@stabrise/scaledp 0.1.0

package/dist/registry/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":[],"sources":["../../src/registry/catalog.ts","../../src/registry/codegen.ts","../../src/registry/index.ts"],"sourcesContent":["/**\n * The stage catalogue: one `StageSpec` per concrete stage.\n *\n * Transcribed from `docs/stages.md`, which is the same table in prose. Every\n * `defaults` entry is the stage's own frozen constant rather than a copy, and\n * every enum's options come from the model registries that already exist\n * (`PADDLE_OCR_PRESETS`, `NER_MODELS`, `DETECTOR_MODELS`) rather than a second\n * hand-maintained list.\n *\n * Importing this module costs no ML runtime: every engine in the library is\n * reached through a dynamic `import()` inside `ocr/ort.ts`, `pdf/pdfjs.ts`,\n * `ocr/paddle-service.ts` and `ocr/tesseract.ts`, so pulling in the twelve stage\n * classes pulls in no `onnxruntime-web`, `pdfjs-dist`, `ppu-paddle-ocr` or\n * `tesseract-wasm`. It does pull in every stage's code, which is why this lives\n * behind its own subpath instead of the root barrel.\n */\n\nimport { FaceDetector, SignatureDetector, YOLO_DETECTOR_DEFAULTS, YoloOnnxDetector } from '../detect/index.js'\nimport { GLINER_NER_DEFAULTS, GlinerNer, modelSizeBytes, NER_MODELS } from '../ner/index.js'\nimport {\n    DBNET_DETECTOR_DEFAULTS,\n    DbnetOnnxDetector,\n    DEFAULT_ORIENTATION_MODEL,\n    DETECTOR_MODELS,\n    LINE_ORIENTATION_DEFAULTS,\n    LineOrientationDetector,\n    PADDLE_DETECTOR_DEFAULTS,\n    PADDLE_OCR_PRESETS,\n    PADDLE_RECOGNIZER_DEFAULTS,\n    PaddleTextDetector,\n    PaddleTextRecognizer,\n    TESSERACT_OCR_DEFAULTS,\n    TESSERACT_RECOGNIZER_DEFAULTS,\n    TesseractOcr,\n    TesseractRecognizer,\n} from '../ocr/index.js'\nimport { PDF_TO_DOCUMENT_DEFAULTS, PDF_TO_IMAGE_DEFAULTS, PdfToDocument, PdfToImage } from '../pdf/index.js'\nimport { DATA_TO_IMAGE_DEFAULTS, DataToImage } from '../stages/data-to-image.js'\nimport { IMAGE_CROP_BOXES_DEFAULTS, ImageCropBoxes } from '../stages/image-crop-boxes.js'\nimport { IMAGE_DRAW_BOXES_DEFAULTS, ImageDrawBoxes } from '../stages/image-draw-boxes.js'\nimport type { ColumnKind, StageParamOption, StageParamSpec, StageSpec } from './types.js'\n\n/**\n * Widen a stage's typed defaults into an indexable record.\n *\n * `*_DEFAULTS` are typed as their stage's params interface, and an interface has\n * no index signature -- so it is not assignable to `Record<string, unknown>`\n * even though every key is a string. The values are frozen and read-only either\n * way; only the type changes.\n */\nconst asRecord = <T extends object>(defaults: Readonly<T>): Readonly<Record<string, unknown>> =>\n    defaults as Readonly<Record<string, unknown>>\n\n/* ── Enum option lists, derived from the existing registries ─────────────── */\n\nconst IMAGE_TYPES: readonly StageParamOption[] = Object.freeze([\n    { value: 'png', label: 'PNG', title: 'Lossless; the safe default for OCR' },\n    { value: 'webp', label: 'WebP', title: 'Smaller, lossy' },\n    { value: 'jpeg', label: 'JPEG', title: 'Smallest, lossy; artefacts can cost accuracy' },\n])\n\nconst STRATEGIES: readonly StageParamOption[] = Object.freeze([\n    { value: 'per-box', label: 'Per box', title: 'Word-level boxes, as detected' },\n    { value: 'per-line', label: 'Per line', title: 'Merge boxes that share a line' },\n    { value: 'cross-line', label: 'Cross line', title: 'Merge across line breaks too' },\n])\n\n/**\n * Fields `ImageDrawBoxes` can render above a box.\n *\n * The label is built by reading these off the box or entity, so the valid names\n * are the schema's own field names -- `Box` for a detector or OCR result,\n * `Entity` for NER output. Numbers are formatted to two decimals.\n */\nconst LABEL_FIELDS: readonly StageParamOption[] = Object.freeze([\n    { value: 'text', label: 'text', title: 'Box: the recognised text' },\n    { value: 'score', label: 'score', title: 'Box or entity: confidence, 0-1' },\n    { value: 'angle', label: 'angle', title: 'Box: degrees about its centre' },\n    { value: 'entity_group', label: 'entity_group', title: 'Entity: the label it matched' },\n    { value: 'word', label: 'word', title: 'Entity: the matched text' },\n    { value: 'x', label: 'x', title: 'Box: top-left x' },\n    { value: 'y', label: 'y', title: 'Box: top-left y' },\n    { value: 'width', label: 'width', title: 'Box: the longer side' },\n    { value: 'height', label: 'height', title: 'Box: the shorter side' },\n])\n\nconst OCR_PRESETS: readonly StageParamOption[] = Object.freeze(\n    PADDLE_OCR_PRESETS.map((preset) => ({\n        value: preset.value,\n        label: preset.label,\n        title: preset.scripts.join(', '),\n    }))\n)\n\nconst NER_MODEL_OPTIONS: readonly StageParamOption[] = Object.freeze(\n    NER_MODELS.map((model) => ({\n        value: model.id,\n        label: `${model.name} · ${model.languages.join('/')}${model.private ? ' · private' : ''}`,\n        title: `${model.arch}, ${Math.round(modelSizeBytes(model) / 1e6)} MB, ${model.repo}`,\n        // Private repos need configure({ auth }); offering one without it only\n        // produces a 401 mid-pipeline.\n        disabled: model.private === true,\n    }))\n)\n\nconst DBNET_MODEL_OPTIONS: readonly StageParamOption[] = Object.freeze(\n    DETECTOR_MODELS.filter((model) => model.kind === 'dbnet-onnx' && model.repo).map((model) => ({\n        value: model.repo as string,\n        label: model.name,\n        title: model.notes,\n    }))\n)\n\n/* ── Param builders ──────────────────────────────────────────────────────── */\n\nconst column = (key: string, label: string, extra: Partial<StageParamSpec> = {}): StageParamSpec => ({\n    key,\n    kind: 'column',\n    label,\n    ...extra,\n})\n\n/**\n * The six params every stage inherits from `BaseStageParams`.\n *\n * `inputCol` and `outputCol` are the wiring a builder actually cares about;\n * the rest is plumbing, so it is marked advanced. Stages that read several\n * columns pass `multiInput` -- there `inputCol` is inherited but unused, and\n * showing it as the input would be a lie.\n */\nfunction baseParams(\n    options: { input?: { help?: string; accepts?: ColumnKind[] } | 'unused'; output?: { help?: string } } = {}\n): StageParamSpec[] {\n    const input =\n        options.input === 'unused'\n            ? column('inputCol', 'Input column', {\n                  advanced: true,\n                  help: 'Inherited but unused: this stage reads the columns listed in inputCols.',\n              })\n            : column('inputCol', 'Input column', {\n                  help: options.input?.help,\n                  accepts: options.input?.accepts,\n              })\n\n    return [\n        input,\n        column('outputCol', 'Output column', { help: options.output?.help }),\n        column('pathCol', 'Path column', {\n            advanced: true,\n            help: 'Row field holding the source path, copied onto the output.',\n        }),\n        column('pageCol', 'Page column', {\n            advanced: true,\n            help: 'Row field holding the page index for multi-page inputs.',\n        }),\n        {\n            key: 'keepInputData',\n            kind: 'boolean',\n            label: 'Keep input column',\n            advanced: true,\n            help: 'Keep the input column in the output rows instead of dropping it.',\n        },\n        {\n            key: 'propagateError',\n            kind: 'boolean',\n            label: 'Throw on error',\n            advanced: true,\n            help: 'Throw instead of recording the failure in the output’s exception field.',\n        },\n    ]\n}\n\nconst imageType: StageParamSpec = {\n    key: 'imageType',\n    kind: 'enum',\n    label: 'Image format',\n    options: IMAGE_TYPES,\n}\n\nconst scoreThreshold = (help: string): StageParamSpec => ({\n    key: 'scoreThreshold',\n    kind: 'number',\n    label: 'Score threshold',\n    min: 0,\n    max: 1,\n    step: 0.05,\n    help,\n})\n\nconst lang: StageParamSpec = {\n    key: 'lang',\n    kind: 'stringList',\n    label: 'Languages',\n    required: true,\n    help: 'Tesseract language codes, e.g. eng or eng, deu. Each needs its traineddata file.',\n}\n\nconst keepFormatting: StageParamSpec = {\n    key: 'keepFormatting',\n    kind: 'boolean',\n    label: 'Keep formatting',\n    help: 'Rebuild the original layout with spaces and blank lines.',\n}\n\nconst lineTolerance: StageParamSpec = {\n    key: 'lineTolerance',\n    kind: 'number',\n    label: 'Line tolerance',\n    min: 0,\n    step: 1,\n    help: 'Line-grouping tolerance in pixels; 0 derives it from character height.',\n}\n\n/**\n * `SignatureDetector` and `FaceDetector` are `YoloOnnxDetector` with different\n * pre-set params (`src/detect/index.ts`), so they take the same parameters.