAuBnC,MAAa,0BAAmD,OAAO,OAAO;CAC1E,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,OAAO;CACP,gBAAgB,wBAAwB;CACxC,iBAAiB,wBAAwB;CACzC,aAAa,wBAAwB;CACrC,YAAY;AAChB,CAAC;AAED,SAAS,UAAU,OAA0B;CACzC,OAAO,eAAe,KAAK,KAAK,IAC1B;EAAE,MAAM;EAAS,OAAO,CAAC,EAAE,MAAM,MAAM,CAAC;CAAE,IAC1C;EAAE,MAAM;EAAO,OAAO,CAAC,EAAE,MAAM,aAAa,CAAC;CAAE;AACzD;AAEA,IAAa,oBAAb,cAAuC,MAA+B;CAClE,OAAgB;CAEhB,UAAqE;CACrE,UAA8E;CAE9E,YAAY,UAA4C,CAAC,GAAG;EACxD,MAAM,cAAc,yBAAyB,OAAO,CAAC;CACzD;CAEA,MAAe,OAAsB;EACjC,MAAM,KAAK,WAAW;CAC1B;CAEA,aAA0E;EACtE,IAAI,KAAK,SAAS,OAAO,QAAQ,QAAQ,KAAK,OAAO;EACrD,IAAI,KAAK,SAAS,OAAO,KAAK;EAE9B,KAAK,WAAW,YAAY;GACxB,MAAM,OAAO,UAAU,KAAK,OAAO,KAAK;GAExC,MAAM,SAAQ,MADM,iBAAiB,IAAI,EAAA,CACrB,KAAK,MAAM,EAAE,EAAE,QAAQ;GAC3C,IAAI,CAAC,OAAO,MAAM,IAAI,eAAe,SAAS,KAAK,OAAO,MAAM,aAAa,KAAK,IAAI;GAEtF,MAAM,UAAU,MAAM,cAAc,OAAO,EACvC,oBAAoB,UAAU,CAAC,CAAC,mBACpC,CAAC;GACD,KAAK,UAAU;GACf,OAAO;EACX,EAAA,CAAG;EAIH,KAAK,QAAQ,YAAY;GACrB,KAAK,UAAU;EACnB,CAAC;EACD,OAAO,KAAK;CAChB;CAEA,MAAe,UAAyB;EACpC,MAAM,KAAK,SAAS,QAAQ;EAC5B,KAAK,UAAU;EACf,KAAK,UAAU;CACnB;CAEA,MAAgB,MAAM,OAAgB,KAAmC;EACrE,MAAM,QAAQ;EAId,IAAI,OAAO,WACP,MAAM,IAAI,eAAe,0BAA0B,MAAM,aAAa,KAAK,IAAI;EAEnF,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,eAAe,wCAAwC,KAAK,IAAI;EAG9E,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAC3C,IAAI;EACJ,IAAI;GACA,QAAQ,MAAM,KAAK,OAAO,MAAM;EACpC,UAAU;GACN,OAAO,MAAM;EACjB;EAEA,OAAO,qBAAqB;GACxB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,QAAQ;EACZ,CAAC;CACL;;CAGA,MAAM,OAAO,QAAuD;EAChE,MAAM,UAAU,MAAM,KAAK,WAAW;EAKtC,MAAM,SAAS,UACX,QACA;GAAE,OAAO;GAAkB,QAAQ;EAAiB,GACpD;GACI,SAAS;GACT,MAAM;EACV,CACJ;EACA,MAAM,aAAa,cAAc,YAAY,OAAO,MAAM,GAAG;GACzD,MAAM;GACN,KAAK;GACL,KAAK;EACT,CAAC;EAED,MAAM,EAAE,WAAW,MAAM,OAAO;EAChC,MAAM,YAAY,QAAQ,WAAW;EACrC,MAAM,aAAa,QAAQ,YAAY;EACvC,IAAI,CAAC,aAAa,CAAC,YACf,MAAM,IAAI,eAAe,oCAAoC,KAAK,IAAI;EAM1E,MAAM,UAAS,MAHO,QAAQ,IAAI,GAC7B,YAAY,IAAI,OAAO,WAAW,YAAY;GAAC;GAAG;GAAG;GAAkB;EAAgB,CAAC,EAC7F,CAAC,EAAA,CACsB;EACvB,IAAI,CAAC,QAAQ,MAAM,IAAI,eAAe,sBAAsB,WAAW,WAAW,KAAK,IAAI;EAG3F,MAAM,KAAK,SAAS,kBAAkB,QAAQ,oBAAoB,OAAO;EAkBzE,MAAM,QAXQ,wBACV;GANA,MAAM,OAAO;GACb;GACA;EAIA,GACA;GAAE,OAAO,OAAO;GAAO,QAAQ,OAAO;EAAO,GAC7C,OAAO,OACP;GACI,QAAQ,KAAK,OAAO;GACpB,WAAW,KAAK,OAAO;GACvB,aAAa,KAAK,OAAO;EAC7B,CAGc,CAAC,CAAC,KAAK,SAAS,eAAe,KAAK,QAAQ,EAAE,OAAO,KAAK,MAAM,CAAC,CAAC;EACpF,OAAO,KAAK,OAAO,aAAa,sBAAsB,OAAO,KAAM,IAAI,EAAG,IAAI;CAClF;CAEA,QAAkB,SAAiB,KAA0B;EACzD,OAAO,qBAAqB;GACxB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;AACJ;;;AChMA,MAAa,kBAA4C,OAAO,OAAO;CACnE;EACI,IAAI;EACJ,MAAM;EACN,MAAM;EACN,OAAO;CACX;CACA;EACI,IAAI;EACJ,MAAM;EACN,MAAM;EACN,MAAM;EACN,aAAa;EACb,OAAO;CACX;CACA;EACI,IAAI;EACJ,MAAM;EACN,MAAM;EACN,MAAM;EACN,aAAa;EACb,OAAO;CACX;AACJ,CAAC;AAED,MAAa,sBAAsB;AAEnC,SAAgB,iBAAiB,IAAuC;CACpE,OAAO,gBAAgB,MAAM,MAAM,EAAE,OAAO,EAAE;AAClD;;;;;;;;;;;;;AC/BA,MAAa,4BAA4B;;AAGzC,MAAa,oBAAoB;CAAE,OAAO;CAAK,QAAQ;AAAG;AAI1D,MAAM,SAAqC,CAAC,YAAY,YAAY;AAEpE,IAAa,4BAAb,MAAuC;CAId;CAHrB,UAAqE;CACrE,UAA8E;CAE9E,YAAY,QAAyB,2BAA2B;EAA3C,KAAA,QAAA;CAA4C;CAEjE,aAA0E;EACtE,IAAI,KAAK,SAAS,OAAO,QAAQ,QAAQ,KAAK,OAAO;EACrD,IAAI,KAAK,SAAS,OAAO,KAAK;EAE9B,KAAK,WAAW,YAAY;GAKxB,MAAM,SAAQ,MAJM,iBAAiB;IACjC,MAAM,KAAK;IACX,OAAO,CAAC,EAAE,MAAM,aAAa,CAAC;GAClC,CAAC,EAAA,CACmB;GACpB,IAAI,CAAC,OACD,MAAM,IAAI,eAAe,SAAS,KAAK,MAAM,qBAAqB,iBAAiB;GAEvF,MAAM,UAAU,MAAM,cAAc,KAAK;GACzC,KAAK,UAAU;GACf,OAAO;EACX,EAAA,CAAG;EAEH,KAAK,QAAQ,YAAY;GACrB,KAAK,UAAU;EACnB,CAAC;EACD,OAAO,KAAK;CAChB;CAEA,MAAM,SAAS,QAAiE;EAC5E,MAAM,UAAU,MAAM,KAAK,WAAW;EAItC,MAAM,SAAS,aAAa,kBAAkB,OAAO,kBAAkB,MAAM;EAC7E,UAAU,MAAM,CAAC,CAAC,UAAU,QAAQ,GAAG,GAAG,OAAO,OAAO,OAAO,MAAM;EAErE,MAAM,OAAO,cAAc,YAAY,MAAM,GAAG;GAC5C,MAAM;GACN,KAAK;GACL,KAAK;EACT,CAAC;EAED,MAAM,EAAE,WAAW,MAAM,OAAO;EAChC,MAAM,YAAY,QAAQ,WAAW;EACrC,MAAM,aAAa,QAAQ,YAAY;EACvC,IAAI,CAAC,aAAa,CAAC,YACf,MAAM,IAAI,eAAe,oCAAoC,iBAAiB;EAWlF,MAAM,UAAS,MARO,QAAQ,IAAI,GAC7B,YAAY,IAAI,OAAO,WAAW,MAAM;GACrC;GACA;GACA,kBAAkB;GAClB,kBAAkB;EACtB,CAAC,EACL,CAAC,EAAA,CACsB,WAAW,EAAE;EACpC,IAAI,CAAC,UAAU,OAAO,SAAS,GAC3B,MAAM,IAAI,eAAe,iCAAiC,iBAAiB;EAG/E,OAAQ,OAAO,KAAiB,OAAO,KAChC,OAAO,KACP,OAAO;CAClB;CAEA,MAAM,UAAyB;EAC3B,MAAM,KAAK,SAAS,QAAQ;EAC5B,KAAK,UAAU;EACf,KAAK,UAAU;CACnB;AACJ;;;;;;;;;;;;;;;;;;AC/CA,MAAa,4BAA2D,OAAO,OAAO;CAClF,GAAG;CACH,UAAU;CACV,WAAW,CAAC,SAAS,OAAO;CAC5B,WAAW;CACX,gBAAgB;CAChB,eAAe;CACf,OAAO;CACP,SAAS;CACT,aAAa;CACb,SAAS;CACT,WAAW;AACf,CAAC;AAED,SAASA,UAAQ,QAAwB;CACrC,IAAI,OAAO,WAAW,YAAY,WAAW,MAAM,OAAO,CAAC;CAC3D,OAAQ,OAAqC,UAAU,CAAC;AAC5D;AAEA,IAAa,0BAAb,cAA6C,MAAqC;CAC9E,OAAgB;CAEhB,aAAuD;CAEvD,YAAY,UAAkD,CAAC,GAAG;EAC9D,MACI,cAAc,2BAA2B,SAAS,EAC9C,YAAY,UAAU;GAClB,IAAI,MAAM,WAAW,GACjB,MAAM,IAAI,WAAW,4CAA4C;EAEzE,EACJ,CAAC,CACL;CACJ;CAEA,MAAe,OAAsB;EACjC,KAAK,eAAe,IAAI,0BAA0B,KAAK,OAAO,KAAK;CACvE;;CAGA,MAAyB,OAAO,QAAiB,KAAU,KAAmC;EAC1F,MAAM,EAAE,WAAW,WAAW,gBAAgB,SAAS,aAAa,YAAY,KAAK;EACrF,MAAM,CAAC,UAAU,UAAU;EAC3B,MAAM,QAAQ,IAAI;EAElB,IAAI,OAAO,WACP,MAAM,IAAI,eAAe,0BAA0B,MAAM,aAAa,KAAK,IAAI;EAEnF,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,eAAe,wCAAwC,KAAK,IAAI;EAG9E,MAAM,KAAK,KAAK;EAChB,MAAM,aAAa,KAAK;EACxB,MAAM,QAAQA,UAAQ,IAAI,OAAO;EAEjC,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAC3C,IAAI;GACA,MAAM,SAAS,aAAa,OAAO,OAAO,OAAO,MAAM;GACvD,MAAM,QAAQ,UAAU,MAAM;GAC9B,MAAM,UAAU,QAAQ,GAAG,CAAC;GAE5B,MAAM,eAAkC,CAAC;GACzC,IAAI,UAAU;GAEd,KAAK,MAAM,OAAO,OAAO;IACrB,IAAI,QAAQ,eAAe;IAE3B,IAAI,eAAe,CAAC,UAAU,GAAG,GAAG;KAChC,aAAa,KAAK,UAAU;KAC5B;IACJ;IAEA,MAAM,OAAO,QAAQ,QAAQ,KAAK,EAAE,QAAQ,CAAC;IAC7C,MAAM,cAAc,MAAM,WAAW,SAAS,IAAI;IAClD,aAAa,KAAK,WAAW;IAE7B,IAAI,gBAAgB,gBAAgB,SAAS;KACzC,WAAW,OAAO,QAAQ,GAAG;KAC7B;IACJ;GACJ;GAEA,MAAM,YACF,UAAU,IACJ,YAAY;IACR,MAAM,MAAM;IACZ,YAAY,MAAM;IAClB,MAAM,MAAM,YAAY,QAAQ,SAAS,KAAK,OAAO,WAAoB;IACzE,WAAW,KAAK,OAAO;IACvB,OAAO,OAAO;IACd,QAAQ,OAAO;GACnB,CAAC,IAGD;GAEV,OAAO,CAAC;IAAE,GAAG;KAAM,YAAY;KAAY,iBAAiB;GAAa,CAAC;EAC9E,UAAU;GACN,OAAO,MAAM;EACjB;CACJ;CAEA,MAAgB,QAAwB;EACpC,MAAM,IAAI,eAAe,yCAAyC,KAAK,IAAI;CAC/E;CAEA,QAAkB,SAAiB,KAAwB;EACvD,OAAO,YAAY;GACf,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,WAAW;EACf,CAAC;CACL;CAEA,MAAe,UAAyB;EACpC,MAAM,KAAK,YAAY,QAAQ;EAC/B,KAAK,aAAa;CACtB;AACJ;;;;;;;;;;;;;AAcA,SAAS,WAAW,KAAwC,QAAyB,KAAgB;CACjG,MAAM,MAAM;CACZ,MAAM,IAAI,KAAK,IAAI,GAAG,KAAK,IAAI,KAAK,MAAM,IAAI,CAAC,IAAI,KAAK,OAAO,QAAQ,CAAC,CAAC;CACzE,MAAM,IAAI,KAAK,IAAI,GAAG,KAAK,IAAI,KAAK,MAAM,IAAI,CAAC,IAAI,KAAK,OAAO,SAAS,CAAC,CAAC;CAC1E,MAAM,QAAQ,KAAK,IAAI,KAAK,MAAM,IAAI,KAAK,IAAI,GAAS,OAAO,QAAQ,CAAC;CACxE,MAAM,SAAS,KAAK,IAAI,KAAK,MAAM,IAAI,MAAM,IAAI,GAAS,OAAO,SAAS,CAAC;CAC3E,IAAI,QAAQ,KAAK,SAAS,GAAG;CAI7B,MAAM,SAAS,aAAa,OAAO,MAAM;CACzC,UAAU,MAAM,CAAC,CAAC,UAAU,QAAQ,GAAG,GAAG,OAAO,QAAQ,GAAG,GAAG,OAAO,MAAM;CAE5E,MAAM,KAAK,IAAI,QAAQ;CACvB,MAAM,KAAK,IAAI,SAAS;CAExB,IAAI,KAAK;CACT,IAAI,UAAU;CACd,IAAI,KAAK,IAAI,IAAI,KAAK,IAAI,GACtB,IAAI,KAAK,GAAG,GAAG,OAAO,MAAM;MACzB;EACH,MAAM,MAAO,IAAI,QAAQ,KAAK,KAAM;EACpC,MAAM,MAAM,KAAK,IAAI,GAAG;EACxB,MAAM,MAAM,KAAK,IAAI,GAAG;EAOxB;GALI,CAAC,CAAC,IAAI,QAAQ,GAAG,CAAC,IAAI,SAAS,CAAC;GAChC,CAAC,IAAI,QAAQ,GAAG,CAAC,IAAI,SAAS,CAAC;GAC/B,CAAC,IAAI,QAAQ,GAAG,IAAI,SAAS,CAAC;GAC9B,CAAC,CAAC,IAAI,QAAQ,GAAG,IAAI,SAAS,CAAC;EAE7B,CAAC,CAAC,SAAS,CAAC,IAAI,KAAK,UAAU;GACjC,MAAM,KAAK,KAAK,MAAM,KAAK,MAAM;GACjC,MAAM,KAAK,KAAK,MAAM,KAAK,MAAM;GACjC,IAAI,UAAU,GAAG,IAAI,OAAO,IAAI,EAAE;QAC7B,IAAI,OAAO,IAAI,EAAE;EAC1B,CAAC;EACD,IAAI,UAAU;CAClB;CACA,IAAI,KAAK;CAET,IAAI,UAAU,IAAI,EAAE;CACpB,IAAI,OAAO,KAAK,EAAE;CAClB,IAAI,UAAU,QAAQ,CAAC,QAAQ,GAAG,CAAC,SAAS,CAAC;CAC7C,IAAI,QAAQ;AAChB;;;ACxNA,MAAa,qBAA2C,OAAO,OAAO;CAClE;EAAE,OAAO;EAAY,OAAO;EAAyB,SAAS;GAAC;GAAS;GAAO;GAAY;EAAU;CAAE;CACvG;EACI,OAAO;EACP,OAAO;EACP,SAAS;GAAC;GAAS;GAAO;GAAY;EAAU;CACpD;CACA;EAAE,OAAO;EAAW,OAAO;EAA+B,SAAS,CAAC,SAAS,KAAK;CAAE;CACpF;EAAE,OAAO;EAAmB,OAAO;EAAwC,SAAS,CAAC,OAAO;CAAE;CAC9F;EACI,OAAO;EACP,OAAO;EACP,SAAS,CAAC,SAAS,UAAU;CACjC;CACA;EAAE,OAAO;EAAsB,OAAO;EAAiB,SAAS,CAAC,UAAU;CAAE;CAC7E;EAAE,OAAO;EAAwB,OAAO;EAA8B,SAAS,CAAC,SAAS,YAAY;CAAE;CACvG;EAAE,OAAO;EAAoB,OAAO;EAAkB,SAAS,CAAC,SAAS,QAAQ;CAAE;CACnF;EAAE,OAAO;EAAmB,OAAO;EAAiB,SAAS,CAAC,SAAS,OAAO;CAAE;CAChF;EAAE,OAAO;EAAoB,OAAO;EAAkB,SAAS,CAAC,SAAS,QAAQ;CAAE;CACnF;EAAE,OAAO;EAAkB,OAAO;EAAgB,SAAS,CAAC,SAAS,MAAM;CAAE;CAC7E;EAAE,OAAO;EAAmB,OAAO;EAAiB,SAAS,CAAC,SAAS,OAAO;CAAE;CAChF;EAAE,OAAO;EAAoB,OAAO;EAAkB,SAAS,CAAC,SAAS,QAAQ;CAAE;CACnF;EAAE,OAAO;EAAgB,OAAO;EAA0B,SAAS,CAAC,OAAO;CAAE;AACjF,CAAC;AAED,MAAa,qBAAqB;AAElC,SAAgB,cAAc,OAAwB;CAClD,OAAO,mBAAmB,MAAM,MAAM,EAAE,UAAU,KAAK;AAC3D;;AAGA,SAAgB,eAAe,OAAqB;CAChD,IAAI,CAAC,cAAc,KAAK,GACpB,MAAM,IAAI,WAAW,uBAAuB,MAAM,4CAA4C;AAEtG;;AAGA,SAAgB,iBAAiB,QAA6B;CAC1D,OAAO,mBAAmB,QAAQ,MAAM,EAAE,QAAQ,SAAS,MAAM,CAAC;AACtE;;;;;;;;;;;;;;;;;;ACpBA,IAAI,gBAAwC;AAE5C,eAAe,UAA2B;CACtC,IAAI,eAAe,OAAO;CAC1B,iBAAiB,YAAY;EAGzB,MAAM,QAAQ;EACd,IAAI;GACA,OAAO,MAAM,OAAO;EACxB,SAAS,OAAO;GACZ,MAAM,IAAI,MACN,qFACA,EAAE,MAAM,CACZ;EACJ;CACJ,EAAA,CAAG;CACH,OAAO;AACX;;AAGA,eAAe,cAAc,QAA+D;CACxF,MAAM,MAAM,MAAM,QAAQ;CAC1B,MAAM,OACF,UAAU,IAAI,gBACR,IAAI,cAAc,UAClB,IAAI;CAEd,MAAM,QAAQ;EAAC;EAAa;EAAe;CAAsB;CAGjE,OAAO;EACH,MAAM;GAAE,MAAM,kBAAkB;GAAU,OAAO,MAAM,KAAK,OAAO,EAAE,MAAM,KAAK,GAAG,EAAE;EAAE;EACvF,OAAO,CAAC,GAAG,KAAK;CACpB;AACJ;;;;;AAMA,eAAe,iBAAiB,QAAsD;CAClF,MAAM,EAAE,MAAM,UAAU,MAAM,cAAc,MAAM;CAClD,MAAM,QAAQ,MAAM,iBAAiB;EACjC,GAAG;EACH,OAAO,KAAK,MAAM,KAAK,OAAO,EAAE,GAAG,EAAE,EAAE;CAC3C,CAAC;CACD,MAAM,MAAmC,CAAC;CAC1C,KAAK,MAAM,CAAC,GAAG,SAAS,MAAM,QAAQ,GAAG;EACrC,MAAM,OAAO,KAAK,MAAM,EAAE,EAAE;EAC5B,IAAI,QAAQ,MAAM,OAAO,IAAI,QAAQ,MAAM;CAC/C;CACA,OAAO;AACX;AAIA,MAAM,2BAAW,IAAI,IAAuC;AAE5D,eAAsB,iBAAiB,SAAS,oBAA+C;CAC3F,MAAM,WAAW,SAAS,IAAI,MAAM;CACpC,IAAI,UAAU,OAAO;CAErB,MAAM,WAAW,YAAY;EACzB,MAAM,MAAM,MAAM,QAAQ;EAC1B,MAAM,QAAQ,MAAM,iBAAiB,MAAM;EAC3C,MAAM,UAAU,IAAI,IAAI,iBAAiB;GACrC;GACA,SAAS;IACL,oBAAoB,CAAC,GAAG,UAAU,CAAC,CAAC,kBAAkB;IACtD,wBAAwB;GAC5B;EACJ,CAAC;EACD,MAAM,QAAQ,WAAW;EACzB,OAAO;CACX,EAAA,CAAG;CAIH,QAAQ,YAAY,SAAS,OAAO,MAAM,CAAC;CAC3C,SAAS,IAAI,QAAQ,OAAO;CAC5B,OAAO;AACX;;;;;;;;;AAUA,MAAM,oCAAoB,IAAI,IAAoE;;;;;;;;;AAUlG,SAAS,gBAAgB,QAA+B;CACpD,OAAO,IAAI,YAAY,OAAO,CAAC,CAAC,OAAO,MAAM,CAAC,CAAC,MAAM,OAAO;AAChE;AAEA,eAAe,sBAAsB,QAAwE;CACzG,MAAM,WAAW,kBAAkB,IAAI,MAAM;CAC7C,IAAI,UAAU,OAAO;CAErB,MAAM,WAAW,YAAY;EAOzB,MAAM,QAAQ;EACd,MAAM,EAAE,MAAM,UAAU,MAAM,cAAc,MAAM;EAGlD,MAAM,SAAS,CAAC,eAAe,sBAAsB,CAAC,CAAC,KAClD,SAAS,KAAK,MAAM,MAAM,QAAQ,IAAI,EAAE,EAAE,QAAQ,EACvD;EACA,MAAM,QAAQ,MAAM,iBAAiB;GAAE,GAAG;GAAM,OAAO,OAAO,KAAK,UAAU,EAAE,KAAK,EAAE;EAAE,CAAC;EACzF,MAAM,QAAQ,MAAM,OAAO;EAC3B,MAAM,OAAO,MAAM,OAAO;EAC1B,IAAI,CAAC,SAAS,CAAC,MACX,MAAM,IAAI,MAAM,qBAAqB,OAAO,kDAAkD;EAGlG,MAAM,aAAa,gBAAgB,IAAI;EACvC,IAAI,WAAW,WAAW,GACtB,MAAM,IAAI,MAAM,qBAAqB,OAAO,qCAAqC;EAMrF,OAAO;GAAE,SAAS,MAAM,cAAc,OAAO,EAAE,gBAAgB,KAAK,CAAC;GAAG;EAAW;CACvF,EAAA,CAAG;CAEH,QAAQ,YAAY,kBAAkB,OAAO,MAAM,CAAC;CACpD,kBAAkB,IAAI,QAAQ,OAAO;CACrC,OAAO;AACX;;;;;;;;;;AAWA,eAAsB,oBAClB,SAAS,oBACT,UAAmC,CAAC,GACH;CACjC,MAAM,MAAM,MAAM,QAAQ;CAC1B,MAAM,EAAE,SAAS,eAAe,MAAM,sBAAsB,MAAM;CAClE,OAAO,IAAI,IAAI,mBAAmB,SAAS;EACvC,sBAAsB;EACtB,mBAAmB;EACnB,yBAAyB,OAAO;EAChC,GAAI,QAAQ,iBAAiB,KAAA,IAAY,CAAC,IAAI,EAAE,cAAc,QAAQ,aAAa;EACnF,GAAI,QAAQ,kBAAkB,KAAA,IAAY,CAAC,IAAI,EAAE,eAAe,QAAQ,cAAc;CAC1F,CAAC;AACL;AAEA,eAAe,mBAAmB,QAA+B;CAC7D,MAAM,UAAU,kBAAkB,IAAI,MAAM;CAC5C,IAAI,CAAC,SAAS;CACd,kBAAkB,OAAO,MAAM;CAC/B,MAAM,QAAQ,MAAM,EAAE,cAAc,QAAQ,QAAQ,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;AAChF;AAEA,eAAsB,eAAe,QAAkC;CACnE,MAAM,EAAE,SAAS,MAAM,cAAc,MAAM;CAC3C,OAAO,SAAS,IAAI;AACxB;;AAGA,eAAsB,WAAW,QAA+B;CAC5D,MAAM,iBAAiB,MAAM;AACjC;AAEA,eAAsB,aAAa,QAA+B;CAC9D,MAAM,EAAE,SAAS,MAAM,cAAc,MAAM;CAC3C,MAAM,UAAU,SAAS,IAAI,MAAM;CACnC,SAAS,OAAO,MAAM;CACtB,MAAM,SAAS,MAAM,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;CAC7D,MAAM,mBAAmB,MAAM;CAC/B,MAAM,MAAM,IAAI;AACpB;;AAGA,eAAsB,wBAAuC;CACzD,MAAM,UAAU,CAAC,GAAG,SAAS,OAAO,CAAC;CACrC,SAAS,MAAM;CACf,MAAM,cAAc,CAAC,GAAG,kBAAkB,KAAK,CAAC;CAChD,MAAM,QAAQ,IAAI,CACd,GAAG,QAAQ,KAAK,MAAM,EAAE,MAAM,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS,CAAC,GACvE,GAAG,YAAY,KAAK,WAAW,mBAAmB,MAAM,CAAC,CAC7D,CAAC;AACL;;;;;;;;;;ACvNA,MAAa,gCAA0D,OAAO,OAAO;CACjF,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,QAAQ;CACR,gBAAgB;AACpB,CAAC;;AAGD,eAAe,SAAS,OAA0C;CAC9D,IAAI,OAAO,oBAAoB,eAAe,iBAAiB,iBAAiB,OAAO;CACvF,IAAI,OAAO,cAAc,eAAe,iBAAiB,WACrD,OAAO,kBAAkB,KAAK;CAGlC,MAAM,QAAQ;CAId,IAAI,OAAO,WACP,MAAM,IAAI,SAAS,0BAA0B,MAAM,aAAa,UAAU;CAE9E,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,SAAS,wCAAwC,UAAU;CAGzE,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;CAC3C,IAAI;EACA,MAAM,SAAS,IAAI,gBAAgB,OAAO,OAAO,OAAO,MAAM;EAC9D,MAAM,MAAM,OAAO,WAAW,IAAI;EAClC,IAAI,CAAC,KAAK,MAAM,IAAI,SAAS,kCAAkC,UAAU;EACzE,IAAI,UAAU,QAAQ,GAAG,CAAC;EAC1B,OAAO;CACX,UAAU;EACN,OAAO,MAAM;CACjB;AACJ;;AAGA,IAAa,qBAAb,cAAwC,MAAgC;CACpE,OAAgB;CAEhB,YAAY,UAA6C,CAAC,GAAG;EACzD,MAAM,cAAc,+BAA+B,SAAS,EAAE,QAAQ,eAAe,CAAC,CAAC;CAC3F;CAEA,MAAe,OAAsB;EACjC,MAAM,iBAAiB,KAAK,OAAO,MAAM;CAC7C;CAEA,MAAgB,MAAM,OAAgB,KAAmC;EACrE,MAAM,UAAU,MAAM,iBAAiB,KAAK,OAAO,MAAM;EACzD,MAAM,SAAS,MAAM,SAAS,KAAK;EACnC,MAAM,EAAE,UAAU,MAAM,QAAQ,OAAO,MAAe;EAEtD,OAAO,qBAAqB;GACxB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GAEN,QAAQ,MAAM,KAAK,MAAM,YAAY;IAAC,EAAE;IAAG,EAAE;IAAG,EAAE,IAAI,EAAE;IAAO,EAAE,IAAI,EAAE;GAAM,GAAG,EAAE,OAAO,EAAE,CAAC,CAAC;EACjG,CAAC;CACL;CAEA,QAAkB,SAAiB,KAA0B;EACzD,OAAO,qBAAqB;GACxB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;AACJ;AAmBA,MAAa,kCAA8D,OAAO,OAAO;CACrF,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,QAAQ;CACR,gBAAgB;CAChB,UAAU;CACV,gBAAgB;CAChB,eAAe;AACnB,CAAC;;AAGD,IAAa,uBAAb,cAA0C,MAAkC;CACxE,OAAgB;CAEhB,YAAY,UAA+C,CAAC,GAAG;EAC3D,MAAM,cAAc,iCAAiC,SAAS,EAAE,QAAQ,eAAe,CAAC,CAAC;CAC7F;CAEA,MAAe,OAAsB;EACjC,MAAM,iBAAiB,KAAK,OAAO,MAAM;CAC7C;CAEA,MAAgB,MAAM,OAAgB,KAA6B;EAC/D,MAAM,EAAE,QAAQ,UAAU,gBAAgB,gBAAgB,kBAAkB,KAAK;EACjF,MAAM,UAAU,MAAM,iBAAiB,MAAM;EAC7C,MAAM,SAAS,MAAM,SAAS,KAAK;EAEnC,MAAM,SAAS,MAAM,QAAQ,UAAU,QAAiB;GACpD,SAAS;GACT;GAGA,SAAS;EACb,CAAC;EAGD,MAAM,UADQ,aAAa,SAAS,OAAO,UAAU,CAAC,EAAA,CAEjD,QAAQ,SAAS,KAAK,cAAc,cAAc,CAAC,CACnD,KAAK,SACF,YACI;GAAC,KAAK,IAAI;GAAG,KAAK,IAAI;GAAG,KAAK,IAAI,IAAI,KAAK,IAAI;GAAO,KAAK,IAAI,IAAI,KAAK,IAAI;EAAM,GAClF;GAAE,MAAM,KAAK;GAAM,OAAO,KAAK;EAAW,CAC9C,CACJ;EAEJ,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,MAAM,iBAAiB,qBAAqB,QAAQ,aAAa,IAAI,YAAY,MAAM;GACvF;EACJ,CAAC;CACL;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;AACJ;;;;;;;;;;;;;;;;ACxHA,MAAa,6BAAqD,OAAO,OAAO;CAC5E,GAAG;CACH,UAAU;CACV,WAAW,CAAC,SAAS,OAAO;CAC5B,WAAW;CACX,eAAe;CACf,QAAQ;CACR,aAAa;CACb,SAAS;CACT,gBAAgB;CAChB,gBAAgB;CAChB,eAAe;CACf,uBAAuB;CACvB,aAAa;CACb,UAAU;CACV,eAAe;CACf,cAAc;AAClB,CAAC;;;;;;;AAQD,MAAM,mBAAmB;AAEzB,SAASC,UAAQ,QAAwB;CACrC,IAAI,OAAO,WAAW,YAAY,WAAW,MAAM,OAAO,CAAC;CAC3D,OAAQ,OAA0B,UAAU,CAAC;AACjD;AAEA,IAAa,mBAAb,cAAsC,MAA8B;CAChE,OAAgB;CAEhB,cAAwD;CACxD,cAAuD;CAEvD,YAAY,UAA2C,CAAC,GAAG;EACvD,MACI,cAAc,4BAA4B,SAAS;GAC/C,YAAY,UAAU;IAClB,IAAI,MAAM,WAAW,GACjB,MAAM,IAAI,WAAW,4CAA4C;GAEzE;GACA,QAAQ;EACZ,CAAC,CACL;CACJ;CAEA,MAAe,OAAsB;EACjC,MAAM,EAAE,QAAQ,cAAc,eAAe,uBAAuB,aAAa,KAAK;EACtF,KAAK,gBAAgB,MAAM,oBAAoB,QAAQ;GAAE;GAAc;EAAc,CAAC;EACtF,IAAI,uBACA,KAAK,gBAAgB,IAAI,0BAA0B,QAAQ;CAEnE;CAEA,MAAgB,MAAM,QAAiB,KAAU,KAAsC;EACnF,MAAM,EACF,WACA,aACA,SACA,gBACA,gBACA,eACA,uBACA,gBACA,KAAK;EACT,MAAM,CAAC,UAAU,UAAU;EAC3B,MAAM,QAAQ,IAAI;EAKlB,IAAI,OAAO,WACP,MAAM,IAAI,SAAS,0BAA0B,MAAM,aAAa,KAAK,IAAI;EAE7E,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,SAAS,wCAAwC,KAAK,IAAI;EAGxE,MAAM,SAAS,IAAI;EACnB,IAAI,WAAW,KAAA,GACX,MAAM,IAAI,SACN,uBAAuB,OAAO,0EAE9B,KAAK,IACT;EAGJ,MAAM,KAAK,KAAK;EAChB,MAAM,cAAc,KAAK;EACzB,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAM3C,MAAM,SAAS,gBAAgB,IAAI,SAAS,OAAO,QAAQ,WAAW;EAGtE,MAAM,OAAc,CAAC;EACrB,MAAM,QAA2B,CAAC;EAElC,IAAI;GACA,KAAK,MAAM,OAAOA,UAAQ,MAAM,GAAG;IAC/B,IAAI,QAAQ,eAAe;IAK3B,IAAI,OAAO,QAAQ,QAAQ,KAAK;KAAE;KAAa;IAAQ,CAAC;IAExD,IAAI,WAAW;IACf,IAAI,yBAAyB,KAAK,aAAa;KAC3C,WAAY,MAAM,KAAK,YAAY,SAAS,IAAI,MAAO;KACvD,IAAI,UAAU,OAAO,UAAU,IAAI;IACvC;IAIA,IAAI,eAAe,CAAC,UAAU,GAAG,KAAK,CAAC,UAAU;IAEjD,KAAK,KAAK,GAAG;IACb,MAAM,KAAK,IAAI;GACnB;EACJ,UAAU;GACN,OAAO,MAAM;EACjB;EAEA,MAAM,OAAO,MAAM,eAAe,aAAa,OAAO,GAAG;EAEzD,MAAM,SAAgB,CAAC;EACvB,KAAK,MAAM,CAAC,GAAG,QAAQ,KAAK,QAAQ,GAAG;GACnC,MAAM,SAAS,KAAK;GACpB,IAAI,CAAC,QAAQ,MAAM;GACnB,IAAI,OAAO,QAAQ,gBAAgB;GACnC,OAAO,KAAK;IAAE,GAAG;IAAK,MAAM,OAAO;IAAM,OAAO,OAAO;GAAM,CAAC;EAClE;EAEA,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,MAAM,IAAI;GACnD,MAAM;GACN,MAAM,iBAAiB,qBAAqB,QAAQ,aAAa,IAAI,YAAY,MAAM;GACvF;EACJ,CAAC;CACL;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;CAEA,MAAe,UAAyB;EACpC,MAAM,KAAK,aAAa,QAAQ;EAChC,KAAK,cAAc;EAGnB,KAAK,cAAc;