\n *\n * One caveat those two carry: they spread their options straight into `super`\n * rather than going through `resolveParams`, so their validators never run --\n * an empty `model` is accepted there and fails later, at init.\n */\nconst yoloParams: readonly StageParamSpec[] = Object.freeze([\n    ...baseParams({ input: { accepts: ['image'] } }),\n    {\n        key: 'model',\n        kind: 'string',\n        label: 'Model',\n        required: true,\n        help: 'Hugging Face repo id, or a URL when self-hosting. No default.',\n    },\n    {\n        key: 'labels',\n        kind: 'stringList',\n        label: 'Labels',\n        help: 'Class index → label. Empty falls back to class_<n>.',\n    },\n    scoreThreshold('Drop detections below this confidence.'),\n    {\n        key: 'iouThreshold',\n        kind: 'number',\n        label: 'IoU threshold',\n        min: 0,\n        max: 1,\n        step: 0.05,\n        help: 'Overlap above which two same-class boxes count as duplicates.',\n    },\n    {\n        key: 'padding',\n        kind: 'number',\n        label: 'Padding',\n        min: 0,\n        max: 1,\n        step: 0.01,\n        help: 'Grow each box by this fraction of its size, to avoid clipping edges.',\n    },\n    {\n        key: 'outputType',\n        kind: 'string',\n        label: 'Output type',\n        advanced: true,\n        help: 'Recorded as DetectorOutput.type.',\n    },\n])\n\n/* ── The catalogue ───────────────────────────────────────────────────────── */\n\nexport const STAGE_SPECS: readonly StageSpec[] = Object.freeze([\n    {\n        type: 'DataToImage',\n        label: 'Image bytes',\n        group: 'Read',\n        subpath: '@stabrise/scaledp',\n        summary: 'Wrap raw image bytes as an Image without re-encoding them.',\n        consumes: ['bytes'],\n        produces: 'image',\n        defaults: asRecord(DATA_TO_IMAGE_DEFAULTS),\n        params: [\n            ...baseParams({ input: { help: 'Raw file bytes, normally the content column.' } }),\n            { ...imageType, help: 'Encoding recorded on the output; the bytes pass through unchanged.' },\n            {\n                key: 'resolution',\n                kind: 'number',\n                label: 'Resolution',\n                min: 0,\n                step: 1,\n                help: 'DPI to record when the row carries no resolution field. 0 leaves it unknown.',\n            },\n        ],\n    },\n    {\n        type: 'PdfToImage',\n        label: 'PDF pages',\n        group: 'Read',\n        subpath: '@stabrise/scaledp/pdf',\n        summary: 'Rasterise every page of a PDF, one output row per page.',\n        consumes: ['bytes'],\n        produces: 'image',\n        peer: 'pdfjs-dist',\n        expands: true,\n        defaults: asRecord(PDF_TO_IMAGE_DEFAULTS),\n        params: [\n            ...baseParams({ input: { help: 'Raw PDF bytes, normally the content column.' } }),\n            {\n                key: 'resolution',\n                kind: 'number',\n                label: 'Resolution (DPI)',\n                min: 36,\n                max: 600,\n                step: 1,\n                help: 'Render DPI. 300 matches ScaleDP; 200 is a good speed/accuracy trade.',\n            },\n            {\n                key: 'pageLimit',\n                kind: 'number',\n                label: 'Page limit',\n                min: 0,\n                step: 1,\n                help: '0 renders every page.',\n            },\n            imageType,\n        ],\n    },\n    {\n        type: 'PdfToDocument',\n        label: 'PDF text layer',\n        group: 'Read',\n        subpath: '@stabrise/scaledp/pdf',\n        summary: 'Read a PDF’s embedded text layer, skipping OCR entirely.',\n        consumes: ['bytes'],\n        produces: 'document',\n        peer: 'pdfjs-dist',\n        expands: true,\n        defaults: asRecord(PDF_TO_DOCUMENT_DEFAULTS),\n        params: [\n            ...baseParams({ input: { help: 'Raw PDF bytes, normally the content column.' } }),\n            {\n                key: 'resolution',\n                kind: 'number',\n                label: 'Resolution (DPI)',\n                min: 36,\n                max: 600,\n                step: 1,\n                help: 'Pixel space the boxes are expressed in. Match PdfToImage to align them.',\n            },\n            {\n                key: 'pageLimit',\n                kind: 'number',\n                label: 'Page limit',\n                min: 0,\n                step: 1,\n                help: '0 reads every page.',\n            },\n            {\n                key: 'splitWords',\n                kind: 'boolean',\n                label: 'Split words',\n                help: 'Split pdf.js line runs into word boxes. Off yields run-level boxes.',\n            },\n        ],\n    },\n    {\n        type: 'PaddleTextDetector',\n        label: 'PaddleOCR detector',\n        group: 'Detect',\n        subpath: '@stabrise/scaledp/ocr',\n        summary: 'Find text regions with the detection half of a PaddleOCR preset. Regions are line-level.',\n        consumes: ['image'],\n        produces: 'boxes',\n        peer: 'ppu-paddle-ocr',\n        cache: { kind: 'paddle-preset', param: 'preset' },\n        defaults: asRecord(PADDLE_DETECTOR_DEFAULTS),\n        params: [\n            ...baseParams({ input: { accepts: ['image'] } }),\n            {\n                key: 'preset',\n                kind: 'enum',\n                label: 'Preset',\n                options: OCR_PRESETS,\n                help: 'Language/script pairing. Detection and recognition share the download.',\n            },\n            scoreThreshold('Drop regions below this confidence. 0 keeps everything.'),\n        ],\n    },\n    {\n        type: 'DbnetOnnxDetector',\n        label: 'DBNet detector',\n        group: 'Detect',\n        subpath: '@stabrise/scaledp/ocr',\n        summary:\n            'The DBNet ONNX detector ScaleDP uses server-side. Finds rotated regions, one box per text line.',\n        consumes: ['image'],\n        produces: 'boxes',\n        peer: 'onnxruntime-web',\n        cache: { kind: 'hf-repo', param: 'model', approxBytes: 4_800_000 },\n        defaults: asRecord(DBNET_DETECTOR_DEFAULTS),\n        params: [\n            ...baseParams({ input: { accepts: ['image'] } }),\n            {\n                key: 'model',\n                kind: 'enum',\n                label: 'Model',\n                options: DBNET_MODEL_OPTIONS,\n                allowCustom: true,\n                help: 'Hugging Face repo id, or a URL when self-hosting.',\n            },\n            scoreThreshold('Mean in-box probability a candidate must reach.'),\n            {\n                key: 'binaryThreshold',\n                kind: 'number',\n                label: 'Binary threshold',\n                min: 0,\n                max: 1,\n                step: 0.05,\n                help: 'Probability above which a pixel counts as text.',\n            },\n            {\n                key: 'unclipRatio',\n                kind: 'number',\n                label: 'Unclip ratio',\n                min: 0.5,\n                max: 5,\n                step: 0.1,\n                help: 'How far to grow each box; DB shrinks text regions during training.',\n            },\n            {\n                key: 'mergeBoxes',\n                kind: 'boolean',\n                label: 'Merge boxes',\n                help: 'Merge overlapping boxes that share a line. Usually changes nothing here — this model already returns one region per line, so no two boxes overlap.',\n            },\n        ],\n    },\n    {\n        type: 'YoloOnnxDetector',\n        label: 'YOLO detector',\n        group: 'Detect',\n        subpath: '@stabrise/scaledp/detect',\n        summary: 'Detect arbitrary objects with a YOLO ONNX model.',\n        consumes: ['image'],\n        produces: 'boxes',\n        peer: 'onnxruntime-web',\n        cache: { kind: 'hf-repo', param: 'model' },\n        defaults: asRecord(YOLO_DETECTOR_DEFAULTS),\n        params: yoloParams,\n    },\n    {\n        type: 'SignatureDetector',\n        label: 