CACvB;AACJ;;;;;;;;;;;;;;;AAqBA,eAAe,eACX,aACA,OACA,KACmC;CACnC,MAAM,MAAkC,IAAI,MAAM,MAAM,MAAM;CAE9D,KAAK,IAAI,QAAQ,GAAG,QAAQ,MAAM,SAAU;EACxC,IAAI,QAAQ,eAAe;EAI3B,IAAI,MAAM;EACV,IAAI,SAAS;EACb,IAAI,QAAQ;EACZ,OAAO,MAAM,MAAM,QAAQ;GACvB,MAAM,OAAO,MAAM;GACnB,IAAI,MAAM,SAAS,SAAS,KAAK,SAAS,kBAAkB;GAC5D,UAAU,KAAK;GACf,QAAQ,KAAK,IAAI,OAAO,KAAK,KAAK;GAClC;EACJ;EAEA,MAAM,QAAQ,aAAa,OAAO,MAAM;EACxC,MAAM,WAAW,UAAU,KAAK;EAChC,MAAM,QAAmE,CAAC;EAC1E,MAAM,2BAAW,IAAI,IAAoB;EACzC,IAAI,SAAS;EACb,KAAK,IAAI,IAAI,OAAO,IAAI,KAAK,KAAK;GAC9B,MAAM,OAAO,MAAM;GACnB,SAAS,UAAU,MAAM,GAAG,MAAM;GAClC,MAAM,KAAK;IAAE,GAAG;IAAG,GAAG;IAAQ,OAAO,KAAK;IAAO,QAAQ,KAAK;GAAO,CAAC;GACtE,SAAS,IAAI,QAAQ,CAAC;GACtB,UAAU,KAAK;EACnB;EAEA,MAAM,UAAU,MAAM,YAAY,IAAI,OAAgB,OAAO,KAAA,GAAW,SAAS;EACjF,KAAK,MAAM,UAAU,SAAS;GAC1B,MAAM,QAAQ,SAAS,IAAI,OAAO,IAAI,CAAC;GACvC,IAAI,UAAU,KAAA,GAAW,IAAI,SAAS;IAAE,MAAM,OAAO,KAAK,KAAK;IAAG,OAAO,OAAO;GAAW;EAC/F;EAEA,QAAQ;CACZ;CAEA,OAAO;AACX;;;;;;;;;;;;;AC9QA,IAAI,gBAA2C;AAE/C,eAAe,eAAmC;CAC9C,IAAI,eAAe,OAAO;CAE1B,iBAAiB,YAAY;EACzB,IAAI;EACJ,IAAI;GACA,MAAM,MAAM,OAAO;EACvB,SAAS,OAAO;GACZ,MAAM,IAAI,MAAM,iFAAiF,EAC7F,MACJ,CAAC;EACL;EAEA,OAAO,IAAI,aAAa,OAAO,IAAI,IAAI,cAAc;CACzD,EAAA,CAAG;CAEH,cAAc,YAAY;EACtB,gBAAgB;CACpB,CAAC;CACD,OAAO;AACX;;AAGA,eAAsB,aAClB,QAC8B;CAC9B,MAAM,SAAS,MAAM,aAAa;CAElC,IAAI;CACJ,IAAI,OAAO,oBAAoB,eAAe,kBAAkB,iBAC5D,SAAS;MACN,IAAI,OAAO,cAAc,eAAe,kBAAkB,WAC7D,SAAS,kBAAkB,MAAM;MAC9B,IAAI,OAAO,gBAAgB,eAAe,kBAAkB,aAC/D,SAAS,kBAAkB,YAAY,MAAM,CAAC;MAC3C;EACH,MAAM,QAAQ;EACd,IAAI,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GACjE,MAAM,IAAI,SAAS,wCAAwC,cAAc;EAE7E,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAC3C,IAAI;GACA,SAAS,kBAAkB,YAAY,MAAM,CAAC;EAClD,UAAU;GACN,OAAO,MAAM;EACjB;CACJ;CAEA,MAAM,OAAO,MAAM,OAAO,cAAc,EAAE,MAAM,YAAY,CAAC;CAC7D,MAAM,EAAE,SAAS,MAAM,OAAO,OAAO,IAAI;CACzC,IAAI,CAAC,MAAM,QAAQ,OAAO;CAC1B,OAAO;EAAE,QAAQ,KAAK;EAAQ,YAAY,KAAK,qBAAqB;CAAE;AAC1E;;AAGA,eAAsB,eAClB,QACoB;CACpB,MAAM,WAAW,MAAM,aAAa,MAAM;CAC1C,OAAO,WAAW,iBAAiB,SAAS,MAAM,IAAI,CAAC;AAC3D;;AAGA,eAAsB,yBAAwC;CAC1D,MAAM,UAAU;CAChB,gBAAgB;CAChB,MAAM,SAAS,MAAM,MAAM,EAAE,UAAU,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;AACnE;;;;;;;;;;;;;;;;ACvEA,MAAa,uBAAuB;AAWpC,MAAa,yBAA6C,OAAO,OAAO;CACpE,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,MAAM,CAAC,KAAK;CACZ,gBAAgB;CAChB,gBAAgB;CAChB,eAAe;AACnB,CAAC;AAKD,IAAI,gBAA2C;AAC/C,IAAI,iBAAgC;AAEpC,SAAS,eAAe,MAAsB;CAE1C,OAAO,IADO,UAAU,CAAC,CAAC,UAAU,WAAA,sEAAA,CAAiC,QAAQ,QAAQ,GACxE,IAAI,KAAK;AAC1B;;;;;;;AAQA,eAAsB,mBAAmB,MAAkC;CACvE,IAAI,iBAAiB,mBAAmB,MAAM,OAAO;CAGrD,IAAI,eAAe;EACf,MAAM,WAAW;EACjB,gBAAgB;EAChB,MAAM,SAAS,MAAM,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;CACjE;CAEA,iBAAiB;CACjB,iBAAiB,YAAY;EACzB,IAAI;EACJ,IAAI;GACA,MAAM,MAAM,OAAO;EACvB,SAAS,OAAO;GACZ,MAAM,IAAI,MACN,yFACA,EAAE,MAAM,CACZ;EACJ;EAEA,MAAM,EAAE,cAAc,UAAU,CAAC,CAAC;EAClC,MAAM,SAAS,IAAI,IAAI,UAAU,YAAY,EAAE,WAAW,UAAU,IAAI,CAAC,CAAC;EAC1E,MAAM,WAAW,MAAM,MAAM,eAAe,IAAI,CAAC;EACjD,IAAI,CAAC,SAAS,IACV,MAAM,IAAI,SACN,mBAAmB,KAAK,gBAAgB,SAAS,OAAO,GAAG,SAAS,cACpE,cACJ;EAEJ,MAAM,OAAO,UAAU,IAAI,WAAW,MAAM,SAAS,YAAY,CAAC,CAAC;EACnE,OAAO;CACX,EAAA,CAAG;CAEH,cAAc,YAAY;EACtB,gBAAgB;EAChB,iBAAiB;CACrB,CAAC;CACD,OAAO;AACX;;AAGA,eAAsB,mBAAkC;CACpD,MAAM,UAAU;CAChB,gBAAgB;CAChB,iBAAiB;CACjB,MAAM,SAAS,MAAM,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;AACjE;AAEA,IAAa,eAAb,cAAkC,MAA0B;CACxD,OAAgB;CAEhB,YAAY,UAAuC,CAAC,GAAG;EACnD,MACI,cAAc,wBAAwB,SAAS,EAC3C,OAAO,UAAU;GACb,IAAI,MAAM,WAAW,GAAG,MAAM,IAAI,WAAW,wBAAwB;EACzE,EACJ,CAAC,CACL;CACJ;;CAGA,IAAY,WAAmB;EAC3B,OAAO,KAAK,OAAO,KAAK,KAAK,GAAG;CACpC;CAEA,MAAe,OAAsB;EACjC,MAAM,mBAAmB,KAAK,QAAQ;CAC1C;CAEA,MAAgB,MAAM,OAAgB,KAA6B;EAC/D,MAAM,QAAQ;EAId,IAAI,OAAO,WACP,MAAM,IAAI,SAAS,0BAA0B,MAAM,aAAa,KAAK,IAAI;EAE7E,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,SAAS,wCAAwC,KAAK,IAAI;EAGxE,MAAM,SAAS,MAAM,mBAAmB,KAAK,QAAQ;EACrD,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAC3C,IAAI;EACJ,IAAI;GACA,MAAM,OAAO,UAAU,YAAY,MAAM,CAAC;GAG1C,QAAQ,MAAM,OAAO,aAAa,MAAM;EAC5C,UAAU;GACN,OAAO,MAAM;EACjB;EAEA,MAAM,EAAE,gBAAgB,gBAAgB,kBAAkB,KAAK;EAC/D,MAAM,SAAgB,MAEjB,QAAQ,SAAS,KAAK,cAAc,kBAAkB,KAAK,KAAK,KAAK,CAAC,CAAC,SAAS,CAAC,CAAC,CAClF,KAAK,SACF,YAAY;GAAC,KAAK,KAAK;GAAM,KAAK,KAAK;GAAK,KAAK,KAAK;GAAO,KAAK,KAAK;EAAM,GAAG;GAC5E,MAAM,KAAK,KAAK,KAAK;GACrB,OAAO,KAAK;EAChB,CAAC,CACL;EAEJ,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,MAAM,iBAAiB,qBAAqB,QAAQ,aAAa,IAAI,YAAY,MAAM;GACvF;EACJ,CAAC;CACL;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;CAEA,MAAe,UAAyB;EACpC,MAAM,iBAAiB;CAC3B;AACJ;;;;;;;;;;;;AChIA,MAAa,gCAA2D,OAAO,OAAO;CAClF,GAAG;CACH,UAAU;CACV,WAAW,CAAC,SAAS,OAAO;CAC5B,WAAW;CACX,eAAe;CACf,MAAM,CAAC,KAAK;CACZ,aAAa;CACb,SAAS;CACT,gBAAgB;CAChB,gBAAgB;CAChB,eAAe;CACf,UAAU;CACV,uBAAuB;CACvB,aAAa;CACb,UAAU;AACd,CAAC;AAED,SAAS,QAAQ,QAAwB;CACrC,IAAI,OAAO,WAAW,YAAY,WAAW,MAAM,OAAO,CAAC;CAC3D,OAAQ,OAA0B,UAAU,CAAC;AACjD;AAEA,IAAa,sBAAb,cAAyC,MAAiC;CACtE,OAAgB;CAEhB,cAAwD;CAExD,YAAY,UAA8C,CAAC,GAAG;EAC1D,MACI,cAAc,+BAA+B,SAAS;GAClD,YAAY,UAAU;IAClB,IAAI,MAAM,WAAW,GACjB,MAAM,IAAI,WAAW,4CAA4C;GAEzE;GACA,OAAO,UAAU;IACb,IAAI,MAAM,WAAW,GAAG,MAAM,IAAI,WAAW,wBAAwB;GACzE;EACJ,CAAC,CACL;CACJ;CAEA,IAAY,WAAmB;EAC3B,OAAO,KAAK,OAAO,KAAK,KAAK,GAAG;CACpC;CAEA,MAAe,OAAsB;EACjC,MAAM,mBAAmB,KAAK,QAAQ;EACtC,IAAI,KAAK,OAAO,uBACZ,KAAK,gBAAgB,IAAI,0BAA0B,KAAK,OAAO,QAAQ;CAE/E;CAEA,MAAgB,MAAM,QAAiB,KAAU,KAAsC;EACnF,MAAM,EACF,WACA,aACA,SACA,gBACA,gBACA,eACA,UACA,uBACA,gBACA,KAAK;EACT,MAAM,CAAC,UAAU,UAAU;EAC3B,MAAM,QAAQ,IAAI;EAElB,IAAI,OAAO,WACP,MAAM,IAAI,SAAS,0BAA0B,MAAM,aAAa,KAAK,IAAI;EAE7E,IAAI,CAAC,SAAS,EAAE,MAAM,gBAAgB,eAAe,MAAM,KAAK,eAAe,GAC3E,MAAM,IAAI,SAAS,wCAAwC,KAAK,IAAI;EAGxE,MAAM,SAAS,IAAI;EACnB,IAAI,WAAW,KAAA,GACX,MAAM,IAAI,SACN,uBAAuB,OAAO,0EAE9B,KAAK,IACT;EAGJ,MAAM,KAAK,KAAK;EAChB,MAAM,SAAS,MAAM,mBAAmB,KAAK,QAAQ;EACrD,MAAM,SAAS,MAAM,YAAY,MAAM,IAAI;EAC3C,MAAM,aAAoB,CAAC;EAE3B,IAAI;GACA,KAAK,MAAM,OAAO,QAAQ,MAAM,GAAG;IAC/B,IAAI,QAAQ,eAAe;IAK3B,MAAM,WAAW,aAAa,KAAK;KAAE;KAAa;IAAQ,CAAC;IAC3D,IAAI,OAAO,QAAQ,QAAQ,KAAK;KAAE;KAAa;IAAQ,CAAC;IAExD,IAAI,WAAW;IACf,IAAI,yBAAyB,KAAK,aAAa;KAC3C,WAAY,MAAM,KAAK,YAAY,SAAS,IAAI,MAAO;KACvD,IAAI,UAAU,OAAO,UAAU,IAAI;IACvC;IAIA,IAAI,eAAe,CAAC,UAAU,GAAG,KAAK,CAAC,UAAU;IAEjD,MAAM,OAAO,UAAU,YAAY,IAAI,CAAC;IACxC,MAAM,QAAQ,MAAM,OAAO,aAAa,MAAM;IAE9C,MAAM,QAAQ,MAAM,QAAQ,SAAS,KAAK,KAAK,KAAK,CAAC;IACrD,IAAI,MAAM,WAAW,GAAG;IASxB,MAAM,QAAQ,MAAM,QAAQ,KAAK,SAAS,MAAM,KAAK,YAAY,CAAC,KAAK,MAAM,UAAU;IASvF,IAAI,QAAQ,gBAAgB;IAE5B,IAAI,aAAa,QAAQ;KACrB,KAAK,MAAM,QAAQ,OACf,WAAW,KACP,QACI,KAAK,MACL,UACA,UACA,aACA,KAAK,KAAK,KAAK,GACf,KAAK,UACT,CACJ;KAEJ;IACJ;IAEA,WAAW,KAAK;KAAE,GAAG;KAAK,MAAM,MAAM,KAAK,SAAS,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,GAAG;KAAG;IAAM,CAAC;GAC5F;EACJ,UAAU;GACN,OAAO,MAAM;EACjB;EAKA,MAAM,SACF,gBAAgB,KAAK,aAAa,SAC5B,aACA,WAAW,KAAK,SAAS;GACrB,GAAG;GACH,GAAG,KAAK,MAAM,IAAI,IAAI,WAAW;GACjC,GAAG,KAAK,MAAM,IAAI,IAAI,WAAW;GACjC,OAAO,KAAK,MAAM,IAAI,QAAQ,WAAW;GACzC,QAAQ,KAAK,MAAM,IAAI,SAAS,WAAW;EAC/C,EAAE;EAEZ,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,MAAM,IAAI;GACnD,MAAM;GACN,MAAM,iBAAiB,qBAAqB,QAAQ,aAAa,IAAI,YAAY,MAAM;GACvF;EACJ,CAAC;CACL;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;CAEA,MAAe,UAAyB;EACpC,MAAM,KAAK,aAAa,QAAQ;EAChC,KAAK,cAAc;CACvB;AACJ;;;;;;;;;AAUA,SAAS,QACL,MACA,UACA,UACA,aACA,MACA,OACG;CACH,MAAM,EAAE,OAAO,QAAQ,QAAQ;CAE/B,MAAM,CAAC,MAAM,KAAK,OAAO,UAAU,WAC7B;EAAC,QAAQ,KAAK;EAAO,SAAS,KAAK;EAAQ,QAAQ,KAAK;EAAM,SAAS,KAAK;CAAG,IAC/E;EAAC,KAAK;EAAM,KAAK;EAAK,KAAK;EAAO,KAAK;CAAM;CAEnD,MAAM,UAAU,GAAW,MAAqB;EAC5C,MAAM,CAAC,IAAI,MAAM,IAAI,GAAG,CAAC;EACzB,OAAO,CAAC,KAAK,aAAa,KAAK,WAAW;CAC9C;CACA,MAAM,UAAmB;EACrB,OAAO,MAAM,GAAG;EAChB,OAAO,OAAO,GAAG;EACjB,OAAO,OAAO,MAAM;EACpB,OAAO,MAAM,MAAM;CACvB;CACA,OAAO,eAAe,SAAS;EAAE;EAAM;CAAM,CAAC;AAClD"}

package/dist/{ort-CXDoPrtw.js → ort-DZEG14nY.js}

@@ -39,15 +39,36 @@ async function loadOrt() {
 function resetOrt() {
 	modulePromise = null;
 }
+/** Providers that exist in every onnxruntime-web build. */
+const ALWAYS_AVAILABLE = /* @__PURE__ */ new Set(["cpu", "wasm"]);
+/**
+* The provider list to retry with, or `null` when there is nothing safer left.
+*
+* Mirrors ppu-paddle-ocr's own fallback, so a stage building its own session
+* behaves the same as one going through `PaddleOcrService`.
+*/
+function wasmFallbackProviders(providers) {
+	if (providers.length === 0 || providers.every((name) => ALWAYS_AVAILABLE.has(name))) return null;
+	return [providers.find((name) => ALWAYS_AVAILABLE.has(name)) ?? "wasm"];
+}
 /** Create an inference session from model bytes. */
 async function createSession(model, options = {}) {
 	const ort = await loadOrt();
 	const bytes = model instanceof Uint8Array ? model : new Uint8Array(model);
-	return ort.InferenceSession.create(bytes, {
-		executionProviders: [...options.executionProviders ?? getConfig().executionProviders],
+	const providers = [...options.executionProviders ?? getConfig().executionProviders];
+	const create = (executionProviders) => ort.InferenceSession.create(bytes, {
+		executionProviders,
 		graphOptimizationLevel: "all",
 		...options.externalData ? { externalData: options.externalData } : {}
 	});
+	try {
+		return await create(providers);
+	} catch (cause) {
+		const fallback = options.fallbackToWasm ? wasmFallbackProviders(providers) : null;
+		if (!fallback) throw cause;
+		console.warn(`[scaledp] executionProviders ${JSON.stringify(providers)} could not run this model (${cause instanceof Error ? cause.message : String(cause)}); falling back to ${JSON.stringify(fallback)}.`);
+		return create(fallback);
+	}
 }
 /** True when the browser exposes a usable WebGPU adapter. */
 async function isWebGpuAvailable() {
@@ -68,6 +89,6 @@ function isCrossOriginIsolated() {
 	return typeof crossOriginIsolated !== "undefined" && crossOriginIsolated === true;
 }
 //#endregion
-export { resetOrt as a, loadOrt as i, isCrossOriginIsolated as n, isWebGpuAvailable as r, createSession as t };
+export { resetOrt as a, loadOrt as i, isCrossOriginIsolated as n, wasmFallbackProviders as o, isWebGpuAvailable as r, createSession as t };
 
-//# sourceMappingURL=ort-CXDoPrtw.js.map
+//# sourceMappingURL=ort-DZEG14nY.js.map

package/dist/ort-DZEG14nY.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ort-DZEG14nY.js","names":[],"sources":["../src/ocr/ort.ts"],"sourcesContent":["/**\n * onnxruntime-web setup, shared by every ONNX-backed stage.\n *\n * onnxruntime-web is an optional peer dependency, imported lazily so the core\n * pulls in no ML runtime.\n */\n\nimport { getConfig, resolveNumThreads } from '../core/config.js'\n\ntype OrtModule = typeof import('onnxruntime-web')\n\nlet modulePromise: Promise<OrtModule> | null = null\n\n/**\n * onnxruntime-web's JS glue and its .wasm binary must come from the same build\n * variant AND the same version -- a mismatch fails at session creation with an\n * opaque error. Deriving the CDN URL from the resolved package version is what\n * keeps them in step; the pdftools prototype hardcoded 1.23.2 in three files\n * while actually running 1.26/1.27.\n */\nfunction cdnWasmPaths(version: string): string {\n    return `https://cdn.jsdelivr.net/npm/onnxruntime-web@${version}/dist/`\n}\n\nexport async function loadOrt(): Promise<OrtModule> {\n    if (modulePromise) return modulePromise\n\n    modulePromise = (async () => {\n        let ort: OrtModule\n        try {\n            ort = await import('onnxruntime-web')\n        } catch (cause) {\n            throw new Error(\n                'onnxruntime-web is required for OCR, NER and detection. Install it: npm i onnxruntime-web',\n                { cause }\n            )\n        }\n\n        const config = getConfig()\n        if (config.ortWasmPaths) {\n            ort.env.wasm.wasmPaths = config.ortWasmPaths\n        } else if (!ort.env.wasm.wasmPaths && ort.env.versions?.web) {\n            // Set eagerly, before any session exists, so a bundled library's own\n            // \"if not already set\" default never wins.\n            ort.env.wasm.wasmPaths = cdnWasmPaths(ort.env.versions.web)\n        }\n\n        // Threading only engages on a cross-origin-isolated page (COOP/COEP).\n        // Without that, ORT silently falls back to its single-threaded build and\n        // this value has no effect -- not an error, just a slower run.\n        ort.env.wasm.numThreads = resolveNumThreads()\n        ort.env.logLevel = 'error'\n        return ort\n    })()\n\n    return modulePromise\n}\n\n/** Reset the cached module. Tests only. */\nexport function resetOrt(): void {\n    modulePromise = null\n}\n\nexport interface SessionOptions {\n    executionProviders?: readonly string[]\n    /** Companion `.onnx_data` files for models with external weights. */\n    externalData?: { path: string; data: ArrayBuffer | Uint8Array }[]\n    /**\n     * Retry on WASM when the preferred providers cannot run the graph.\n     *\n     * Off by default, so a misconfigured provider still fails loudly. Turn it on\n     * for a model known to be picky: WebGPU rewrites convolutions into its\n     * `com.ms.internal.nhwc` domain and a graph it has no kernel for dies at\n     * session creation, not at inference.\n     */\n    fallbackToWasm?: boolean\n}\n\n/** Providers that exist in every onnxruntime-web build. */\nconst ALWAYS_AVAILABLE = new Set(['cpu', 'wasm'])\n\n/**\n * The provider list to retry with, or `null` when there is nothing safer left.\n *\n * Mirrors ppu-paddle-ocr's own fallback, so a stage building its own session\n * behaves the same as one going through `PaddleOcrService`.\n */\nexport function wasmFallbackProviders(providers: readonly string[]): string[] | null {\n    if (providers.length === 0 || providers.every((name) => ALWAYS_AVAILABLE.has(name))) return null\n    return [providers.find((name) => ALWAYS_AVAILABLE.has(name)) ?? 'wasm']\n}\n\n/** Create an inference session from model bytes. */\nexport async function createSession(\n    model: ArrayBuffer | Uint8Array,\n    options: SessionOptions = {}\n): Promise<import('onnxruntime-web').InferenceSession> {\n    const ort = await loadOrt()\n    const bytes = model instanceof Uint8Array ? model : new Uint8Array(model)\n    const providers = [...(options.executionProviders ?? getConfig().executionProviders)] as string[]\n    const create = (executionProviders: string[]) =>\n        ort.InferenceSession.create(bytes, {\n            executionProviders,\n            graphOptimizationLevel: 'all',\n            ...(options.externalData ? { externalData: options.externalData } : {}),\n        })\n\n    try {\n        return await create(providers)\n    } catch (cause) {\n        const fallback = options.fallbackToWasm ? wasmFallbackProviders(providers) : null\n        if (!fallback) throw cause\n        console.warn(\n            `[scaledp] executionProviders ${JSON.stringify(providers)} could not run this model ` +\n                `(${cause instanceof Error ? cause.message : String(cause)}); ` +\n                `falling back to ${JSON.stringify(fallback)}.`\n        )\n        return create(fallback)\n    }\n}\n\n/** True when the browser exposes a usable WebGPU adapter. */\nexport async function isWebGpuAvailable(): Promise<boolean> {\n    const gpu = (navigator as { gpu?: { requestAdapter(): Promise<unknown> } }).gpu\n    if (!gpu) return false\n    try {\n        return (await gpu.requestAdapter()) !== null\n    } catch {\n        return false\n    }\n}\n\n/**\n * True when the page is cross-origin isolated, which is what SharedArrayBuffer\n * and therefore multi-threaded WASM require. A library cannot set COOP/COEP for\n * its consumer, so this is reported rather than enforced.\n */\nexport function isCrossOriginIsolated(): boolean {\n    return typeof crossOriginIsolated !== 'undefined' && crossOriginIsolated === true\n}\n"],"mappings":";;;;;;;;AAWA,IAAI,gBAA2C;;;;;;;;AAS/C,SAAS,aAAa,SAAyB;CAC3C,OAAO,gDAAgD,QAAQ;AACnE;AAEA,eAAsB,UAA8B;CAChD,IAAI,eAAe,OAAO;CAE1B,iBAAiB,YAAY;EACzB,IAAI;EACJ,IAAI;GACA,MAAM,MAAM,OAAO;EACvB,SAAS,OAAO;GACZ,MAAM,IAAI,MACN,6FACA,EAAE,MAAM,CACZ;EACJ;EAEA,MAAM,SAAS,UAAU;EACzB,IAAI,OAAO,cACP,IAAI,IAAI,KAAK,YAAY,OAAO;OAC7B,IAAI,CAAC,IAAI,IAAI,KAAK,aAAa,IAAI,IAAI,UAAU,KAGpD,IAAI,IAAI,KAAK,YAAY,aAAa,IAAI,IAAI,SAAS,GAAG;EAM9D,IAAI,IAAI,KAAK,aAAa,kBAAkB;EAC5C,IAAI,IAAI,WAAW;EACnB,OAAO;CACX,EAAA,CAAG;CAEH,OAAO;AACX;;AAGA,SAAgB,WAAiB;CAC7B,gBAAgB;AACpB;;AAkBA,MAAM,mCAAmB,IAAI,IAAI,CAAC,OAAO,MAAM,CAAC;;;;;;;AAQhD,SAAgB,sBAAsB,WAA+C;CACjF,IAAI,UAAU,WAAW,KAAK,UAAU,OAAO,SAAS,iBAAiB,IAAI,IAAI,CAAC,GAAG,OAAO;CAC5F,OAAO,CAAC,UAAU,MAAM,SAAS,iBAAiB,IAAI,IAAI,CAAC,KAAK,MAAM;AAC1E;;AAGA,eAAsB,cAClB,OACA,UAA0B,CAAC,GACwB;CACnD,MAAM,MAAM,MAAM,QAAQ;CAC1B,MAAM,QAAQ,iBAAiB,aAAa,QAAQ,IAAI,WAAW,KAAK;CACxE,MAAM,YAAY,CAAC,GAAI,QAAQ,sBAAsB,UAAU,CAAC,CAAC,kBAAmB;CACpF,MAAM,UAAU,uBACZ,IAAI,iBAAiB,OAAO,OAAO;EAC/B;EACA,wBAAwB;EACxB,GAAI,QAAQ,eAAe,EAAE,cAAc,QAAQ,aAAa,IAAI,CAAC;CACzE,CAAC;CAEL,IAAI;EACA,OAAO,MAAM,OAAO,SAAS;CACjC,SAAS,OAAO;EACZ,MAAM,WAAW,QAAQ,iBAAiB,sBAAsB,SAAS,IAAI;EAC7E,IAAI,CAAC,UAAU,MAAM;EACrB,QAAQ,KACJ,gCAAgC,KAAK,UAAU,SAAS,EAAE,6BAClD,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE,qBACxC,KAAK,UAAU,QAAQ,EAAE,EACpD;EACA,OAAO,OAAO,QAAQ;CAC1B;AACJ;;AAGA,eAAsB,oBAAsC;CACxD,MAAM,MAAO,UAA+D;CAC5E,IAAI,CAAC,KAAK,OAAO;CACjB,IAAI;EACA,OAAQ,MAAM,IAAI,eAAe,MAAO;CAC5C,QAAQ;EACJ,OAAO;CACX;AACJ;;;;;;AAOA,SAAgB,wBAAiC;CAC7C,OAAO,OAAO,wBAAwB,eAAe,wBAAwB;AACjF"}

package/dist/pdf/index.js

@@ -1,2 +1,2 @@
-import { _ as extractTextBoxes, a as relativeCharWidth, c as splitRunsIntoWords, d as PdfToImage, f as renderPage, g as TEXT_LAYER_SCORE, h as resetPdfjs, i as cssFontFromPdfName, l as PDF_TO_IMAGE_DEFAULTS, m as loadPdfjs, n as PdfToDocument, o as resetMeasurementContext, p as documentOptions, r as hasUsableTextLayer, s as splitRunIntoWords, t as PDF_TO_DOCUMENT_DEFAULTS, u as POINTS_PER_INCH, v as isTextItem, y as textItemToBox } from "../pdf-BQl0dneD.js";
+import { _ as extractTextBoxes, a as relativeCharWidth, c as splitRunsIntoWords, d as PdfToImage, f as renderPage, g as TEXT_LAYER_SCORE, h as resetPdfjs, i as cssFontFromPdfName, l as PDF_TO_IMAGE_DEFAULTS, m as loadPdfjs, n as PdfToDocument, o as resetMeasurementContext, p as documentOptions, r as hasUsableTextLayer, s as splitRunIntoWords, t as PDF_TO_DOCUMENT_DEFAULTS, u as POINTS_PER_INCH, v as isTextItem, y as textItemToBox } from "../pdf-qbdnOnqZ.js";
 export { PDF_TO_DOCUMENT_DEFAULTS, PDF_TO_IMAGE_DEFAULTS, POINTS_PER_INCH, PdfToDocument, PdfToImage, TEXT_LAYER_SCORE, cssFontFromPdfName, documentOptions, extractTextBoxes, hasUsableTextLayer, isTextItem, loadPdfjs, relativeCharWidth, renderPage, resetMeasurementContext, resetPdfjs, splitRunIntoWords, splitRunsIntoWords, textItemToBox };

package/dist/{pdf-BQl0dneD.js → pdf-qbdnOnqZ.js}

@@ -1,8 +1,8 @@
 import { c as ImageError, h as getConfig, i as Stage } from "./pipeline-DACqGkpN.js";
-import { c as encodeImage, i as createCanvas, r as context2d } from "./image-CAH2rLv9.js";
+import { c as encodeImage, i as createCanvas, r as context2d } from "./image-DRBsbv7G.js";
 import { i as resolveParams, t as BASE_STAGE_DEFAULTS } from "./params-DapwK9Ns.js";
 import { n as createDocument, t as createImage } from "./image-DoZDJkcR.js";
-import { r as toBytes } from "./data-to-image-DoZ4jQ3R.js";
+import { r as toBytes } from "./data-to-image-CRvh92wt.js";
 //#region src/pdf/extract-text.ts
 /** Glyphs sit above the baseline by roughly three quarters of the line height. */
 const ASCENT_RATIO = .75;
@@ -414,4 +414,4 @@ function hasUsableTextLayer(document, minimumBoxes = 1) {
 //#endregion
 export { extractTextBoxes as _, relativeCharWidth as a, splitRunsIntoWords as c, PdfToImage as d, renderPage as f, TEXT_LAYER_SCORE as g, resetPdfjs as h, cssFontFromPdfName as i, PDF_TO_IMAGE_DEFAULTS as l, loadPdfjs as m, PdfToDocument as n, resetMeasurementContext as o, documentOptions as p, hasUsableTextLayer as r, splitRunIntoWords as s, PDF_TO_DOCUMENT_DEFAULTS as t, POINTS_PER_INCH as u, isTextItem as v, textItemToBox as y };
 
-//# sourceMappingURL=pdf-BQl0dneD.js.map
+//# sourceMappingURL=pdf-qbdnOnqZ.js.map

package/dist/{pdf-BQl0dneD.js.map → pdf-qbdnOnqZ.js.map}

@@ -1 +1 @@
-{"version":3,"file":"pdf-BQl0dneD.js","names":[],"sources":["../src/pdf/extract-text.ts","../src/pdf/pdfjs.ts","../src/pdf/pdf-to-image.ts","../src/pdf/split-words.ts","../src/pdf/pdf-to-document.ts"],"sourcesContent":["/**\n * Word-level text extraction from a PDF's embedded text layer.\n *\n * Ported from the pdftools prototype's `extractTextFromPage`, whose rotation\n * handling is the hard-won part and is reproduced here with its reasoning.\n *\n * A pdf.js text item carries a full transform matrix [a, b, c, d, e, f]. Naively\n * reading `a` as a width scale breaks on rotated text, so the reading direction\n * and the \"up\" direction are recovered as *unit* vectors -- otherwise the font\n * size gets counted twice. The four corners are then pushed through\n * `convertToViewportPoint` and reduced to an axis-aligned box.\n *\n * No semantic `angle` is emitted, deliberately. A rotated glyph matrix is\n * indistinguishable from page-level /Rotate compensation or an embedded\n * FontMatrix, so composing the corners geometrically and taking their bounding\n * box is correct regardless of which caused it. `readDirX`/`readDirY` carry the\n * direction that word-splitting needs.\n */\n\nimport type { Box } from '../schemas/box.js'\n\n/** A box plus the reading direction, which splitting needs but ScaleDP's Box lacks. */\nexport interface TextBox extends Box {\n    /** Unit vector, in viewport space, along which reading progresses. */\n    readDirX: number\n    readDirY: number\n    /** pdf.js font identifier, e.g. 'g_d0_f1'. */\n    fontName: string\n}\n\n/** Glyphs sit above the baseline by roughly three quarters of the line height. */\nconst ASCENT_RATIO = 0.75\n\n/** Confidence assigned to text read from a PDF's own text layer. */\nexport const TEXT_LAYER_SCORE = 0.99\n\ninterface TextItemLike {\n    str: string\n    transform: number[]\n    width: number\n    height: number\n    fontName?: string\n}\n\ninterface ViewportLike {\n    convertToViewportPoint(x: number, y: number): number[]\n}\n\nexport function isTextItem(item: unknown): item is TextItemLike {\n    return (\n        typeof item === 'object' &&\n        item !== null &&\n        'str' in item &&\n        'transform' in item &&\n        Array.isArray((item as TextItemLike).transform)\n    )\n}\n\n/** Convert one pdf.js text item into a viewport-space box. */\nexport function textItemToBox(item: TextItemLike, viewport: ViewportLike): TextBox | null {\n    if (item.str.length === 0) return null\n\n    const [a = 0, b = 0, c = 0, d = 0, e = 0, f = 0] = item.transform\n\n    // Unit direction vectors. Dividing out the magnitudes is what stops the\n    // font size being applied twice, since item.width/height already include it.\n    const abMag = Math.hypot(a, b) || 1\n    const cdMag = Math.hypot(c, d) || 1\n    const dirX = a / abMag\n    const dirY = b / abMag\n    const upX = c / cdMag\n    const upY = d / cdMag\n\n    // The transform's origin is the baseline; shift up to the glyph tops.