'Signature detector',\n        group: 'Detect',\n        subpath: '@stabrise/scaledp/detect',\n        summary: 'YOLO pre-set for StabRise/signature_detection.',\n        consumes: ['image'],\n        produces: 'boxes',\n        peer: 'onnxruntime-web',\n        cache: { kind: 'hf-repo', param: 'model' },\n        defaults: asRecord({\n            ...YOLO_DETECTOR_DEFAULTS,\n            model: 'StabRise/signature_detection',\n            labels: ['signature'],\n            outputCol: 'signatures',\n            outputType: 'signature',\n            scoreThreshold: 0.2,\n        }),\n        params: yoloParams,\n    },\n    {\n        type: 'FaceDetector',\n        label: 'Face detector',\n        group: 'Detect',\n        subpath: '@stabrise/scaledp/detect',\n        summary: 'YOLO pre-set for StabRise/face_detection.',\n        consumes: ['image'],\n        produces: 'boxes',\n        peer: 'onnxruntime-web',\n        cache: { kind: 'hf-repo', param: 'model' },\n        defaults: asRecord({\n            ...YOLO_DETECTOR_DEFAULTS,\n            model: 'StabRise/face_detection',\n            labels: ['face'],\n            outputCol: 'faces',\n            outputType: 'face',\n            scoreThreshold: 0.2,\n        }),\n        params: yoloParams,\n    },\n    {\n        type: 'LineOrientationDetector',\n        label: 'Line orientation',\n        group: 'Transform',\n        subpath: '@stabrise/scaledp/ocr',\n        summary: 'Classify each detected region 0°/180° and turn the inverted ones.',\n        consumes: ['image', 'boxes'],\n        produces: 'image',\n        alsoProduces: [{ param: 'orientationCol', kind: 'orientations' }],\n        peer: 'onnxruntime-web',\n        cache: { kind: 'hf-repo', param: 'model', approxBytes: 9_000_000 },\n        expands: true,\n        defaults: asRecord(LINE_ORIENTATION_DEFAULTS),\n        params: [\n            ...baseParams({\n                input: 'unused',\n                output: { help: 'The corrected page image.' },\n            }),\n            {\n                key: 'inputCols',\n                kind: 'columns',\n                label: 'Input columns',\n                arity: 2,\n                accepts: ['image', 'boxes'],\n                help: 'The page image, then the detector output whose regions to classify.',\n            },\n            column('orientationCol', 'Orientation column', {\n                help: 'Per-region 0_degree / 180_degree labels.',\n            }),\n            {\n                key: 'model',\n                kind: 'string',\n                label: 'Model',\n                help: `Hugging Face repo id. Defaults to ${DEFAULT_ORIENTATION_MODEL}.`,\n            },\n            {\n                key: 'correct',\n                kind: 'boolean',\n                label: 'Correct',\n                help: 'Turn the inverted regions. Off classifies only, leaving the page as-is.',\n            },\n            {\n                key: 'onlyRotated',\n                kind: 'boolean',\n                label: 'Only rotated',\n                help: 'Classify only already-rotated boxes, where the signal is. Off also catches upside-down horizontal text, at a false-positive cost.',\n            },\n            {\n                key: 'padding',\n                kind: 'number',\n                label: 'Padding',\n                min: 0,\n                step: 1,\n                help: 'Grow each box before cropping, so glyph edges are not clipped.',\n            },\n            imageType,\n        ],\n    },\n    {\n        type: 'PaddleTextRecognizer',\n        label: 'PaddleOCR',\n        group: 'Recognise',\n        subpath: '@stabrise/scaledp/ocr',\n        summary: 'Detect and read a page in one pass. Ignores any separate detector’s boxes.',\n        consumes: ['image'],\n        produces: 'document',\n        peer: 'ppu-paddle-ocr',\n        cache: { kind: 'paddle-preset', param: 'preset' },\n        defaults: asRecord(PADDLE_RECOGNIZER_DEFAULTS),\n        params: [\n            ...baseParams({ input: { accepts: ['image'] } }),\n            {\n                key: 'preset',\n                kind: 'enum',\n                label: 'Preset',\n                options: OCR_PRESETS,\n                help: 'Language/script pairing. Pick the one matching your documents.',\n            },\n            scoreThreshold('Drop words below this confidence.'),\n            {\n                key: 'strategy',\n                kind: 'enum',\n                label: 'Strategy',\n                help: 'How the detected regions are grouped. It cannot subdivide them: the boxes are whatever the preset’s detector found, which is line-level.',\n                options: STRATEGIES,\n            },\n            keepFormatting,\n            lineTolerance,\n        ],\n    },\n    {\n        type: 'TesseractOcr',\n        label: 'Tesseract',\n        group: 'Recognise',\n        subpath: '@stabrise/scaledp/ocr',\n        summary:\n            'Read a whole page with Tesseract, detecting layout itself. The one stage here that returns word-level boxes.',\n        consumes: ['image'],\n        produces: 'document',\n        peer: 'tesseract-wasm',\n        defaults: asRecord(TESSERACT_OCR_DEFAULTS),\n        params: [\n            ...baseParams({ input: { accepts: ['image'] } }),\n            lang,\n            scoreThreshold('Drop words below this confidence.'),\n            keepFormatting,\n            lineTolerance,\n        ],\n    },\n    {\n        type: 'TesseractRecognizer',\n        label: 'Tesseract on boxes',\n        group: 'Recognise',\n        subpath: '@stabrise/scaledp/ocr',\n        summary:\n            'Read exactly the regions a detector found, cropping and turning each one. Returns a box per region, or per word with “Box level”.',\n        consumes: ['image', 'boxes'],\n        produces: 'document',\n        peer: 'tesseract-wasm',\n        cache: { kind: 'hf-repo', param: 'oriModel', approxBytes: 9_000_000 },\n        defaults: asRecord(TESSERACT_RECOGNIZER_DEFAULTS),\n        params: [\n            ...baseParams({ input: 'unused' }),\n            {\n                key: 'inputCols',\n                kind: 'columns',\n                label: 'Input columns',\n                arity: 2,\n                accepts: ['image', 'boxes'],\n                help: 'The page image, then the detector output whose regions to read.',\n            },\n            lang,\n            {\n                key: 'boxLevel',\n                kind: 'enum',\n                label: 'Box level',\n                options: [\n                    {\n                        value: 'region',\n                        label: 'One box per region',\n                        title: 'ScaleDP’s behaviour: each detected region keeps its own box, carrying everything read inside it',\n                    },\n                    {\n                        value: 'word',\n                        label: 'One box per word',\n                        title: 'Tesseract’s own word boxes, mapped back into page coordinates',\n                    },\n                ],\n                help: 'The detectors here are line-level, so “region” gives line boxes. Pick “word” to get one box per word instead.',\n            },\n            {\n                key: 'scaleFactor',\n                kind: 'number',\n                label: 'Scale factor',\n                min: 0.1,\n                max: 8,\n                step: 0.1,\n                help: 'Resize the page by this factor before cropping.',\n            },\n            {\n                key: 'padding',\n                kind: 'number',\n                label: 'Padding',\n                min: 0,\n                step: 1,\n                help: 'Grow each box before cropping. ScaleDP hardcodes 5.',\n            },\n            scoreThreshold('Drop words below this confidence.'),\n            keepFormatting,\n            lineTolerance,\n            {\n                key: 'detectLineOrientation',\n                kind: 'boolean',\n                label: 'Detect line orientation',\n                help: 'Classify each crop 0°/180° and turn the inverted ones.',\n            },\n            {\n                key: 'onlyRotated',\n                kind: 'boolean',\n                label: 'Only rotated',\n                help: 'Read only rotated or inverted boxes. On, an ordinary page returns nothing.',\n            },\n            {\n                key: 'oriModel',\n                kind: 'string',\n                label: 'Orientation model',\n                advanced: true,\n                help: `Hugging Face repo id. Defaults to ${DEFAULT_ORIENTATION_MODEL}.