\n    const ascent = item.height * ASCENT_RATIO\n    const startX = e + upX * ascent\n    const startY = f + upY * ascent\n\n    const corners = [\n        viewport.convertToViewportPoint(startX, startY),\n        viewport.convertToViewportPoint(startX + dirX * item.width, startY + dirY * item.width),\n        viewport.convertToViewportPoint(startX - upX * item.height, startY - upY * item.height),\n        viewport.convertToViewportPoint(\n            startX + dirX * item.width - upX * item.height,\n            startY + dirY * item.width - upY * item.height\n        ),\n    ]\n\n    const xs = corners.map((p) => p[0] as number)\n    const ys = corners.map((p) => p[1] as number)\n    const x = Math.min(...xs)\n    const y = Math.min(...ys)\n\n    // Reading direction in viewport space, taken from the start and end points\n    // rather than from the matrix, so the viewport's own flip is accounted for.\n    const [startScreenX = 0, startScreenY = 0] = corners[0] ?? []\n    const [endScreenX = 0, endScreenY = 0] = corners[1] ?? []\n    const readMag = Math.hypot(endScreenX - startScreenX, endScreenY - startScreenY) || 1\n\n    return {\n        text: item.str,\n        score: TEXT_LAYER_SCORE,\n        x: Math.floor(x),\n        y: Math.floor(y),\n        width: Math.max(1, Math.ceil(Math.max(...xs) - x)),\n        height: Math.max(1, Math.ceil(Math.max(...ys) - y)),\n        angle: 0,\n        readDirX: (endScreenX - startScreenX) / readMag,\n        readDirY: (endScreenY - startScreenY) / readMag,\n        fontName: item.fontName ?? '',\n    }\n}\n\n/** Extract every text item on a page as a viewport-space box. */\nexport async function extractTextBoxes(\n    page: { getTextContent(): Promise<{ items: unknown[] }> },\n    viewport: ViewportLike\n): Promise<TextBox[]> {\n    const content = await page.getTextContent()\n    const boxes: TextBox[] = []\n    for (const item of content.items) {\n        if (!isTextItem(item)) continue\n        const box = textItemToBox(item, viewport)\n        if (box) boxes.push(box)\n    }\n    return boxes\n}\n","/**\n * Lazy pdf.js loader.\n *\n * pdfjs-dist is an optional peer dependency, so it is imported only when a PDF\n * stage actually runs. Every asset path comes from `configure()` -- unlike the\n * pdftools prototype, which hardcoded `/pdf.worker.min.mjs`, a path only its\n * own Next app could serve.\n */\n\nimport { getConfig } from '../core/config.js'\n\ntype PdfjsModule = typeof import('pdfjs-dist')\n\nlet modulePromise: Promise<PdfjsModule> | null = null\n\nexport async function loadPdfjs(): Promise<PdfjsModule> {\n    if (modulePromise) return modulePromise\n\n    modulePromise = (async () => {\n        let pdfjs: PdfjsModule\n        try {\n            pdfjs = await import('pdfjs-dist')\n        } catch (cause) {\n            throw new Error('pdfjs-dist is required for PDF support. Install it: npm i pdfjs-dist', { cause })\n        }\n\n        const { workerSrc } = getConfig().pdf\n        if (workerSrc) pdfjs.GlobalWorkerOptions.workerSrc = workerSrc\n        return pdfjs\n    })()\n\n    return modulePromise\n}\n\n/** Reset the cached module. Tests only. */\nexport function resetPdfjs(): void {\n    modulePromise = null\n}\n\n/**\n * Turn pdf.js's worker-setup failure into something actionable.\n *\n * When `workerSrc` is unset or 404s, pdf.js reports \"Setting up fake worker\n * failed\" with a bare module URL, which says nothing about what to do. The\n * worker is not bundled with this library on purpose -- it has to be served by\n * the consuming application -- so the fix is always the same two steps.\n */\nexport function describePdfError(error: unknown): Error {\n    const message = error instanceof Error ? error.message : String(error)\n    if (!/fake worker|worker/i.test(message)) {\n        return error instanceof Error ? error : new Error(message)\n    }\n\n    const { workerSrc } = getConfig().pdf\n    const cause = workerSrc\n        ? `pdf.js could not load its worker from \"${workerSrc}\".`\n        : 'pdf.js has no worker configured.'\n\n    return new Error(\n        `${cause}\\n` +\n            'Copy it out of the package and point the config at it:\\n' +\n            '  cp node_modules/pdfjs-dist/build/pdf.worker.min.mjs public/\\n' +\n            \"  configure({ pdf: { workerSrc: '/pdf.worker.min.mjs' } })\\n\" +\n            `Original error: ${message}`,\n        { cause: error }\n    )\n}\n\n/** Document-level options assembled from the global config. */\nexport function documentOptions(data: Uint8Array): Record<string, unknown> {\n    const { cMapUrl, standardFontDataUrl, wasmUrl } = getConfig().pdf\n    // pdf.js takes ownership of the buffer it is given and detaches it, so hand\n    // over a copy: callers routinely reuse the row's `content` afterwards.\n    const owned = new Uint8Array(data.byteLength)\n    owned.set(data)\n\n    const options: Record<string, unknown> = { data: owned }\n    if (cMapUrl) {\n        options.cMapUrl = cMapUrl\n        options.cMapPacked = true\n    }\n    if (standardFontDataUrl) options.standardFontDataUrl = standardFontDataUrl\n    if (wasmUrl) options.wasmUrl = wasmUrl\n    return options\n}\n","/**\n * Port of `scaledp/pdf/PdfDataToImage.py`: a PDF into one `Image` row per page.\n *\n * Python renders with PyMuPDF at a DPI; pdf.js works in scale factors, so the\n * DPI converts through the PDF unit of 72 points per inch.\n */\n\nimport { ImageError } from '../core/errors.js'\nimport { createCanvas, encodeImage } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { createImage, type ImageFormat, type ScaleDpImage } from '../schemas/image.js'\nimport { toBytes } from '../stages/data-to-image.js'\nimport { describePdfError, documentOptions, loadPdfjs } from './pdfjs.js'\n\n/** PDF user space is defined in points; 72 of them make an inch. */\nexport const POINTS_PER_INCH = 72\n\nexport interface PdfToImageParams extends BaseStageParams {\n    /** Render DPI. 300 matches ScaleDP's default and suits OCR. */\n    resolution: number\n    /** Maximum pages to render; 0 renders all of them. */\n    pageLimit: number\n    imageType: ImageFormat\n}\n\nexport const PDF_TO_IMAGE_DEFAULTS: PdfToImageParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'content',\n    outputCol: 'image',\n    resolution: 300,\n    pageLimit: 0,\n    imageType: 'png' as ImageFormat,\n})\n\nconst MIME: Record<ImageFormat, 'image/png' | 'image/webp' | 'image/jpeg'> = {\n    png: 'image/png',\n    webp: 'image/webp',\n    jpeg: 'image/jpeg',\n}\n\nexport class PdfToImage extends Stage<PdfToImageParams> {\n    readonly name = 'PdfToImage'\n\n    constructor(options: Partial<PdfToImageParams> = {}) {\n        super(\n            resolveParams(PDF_TO_IMAGE_DEFAULTS, options, {\n                resolution: (value) => {\n                    if (!Number.isFinite(value) || value <= 0) {\n                        throw new RangeError(`resolution must be positive, received ${value}`)\n                    }\n                },\n                pageLimit: (value) => {\n                    if (!Number.isInteger(value) || value < 0) {\n                        throw new RangeError(`pageLimit must be a non-negative integer, received ${value}`)\n                    }\n                },\n            })\n        )\n    }\n\n    /** One input PDF becomes N rows, each carrying its page index. */\n    protected override async expand(input: unknown, row: Row, ctx: StageContext): Promise<Row[]> {\n        const { outputCol, pageCol, pathCol, resolution, pageLimit, imageType } = this.params\n        const path = String(row[pathCol] ?? 'memory')\n\n        const pdfjs = await loadPdfjs()\n        const task = pdfjs.getDocument(documentOptions(toBytes(input)))\n\n        try {\n            // pdf.js defers worker setup, so a missing worker surfaces on first\n            // page access rather than from task.promise. Wrap the whole\n            // operation so the failure is described wherever it lands.\n            const document = await task.promise\n            const pageCount = pageLimit > 0 ? Math.min(pageLimit, document.numPages) : document.numPages\n            const rows: Row[] = []\n\n            for (let index = 0; index < pageCount; index++) {\n                ctx.signal?.throwIfAborted()\n                const image = await renderPage(document, index + 1, {\n                    resolution,\n                    imageType,\n                    path,\n                })\n                rows.push({ ...row, [pageCol]: index, [outputCol]: image })\n            }\n            return rows\n        } catch (error) {\n            throw describePdfError(error)\n        } finally {\n            // destroy() lives on the loading task, not the document proxy, and\n            // is what releases the pdf.js worker's copy of the file.\n            await task.destroy()\n        }\n    }\n\n    protected async apply(): Promise<never> {\n        throw new ImageError('unreachable: expand handles every row', this.name)\n    }\n\n    protected onError(message: string, row: Row): ScaleDpImage {\n        return createImage({ path: String(row[this.params.pathCol] ?? 'memory'), exception: message })\n    }\n}\n\n/** Rasterise a single 1-based page to encoded image bytes. */\nexport async function renderPage(\n    document: Awaited<ReturnType<typeof import('pdfjs-dist').getDocument>['promise']>,\n    pageNumber: number,\n    opts: { resolution: number; imageType: ImageFormat; path: string }\n): Promise<ScaleDpImage> {\n    const page = await document.getPage(pageNumber)\n    try {\n        const viewport = page.getViewport({ scale: opts.resolution / POINTS_PER_INCH })\n        const canvas = createCanvas(viewport.width, viewport.height)\n\n        // pdf.js >= 5 takes `canvas`, not `canvasContext`. Passing the context\n        // still works but is the documented legacy path.\n        await page.render({\n            canvas: canvas as unknown as HTMLCanvasElement,\n            viewport,\n        }).promise\n\n        return createImage({\n            path: opts.path,\n            resolution: opts.resolution,\n            data: await encodeImage(canvas, MIME[opts.imageType]),\n            imageType: opts.imageType,\n            width: canvas.width,\n            height: canvas.height,\n        })\n    } finally {\n        page.cleanup()\n    }\n}\n","/**\n * Split a pdf.js text run into word-level boxes.\n *\n * pdf.js emits runs at line granularity (\"Client: Raja Raman\"), but detection\n * boxes and NER offsets both want words. Each word is measured with Canvas 2D\n * `measureText` using a font reconstructed from the pdf.js font name, then the\n * measured widths are scaled so they sum to the run's actual width -- the\n * substitute font is never metrically identical to the embedded one, so the\n * measurements are only useful as *proportions*.\n *\n * Walking along `readDirX`/`readDirY` rather than assuming left-to-right is what\n * makes rotated, bottom-to-top and right-to-left runs come out correctly.\n */\n\nimport { context2d, createCanvas } from '../core/image.js'\nimport type { Box } from '../schemas/box.js'\nimport type { TextBox } from './extract-text.js'\n\n/** Widen each word slightly so glyph overhang is not clipped. */\nconst WORD_PADDING_RATIO = 0.1\n\nlet measureCtx: OffscreenCanvasRenderingContext2D | null = null\n\nfunction measurementContext(): OffscreenCanvasRenderingContext2D | null {\n    if (measureCtx) return measureCtx\n    try {\n        measureCtx = context2d(createCanvas(1, 1))\n        return measureCtx\n    } catch {\n        // No canvas (e.g. a non-browser test run): fall back to glyph heuristics.\n        return null\n    }\n}\n\n/**\n * Rebuild a CSS font string from a pdf.js font name.\n *\n * Names look like `AAAAAA+Helvetica-BoldOblique`: a six-letter subset prefix,\n * then the real family and style suffixes.\n */\nexport function cssFontFromPdfName(fontName: string, size: number): string {\n    const name = fontName.replace(/^[A-Z]{6}\\+/, '')\n    const lower = name.toLowerCase()\n    const weight = /bold|black|heavy|semibold/.test(lower) ? 'bold' : 'normal'\n    const style = /italic|oblique/.test(lower) ? 'italic' : 'normal'\n    const family = /serif|times|georgia|garamond|roman/.test(lower)\n        ? 'serif'\n        : /mono|courier|consol/.test(lower)\n          ? 'monospace'\n          : 'sans-serif'\n    return `${style} ${weight} ${Math.max(1, Math.round(size))}px ${family}`\n}\n\n/**\n * Relative advance width per character, used when no canvas is available.\n * Buckets rather than real metrics -- enough to keep proportions sane.\n */\nexport function relativeCharWidth(char: string): number {\n    if (\"iljI|.,:;'`!\".includes(char)) return 0.6\n    if ('ftr()[]{}-'.includes(char)) return 0.8\n    if ('MWmw@%'.includes(char)) return 1.6\n    if (char === ' ') return 0.6\n    if (char >= 'A' && char <= 'Z') return 1.3\n    return 1.0\n}\n\nfunction measureWord(word: string, font: string): number {\n    const ctx = measurementContext()\n    if (ctx) {\n        ctx.font = font\n        return ctx.measureText(word).width\n    }\n    let total = 0\n    for (const char of word) total += relativeCharWidth(char)\n    return total\n}\n\n/**\n * Split one run into word boxes.\n *\n * Returns the run itself when it holds a single word, so the common case costs\n * nothing.\n */\nexport function splitRunIntoWords(run: TextBox): Box[] {\n    const trimmed = run.text.trim()\n    if (trimmed.length === 0) return []\n\n    const tokens = trimmed.split(/(\\s+)/).filter((t) => t.length > 0)\n    const words = tokens.filter((t) => !/^\\s+$/.test(t))\n    if (words.length <= 1) {\n        return [{ ...run, text: trimmed }]\n    }\n\n    const font = cssFontFromPdfName(run.fontName, run.height)\n    const measured = tokens.map((token) => measureWord(token, font))\n    const totalMeasured = measured.reduce((sum, w) => sum + w, 0) || 1\n\n    // The substitute font's absolute metrics are meaningless; only the ratios\n    // matter, so normalise them onto the run's real extent.\n    const runLength = Math.hypot(run.width * run.readDirX, run.height * run.readDirY) || run.width\n    const scale = runLength / totalMeasured\n\n    // Walking starts at whichever corner the reading direction comes *from*, so\n    // a right-to-left or bottom-to-top run starts at the opposite edge.\n    let cursorX = run.readDirX >= 0 ? run.x : run.x + run.width\n    let cursorY = run.readDirY >= 0 ? run.y : run.y + run.height\n\n    const out: Box[] = []\n    for (const [i, token] of tokens.entries()) {\n        const advance = (measured[i] as number) * scale\n        if (!