`,\n            },\n        ],\n    },\n    {\n        type: 'GlinerNer',\n        label: 'GLiNER entities',\n        group: 'Understand',\n        subpath: '@stabrise/scaledp/ner',\n        summary: 'Find named entities in a document, scored against labels you write.',\n        consumes: ['document'],\n        produces: 'ner',\n        peer: '@huggingface/transformers',\n        cache: { kind: 'ner-id', param: 'model' },\n        defaults: asRecord(GLINER_NER_DEFAULTS),\n        params: [\n            ...baseParams({ input: { accepts: ['document'] } }),\n            {\n                key: 'model',\n                kind: 'enum',\n                label: 'Model',\n                options: NER_MODEL_OPTIONS,\n                help: 'Registry id. Private repos need configure({ auth }).',\n            },\n            {\n                key: 'labels',\n                kind: 'stringList',\n                label: 'Labels',\n                required: true,\n                help: 'GLiNER scores a label by its prompt text, so other wording asks a different question.',\n            },\n            {\n                key: 'threshold',\n                kind: 'number',\n                label: 'Threshold',\n                min: 0,\n                max: 1,\n                step: 0.05,\n                help: 'Minimum score an entity must reach.',\n            },\n            {\n                key: 'whiteList',\n                kind: 'stringList',\n                label: 'Only these groups',\n                help: 'Keep only these entity groups; empty keeps everything.',\n            },\n            {\n                key: 'chunkLength',\n                kind: 'number',\n                label: 'Chunk length',\n                min: 1,\n                step: 1,\n                advanced: true,\n                help: 'Tokens per inference window.',\n            },\n            {\n                key: 'chunkStride',\n                kind: 'number',\n                label: 'Chunk stride',\n                min: 1,\n                step: 1,\n                advanced: true,\n                help: 'Window step. Below chunk length, windows overlap.',\n            },\n            {\n                key: 'normaliseCasing',\n                kind: 'boolean',\n                label: 'Normalise casing',\n                help: 'Title-case runs of capitals first. GLiNER1 models are cased; scans are often all caps.',\n            },\n        ],\n    },\n    {\n        type: 'ImageDrawBoxes',\n        label: 'Draw boxes',\n        group: 'Transform',\n        subpath: '@stabrise/scaledp',\n        summary: 'Annotate a page with any boxes or entities found so far.',\n        consumes: ['image', 'boxes'],\n        produces: 'image',\n        terminal: true,\n        defaults: asRecord(IMAGE_DRAW_BOXES_DEFAULTS),\n        params: [\n            ...baseParams({ input: 'unused' }),\n            {\n                key: 'inputCols',\n                kind: 'columns',\n                label: 'Input columns',\n                minArity: 2,\n                accepts: ['image', 'boxes'],\n                help: 'The image first, then one or more box, document or entity columns.',\n            },\n            {\n                key: 'color',\n                kind: 'color',\n                label: 'Colour',\n                help: 'One colour for every box. Unset colours by group instead.',\n            },\n            {\n                key: 'filled',\n                kind: 'boolean',\n                label: 'Filled',\n                help: 'Fill boxes as well as outlining them.',\n            },\n            { key: 'lineWidth', kind: 'number', label: 'Line width', min: 0, max: 20, step: 1 },\n            { key: 'textSize', kind: 'number', label: 'Text size', min: 4, max: 72, step: 1 },\n            {\n                key: 'displayDataList',\n                kind: 'stringList',\n                label: 'Label fields',\n                options: LABEL_FIELDS,\n                help: 'Rendered above each box, joined by \":\". Pick \"text\" to read the OCR output back off the page.',\n            },\n            {\n                key: 'padding',\n                kind: 'number',\n                label: 'Padding',\n                min: 0,\n                step: 1,\n                help: 'Grow each box by this many pixels before drawing.',\n            },\n            {\n                key: 'whiteList',\n                kind: 'stringList',\n                label: 'Only these groups',\n                help: 'Draw only these entity groups; empty draws all.',\n            },\n            {\n                key: 'blackList',\n                kind: 'stringList',\n                label: 'Never these groups',\n                help: 'Never draw these entity groups.',\n            },\n            imageType,\n        ],\n    },\n    {\n        type: 'ImageCropBoxes',\n        label: 'Crop boxes',\n        group: 'Transform',\n        subpath: '@stabrise/scaledp',\n        summary: 'Cut each detected region out of the page, one output row per crop.',\n        consumes: ['image', 'boxes'],\n        produces: 'image',\n        alsoProduces: [{ param: 'boxCol', kind: 'box' }],\n        expands: true,\n        defaults: asRecord(IMAGE_CROP_BOXES_DEFAULTS),\n        params: [\n            ...baseParams({ input: 'unused' }),\n            {\n                key: 'inputCols',\n                kind: 'columns',\n                label: 'Input columns',\n                arity: 2,\n                accepts: ['image', 'boxes'],\n                help: 'The page image, then the box column to cut from it.',\n            },\n            column('boxCol', 'Box column', { help: 'Where each crop’s source box is written.' }),\n            {\n                key: 'padding',\n                kind: 'number',\n                label: 'Padding',\n                min: 0,\n                step: 1,\n                help: 'Grow each box by this many pixels before cropping.',\n            },\n            {\n                key: 'limit',\n                kind: 'number',\n                label: 'Limit',\n                min: 0,\n                step: 1,\n                help: 'Maximum crops per page; 0 means all of them.',\n            },\n            {\n                key: 'autoRotate',\n                kind: 'boolean',\n                label: 'Auto-rotate',\n                help: 'Turn portrait crops a quarter turn, so text reads horizontally.',\n            },\n            {\n                key: 'returnEmpty',\n                kind: 'boolean',\n                label: 'Return empty',\n                help: 'Emit the whole page when nothing was detected, instead of failing.',\n            },\n            imageType,\n        ],\n    },\n])\n\n/** Name → constructor, for building a stage from a `StageDescriptor`. */\nexport const STAGE_CLASSES = Object.freeze({\n    DataToImage,\n    DbnetOnnxDetector,\n    FaceDetector,\n    GlinerNer,\n    ImageCropBoxes,\n    ImageDrawBoxes,\n    LineOrientationDetector,\n    PaddleTextDetector,\n    PaddleTextRecognizer,\n    PdfToDocument,\n    PdfToImage,\n    SignatureDetector,\n    TesseractOcr,\n    TesseractRecognizer,\n    YoloOnnxDetector,\n}) as Readonly<Record<string, new (options?: never) => import('../core/pipeline.js').Stage>>\n","/**\n * A pipeline, written back out as the TypeScript that builds it.\n *\n * The registry already knows which subpath each stage is imported from and what\n * its defaults are, which is everything the source needs. Emitting only the\n * options that differ from the defaults keeps the result readable as the set of\n * decisions someone actually made -- the same reason the JSON form stores only\n * those.\n */\n\nimport type { StageDescriptor } from '../core/pipeline.js'\nimport { STAGE_SPECS } from './catalog.js'\n\nexport interface PipelineCodeOptions {\n    /** Name bound to the pipeline. */\n    variable?: string\n    /** Emit the import block above it. */\n    imports?: boolean\n    /** Spaces per indent level. */\n    indent?: number\n}\n\n/** Structural equality, deep enough for param values: scalars and string lists. */\nfunction same(a: unknown, b: unknown): boolean {\n    if (Array.isArray(a) && Array.isArray(b)) {\n        return a.length === b.length && a.every((value, index) => value === b[index])\n    }\n    return a === b\n}\n\n/** A value as TypeScript source, in the quote style the library is written in. */\nfunction literal(value: unknown): string {\n    if (typeof value === 'string') return `'${value.replace(/\\\\/g, '\\\\\\\\').replace(/'/g, \"\\\\'\")}'`\n    if (typeof value === 'number' || typeof value === 'boolean' || value === null) return String(value)\n    if (Array.isArray(value)) return `[${value.map(literal).join(', ')}]`\n    return JSON.stringify(value) ?? 'undefined'\n}\n\n/** The options that differ from the stage's defaults, as `key: value` source. */\nfunction changedOptions(descriptor: StageDescriptor): string[] {\n    const spec = STAGE_SPECS.find((candidate) => candidate.type === descriptor.type)\n    const options = descriptor.options ?? {}\n\n    return Object.keys(options)\n        .filter((key) => !spec || !same(spec.defaults[key], options[key]))\n        .map((key) => `${key}: ${literal(options[key])}`)\n}\n\n/**\n * Import lines, grouped by subpath.\n *\n * The root comes first because `Pipeline` lives there and every pipeline needs\n * it; the engine subpaths follow in alphabetical order, which is what biome's\n * import sorting would produce anyway.\n */\nfunction importLines(descriptors: readonly StageDescriptor[], root: string): string[] {\n    const bySubpath = new Map<string, Set<string>>()\n    bySubpath.set(root, new Set(['Pipeline']))\n\n    for (const descriptor of descriptors) {\n        const spec = STAGE_SPECS.find((candidate) => candidate.type === descriptor.type)\n        if (!spec) continue\n        const names = bySubpath.get(spec.subpath) ?? new Set<string>()\n        names.add(descriptor.type)\n        bySubpath.set(spec.subpath, names)\n    }\n\n    return [...bySubpath.entries()]\n        .sort(([a], [b]) => (a === root ? -1 : b === root ? 1 : a.localeCompare(b)))\n        .map(([subpath, names]) => `import { ${[...names].sort().join(', ')} } from '${subpath}'`)\n}\n\nexport function pipelineCode(\n    descriptors: readonly StageDescriptor[],\n    options: PipelineCodeOptions = {}\n): string {\n    const variable = options.variable ?? 'pipeline'\n    const pad = ' '.repeat(options.indent ?? 4)\n    const root = STAGE_SPECS.find((spec) => spec.subpath.endsWith('/scaledp'))?.subpath ?? '@stabrise/scaledp'\n\n    const stages = descriptors.map((descriptor) => {\n        const entries = changedOptions(descriptor)\n        if (entries.length === 0) return `${pad}new ${descriptor.type}(),`\n\n        const inline = `${pad}new ${descriptor.type}({ ${entries.join(', ')} }),`\n        if (inline.length <= 100) return inline\n\n        // One option per line once the call outgrows a sensible line length,\n        // which is what a formatter would do to it anyway.\n        return [\n            `${pad}new ${descriptor.type}({`,\n            ...entries.map((entry) => `${pad}${pad}${entry},`),\n            `${pad}}),`,\n        ].join('\\n')\n    })\n\n    const body = [`const ${variable} = new Pipeline([`, ...stages, '])']\n    if (options.imports === false) return body.join('\\n')\n\n    return [...importLines(descriptors, root), '', ...body].join('\\n')\n}\n","/**\n * @stabrise/scaledp/registry -- describe, build and serialise pipelines as data.\n *\n * A pipeline is an ordered array of stages, each a class with a params object.\n * That is fine to write by hand, but a UI that assembles one needs two things\n * the classes themselves do not carry at run time: a description of each\n * stage's parameters, and a way back from JSON to live objects.\n *\n *     import { STAGE_SPECS, pipelineFromDescriptors } from '@stabrise/scaledp/registry'\n *\n *     const stages = [\n *         { type: 'PdfToImage', options: { resolution: 200 } },\n *         { type: 'PaddleTextRecognizer', options: { keepFormatting: true } },\n *     ]\n *     const rows = await pipelineFromDescriptors(stages).transform(file)\n *\n * `StageDescriptor[]` is plain JSON, and the same shape the worker protocol\n * already sends across the boundary -- so a saved pipeline, a posted message and\n * a pipeline built here are all the same data.\n *\n * This subpath pulls in every stage class. It pulls in no ML runtime: engines\n * are reached through dynamic `import()` and stay lazy.\n */\n\nimport { Pipeline, type Stage, type StageDescriptor } from '../core/pipeline.js'\nimport { STAGE_CLASSES, STAGE_SPECS } from './catalog.js'\n\n// Re-exported so a consumer building pipelines as data needs only this subpath.\nexport type { StageDescriptor } from '../core/pipeline.js'\nexport type { PipelineCodeOptions } from './codegen.js'\nexport { pipelineCode } from './codegen.js'\nexport type {\n    ColumnKind,\n    StageCacheSpec,\n    StageParamKind,\n    StageParamOption,\n    StageParamSpec,\n    StageSpec,\n} from './types.js'\nexport { STAGE_CLASSES, STAGE_SPECS }\n\n/** Metadata for one stage, by exported class name. */\nexport function getStageSpec(type: string): (typeof STAGE_SPECS)[number] | undefined {\n    return STAGE_SPECS.find((spec) => spec.type === type)\n}\n\n/**\n * Build one stage from its serialised form.\n *\n * Throws on an unknown type, and lets the stage's own constructor validators\n * throw on a bad param -- an unknown OCR preset should fail here, where the\n * message can name the offending field, rather than several stages later.\n */\nexport function createStage(descriptor: StageDescriptor): Stage {\n    const Ctor = STAGE_CLASSES[descriptor.type]\n    if (!Ctor) {\n        throw new Error(\n            `Unknown stage \"${descriptor.type}\". See STAGE_SPECS for the stages this build knows about.`\n        )\n    }\n    return new Ctor(descriptor.options as never)\n}\n\n/** Build a whole pipeline from its serialised form. */\nexport function pipelineFromDescriptors(descriptors: readonly StageDescriptor[]): Pipeline {\n    return new Pipeline(descriptors.map(createStage))\n}\n\n/**\n * The serialised form of a live stage.\n *\n * `stage.params` is fully resolved -- defaults merged in -- so this round-trips\n * exactly but is more verbose than the options originally passed.\n */\nexport function describeStage(stage: Stage): StageDescriptor {\n    return { type: stage.name, options: { ...stage.params } }\n}\n\n/** The serialised form of a live pipeline. */\nexport function describePipeline(pipeline: Pipeline): StageDescriptor[] {\n    return pipeline.stages.map(describeStage)\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAkDA,MAAM,YAA8B,aAChC;AAIJ,MAAM,cAA2C,OAAO,OAAO;CAC3D;EAAE,OAAO;EAAO,OAAO;EAAO,OAAO;CAAqC;CAC1E;EAAE,OAAO;EAAQ,OAAO;EAAQ,OAAO;CAAiB;CACxD;EAAE,OAAO;EAAQ,OAAO;EAAQ,OAAO;CAA+C;AAC1F,CAAC;AAED,MAAM,aAA0C,OAAO,OAAO;CAC1D;EAAE,OAAO;EAAW,OAAO;EAAW,OAAO;CAAgC;CAC7E;EAAE,OAAO;EAAY,OAAO;EAAY,OAAO;CAAgC;CAC/E;EAAE,OAAO;EAAc,OAAO;EAAc,OAAO;CAA+B;AACtF,CAAC;;;;;;;;AASD,MAAM,eAA4C,OAAO,OAAO;CAC5D;EAAE,OAAO;EAAQ,OAAO;EAAQ,OAAO;CAA2B;CAClE;EAAE,OAAO;EAAS,OAAO;EAAS,OAAO;CAAiC;CAC1E;EAAE,OAAO;EAAS,OAAO;EAAS,OAAO;CAAgC;CACzE;EAAE,OAAO;EAAgB,OAAO;EAAgB,OAAO;CAA+B;CACtF;EAAE,OAAO;EAAQ,OAAO;EAAQ,OAAO;CAA2B;CAClE;EAAE,OAAO;EAAK,OAAO;EAAK,OAAO;CAAkB;CACnD;EAAE,OAAO;EAAK,OAAO;EAAK,OAAO;CAAkB;CACnD;EAAE,OAAO;EAAS,OAAO;EAAS,OAAO;CAAuB;CAChE;EAAE,OAAO;EAAU,OAAO;EAAU,OAAO;CAAwB;AACvE,CAAC;AAED,MAAM,cAA2C,OAAO,OACpD,mBAAmB,KAAK,YAAY;CAChC,OAAO,OAAO;CACd,OAAO,OAAO;CACd,OAAO,OAAO,QAAQ,KAAK,IAAI;AACnC,EAAE,CACN;AAEA,MAAM,oBAAiD,OAAO,OAC1D,WAAW,KAAK,WAAW;CACvB,OAAO,MAAM;CACb,OAAO,GAAG,MAAM,KAAK,KAAK,MAAM,UAAU,KAAK,GAAG,IAAI,MAAM,UAAU,eAAe;CACrF,OAAO,GAAG,MAAM,KAAK,IAAI,KAAK,MAAM,eAAe,KAAK,IAAI,GAAG,EAAE,OAAO,MAAM;CAG9E,UAAU,MAAM,YAAY;AAChC,EAAE,CACN;AAEA,MAAM,sBAAmD,OAAO,OAC5D,gBAAgB,QAAQ,UAAU,MAAM,SAAS,gBAAgB,MAAM,IAAI,CAAC,CAAC,KAAK,WAAW;CACzF,OAAO,MAAM;CACb,OAAO,MAAM;CACb,OAAO,MAAM;AACjB,EAAE,CACN;AAIA,MAAM,UAAU,KAAa,OAAe,QAAiC,CAAC,OAAuB;CACjG;CACA,MAAM;CACN;CACA,GAAG;AACP;;;;;;;;;AAUA,SAAS,WACL,UAAwG,CAAC,GACzF;CAYhB,OAAO;EAVH,QAAQ,UAAU,WACZ,OAAO,YAAY,gBAAgB;GAC/B,UAAU;GACV,MAAM;EACV,CAAC,IACD,OAAO,YAAY,gBAAgB;GAC/B,MAAM,QAAQ,OAAO;GACrB,SAAS,QAAQ,OAAO;EAC5B,CAAC;EAIP,OAAO,aAAa,iBAAiB,EAAE,MAAM,QAAQ,QAAQ,KAAK,CAAC;EACnE,OAAO,WAAW,eAAe;GAC7B,UAAU;GACV,MAAM;EACV,CAAC;EACD,OAAO,WAAW,eAAe;GAC7B,UAAU;GACV,MAAM;EACV,CAAC;EACD;GACI,KAAK;GACL,MAAM;GACN,OAAO;GACP,UAAU;GACV,MAAM;EACV;EACA;GACI,KAAK;GACL,MAAM;GACN,OAAO;GACP,UAAU;GACV,MAAM;EACV;CACJ;AACJ;AAEA,MAAM,YAA4B;CAC9B,KAAK;CACL,MAAM;CACN,OAAO;CACP,SAAS;AACb;AAEA,MAAM,kBAAkB,UAAkC;CACtD,KAAK;CACL,MAAM;CACN,OAAO;CACP,KAAK;CACL,KAAK;CACL,MAAM;CACN;AACJ;AAEA,MAAM,OAAuB;CACzB,KAAK;CACL,MAAM;CACN,OAAO;CACP,UAAU;CACV,MAAM;AACV;AAEA,MAAM,iBAAiC;CACnC,KAAK;CACL,MAAM;CACN,OAAO;CACP,MAAM;AACV;AAEA,MAAM,gBAAgC;CAClC,KAAK;CACL,MAAM;CACN,OAAO;CACP,KAAK;CACL,MAAM;CACN,MAAM;AACV;;;;;;;;;AAUA,MAAM,aAAwC,OAAO,OAAO;CACxD,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;CAC/C;EACI,KAAK;EACL,MAAM;EACN,OAAO;EACP,UAAU;EACV,MAAM;CACV;CACA;EACI,KAAK;EACL,MAAM;EACN,OAAO;EACP,MAAM;CACV;CACA,eAAe,wCAAwC;CACvD;EACI,KAAK;EACL,MAAM;EACN,OAAO;EACP,KAAK;EACL,KAAK;EACL,MAAM;EACN,MAAM;CACV;CACA;EACI,KAAK;EACL,MAAM;EACN,OAAO;EACP,KAAK;EACL,KAAK;EACL,MAAM;EACN,MAAM;CACV;CACA;EACI,KAAK;EACL,MAAM;EACN,OAAO;EACP,UAAU;EACV,MAAM;CACV;AACJ,CAAC;AAID,MAAa,cAAoC,OAAO,OAAO;CAC3D;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,UAAU,SAAS,sBAAsB;EACzC,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,MAAM,+CAA+C,EAAE,CAAC;GACjF;IAAE,GAAG;IAAW,MAAM;GAAqE;GAC3F;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,SAAS;EACT,UAAU,SAAS,qBAAqB;EACxC,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,MAAM,8CAA8C,EAAE,CAAC;GAChF;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,SAAS;EACT,UAAU,SAAS,wBAAwB;EAC3C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,MAAM,8CAA8C,EAAE,CAAC;GAChF;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAiB,OAAO;EAAS;EAChD,UAAU,SAAS,wBAAwB;EAC3C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;GAC/C;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS;IACT,MAAM;GACV;GACA,eAAe,yDAAyD;EAC5E;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SACI;EACJ,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;GAAS,aAAa;EAAU;EACjE,UAAU,SAAS,uBAAuB;EAC1C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;GAC/C;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS;IACT,aAAa;IACb,MAAM;GACV;GACA,eAAe,iDAAiD;GAChE;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;EAAQ;EACzC,UAAU,SAAS,sBAAsB;EACzC,QAAQ;CACZ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;EAAQ;EACzC,UAAU,SAAS;GACf,GAAG;GACH,OAAO;GACP,QAAQ,CAAC,WAAW;GACpB,WAAW;GACX,YAAY;GACZ,gBAAgB;EACpB,CAAC;EACD,QAAQ;CACZ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;EAAQ;EACzC,UAAU,SAAS;GACf,GAAG;GACH,OAAO;GACP,QAAQ,CAAC,MAAM;GACf,WAAW;GACX,YAAY;GACZ,gBAAgB;EACpB,CAAC;EACD,QAAQ;CACZ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,SAAS,OAAO;EAC3B,UAAU;EACV,cAAc,CAAC;GAAE,OAAO;GAAkB,MAAM;EAAe,CAAC;EAChE,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;GAAS,aAAa;EAAU;EACjE,SAAS;EACT,UAAU,SAAS,yBAAyB;EAC5C,QAAQ;GACJ,GAAG,WAAW;IACV,OAAO;IACP,QAAQ,EAAE,MAAM,4BAA4B;GAChD,CAAC;GACD;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,OAAO;IACP,SAAS,CAAC,SAAS,OAAO;IAC1B,MAAM;GACV;GACA,OAAO,kBAAkB,sBAAsB,EAC3C,MAAM,2CACV,CAAC;GACD;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM,qCAAqC,0BAA0B;GACzE;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAiB,OAAO;EAAS;EAChD,UAAU,SAAS,0BAA0B;EAC7C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;GAC/C;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS;IACT,MAAM;GACV;GACA,eAAe,mCAAmC;GAClD;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;IACN,SAAS;GACb;GACA;GACA;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SACI;EACJ,UAAU,CAAC,OAAO;EAClB,UAAU;EACV,MAAM;EACN,UAAU,SAAS,sBAAsB;EACzC,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,EAAE,CAAC;GAC/C;GACA,eAAe,mCAAmC;GAClD;GACA;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SACI;EACJ,UAAU,CAAC,SAAS,OAAO;EAC3B,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAW,OAAO;GAAY,aAAa;EAAU;EACpE,UAAU,SAAS,6BAA6B;EAChD,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,SAAS,CAAC;GACjC;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,OAAO;IACP,SAAS,CAAC,SAAS,OAAO;IAC1B,MAAM;GACV;GACA;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS,CACL;KACI,OAAO;KACP,OAAO;KACP,OAAO;IACX,GACA;KACI,OAAO;KACP,OAAO;KACP,OAAO;IACX,CACJ;IACA,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA,eAAe,mCAAmC;GAClD;GACA;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,UAAU;IACV,MAAM,qCAAqC,0BAA0B;GACzE;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,UAAU;EACrB,UAAU;EACV,MAAM;EACN,OAAO;GAAE,MAAM;GAAU,OAAO;EAAQ;EACxC,UAAU,SAAS,mBAAmB;EACtC,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,EAAE,SAAS,CAAC,UAAU,EAAE,EAAE,CAAC;GAClD;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS;IACT,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,UAAU;IACV,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,UAAU;IACV,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,UAAU;IACV,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,SAAS,OAAO;EAC3B,UAAU;EACV,UAAU;EACV,UAAU,SAAS,yBAAyB;EAC5C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,SAAS,CAAC;GACjC;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,UAAU;IACV,SAAS,CAAC,SAAS,OAAO;IAC1B,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IAAE,KAAK;IAAa,MAAM;IAAU,OAAO;IAAc,KAAK;IAAG,KAAK;IAAI,MAAM;GAAE;GAClF;IAAE,KAAK;IAAY,MAAM;IAAU,OAAO;IAAa,KAAK;IAAG,KAAK;IAAI,MAAM;GAAE;GAChF;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,SAAS;IACT,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;EACJ;CACJ;CACA;EACI,MAAM;EACN,OAAO;EACP,OAAO;EACP,SAAS;EACT,SAAS;EACT,UAAU,CAAC,SAAS,OAAO;EAC3B,UAAU;EACV,cAAc,CAAC;GAAE,OAAO;GAAU,MAAM;EAAM,CAAC;EAC/C,SAAS;EACT,UAAU,SAAS,yBAAyB;EAC5C,QAAQ;GACJ,GAAG,WAAW,EAAE,OAAO,SAAS,CAAC;GACjC;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,OAAO;IACP,SAAS,CAAC,SAAS,OAAO;IAC1B,MAAM;GACV;GACA,OAAO,UAAU,cAAc,EAAE,MAAM,2CAA2C,CAAC;GACnF;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;IACI,KAAK;IACL,MAAM;IACN,OAAO;IACP,MAAM;GACV;GACA;EACJ;CACJ;AACJ,CAAC;;AAGD,MAAa,gBAAgB,OAAO,OAAO;CACvC;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;AACJ,CAAC;;;;ACz1BD,SAAS,KAAK,GAAY,GAAqB;CAC3C,IAAI,MAAM,QAAQ,CAAC,KAAK,MAAM,QAAQ,CAAC,GACnC,OAAO,EAAE,WAAW,EAAE,UAAU,EAAE,OAAO,OAAO,UAAU,UAAU,EAAE,MAAM;CAEhF,OAAO,MAAM;AACjB;;AAGA,SAAS,QAAQ,OAAwB;CACrC,IAAI,OAAO,UAAU,UAAU,OAAO,IAAI,MAAM,QAAQ,OAAO,MAAM,CAAC,CAAC,QAAQ,MAAM,KAAK,EAAE;CAC5F,IAAI,OAAO,UAAU,YAAY,OAAO,UAAU,aAAa,UAAU,MAAM,OAAO,OAAO,KAAK;CAClG,IAAI,MAAM,QAAQ,KAAK,GAAG,OAAO,IAAI,MAAM,IAAI,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE;CACnE,OAAO,KAAK,UAAU,KAAK,KAAK;AACpC;;AAGA,SAAS,eAAe,YAAuC;CAC3D,MAAM,OAAO,YAAY,MAAM,cAAc,UAAU,SAAS,WAAW,IAAI;CAC/E,MAAM,UAAU,WAAW,WAAW,CAAC;CAEvC,OAAO,OAAO,KAAK,OAAO,CAAC,CACtB,QAAQ,QAAQ,CAAC,QAAQ,CAAC,KAAK,KAAK,SAAS,MAAM,QAAQ,IAAI,CAAC,CAAC,CACjE,KAAK,QAAQ,GAAG,IAAI,IAAI,QAAQ,QAAQ,IAAI,GAAG;AACxD;;;;;;;;AASA,SAAS,YAAY,aAAyC,MAAwB;CAClF,MAAM,4BAAY,IAAI,IAAyB;CAC/C,UAAU,IAAI,sBAAM,IAAI,IAAI,CAAC,UAAU,CAAC,CAAC;CAEzC,KAAK,MAAM,cAAc,aAAa;EAClC,MAAM,OAAO,YAAY,MAAM,cAAc,UAAU,SAAS,WAAW,IAAI;EAC/E,IAAI,CAAC,MAAM;EACX,MAAM,QAAQ,UAAU,IAAI,KAAK,OAAO,qBAAK,IAAI,IAAY;EAC7D,MAAM,IAAI,WAAW,IAAI;EACzB,UAAU,IAAI,KAAK,SAAS,KAAK;CACrC;CAEA,OAAO,CAAC,GAAG,UAAU,QAAQ,CAAC,CAAC,CAC1B,MAAM,CAAC,IAAI,CAAC,OAAQ,MAAM,OAAO,KAAK,MAAM,OAAO,IAAI,EAAE,cAAc,CAAC,CAAE,CAAC,CAC3E,KAAK,CAAC,SAAS,WAAW,YAAY,CAAC,GAAG,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,IAAI,EAAE,WAAW,QAAQ,EAAE;AACjG;AAEA,SAAgB,aACZ,aACA,UAA+B,CAAC,GAC1B;CACN,MAAM,WAAW,QAAQ,YAAY;CACrC,MAAM,MAAM,IAAI,OAAO,QAAQ,UAAU,CAAC;CAC1C,MAAM,OAAO,YAAY,MAAM,SAAS,KAAK,QAAQ,SAAS,UAAU,CAAC,CAAC,EAAE,WAAW;CAEvF,MAAM,SAAS,YAAY,KAAK,eAAe;EAC3C,MAAM,UAAU,eAAe,UAAU;EACzC,IAAI,QAAQ,WAAW,GAAG,OAAO,GAAG,IAAI,MAAM,WAAW,KAAK;EAE9D,MAAM,SAAS,GAAG,IAAI,MAAM,WAAW,KAAK,KAAK,QAAQ,KAAK,IAAI,EAAE;EACpE,IAAI,OAAO,UAAU,KAAK,OAAO;EAIjC,OAAO;GACH,GAAG,IAAI,MAAM,WAAW,KAAK;GAC7B,GAAG,QAAQ,KAAK,UAAU,GAAG,MAAM,MAAM,MAAM,EAAE;GACjD,GAAG,IAAI;EACX,CAAC,CAAC,KAAK,IAAI;CACf,CAAC;CAED,MAAM,OAAO;EAAC,SAAS,SAAS;EAAoB,GAAG;EAAQ;CAAI;CACnE,IAAI,QAAQ,YAAY,OAAO,OAAO,KAAK,KAAK,IAAI;CAEpD,OAAO;EAAC,GAAG,YAAY,aAAa,IAAI;EAAG;EAAI,GAAG;CAAI,CAAC,CAAC,KAAK,IAAI;AACrE;;;;;;;;;;;;;;;;;;;;;;;;;;;AC1DA,SAAgB,aAAa,MAAwD;CACjF,OAAO,YAAY,MAAM,SAAS,KAAK,SAAS,IAAI;AACxD;;;;;;;;AASA,SAAgB,YAAY,YAAoC;CAC5D,MAAM,OAAO,cAAc,WAAW;CACtC,IAAI,CAAC,MACD,MAAM,IAAI,MACN,kBAAkB,WAAW,KAAK,0DACtC;CAEJ,OAAO,IAAI,KAAK,WAAW,OAAgB;AAC/C;;AAGA,SAAgB,wBAAwB,aAAmD;CACvF,OAAO,IAAI,SAAS,YAAY,IAAI,WAAW,CAAC;AACpD;;;;;;;AAQA,SAAgB,cAAc,OAA+B;CACzD,OAAO;EAAE,MAAM,MAAM;EAAM,SAAS,EAAE,GAAG,MAAM,OAAO;CAAE;AAC5D;;AAGA,SAAgB,iBAAiB,UAAuC;CACpE,OAAO,SAAS,OAAO,IAAI,aAAa;AAC5C"}

package/dist/text-ahMLpxN9.js

@@ -0,0 +1,109 @@
+//#region src/core/text.ts
+/**
+* Arrange items into groups whose successive elements differ by no more than
+* `maxGap`. Unlike Python's `cluster`, this does not sort the caller's array
+* in place.