/^\\s+$/.test(token)) {\n            const pad = run.height * WORD_PADDING_RATIO\n            const spanX = Math.abs(run.readDirX) > Math.abs(run.readDirY) ? advance : run.width\n            const spanY = Math.abs(run.readDirY) > Math.abs(run.readDirX) ? advance : run.height\n\n            const left = run.readDirX >= 0 ? cursorX : cursorX - spanX\n            const top = run.readDirY >= 0 ? cursorY : cursorY - spanY\n\n            out.push({\n                text: token,\n                score: run.score,\n                x: Math.floor(left - pad),\n                y: Math.floor(top - pad),\n                width: Math.max(1, Math.ceil(spanX + pad * 2)),\n                height: Math.max(1, Math.ceil(spanY + pad * 2)),\n                angle: run.angle,\n            })\n        }\n        cursorX += run.readDirX * advance\n        cursorY += run.readDirY * advance\n    }\n    return out\n}\n\n/** Split every run on a page into word boxes. */\nexport function splitRunsIntoWords(runs: readonly TextBox[]): Box[] {\n    return runs.flatMap(splitRunIntoWords)\n}\n\n/** Reset the cached measurement canvas. Tests only. */\nexport function resetMeasurementContext(): void {\n    measureCtx = null\n}\n","/**\n * Port of `scaledp/pdf/PdfDataToText.py`: a PDF's embedded text layer into one\n * `Document` row per page, with word-level boxes.\n *\n * Coordinates are emitted in the same pixel space `PdfToImage` renders at, so\n * boxes from this stage and boxes from OCR are directly comparable. Python\n * leaves PdfDataToText in PDF points and scales only in PdfDataToDocument; a\n * single consistent space is more useful and avoids a class of silent mismatch.\n *\n * The output feeds the `bypassCol` optimisation: a page that already has a text\n * layer does not need OCR.\n */\n\nimport { ImageError } from '../core/errors.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport { toBytes } from '../stages/data-to-image.js'\nimport { extractTextBoxes } from './extract-text.js'\nimport { POINTS_PER_INCH } from './pdf-to-image.js'\nimport { describePdfError, documentOptions, loadPdfjs } from './pdfjs.js'\nimport { splitRunsIntoWords } from './split-words.js'\n\nexport interface PdfToDocumentParams extends BaseStageParams {\n    /** Pixel space the boxes are expressed in; match PdfToImage to align them. */\n    resolution: number\n    pageLimit: number\n    /** Split pdf.js line runs into word boxes. Off yields run-level boxes. */\n    splitWords: boolean\n}\n\nexport const PDF_TO_DOCUMENT_DEFAULTS: PdfToDocumentParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'content',\n    outputCol: 'document',\n    keepInputData: true,\n    resolution: 300,\n    pageLimit: 0,\n    splitWords: true,\n})\n\nexport class PdfToDocument extends Stage<PdfToDocumentParams> {\n    readonly name = 'PdfToDocument'\n\n    constructor(options: Partial<PdfToDocumentParams> = {}) {\n        super(resolveParams(PDF_TO_DOCUMENT_DEFAULTS, options))\n    }\n\n    protected override async expand(input: unknown, row: Row, ctx: StageContext): Promise<Row[]> {\n        const { outputCol, pageCol, pathCol, resolution, pageLimit, splitWords } = this.params\n        const path = String(row[pathCol] ?? 'memory')\n\n        const pdfjs = await loadPdfjs()\n        const task = pdfjs.getDocument(documentOptions(toBytes(input)))\n\n        try {\n            // pdf.js defers worker setup, so a missing worker surfaces on first\n            // page access rather than from task.promise.\n            const pdf = await task.promise\n            const pageCount = pageLimit > 0 ? Math.min(pageLimit, pdf.numPages) : pdf.numPages\n            const rows: Row[] = []\n\n            for (let index = 0; index < pageCount; index++) {\n                ctx.signal?.throwIfAborted()\n                const page = await pdf.getPage(index + 1)\n                try {\n                    const viewport = page.getViewport({ scale: resolution / POINTS_PER_INCH })\n                    const runs = await extractTextBoxes(page, viewport)\n                    const bboxes = splitWords ? splitRunsIntoWords(runs) : runs\n\n                    rows.push({\n                        ...row,\n                        [pageCol]: index,\n                        [outputCol]: createDocument({\n                            path,\n                            type: 'pdf',\n                            text: runs.map((r) => r.text).join('\\n'),\n                            bboxes,\n                        }),\n                    })\n                } finally {\n                    page.cleanup()\n                }\n            }\n            return rows\n        } catch (error) {\n            throw describePdfError(error)\n        } finally {\n            await task.destroy()\n        }\n    }\n\n    protected async apply(): Promise<never> {\n        throw new ImageError('unreachable: expand handles every row', this.name)\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'pdf',\n            exception: message,\n        })\n    }\n}\n\n/** True when a page's text layer is substantive enough to skip OCR. */\nexport function hasUsableTextLayer(document: Document, minimumBoxes = 1): boolean {\n    return document.exception === '' && document.bboxes.length >= minimumBoxes\n}\n"],"mappings":";;;;;;;AA+BA,MAAM,eAAe;;AAGrB,MAAa,mBAAmB;AAchC,SAAgB,WAAW,MAAqC;CAC5D,OACI,OAAO,SAAS,YAChB,SAAS,QACT,SAAS,QACT,eAAe,QACf,MAAM,QAAS,KAAsB,SAAS;AAEtD;;AAGA,SAAgB,cAAc,MAAoB,UAAwC;CACtF,IAAI,KAAK,IAAI,WAAW,GAAG,OAAO;CAElC,MAAM,CAAC,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,KAAK,KAAK;CAIxD,MAAM,QAAQ,KAAK,MAAM,GAAG,CAAC,KAAK;CAClC,MAAM,QAAQ,KAAK,MAAM,GAAG,CAAC,KAAK;CAClC,MAAM,OAAO,IAAI;CACjB,MAAM,OAAO,IAAI;CACjB,MAAM,MAAM,IAAI;CAChB,MAAM,MAAM,IAAI;CAGhB,MAAM,SAAS,KAAK,SAAS;CAC7B,MAAM,SAAS,IAAI,MAAM;CACzB,MAAM,SAAS,IAAI,MAAM;CAEzB,MAAM,UAAU;EACZ,SAAS,uBAAuB,QAAQ,MAAM;EAC9C,SAAS,uBAAuB,SAAS,OAAO,KAAK,OAAO,SAAS,OAAO,KAAK,KAAK;EACtF,SAAS,uBAAuB,SAAS,MAAM,KAAK,QAAQ,SAAS,MAAM,KAAK,MAAM;EACtF,SAAS,uBACL,SAAS,OAAO,KAAK,QAAQ,MAAM,KAAK,QACxC,SAAS,OAAO,KAAK,QAAQ,MAAM,KAAK,MAC5C;CACJ;CAEA,MAAM,KAAK,QAAQ,KAAK,MAAM,EAAE,EAAY;CAC5C,MAAM,KAAK,QAAQ,KAAK,MAAM,EAAE,EAAY;CAC5C,MAAM,IAAI,KAAK,IAAI,GAAG,EAAE;CACxB,MAAM,IAAI,KAAK,IAAI,GAAG,EAAE;CAIxB,MAAM,CAAC,eAAe,GAAG,eAAe,KAAK,QAAQ,MAAM,CAAC;CAC5D,MAAM,CAAC,aAAa,GAAG,aAAa,KAAK,QAAQ,MAAM,CAAC;CACxD,MAAM,UAAU,KAAK,MAAM,aAAa,cAAc,aAAa,YAAY,KAAK;CAEpF,OAAO;EACH,MAAM,KAAK;EACX,OAAO;EACP,GAAG,KAAK,MAAM,CAAC;EACf,GAAG,KAAK,MAAM,CAAC;EACf,OAAO,KAAK,IAAI,GAAG,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC;EACjD,QAAQ,KAAK,IAAI,GAAG,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC;EAClD,OAAO;EACP,WAAW,aAAa,gBAAgB;EACxC,WAAW,aAAa,gBAAgB;EACxC,UAAU,KAAK,YAAY;CAC/B;AACJ;;AAGA,eAAsB,iBAClB,MACA,UACkB;CAClB,MAAM,UAAU,MAAM,KAAK,eAAe;CAC1C,MAAM,QAAmB,CAAC;CAC1B,KAAK,MAAM,QAAQ,QAAQ,OAAO;EAC9B,IAAI,CAAC,WAAW,IAAI,GAAG;EACvB,MAAM,MAAM,cAAc,MAAM,QAAQ;EACxC,IAAI,KAAK,MAAM,KAAK,GAAG;CAC3B;CACA,OAAO;AACX;;;;;;;;;;;ACjHA,IAAI,gBAA6C;AAEjD,eAAsB,YAAkC;CACpD,IAAI,eAAe,OAAO;CAE1B,iBAAiB,YAAY;EACzB,IAAI;EACJ,IAAI;GACA,QAAQ,MAAM,OAAO;EACzB,SAAS,OAAO;GACZ,MAAM,IAAI,MAAM,wEAAwE,EAAE,MAAM,CAAC;EACrG;EAEA,MAAM,EAAE,cAAc,UAAU,CAAC,CAAC;EAClC,IAAI,WAAW,MAAM,oBAAoB,YAAY;EACrD,OAAO;CACX,EAAA,CAAG;CAEH,OAAO;AACX;;AAGA,SAAgB,aAAmB;CAC/B,gBAAgB;AACpB;;;;;;;;;AAUA,SAAgB,iBAAiB,OAAuB;CACpD,MAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;CACrE,IAAI,CAAC,sBAAsB,KAAK,OAAO,GACnC,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO;CAG7D,MAAM,EAAE,cAAc,UAAU,CAAC,CAAC;CAClC,MAAM,QAAQ,YACR,0CAA0C,UAAU,MACpD;CAEN,OAAO,IAAI,MACP,GAAG,MAAM;;;kBAIc,WACvB,EAAE,OAAO,MAAM,CACnB;AACJ;;AAGA,SAAgB,gBAAgB,MAA2C;CACvE,MAAM,EAAE,SAAS,qBAAqB,YAAY,UAAU,CAAC,CAAC;CAG9D,MAAM,QAAQ,IAAI,WAAW,KAAK,UAAU;CAC5C,MAAM,IAAI,IAAI;CAEd,MAAM,UAAmC,EAAE,MAAM,MAAM;CACvD,IAAI,SAAS;EACT,QAAQ,UAAU;EAClB,QAAQ,aAAa;CACzB;CACA,IAAI,qBAAqB,QAAQ,sBAAsB;CACvD,IAAI,SAAS,QAAQ,UAAU;CAC/B,OAAO;AACX;;;;;;;;;;ACpEA,MAAa,kBAAkB;AAU/B,MAAa,wBAA0C,OAAO,OAAO;CACjE,GAAG;CACH,UAAU;CACV,WAAW;CACX,YAAY;CACZ,WAAW;CACX,WAAW;AACf,CAAC;AAED,MAAM,OAAuE;CACzE,KAAK;CACL,MAAM;CACN,MAAM;AACV;AAEA,IAAa,aAAb,cAAgC,MAAwB;CACpD,OAAgB;CAEhB,YAAY,UAAqC,CAAC,GAAG;EACjD,MACI,cAAc,uBAAuB,SAAS;GAC1C,aAAa,UAAU;IACnB,IAAI,CAAC,OAAO,SAAS,KAAK,KAAK,SAAS,GACpC,MAAM,IAAI,WAAW,yCAAyC,OAAO;GAE7E;GACA,YAAY,UAAU;IAClB,IAAI,CAAC,OAAO,UAAU,KAAK,KAAK,QAAQ,GACpC,MAAM,IAAI,WAAW,sDAAsD,OAAO;GAE1F;EACJ,CAAC,CACL;CACJ;;CAGA,MAAyB,OAAO,OAAgB,KAAU,KAAmC;EACzF,MAAM,EAAE,WAAW,SAAS,SAAS,YAAY,WAAW,cAAc,KAAK;EAC/E,MAAM,OAAO,OAAO,IAAI,YAAY,QAAQ;EAG5C,MAAM,QAAO,MADO,UAAU,EAAA,CACX,YAAY,gBAAgB,QAAQ,KAAK,CAAC,CAAC;EAE9D,IAAI;GAIA,MAAM,WAAW,MAAM,KAAK;GAC5B,MAAM,YAAY,YAAY,IAAI,KAAK,IAAI,WAAW,SAAS,QAAQ,IAAI,SAAS;GACpF,MAAM,OAAc,CAAC;GAErB,KAAK,IAAI,QAAQ,GAAG,QAAQ,WAAW,SAAS;IAC5C,IAAI,QAAQ,eAAe;IAC3B,MAAM,QAAQ,MAAM,WAAW,UAAU,QAAQ,GAAG;KAChD;KACA;KACA;IACJ,CAAC;IACD,KAAK,KAAK;KAAE,GAAG;MAAM,UAAU;MAAQ,YAAY;IAAM,CAAC;GAC9D;GACA,OAAO;EACX,SAAS,OAAO;GACZ,MAAM,iBAAiB,KAAK;EAChC,UAAU;GAGN,MAAM,KAAK,QAAQ;EACvB;CACJ;CAEA,MAAgB,QAAwB;EACpC,MAAM,IAAI,WAAW,yCAAyC,KAAK,IAAI;CAC3E;CAEA,QAAkB,SAAiB,KAAwB;EACvD,OAAO,YAAY;GAAE,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GAAG,WAAW;EAAQ,CAAC;CACjG;AACJ;;AAGA,eAAsB,WAClB,UACA,YACA,MACqB;CACrB,MAAM,OAAO,MAAM,SAAS,QAAQ,UAAU;CAC9C,IAAI;EACA,MAAM,WAAW,KAAK,YAAY,EAAE,OAAO,KAAK,aAAA,GAA6B,CAAC;EAC9E,MAAM,SAAS,aAAa,SAAS,OAAO,SAAS,MAAM;EAI3D,MAAM,KAAK,OAAO;GACN;GACR;EACJ,CAAC,CAAC,CAAC;EAEH,OAAO,YAAY;GACf,MAAM,KAAK;GACX,YAAY,KAAK;GACjB,MAAM,MAAM,YAAY,QAAQ,KAAK,KAAK,UAAU;GACpD,WAAW,KAAK;GAChB,OAAO,OAAO;GACd,QAAQ,OAAO;EACnB,CAAC;CACL,UAAU;EACN,KAAK,QAAQ;CACjB;AACJ;;;;;;;;;;;;;;;;;ACnHA,MAAM,qBAAqB;AAE3B,IAAI,aAAuD;AAE3D,SAAS,qBAA+D;CACpE,IAAI,YAAY,OAAO;CACvB,IAAI;EACA,aAAa,UAAU,aAAa,GAAG,CAAC,CAAC;EACzC,OAAO;CACX,QAAQ;EAEJ,OAAO;CACX;AACJ;;;;;;;AAQA,SAAgB,mBAAmB,UAAkB,MAAsB;CAEvE,MAAM,QADO,SAAS,QAAQ,eAAe,EAC5B,CAAC,CAAC,YAAY;CAC/B,MAAM,SAAS,4BAA4B,KAAK,KAAK,IAAI,SAAS;CAClE,MAAM,QAAQ,iBAAiB,KAAK,KAAK,IAAI,WAAW;CACxD,MAAM,SAAS,qCAAqC,KAAK,KAAK,IACxD,UACA,sBAAsB,KAAK,KAAK,IAC9B,cACA;CACR,OAAO,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,IAAI,GAAG,KAAK,MAAM,IAAI,CAAC,EAAE,KAAK;AACpE;;;;;AAMA,SAAgB,kBAAkB,MAAsB;CACpD,IAAI,eAAe,SAAS,IAAI,GAAG,OAAO;CAC1C,IAAI,aAAa,SAAS,IAAI,GAAG,OAAO;CACxC,IAAI,SAAS,SAAS,IAAI,GAAG,OAAO;CACpC,IAAI,SAAS,KAAK,OAAO;CACzB,IAAI,QAAQ,OAAO,QAAQ,KAAK,OAAO;CACvC,OAAO;AACX;AAEA,SAAS,YAAY,MAAc,MAAsB;CACrD,MAAM,MAAM,mBAAmB;CAC/B,IAAI,KAAK;EACL,IAAI,OAAO;EACX,OAAO,IAAI,YAAY,IAAI,CAAC,CAAC;CACjC;CACA,IAAI,QAAQ;CACZ,KAAK,MAAM,QAAQ,MAAM,SAAS,kBAAkB,IAAI;CACxD,OAAO;AACX;;;;;;;AAQA,SAAgB,kBAAkB,KAAqB;CACnD,MAAM,UAAU,IAAI,KAAK,KAAK;CAC9B,IAAI,QAAQ,WAAW,GAAG,OAAO,CAAC;CAElC,MAAM,SAAS,QAAQ,MAAM,OAAO,CAAC,CAAC,QAAQ,MAAM,EAAE,SAAS,CAAC;CAEhE,IADc,OAAO,QAAQ,MAAM,CAAC,QAAQ,KAAK,CAAC,CAC1C,CAAC,CAAC,UAAU,GAChB,OAAO,CAAC;EAAE,GAAG;EAAK,MAAM;CAAQ,CAAC;CAGrC,MAAM,OAAO,mBAAmB,IAAI,UAAU,IAAI,MAAM;CACxD,MAAM,WAAW,OAAO,KAAK,UAAU,YAAY,OAAO,IAAI,CAAC;CAC/D,MAAM,gBAAgB,SAAS,QAAQ,KAAK,MAAM,MAAM,GAAG,CAAC,KAAK;CAKjE,MAAM,SADY,KAAK,MAAM,IAAI,QAAQ,IAAI,UAAU,IAAI,SAAS,IAAI,QAAQ,KAAK,IAAI,SAC/D;CAI1B,IAAI,UAAU,IAAI,YAAY,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI;CACtD,IAAI,UAAU,IAAI,YAAY,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI;CAEtD,MAAM,MAAa,CAAC;CACpB,KAAK,MAAM,CAAC,GAAG,UAAU,OAAO,QAAQ,GAAG;EACvC,MAAM,UAAW,SAAS,KAAgB;EAC1C,IAAI,CAAC,QAAQ,KAAK,KAAK,GAAG;GACtB,MAAM,MAAM,IAAI,SAAS;GACzB,MAAM,QAAQ,KAAK,IAAI,IAAI,QAAQ,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,UAAU,IAAI;GAC9E,MAAM,QAAQ,KAAK,IAAI,IAAI,QAAQ,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,UAAU,IAAI;GAE9E,MAAM,OAAO,IAAI,YAAY,IAAI,UAAU,UAAU;GACrD,MAAM,MAAM,IAAI,YAAY,IAAI,UAAU,UAAU;GAEpD,IAAI,KAAK;IACL,MAAM;IACN,OAAO,IAAI;IACX,GAAG,KAAK,MAAM,OAAO,GAAG;IACxB,GAAG,KAAK,MAAM,MAAM,GAAG;IACvB,OAAO,KAAK,IAAI,GAAG,KAAK,KAAK,QAAQ,MAAM,CAAC,CAAC;IAC7C,QAAQ,KAAK,IAAI,GAAG,KAAK,KAAK,QAAQ,MAAM,CAAC,CAAC;IAC9C,OAAO,IAAI;GACf,CAAC;EACL;EACA,WAAW,IAAI,WAAW;EAC1B,WAAW,IAAI,WAAW;CAC9B;CACA,OAAO;AACX;;AAGA,SAAgB,mBAAmB,MAAiC;CAChE,OAAO,KAAK,QAAQ,iBAAiB;AACzC;;AAGA,SAAgB,0BAAgC;CAC5C,aAAa;AACjB;;;;;;;;;;;;;;;AC/GA,MAAa,2BAAgD,OAAO,OAAO;CACvE,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,YAAY;CACZ,WAAW;CACX,YAAY;AAChB,CAAC;AAED,IAAa,gBAAb,cAAmC,MAA2B;CAC1D,OAAgB;CAEhB,YAAY,UAAwC,CAAC,GAAG;EACpD,MAAM,cAAc,0BAA0B,OAAO,CAAC;CAC1D;CAEA,MAAyB,OAAO,OAAgB,KAAU,KAAmC;EACzF,MAAM,EAAE,WAAW,SAAS,SAAS,YAAY,WAAW,eAAe,KAAK;EAChF,MAAM,OAAO,OAAO,IAAI,YAAY,QAAQ;EAG5C,MAAM,QAAO,MADO,UAAU,EAAA,CACX,YAAY,gBAAgB,QAAQ,KAAK,CAAC,CAAC;EAE9D,IAAI;GAGA,MAAM,MAAM,MAAM,KAAK;GACvB,MAAM,YAAY,YAAY,IAAI,KAAK,IAAI,WAAW,IAAI,QAAQ,IAAI,IAAI;GAC1E,MAAM,OAAc,CAAC;GAErB,KAAK,IAAI,QAAQ,GAAG,QAAQ,WAAW,SAAS;IAC5C,IAAI,QAAQ,eAAe;IAC3B,MAAM,OAAO,MAAM,IAAI,QAAQ,QAAQ,CAAC;IACxC,IAAI;KAEA,MAAM,OAAO,MAAM,iBAAiB,MADnB,KAAK,YAAY,EAAE,OAAO,aAAA,GAA6B,CAC9B,CAAQ;KAClD,MAAM,SAAS,aAAa,mBAAmB,IAAI,IAAI;KAEvD,KAAK,KAAK;MACN,GAAG;OACF,UAAU;OACV,YAAY,eAAe;OACxB;OACA,MAAM;OACN,MAAM,KAAK,KAAK,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,IAAI;OACvC;MACJ,CAAC;KACL,CAAC;IACL,UAAU;KACN,KAAK,QAAQ;IACjB;GACJ;GACA,OAAO;EACX,SAAS,OAAO;GACZ,MAAM,iBAAiB,KAAK;EAChC,UAAU;GACN,MAAM,KAAK,QAAQ;EACvB;CACJ;CAEA,MAAgB,QAAwB;EACpC,MAAM,IAAI,WAAW,yCAAyC,KAAK,IAAI;CAC3E;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;AACJ;;AAGA,SAAgB,mBAAmB,UAAoB,eAAe,GAAY;CAC9E,OAAO,SAAS,cAAc,MAAM,SAAS,OAAO,UAAU;AAClE"}