+*/
+function cluster(items, maxGap, key = (x) => x) {
+	if (items.length === 0) return [];
+	const sorted = [...items].sort((a, b) => key(a) - key(b));
+	const groups = [[sorted[0]]];
+	for (const item of sorted.slice(1)) {
+		const group = groups[groups.length - 1];
+		const previous = group[group.length - 1];
+		if (Math.abs(key(item) - key(previous)) <= maxGap) group.push(item);
+		else groups.push([item]);
+	}
+	return groups;
+}
+/**
+* A robust central value: drop the first and last quartile, then take the mode.
+*
+* Trimming the tails is what makes this survive OCR outliers -- a single giant
+* heading would otherwise skew a plain mean. Always returns a truncated
+* integer, matching Python's `int(...)` wrapper.
+*/
+function getSize(items, key = (x) => x) {
+	if (items.length === 0) return 0;
+	const values = items.map(key);
+	if (values.length === 1) return values[0];
+	values.sort((a, b) => a - b);
+	if (values.length <= 4) return Math.trunc(values.reduce((sum, v) => sum + v, 0) / values.length);
+	const trim = Math.trunc(values.length / 4);
+	const middle = values.slice(trim, values.length - trim);
+	const counts = /* @__PURE__ */ new Map();
+	for (const value of middle) counts.set(value, (counts.get(value) ?? 0) + 1);
+	let best = middle[0];
+	let bestCount = 0;
+	for (const [value, count] of counts) if (count > bestCount) {
+		best = value;
+		bestCount = count;
+	}
+	return Math.trunc(best);
+}
+/**
+* Median-ish width of a single character, used to size inter-word gaps.
+* Takes grouped lines, matching Python's signature.
+*/
+function getCharacterWidth(lines) {
+	const widths = [];
+	for (const line of lines) for (const box of line) if (box.text.length > 0) widths.push(Math.trunc(box.width / box.text.length));
+	return getSize(widths);
+}
+/**
+* Group boxes into text lines, each sorted left to right.
+*
+* `lineTolerance` of 0 means "derive it": a third of the typical character
+* height -- tight enough to keep adjacent lines apart, loose enough to tolerate
+* baseline jitter.
+*/
+function groupBoxesIntoLines(boxes, lineTolerance = 0) {
+	if (boxes.length === 0) return [];
+	const characterHeight = getSize(boxes, (b) => b.height);
+	return cluster(boxes, lineTolerance === 0 ? characterHeight / 3 : lineTolerance, (b) => Math.trunc(b.y)).map((line) => [...line].sort((a, b) => Math.trunc(a.x) - Math.trunc(b.x)));
+}
+/**
+* Render grouped lines back to text, preserving horizontal spacing and blank
+* lines.
+*
+* Each line is indented from x = 0, so the left margin is preserved rather than
+* every line being flushed left. Vertical gaps wider than two character heights
+* become blank lines. The running `y` starts at 0, which means the first line is
+* measured against the top of the page -- that is Python's behaviour and it is
+* what makes a top margin survive.
+*/
+function linesToFormattedText(lines, characterHeight) {
+	const output = [];
+	const spaceWidth = getCharacterWidth(lines) || 1;
+	let y = 0;
+	for (const line of lines) {
+		const first = line[0];
+		if (!first) continue;
+		const lineDiffs = characterHeight > 0 ? Math.trunc((first.y - y) / (characterHeight * 2)) : 0;
+		y = first.y;
+		for (let i = 0; i < lineDiffs - 1; i++) output.push("");
+		let text = "";
+		let previousRight = 0;
+		for (const box of line) {
+			const spaces = Math.max(Math.trunc((box.x - previousRight) / spaceWidth), 1);
+			text += " ".repeat(spaces) + box.text;
+			previousRight = box.x + box.width;
+		}
+		output.push(text);
+	}
+	return output.join("\n");
+}
+/** Rebuild layout-preserving text straight from a flat box list. */
+function boxesToFormattedText(boxes, lineTolerance = 0) {
+	if (boxes.length === 0) return "";
+	const characterHeight = getSize(boxes, (b) => b.height);
+	return linesToFormattedText(groupBoxesIntoLines(boxes, lineTolerance), characterHeight);
+}
+/** Plain reading-order join, the default when `keepFormatting` is off. */
+function boxesToText(boxes, separator = " ") {
+	return boxes.map((b) => b.text).join(separator);
+}
+//#endregion
+export { getSize as a, getCharacterWidth as i, boxesToText as n, groupBoxesIntoLines as o, cluster as r, linesToFormattedText as s, boxesToFormattedText as t };
+
+//# sourceMappingURL=text-ahMLpxN9.js.map

package/dist/text-ahMLpxN9.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"text-ahMLpxN9.js","names":[],"sources":["../src/core/text.ts"],"sourcesContent":["/**\n * Layout-preserving text reconstruction from OCR boxes.\n *\n * Port of `BaseOcr.box_to_formatted_text` / `to_formatted_text` /\n * `get_character_width` plus `cluster` and `get_size` from\n * `scaledp/utils/__init__.py`. Pure geometry -- no engine involved.\n *\n * Python truncates with `int()` throughout rather than rounding, and the\n * results feed NER character offsets, so every `Math.trunc` below is\n * deliberate. Verified against the Python implementation in\n * test/unit/text-parity.test.ts.\n */\n\nimport type { Box } from '../schemas/box.js'\n\n/**\n * Arrange items into groups whose successive elements differ by no more than\n * `maxGap`. Unlike Python's `cluster`, this does not sort the caller's array\n * in place.\n */\nexport function cluster<T>(\n    items: readonly T[],\n    maxGap: number,\n    key: (item: T) => number = (x) => x as unknown as number\n): T[][] {\n    if (items.length === 0) return []\n    const sorted = [...items].sort((a, b) => key(a) - key(b))\n\n    const groups: T[][] = [[sorted[0] as T]]\n    for (const item of sorted.slice(1)) {\n        const group = groups[groups.length - 1] as T[]\n        const previous = group[group.length - 1] as T\n        if (Math.abs(key(item) - key(previous)) <= maxGap) group.push(item)\n        else groups.push([item])\n    }\n    return groups\n}\n\n/**\n * A robust central value: drop the first and last quartile, then take the mode.\n *\n * Trimming the tails is what makes this survive OCR outliers -- a single giant\n * heading would otherwise skew a plain mean. Always returns a truncated\n * integer, matching Python's `int(...)` wrapper.