+{"version":3,"file":"pdf-qbdnOnqZ.js","names":[],"sources":["../src/pdf/extract-text.ts","../src/pdf/pdfjs.ts","../src/pdf/pdf-to-image.ts","../src/pdf/split-words.ts","../src/pdf/pdf-to-document.ts"],"sourcesContent":["/**\n * Word-level text extraction from a PDF's embedded text layer.\n *\n * Ported from the pdftools prototype's `extractTextFromPage`, whose rotation\n * handling is the hard-won part and is reproduced here with its reasoning.\n *\n * A pdf.js text item carries a full transform matrix [a, b, c, d, e, f]. Naively\n * reading `a` as a width scale breaks on rotated text, so the reading direction\n * and the \"up\" direction are recovered as *unit* vectors -- otherwise the font\n * size gets counted twice. The four corners are then pushed through\n * `convertToViewportPoint` and reduced to an axis-aligned box.\n *\n * No semantic `angle` is emitted, deliberately. A rotated glyph matrix is\n * indistinguishable from page-level /Rotate compensation or an embedded\n * FontMatrix, so composing the corners geometrically and taking their bounding\n * box is correct regardless of which caused it. `readDirX`/`readDirY` carry the\n * direction that word-splitting needs.\n */\n\nimport type { Box } from '../schemas/box.js'\n\n/** A box plus the reading direction, which splitting needs but ScaleDP's Box lacks. */\nexport interface TextBox extends Box {\n    /** Unit vector, in viewport space, along which reading progresses. */\n    readDirX: number\n    readDirY: number\n    /** pdf.js font identifier, e.g. 'g_d0_f1'. */\n    fontName: string\n}\n\n/** Glyphs sit above the baseline by roughly three quarters of the line height. */\nconst ASCENT_RATIO = 0.75\n\n/** Confidence assigned to text read from a PDF's own text layer. */\nexport const TEXT_LAYER_SCORE = 0.99\n\ninterface TextItemLike {\n    str: string\n    transform: number[]\n    width: number\n    height: number\n    fontName?: string\n}\n\ninterface ViewportLike {\n    convertToViewportPoint(x: number, y: number): number[]\n}\n\nexport function isTextItem(item: unknown): item is TextItemLike {\n    return (\n        typeof item === 'object' &&\n        item !== null &&\n        'str' in item &&\n        'transform' in item &&\n        Array.isArray((item as TextItemLike).transform)\n    )\n}\n\n/** Convert one pdf.js text item into a viewport-space box. */\nexport function textItemToBox(item: TextItemLike, viewport: ViewportLike): TextBox | null {\n    if (item.str.length === 0) return null\n\n    const [a = 0, b = 0, c = 0, d = 0, e = 0, f = 0] = item.transform\n\n    // Unit direction vectors. Dividing out the magnitudes is what stops the\n    // font size being applied twice, since item.width/height already include it.\n    const abMag = Math.hypot(a, b) || 1\n    const cdMag = Math.hypot(c, d) || 1\n    const dirX = a / abMag\n    const dirY = b / abMag\n    const upX = c / cdMag\n    const upY = d / cdMag\n\n    // The transform's origin is the baseline; shift up to the glyph tops.\n    const ascent = item.height * ASCENT_RATIO\n    const startX = e + upX * ascent\n    const startY = f + upY * ascent\n\n    const corners = [\n        viewport.convertToViewportPoint(startX, startY),\n        viewport.convertToViewportPoint(startX + dirX * item.width, startY + dirY * item.width),\n        viewport.convertToViewportPoint(startX - upX * item.height, startY - upY * item.height),\n        viewport.convertToViewportPoint(\n            startX + dirX * item.width - upX * item.height,\n            startY + dirY * item.width - upY * item.height\n        ),\n    ]\n\n    const xs = corners.map((p) => p[0] as number)\n    const ys = corners.map((p) => p[1] as number)\n    const x = Math.min(...xs)\n    const y = Math.min(...ys)\n\n    // Reading direction in viewport space, taken from the start and end points\n    // rather than from the matrix, so the viewport's own flip is accounted for.\n    const [startScreenX = 0, startScreenY = 0] = corners[0] ?? []\n    const [endScreenX = 0, endScreenY = 0] = corners[1] ?? []\n    const readMag = Math.hypot(endScreenX - startScreenX, endScreenY - startScreenY) || 1\n\n    return {\n        text: item.str,\n        score: TEXT_LAYER_SCORE,\n        x: Math.floor(x),\n        y: Math.floor(y),\n        width: Math.max(1, Math.ceil(Math.max(...xs) - x)),\n        height: Math.max(1, Math.ceil(Math.max(...ys) - y)),\n        angle: 0,\n        readDirX: (endScreenX - startScreenX) / readMag,\n        readDirY: (endScreenY - startScreenY) / readMag,\n        fontName: item.fontName ?? '',\n    }\n}\n\n/** Extract every text item on a page as a viewport-space box. */\nexport async function extractTextBoxes(\n    page: { getTextContent(): Promise<{ items: unknown[] }> },\n    viewport: ViewportLike\n): Promise<TextBox[]> {\n    const content = await page.getTextContent()\n    const boxes: TextBox[] = []\n    for (const item of content.items) {\n        if (!isTextItem(item)) continue\n        const box = textItemToBox(item, viewport)\n        if (box) boxes.push(box)\n    }\n    return boxes\n}\n","/**\n * Lazy pdf.js loader.\n *\n * pdfjs-dist is an optional peer dependency, so it is imported only when a PDF\n * stage actually runs. Every asset path comes from `configure()` -- unlike the\n * pdftools prototype, which hardcoded `/pdf.worker.min.mjs`, a path only its\n * own Next app could serve.\n */\n\nimport { getConfig } from '../core/config.js'\n\ntype PdfjsModule = typeof import('pdfjs-dist')\n\nlet modulePromise: Promise<PdfjsModule> | null = null\n\nexport async function loadPdfjs(): Promise<PdfjsModule> {\n    if (modulePromise) return modulePromise\n\n    modulePromise = (async () => {\n        let pdfjs: PdfjsModule\n        try {\n            pdfjs = await import('pdfjs-dist')\n        } catch (cause) {\n            throw new Error('pdfjs-dist is required for PDF support. Install it: npm i pdfjs-dist', { cause })\n        }\n\n        const { workerSrc } = getConfig().pdf\n        if (workerSrc) pdfjs.GlobalWorkerOptions.workerSrc = workerSrc\n        return pdfjs\n    })()\n\n    return modulePromise\n}\n\n/** Reset the cached module. Tests only. */\nexport function resetPdfjs(): void {\n    modulePromise = null\n}\n\n/**\n * Turn pdf.js's worker-setup failure into something actionable.\n *\n * When `workerSrc` is unset or 404s, pdf.js reports \"Setting up fake worker\n * failed\" with a bare module URL, which says nothing about what to do. The\n * worker is not bundled with this library on purpose -- it has to be served by\n * the consuming application -- so the fix is always the same two steps.\n */\nexport function describePdfError(error: unknown): Error {\n    const message = error instanceof Error ? error.message : String(error)\n    if (!/fake worker|worker/i.test(message)) {\n        return error instanceof Error ? error : new Error(message)\n    }\n\n    const { workerSrc } = getConfig().pdf\n    const cause = workerSrc\n        ? `pdf.js could not load its worker from \"${workerSrc}\".`\n        : 'pdf.js has no worker configured.'\n\n    return new Error(\n        `${cause}\\n` +\n            'Copy it out of the package and point the config at it:\\n' +\n            '  cp node_modules/pdfjs-dist/build/pdf.worker.min.mjs public/\\n' +\n            \"  configure({ pdf: { workerSrc: '/pdf.worker.min.mjs' } })\\n\" +\n            `Original error: ${message}`,\n        { cause: error }\n    )\n}\n\n/** Document-level options assembled from the global config. */\nexport function documentOptions(data: Uint8Array): Record<string, unknown> {\n    const { cMapUrl, standardFontDataUrl, wasmUrl } = getConfig().pdf\n    // pdf.js takes ownership of the buffer it is given and detaches it, so hand\n    // over a copy: callers routinely reuse the row's `content` afterwards.\n    const owned = new Uint8Array(data.byteLength)\n    owned.set(data)\n\n    const options: Record<string, unknown> = { data: owned }\n    if (cMapUrl) {\n        options.cMapUrl = cMapUrl\n        options.cMapPacked = true\n    }\n    if (standardFontDataUrl) options.standardFontDataUrl = standardFontDataUrl\n    if (wasmUrl) options.wasmUrl = wasmUrl\n    return options\n}\n","/**\n * Port of `scaledp/pdf/PdfDataToImage.py`: a PDF into one `Image` row per page.\n *\n * Python renders with PyMuPDF at a DPI; pdf.js works in scale factors, so the\n * DPI converts through the PDF unit of 72 points per inch.\n */\n\nimport { ImageError } from '../core/errors.js'\nimport { createCanvas, encodeImage } from '../core/image.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { createImage, type ImageFormat, type ScaleDpImage } from '../schemas/image.js'\nimport { toBytes } from '../stages/data-to-image.js'\nimport { describePdfError, documentOptions, loadPdfjs } from './pdfjs.js'\n\n/** PDF user space is defined in points; 72 of them make an inch. */\nexport const POINTS_PER_INCH = 72\n\nexport interface PdfToImageParams extends BaseStageParams {\n    /** Render DPI. 300 matches ScaleDP's default and suits OCR. */\n    resolution: number\n    /** Maximum pages to render; 0 renders all of them. */\n    pageLimit: number\n    imageType: ImageFormat\n}\n\nexport const PDF_TO_IMAGE_DEFAULTS: PdfToImageParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'content',\n    outputCol: 'image',\n    resolution: 300,\n    pageLimit: 0,\n    imageType: 'png' as ImageFormat,\n})\n\nconst MIME: Record<ImageFormat, 'image/png' | 'image/webp' | 'image/jpeg'> = {\n    png: 'image/png',\n    webp: 'image/webp',\n    jpeg: 'image/jpeg',\n}\n\nexport class PdfToImage extends Stage<PdfToImageParams> {\n    readonly name = 'PdfToImage'\n\n    constructor(options: Partial<PdfToImageParams> = {}) {\n        super(\n            resolveParams(PDF_TO_IMAGE_DEFAULTS, options, {\n                resolution: (value) => {\n                    if (!Number.isFinite(value) || value <= 0) {\n                        throw new RangeError(`resolution must be positive, received ${value}`)\n                    }\n                },\n                pageLimit: (value) => {\n                    if (!Number.isInteger(value) || value < 0) {\n                        throw new RangeError(`pageLimit must be a non-negative integer, received ${value}`)\n                    }\n                },\n            })\n        )\n    }\n\n    /** One input PDF becomes N rows, each carrying its page index. */\n    protected override async expand(input: unknown, row: Row, ctx: StageContext): Promise<Row[]> {\n        const { outputCol, pageCol, pathCol, resolution, pageLimit, imageType } = this.params\n        const path = String(row[pathCol] ?? 'memory')\n\n        const pdfjs = await loadPdfjs()\n        const task = pdfjs.getDocument(documentOptions(toBytes(input)))\n\n        try {\n            // pdf.js defers worker setup, so a missing worker surfaces on first\n            // page access rather than from task.promise. Wrap the whole\n            // operation so the failure is described wherever it lands.\n            const document = await task.promise\n            const pageCount = pageLimit > 0 ? Math.min(pageLimit, document.numPages) : document.numPages\n            const rows: Row[] = []\n\n            for (let index = 0; index < pageCount; index++) {\n                ctx.signal?.throwIfAborted()\n                const image = await renderPage(document, index + 1, {\n                    resolution,\n                    imageType,\n                    path,\n                })\n                rows.push({ ...row, [pageCol]: index, [outputCol]: image })\n            }\n            return rows\n        } catch (error) {\n            throw describePdfError(error)\n        } finally {\n            // destroy() lives on the loading task, not the document proxy, and\n            // is what releases the pdf.js worker's copy of the file.\n            await task.destroy()\n        }\n    }\n\n    protected async apply(): Promise<never> {\n        throw new ImageError('unreachable: expand handles every row', this.name)\n    }\n\n    protected onError(message: string, row: Row): ScaleDpImage {\n        return createImage({ path: String(row[this.params.pathCol] ?? 'memory'), exception: message })\n    }\n}\n\n/** Rasterise a single 1-based page to encoded image bytes. */\nexport async function renderPage(\n    document: Awaited<ReturnType<typeof import('pdfjs-dist').getDocument>['promise']>,\n    pageNumber: number,\n    opts: { resolution: number; imageType: ImageFormat; path: string }\n): Promise<ScaleDpImage> {\n    const page = await document.getPage(pageNumber)\n    try {\n        const viewport = page.getViewport({ scale: opts.resolution / POINTS_PER_INCH })\n        const canvas = createCanvas(viewport.width, viewport.height)\n\n        // pdf.js >= 5 takes `canvas`, not `canvasContext`. Passing the context\n        // still works but is the documented legacy path.\n        await page.render({\n            canvas: canvas as unknown as HTMLCanvasElement,\n            viewport,\n        }).promise\n\n        return createImage({\n            path: opts.path,\n            resolution: opts.resolution,\n            data: await encodeImage(canvas, MIME[opts.imageType]),\n            imageType: opts.imageType,\n            width: canvas.width,\n            height: canvas.height,\n        })\n    } finally {\n        page.cleanup()\n    }\n}\n","/**\n * Split a pdf.js text run into word-level boxes.\n *\n * pdf.js emits runs at line granularity (\"Client: Raja Raman\"), but detection\n * boxes and NER offsets both want words. Each word is measured with Canvas 2D\n * `measureText` using a font reconstructed from the pdf.js font name, then the\n * measured widths are scaled so they sum to the run's actual width -- the\n * substitute font is never metrically identical to the embedded one, so the\n * measurements are only useful as *proportions*.\n *\n * Walking along `readDirX`/`readDirY` rather than assuming left-to-right is what\n * makes rotated, bottom-to-top and right-to-left runs come out correctly.\n */\n\nimport { context2d, createCanvas } from '../core/image.js'\nimport type { Box } from '../schemas/box.js'\nimport type { TextBox } from './extract-text.js'\n\n/** Widen each word slightly so glyph overhang is not clipped. */\nconst WORD_PADDING_RATIO = 0.1\n\nlet measureCtx: OffscreenCanvasRenderingContext2D | null = null\n\nfunction measurementContext(): OffscreenCanvasRenderingContext2D | null {\n    if (measureCtx) return measureCtx\n    try {\n        measureCtx = context2d(createCanvas(1, 1))\n        return measureCtx\n    } catch {\n        // No canvas (e.g. a non-browser test run): fall back to glyph heuristics.\n        return null\n    }\n}\n\n/**\n * Rebuild a CSS font string from a pdf.js font name.\n *\n * Names look like `AAAAAA+Helvetica-BoldOblique`: a six-letter subset prefix,\n * then the real family and style suffixes.\n */\nexport function cssFontFromPdfName(fontName: string, size: number): string {\n    const name = fontName.replace(/^[A-Z]{6}\\+/, '')\n    const lower = name.toLowerCase()\n    const weight = /bold|black|heavy|semibold/.test(lower) ? 'bold' : 'normal'\n    const style = /italic|oblique/.test(lower) ? 'italic' : 'normal'\n    const family = /serif|times|georgia|garamond|roman/.test(lower)\n        ? 'serif'\n        : /mono|courier|consol/.test(lower)\n          ? 'monospace'\n          : 'sans-serif'\n    return `${style} ${weight} ${Math.max(1, Math.round(size))}px ${family}`\n}\n\n/**\n * Relative advance width per character, used when no canvas is available.\n * Buckets rather than real metrics -- enough to keep proportions sane.\n */\nexport function relativeCharWidth(char: string): number {\n    if (\"iljI|.,:;'`!\".includes(char)) return 0.6\n    if ('ftr()[]{}-'.includes(char)) return 0.8\n    if ('MWmw@%'.includes(char)) return 1.6\n    if (char === ' ') return 0.6\n    if (char >= 'A' && char <= 'Z') return 1.3\n    return 1.0\n}\n\nfunction measureWord(word: string, font: string): number {\n    const ctx = measurementContext()\n    if (ctx) {\n        ctx.font = font\n        return ctx.measureText(word).width\n    }\n    let total = 0\n    for (const char of word) total += relativeCharWidth(char)\n    return total\n}\n\n/**\n * Split one run into word boxes.\n *\n * Returns the run itself when it holds a single word, so the common case costs\n * nothing.\n */\nexport function splitRunIntoWords(run: TextBox): Box[] {\n    const trimmed = run.text.trim()\n    if (trimmed.length === 0) return []\n\n    const tokens = trimmed.split(/(\\s+)/).filter((t) => t.length > 0)\n    const words = tokens.filter((t) => !/^\\s+$/.test(t))\n    if (words.length <= 1) {\n        return [{ ...run, text: trimmed }]\n    }\n\n    const font = cssFontFromPdfName(run.fontName, run.height)\n    const measured = tokens.map((token) => measureWord(token, font))\n    const totalMeasured = measured.reduce((sum, w) => sum + w, 0) || 1\n\n    // The substitute font's absolute metrics are meaningless; only the ratios\n    // matter, so normalise them onto the run's real extent.\n    const runLength = Math.hypot(run.width * run.readDirX, run.height * run.readDirY) || run.width\n    const scale = runLength / totalMeasured\n\n    // Walking starts at whichever corner the reading direction comes *from*, so\n    // a right-to-left or bottom-to-top run starts at the opposite edge.\n    let cursorX = run.readDirX >= 0 ? run.x : run.x + run.width\n    let cursorY = run.readDirY >= 0 ? run.y : run.y + run.height\n\n    const out: Box[] = []\n    for (const [i, token] of tokens.entries()) {\n        const advance = (measured[i] as number) * scale\n        if (!/^\\s+$/.test(token)) {\n            const pad = run.height * WORD_PADDING_RATIO\n            const spanX = Math.abs(run.readDirX) > Math.abs(run.readDirY) ? advance : run.width\n            const spanY = Math.abs(run.readDirY) > Math.abs(run.readDirX) ? advance : run.height\n\n            const left = run.readDirX >= 0 ? cursorX : cursorX - spanX\n            const top = run.readDirY >= 0 ? cursorY : cursorY - spanY\n\n            out.push({\n                text: token,\n                score: run.score,\n                x: Math.floor(left - pad),\n                y: Math.floor(top - pad),\n                width: Math.max(1, Math.ceil(spanX + pad * 2)),\n                height: Math.max(1, Math.ceil(spanY + pad * 2)),\n                angle: run.angle,\n            })\n        }\n        cursorX += run.readDirX * advance\n        cursorY += run.readDirY * advance\n    }\n    return out\n}\n\n/** Split every run on a page into word boxes. */\nexport function splitRunsIntoWords(runs: readonly TextBox[]): Box[] {\n    return runs.flatMap(splitRunIntoWords)\n}\n\n/** Reset the cached measurement canvas. Tests only. */\nexport function resetMeasurementContext(): void {\n    measureCtx = null\n}\n","/**\n * Port of `scaledp/pdf/PdfDataToText.py`: a PDF's embedded text layer into one\n * `Document` row per page, with word-level boxes.\n *\n * Coordinates are emitted in the same pixel space `PdfToImage` renders at, so\n * boxes from this stage and boxes from OCR are directly comparable. Python\n * leaves PdfDataToText in PDF points and scales only in PdfDataToDocument; a\n * single consistent space is more useful and avoids a class of silent mismatch.\n *\n * The output feeds the `bypassCol` optimisation: a page that already has a text\n * layer does not need OCR.\n */\n\nimport { ImageError } from '../core/errors.js'\nimport { BASE_STAGE_DEFAULTS, type BaseStageParams, resolveParams } from '../core/params.js'\nimport { type Row, Stage, type StageContext } from '../core/pipeline.js'\nimport { createDocument, type Document } from '../schemas/document.js'\nimport { toBytes } from '../stages/data-to-image.js'\nimport { extractTextBoxes } from './extract-text.js'\nimport { POINTS_PER_INCH } from './pdf-to-image.js'\nimport { describePdfError, documentOptions, loadPdfjs } from './pdfjs.js'\nimport { splitRunsIntoWords } from './split-words.js'\n\nexport interface PdfToDocumentParams extends BaseStageParams {\n    /** Pixel space the boxes are expressed in; match PdfToImage to align them. */\n    resolution: number\n    pageLimit: number\n    /** Split pdf.js line runs into word boxes. Off yields run-level boxes. */\n    splitWords: boolean\n}\n\nexport const PDF_TO_DOCUMENT_DEFAULTS: PdfToDocumentParams = Object.freeze({\n    ...BASE_STAGE_DEFAULTS,\n    inputCol: 'content',\n    outputCol: 'document',\n    keepInputData: true,\n    resolution: 300,\n    pageLimit: 0,\n    splitWords: true,\n})\n\nexport class PdfToDocument extends Stage<PdfToDocumentParams> {\n    readonly name = 'PdfToDocument'\n\n    constructor(options: Partial<PdfToDocumentParams> = {}) {\n        super(resolveParams(PDF_TO_DOCUMENT_DEFAULTS, options))\n    }\n\n    protected override async expand(input: unknown, row: Row, ctx: StageContext): Promise<Row[]> {\n        const { outputCol, pageCol, pathCol, resolution, pageLimit, splitWords } = this.params\n        const path = String(row[pathCol] ?? 'memory')\n\n        const pdfjs = await loadPdfjs()\n        const task = pdfjs.getDocument(documentOptions(toBytes(input)))\n\n        try {\n            // pdf.js defers worker setup, so a missing worker surfaces on first\n            // page access rather than from task.promise.\n            const pdf = await task.promise\n            const pageCount = pageLimit > 0 ? Math.min(pageLimit, pdf.numPages) : pdf.numPages\n            const rows: Row[] = []\n\n            for (let index = 0; index < pageCount; index++) {\n                ctx.signal?.throwIfAborted()\n                const page = await pdf.getPage(index + 1)\n                try {\n                    const viewport = page.getViewport({ scale: resolution / POINTS_PER_INCH })\n                    const runs = await extractTextBoxes(page, viewport)\n                    const bboxes = splitWords ? splitRunsIntoWords(runs) : runs\n\n                    rows.push({\n                        ...row,\n                        [pageCol]: index,\n                        [outputCol]: createDocument({\n                            path,\n                            type: 'pdf',\n                            text: runs.map((r) => r.text).join('\\n'),\n                            bboxes,\n                        }),\n                    })\n                } finally {\n                    page.cleanup()\n                }\n            }\n            return rows\n        } catch (error) {\n            throw describePdfError(error)\n        } finally {\n            await task.destroy()\n        }\n    }\n\n    protected async apply(): Promise<never> {\n        throw new ImageError('unreachable: expand handles every row', this.name)\n    }\n\n    protected onError(message: string, row: Row): Document {\n        return createDocument({\n            path: String(row[this.params.pathCol] ?? 'memory'),\n            type: 'pdf',\n            exception: message,\n        })\n    }\n}\n\n/** True when a page's text layer is substantive enough to skip OCR. */\nexport function hasUsableTextLayer(document: Document, minimumBoxes = 1): boolean {\n    return document.exception === '' && document.bboxes.length >= minimumBoxes\n}\n"],"mappings":";;;;;;;AA+BA,MAAM,eAAe;;AAGrB,MAAa,mBAAmB;AAchC,SAAgB,WAAW,MAAqC;CAC5D,OACI,OAAO,SAAS,YAChB,SAAS,QACT,SAAS,QACT,eAAe,QACf,MAAM,QAAS,KAAsB,SAAS;AAEtD;;AAGA,SAAgB,cAAc,MAAoB,UAAwC;CACtF,IAAI,KAAK,IAAI,WAAW,GAAG,OAAO;CAElC,MAAM,CAAC,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,KAAK,KAAK;CAIxD,MAAM,QAAQ,KAAK,MAAM,GAAG,CAAC,KAAK;CAClC,MAAM,QAAQ,KAAK,MAAM,GAAG,CAAC,KAAK;CAClC,MAAM,OAAO,IAAI;CACjB,MAAM,OAAO,IAAI;CACjB,MAAM,MAAM,IAAI;CAChB,MAAM,MAAM,IAAI;CAGhB,MAAM,SAAS,KAAK,SAAS;CAC7B,MAAM,SAAS,IAAI,MAAM;CACzB,MAAM,SAAS,IAAI,MAAM;CAEzB,MAAM,UAAU;EACZ,SAAS,uBAAuB,QAAQ,MAAM;EAC9C,SAAS,uBAAuB,SAAS,OAAO,KAAK,OAAO,SAAS,OAAO,KAAK,KAAK;EACtF,SAAS,uBAAuB,SAAS,MAAM,KAAK,QAAQ,SAAS,MAAM,KAAK,MAAM;EACtF,SAAS,uBACL,SAAS,OAAO,KAAK,QAAQ,MAAM,KAAK,QACxC,SAAS,OAAO,KAAK,QAAQ,MAAM,KAAK,MAC5C;CACJ;CAEA,MAAM,KAAK,QAAQ,KAAK,MAAM,EAAE,EAAY;CAC5C,MAAM,KAAK,QAAQ,KAAK,MAAM,EAAE,EAAY;CAC5C,MAAM,IAAI,KAAK,IAAI,GAAG,EAAE;CACxB,MAAM,IAAI,KAAK,IAAI,GAAG,EAAE;CAIxB,MAAM,CAAC,eAAe,GAAG,eAAe,KAAK,QAAQ,MAAM,CAAC;CAC5D,MAAM,CAAC,aAAa,GAAG,aAAa,KAAK,QAAQ,MAAM,CAAC;CACxD,MAAM,UAAU,KAAK,MAAM,aAAa,cAAc,aAAa,YAAY,KAAK;CAEpF,OAAO;EACH,MAAM,KAAK;EACX,OAAO;EACP,GAAG,KAAK,MAAM,CAAC;EACf,GAAG,KAAK,MAAM,CAAC;EACf,OAAO,KAAK,IAAI,GAAG,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC;EACjD,QAAQ,KAAK,IAAI,GAAG,KAAK,KAAK,KAAK,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC;EAClD,OAAO;EACP,WAAW,aAAa,gBAAgB;EACxC,WAAW,aAAa,gBAAgB;EACxC,UAAU,KAAK,YAAY;CAC/B;AACJ;;AAGA,eAAsB,iBAClB,MACA,UACkB;CAClB,MAAM,UAAU,MAAM,KAAK,eAAe;CAC1C,MAAM,QAAmB,CAAC;CAC1B,KAAK,MAAM,QAAQ,QAAQ,OAAO;EAC9B,IAAI,CAAC,WAAW,IAAI,GAAG;EACvB,MAAM,MAAM,cAAc,MAAM,QAAQ;EACxC,IAAI,KAAK,MAAM,KAAK,GAAG;CAC3B;CACA,OAAO;AACX;;;;;;;;;;;ACjHA,IAAI,gBAA6C;AAEjD,eAAsB,YAAkC;CACpD,IAAI,eAAe,OAAO;CAE1B,iBAAiB,YAAY;EACzB,IAAI;EACJ,IAAI;GACA,QAAQ,MAAM,OAAO;EACzB,SAAS,OAAO;GACZ,MAAM,IAAI,MAAM,wEAAwE,EAAE,MAAM,CAAC;EACrG;EAEA,MAAM,EAAE,cAAc,UAAU,CAAC,CAAC;EAClC,IAAI,WAAW,MAAM,oBAAoB,YAAY;EACrD,OAAO;CACX,EAAA,CAAG;CAEH,OAAO;AACX;;AAGA,SAAgB,aAAmB;CAC/B,gBAAgB;AACpB;;;;;;;;;AAUA,SAAgB,iBAAiB,OAAuB;CACpD,MAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;CACrE,IAAI,CAAC,sBAAsB,KAAK,OAAO,GACnC,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO;CAG7D,MAAM,EAAE,cAAc,UAAU,CAAC,CAAC;CAClC,MAAM,QAAQ,YACR,0CAA0C,UAAU,MACpD;CAEN,OAAO,IAAI,MACP,GAAG,MAAM;;;kBAIc,WACvB,EAAE,OAAO,MAAM,CACnB;AACJ;;AAGA,SAAgB,gBAAgB,MAA2C;CACvE,MAAM,EAAE,SAAS,qBAAqB,YAAY,UAAU,CAAC,CAAC;CAG9D,MAAM,QAAQ,IAAI,WAAW,KAAK,UAAU;CAC5C,MAAM,IAAI,IAAI;CAEd,MAAM,UAAmC,EAAE,MAAM,MAAM;CACvD,IAAI,SAAS;EACT,QAAQ,UAAU;EAClB,QAAQ,aAAa;CACzB;CACA,IAAI,qBAAqB,QAAQ,sBAAsB;CACvD,IAAI,SAAS,QAAQ,UAAU;CAC/B,OAAO;AACX;;;;;;;;;;ACpEA,MAAa,kBAAkB;AAU/B,MAAa,wBAA0C,OAAO,OAAO;CACjE,GAAG;CACH,UAAU;CACV,WAAW;CACX,YAAY;CACZ,WAAW;CACX,WAAW;AACf,CAAC;AAED,MAAM,OAAuE;CACzE,KAAK;CACL,MAAM;CACN,MAAM;AACV;AAEA,IAAa,aAAb,cAAgC,MAAwB;CACpD,OAAgB;CAEhB,YAAY,UAAqC,CAAC,GAAG;EACjD,MACI,cAAc,uBAAuB,SAAS;GAC1C,aAAa,UAAU;IACnB,IAAI,CAAC,OAAO,SAAS,KAAK,KAAK,SAAS,GACpC,MAAM,IAAI,WAAW,yCAAyC,OAAO;GAE7E;GACA,YAAY,UAAU;IAClB,IAAI,CAAC,OAAO,UAAU,KAAK,KAAK,QAAQ,GACpC,MAAM,IAAI,WAAW,sDAAsD,OAAO;GAE1F;EACJ,CAAC,CACL;CACJ;;CAGA,MAAyB,OAAO,OAAgB,KAAU,KAAmC;EACzF,MAAM,EAAE,WAAW,SAAS,SAAS,YAAY,WAAW,cAAc,KAAK;EAC/E,MAAM,OAAO,OAAO,IAAI,YAAY,QAAQ;EAG5C,MAAM,QAAO,MADO,UAAU,EAAA,CACX,YAAY,gBAAgB,QAAQ,KAAK,CAAC,CAAC;EAE9D,IAAI;GAIA,MAAM,WAAW,MAAM,KAAK;GAC5B,MAAM,YAAY,YAAY,IAAI,KAAK,IAAI,WAAW,SAAS,QAAQ,IAAI,SAAS;GACpF,MAAM,OAAc,CAAC;GAErB,KAAK,IAAI,QAAQ,GAAG,QAAQ,WAAW,SAAS;IAC5C,IAAI,QAAQ,eAAe;IAC3B,MAAM,QAAQ,MAAM,WAAW,UAAU,QAAQ,GAAG;KAChD;KACA;KACA;IACJ,CAAC;IACD,KAAK,KAAK;KAAE,GAAG;MAAM,UAAU;MAAQ,YAAY;IAAM,CAAC;GAC9D;GACA,OAAO;EACX,SAAS,OAAO;GACZ,MAAM,iBAAiB,KAAK;EAChC,UAAU;GAGN,MAAM,KAAK,QAAQ;EACvB;CACJ;CAEA,MAAgB,QAAwB;EACpC,MAAM,IAAI,WAAW,yCAAyC,KAAK,IAAI;CAC3E;CAEA,QAAkB,SAAiB,KAAwB;EACvD,OAAO,YAAY;GAAE,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GAAG,WAAW;EAAQ,CAAC;CACjG;AACJ;;AAGA,eAAsB,WAClB,UACA,YACA,MACqB;CACrB,MAAM,OAAO,MAAM,SAAS,QAAQ,UAAU;CAC9C,IAAI;EACA,MAAM,WAAW,KAAK,YAAY,EAAE,OAAO,KAAK,aAAA,GAA6B,CAAC;EAC9E,MAAM,SAAS,aAAa,SAAS,OAAO,SAAS,MAAM;EAI3D,MAAM,KAAK,OAAO;GACN;GACR;EACJ,CAAC,CAAC,CAAC;EAEH,OAAO,YAAY;GACf,MAAM,KAAK;GACX,YAAY,KAAK;GACjB,MAAM,MAAM,YAAY,QAAQ,KAAK,KAAK,UAAU;GACpD,WAAW,KAAK;GAChB,OAAO,OAAO;GACd,QAAQ,OAAO;EACnB,CAAC;CACL,UAAU;EACN,KAAK,QAAQ;CACjB;AACJ;;;;;;;;;;;;;;;;;ACnHA,MAAM,qBAAqB;AAE3B,IAAI,aAAuD;AAE3D,SAAS,qBAA+D;CACpE,IAAI,YAAY,OAAO;CACvB,IAAI;EACA,aAAa,UAAU,aAAa,GAAG,CAAC,CAAC;EACzC,OAAO;CACX,QAAQ;EAEJ,OAAO;CACX;AACJ;;;;;;;AAQA,SAAgB,mBAAmB,UAAkB,MAAsB;CAEvE,MAAM,QADO,SAAS,QAAQ,eAAe,EAC5B,CAAC,CAAC,YAAY;CAC/B,MAAM,SAAS,4BAA4B,KAAK,KAAK,IAAI,SAAS;CAClE,MAAM,QAAQ,iBAAiB,KAAK,KAAK,IAAI,WAAW;CACxD,MAAM,SAAS,qCAAqC,KAAK,KAAK,IACxD,UACA,sBAAsB,KAAK,KAAK,IAC9B,cACA;CACR,OAAO,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,IAAI,GAAG,KAAK,MAAM,IAAI,CAAC,EAAE,KAAK;AACpE;;;;;AAMA,SAAgB,kBAAkB,MAAsB;CACpD,IAAI,eAAe,SAAS,IAAI,GAAG,OAAO;CAC1C,IAAI,aAAa,SAAS,IAAI,GAAG,OAAO;CACxC,IAAI,SAAS,SAAS,IAAI,GAAG,OAAO;CACpC,IAAI,SAAS,KAAK,OAAO;CACzB,IAAI,QAAQ,OAAO,QAAQ,KAAK,OAAO;CACvC,OAAO;AACX;AAEA,SAAS,YAAY,MAAc,MAAsB;CACrD,MAAM,MAAM,mBAAmB;CAC/B,IAAI,KAAK;EACL,IAAI,OAAO;EACX,OAAO,IAAI,YAAY,IAAI,CAAC,CAAC;CACjC;CACA,IAAI,QAAQ;CACZ,KAAK,MAAM,QAAQ,MAAM,SAAS,kBAAkB,IAAI;CACxD,OAAO;AACX;;;;;;;AAQA,SAAgB,kBAAkB,KAAqB;CACnD,MAAM,UAAU,IAAI,KAAK,KAAK;CAC9B,IAAI,QAAQ,WAAW,GAAG,OAAO,CAAC;CAElC,MAAM,SAAS,QAAQ,MAAM,OAAO,CAAC,CAAC,QAAQ,MAAM,EAAE,SAAS,CAAC;CAEhE,IADc,OAAO,QAAQ,MAAM,CAAC,QAAQ,KAAK,CAAC,CAC1C,CAAC,CAAC,UAAU,GAChB,OAAO,CAAC;EAAE,GAAG;EAAK,MAAM;CAAQ,CAAC;CAGrC,MAAM,OAAO,mBAAmB,IAAI,UAAU,IAAI,MAAM;CACxD,MAAM,WAAW,OAAO,KAAK,UAAU,YAAY,OAAO,IAAI,CAAC;CAC/D,MAAM,gBAAgB,SAAS,QAAQ,KAAK,MAAM,MAAM,GAAG,CAAC,KAAK;CAKjE,MAAM,SADY,KAAK,MAAM,IAAI,QAAQ,IAAI,UAAU,IAAI,SAAS,IAAI,QAAQ,KAAK,IAAI,SAC/D;CAI1B,IAAI,UAAU,IAAI,YAAY,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI;CACtD,IAAI,UAAU,IAAI,YAAY,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI;CAEtD,MAAM,MAAa,CAAC;CACpB,KAAK,MAAM,CAAC,GAAG,UAAU,OAAO,QAAQ,GAAG;EACvC,MAAM,UAAW,SAAS,KAAgB;EAC1C,IAAI,CAAC,QAAQ,KAAK,KAAK,GAAG;GACtB,MAAM,MAAM,IAAI,SAAS;GACzB,MAAM,QAAQ,KAAK,IAAI,IAAI,QAAQ,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,UAAU,IAAI;GAC9E,MAAM,QAAQ,KAAK,IAAI,IAAI,QAAQ,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,UAAU,IAAI;GAE9E,MAAM,OAAO,IAAI,YAAY,IAAI,UAAU,UAAU;GACrD,MAAM,MAAM,IAAI,YAAY,IAAI,UAAU,UAAU;GAEpD,IAAI,KAAK;IACL,MAAM;IACN,OAAO,IAAI;IACX,GAAG,KAAK,MAAM,OAAO,GAAG;IACxB,GAAG,KAAK,MAAM,MAAM,GAAG;IACvB,OAAO,KAAK,IAAI,GAAG,KAAK,KAAK,QAAQ,MAAM,CAAC,CAAC;IAC7C,QAAQ,KAAK,IAAI,GAAG,KAAK,KAAK,QAAQ,MAAM,CAAC,CAAC;IAC9C,OAAO,IAAI;GACf,CAAC;EACL;EACA,WAAW,IAAI,WAAW;EAC1B,WAAW,IAAI,WAAW;CAC9B;CACA,OAAO;AACX;;AAGA,SAAgB,mBAAmB,MAAiC;CAChE,OAAO,KAAK,QAAQ,iBAAiB;AACzC;;AAGA,SAAgB,0BAAgC;CAC5C,aAAa;AACjB;;;;;;;;;;;;;;;AC/GA,MAAa,2BAAgD,OAAO,OAAO;CACvE,GAAG;CACH,UAAU;CACV,WAAW;CACX,eAAe;CACf,YAAY;CACZ,WAAW;CACX,YAAY;AAChB,CAAC;AAED,IAAa,gBAAb,cAAmC,MAA2B;CAC1D,OAAgB;CAEhB,YAAY,UAAwC,CAAC,GAAG;EACpD,MAAM,cAAc,0BAA0B,OAAO,CAAC;CAC1D;CAEA,MAAyB,OAAO,OAAgB,KAAU,KAAmC;EACzF,MAAM,EAAE,WAAW,SAAS,SAAS,YAAY,WAAW,eAAe,KAAK;EAChF,MAAM,OAAO,OAAO,IAAI,YAAY,QAAQ;EAG5C,MAAM,QAAO,MADO,UAAU,EAAA,CACX,YAAY,gBAAgB,QAAQ,KAAK,CAAC,CAAC;EAE9D,IAAI;GAGA,MAAM,MAAM,MAAM,KAAK;GACvB,MAAM,YAAY,YAAY,IAAI,KAAK,IAAI,WAAW,IAAI,QAAQ,IAAI,IAAI;GAC1E,MAAM,OAAc,CAAC;GAErB,KAAK,IAAI,QAAQ,GAAG,QAAQ,WAAW,SAAS;IAC5C,IAAI,QAAQ,eAAe;IAC3B,MAAM,OAAO,MAAM,IAAI,QAAQ,QAAQ,CAAC;IACxC,IAAI;KAEA,MAAM,OAAO,MAAM,iBAAiB,MADnB,KAAK,YAAY,EAAE,OAAO,aAAA,GAA6B,CAC9B,CAAQ;KAClD,MAAM,SAAS,aAAa,mBAAmB,IAAI,IAAI;KAEvD,KAAK,KAAK;MACN,GAAG;OACF,UAAU;OACV,YAAY,eAAe;OACxB;OACA,MAAM;OACN,MAAM,KAAK,KAAK,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,IAAI;OACvC;MACJ,CAAC;KACL,CAAC;IACL,UAAU;KACN,KAAK,QAAQ;IACjB;GACJ;GACA,OAAO;EACX,SAAS,OAAO;GACZ,MAAM,iBAAiB,KAAK;EAChC,UAAU;GACN,MAAM,KAAK,QAAQ;EACvB;CACJ;CAEA,MAAgB,QAAwB;EACpC,MAAM,IAAI,WAAW,yCAAyC,KAAK,IAAI;CAC3E;CAEA,QAAkB,SAAiB,KAAoB;EACnD,OAAO,eAAe;GAClB,MAAM,OAAO,IAAI,KAAK,OAAO,YAAY,QAAQ;GACjD,MAAM;GACN,WAAW;EACf,CAAC;CACL;AACJ;;AAGA,SAAgB,mBAAmB,UAAoB,eAAe,GAAY;CAC9E,OAAO,SAAS,cAAc,MAAM,SAAS,OAAO,UAAU;AAClE"}