\n */\nexport function getSize<T>(\n    items: readonly T[],\n    key: (item: T) => number = (x) => x as unknown as number\n): number {\n    if (items.length === 0) return 0\n\n    const values = items.map(key)\n    if (values.length === 1) return values[0] as number\n\n    values.sort((a, b) => a - b)\n    if (values.length <= 4) {\n        return Math.trunc(values.reduce((sum, v) => sum + v, 0) / values.length)\n    }\n\n    const trim = Math.trunc(values.length / 4)\n    const middle = values.slice(trim, values.length - trim)\n\n    // statistics.mode returns the first most-common value, so `>` (not `>=`)\n    // keeps the earliest winner on ties.\n    const counts = new Map<number, number>()\n    for (const value of middle) counts.set(value, (counts.get(value) ?? 0) + 1)\n    let best = middle[0] as number\n    let bestCount = 0\n    for (const [value, count] of counts) {\n        if (count > bestCount) {\n            best = value\n            bestCount = count\n        }\n    }\n    return Math.trunc(best)\n}\n\n/**\n * Median-ish width of a single character, used to size inter-word gaps.\n * Takes grouped lines, matching Python's signature.\n */\nexport function getCharacterWidth(lines: readonly (readonly Box[])[]): number {\n    const widths: number[] = []\n    for (const line of lines) {\n        for (const box of line) {\n            if (box.text.length > 0) widths.push(Math.trunc(box.width / box.text.length))\n        }\n    }\n    return getSize(widths)\n}\n\n/**\n * Group boxes into text lines, each sorted left to right.\n *\n * `lineTolerance` of 0 means \"derive it\": a third of the typical character\n * height -- tight enough to keep adjacent lines apart, loose enough to tolerate\n * baseline jitter.\n */\nexport function groupBoxesIntoLines(boxes: readonly Box[], lineTolerance = 0): Box[][] {\n    if (boxes.length === 0) return []\n    const characterHeight = getSize(boxes, (b) => b.height)\n    const tolerance = lineTolerance === 0 ? characterHeight / 3 : lineTolerance\n    return cluster(boxes, tolerance, (b) => Math.trunc(b.y)).map((line) =>\n        [...line].sort((a, b) => Math.trunc(a.x) - Math.trunc(b.x))\n    )\n}\n\n/**\n * Render grouped lines back to text, preserving horizontal spacing and blank\n * lines.\n *\n * Each line is indented from x = 0, so the left margin is preserved rather than\n * every line being flushed left. Vertical gaps wider than two character heights\n * become blank lines. The running `y` starts at 0, which means the first line is\n * measured against the top of the page -- that is Python's behaviour and it is\n * what makes a top margin survive.\n */\nexport function linesToFormattedText(lines: readonly (readonly Box[])[], characterHeight: number): string {\n    const output: string[] = []\n    const spaceWidth = getCharacterWidth(lines) || 1\n    let y = 0\n\n    for (const line of lines) {\n        const first = line[0]\n        if (!first) continue\n\n        const lineDiffs = characterHeight > 0 ? Math.trunc((first.y - y) / (characterHeight * 2)) : 0\n        y = first.y\n        for (let i = 0; i < lineDiffs - 1; i++) output.push('')\n\n        let text = ''\n        let previousRight = 0\n        for (const box of line) {\n            const spaces = Math.max(Math.trunc((box.x - previousRight) / spaceWidth), 1)\n            text += ' '.repeat(spaces) + box.text\n            previousRight = box.x + box.width\n        }\n        output.push(text)\n    }\n    return output.join('\\n')\n}\n\n/** Rebuild layout-preserving text straight from a flat box list. */\nexport function boxesToFormattedText(boxes: readonly Box[], lineTolerance = 0): string {\n    if (boxes.length === 0) return ''\n    const characterHeight = getSize(boxes, (b) => b.height)\n    return linesToFormattedText(groupBoxesIntoLines(boxes, lineTolerance), characterHeight)\n}\n\n/** Plain reading-order join, the default when `keepFormatting` is off. */\nexport function boxesToText(boxes: readonly Box[], separator = ' '): string {\n    return boxes.map((b) => b.text).join(separator)\n}\n"],"mappings":";;;;;;AAoBA,SAAgB,QACZ,OACA,QACA,OAA4B,MAAM,GAC7B;CACL,IAAI,MAAM,WAAW,GAAG,OAAO,CAAC;CAChC,MAAM,SAAS,CAAC,GAAG,KAAK,CAAC,CAAC,MAAM,GAAG,MAAM,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC;CAExD,MAAM,SAAgB,CAAC,CAAC,OAAO,EAAO,CAAC;CACvC,KAAK,MAAM,QAAQ,OAAO,MAAM,CAAC,GAAG;EAChC,MAAM,QAAQ,OAAO,OAAO,SAAS;EACrC,MAAM,WAAW,MAAM,MAAM,SAAS;EACtC,IAAI,KAAK,IAAI,IAAI,IAAI,IAAI,IAAI,QAAQ,CAAC,KAAK,QAAQ,MAAM,KAAK,IAAI;OAC7D,OAAO,KAAK,CAAC,IAAI,CAAC;CAC3B;CACA,OAAO;AACX;;;;;;;;AASA,SAAgB,QACZ,OACA,OAA4B,MAAM,GAC5B;CACN,IAAI,MAAM,WAAW,GAAG,OAAO;CAE/B,MAAM,SAAS,MAAM,IAAI,GAAG;CAC5B,IAAI,OAAO,WAAW,GAAG,OAAO,OAAO;CAEvC,OAAO,MAAM,GAAG,MAAM,IAAI,CAAC;CAC3B,IAAI,OAAO,UAAU,GACjB,OAAO,KAAK,MAAM,OAAO,QAAQ,KAAK,MAAM,MAAM,GAAG,CAAC,IAAI,OAAO,MAAM;CAG3E,MAAM,OAAO,KAAK,MAAM,OAAO,SAAS,CAAC;CACzC,MAAM,SAAS,OAAO,MAAM,MAAM,OAAO,SAAS,IAAI;CAItD,MAAM,yBAAS,IAAI,IAAoB;CACvC,KAAK,MAAM,SAAS,QAAQ,OAAO,IAAI,QAAQ,OAAO,IAAI,KAAK,KAAK,KAAK,CAAC;CAC1E,IAAI,OAAO,OAAO;CAClB,IAAI,YAAY;CAChB,KAAK,MAAM,CAAC,OAAO,UAAU,QACzB,IAAI,QAAQ,WAAW;EACnB,OAAO;EACP,YAAY;CAChB;CAEJ,OAAO,KAAK,MAAM,IAAI;AAC1B;;;;;AAMA,SAAgB,kBAAkB,OAA4C;CAC1E,MAAM,SAAmB,CAAC;CAC1B,KAAK,MAAM,QAAQ,OACf,KAAK,MAAM,OAAO,MACd,IAAI,IAAI,KAAK,SAAS,GAAG,OAAO,KAAK,KAAK,MAAM,IAAI,QAAQ,IAAI,KAAK,MAAM,CAAC;CAGpF,OAAO,QAAQ,MAAM;AACzB;;;;;;;;AASA,SAAgB,oBAAoB,OAAuB,gBAAgB,GAAY;CACnF,IAAI,MAAM,WAAW,GAAG,OAAO,CAAC;CAChC,MAAM,kBAAkB,QAAQ,QAAQ,MAAM,EAAE,MAAM;CAEtD,OAAO,QAAQ,OADG,kBAAkB,IAAI,kBAAkB,IAAI,gBAC5B,MAAM,KAAK,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,SAC1D,CAAC,GAAG,IAAI,CAAC,CAAC,MAAM,GAAG,MAAM,KAAK,MAAM,EAAE,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC,CAAC,CAC9D;AACJ;;;;;;;;;;;AAYA,SAAgB,qBAAqB,OAAoC,iBAAiC;CACtG,MAAM,SAAmB,CAAC;CAC1B,MAAM,aAAa,kBAAkB,KAAK,KAAK;CAC/C,IAAI,IAAI;CAER,KAAK,MAAM,QAAQ,OAAO;EACtB,MAAM,QAAQ,KAAK;EACnB,IAAI,CAAC,OAAO;EAEZ,MAAM,YAAY,kBAAkB,IAAI,KAAK,OAAO,MAAM,IAAI,MAAM,kBAAkB,EAAE,IAAI;EAC5F,IAAI,MAAM;EACV,KAAK,IAAI,IAAI,GAAG,IAAI,YAAY,GAAG,KAAK,OAAO,KAAK,EAAE;EAEtD,IAAI,OAAO;EACX,IAAI,gBAAgB;EACpB,KAAK,MAAM,OAAO,MAAM;GACpB,MAAM,SAAS,KAAK,IAAI,KAAK,OAAO,IAAI,IAAI,iBAAiB,UAAU,GAAG,CAAC;GAC3E,QAAQ,IAAI,OAAO,MAAM,IAAI,IAAI;GACjC,gBAAgB,IAAI,IAAI,IAAI;EAChC;EACA,OAAO,KAAK,IAAI;CACpB;CACA,OAAO,OAAO,KAAK,IAAI;AAC3B;;AAGA,SAAgB,qBAAqB,OAAuB,gBAAgB,GAAW;CACnF,IAAI,MAAM,WAAW,GAAG,OAAO;CAC/B,MAAM,kBAAkB,QAAQ,QAAQ,MAAM,EAAE,MAAM;CACtD,OAAO,qBAAqB,oBAAoB,OAAO,aAAa,GAAG,eAAe;AAC1F;;AAGA,SAAgB,YAAY,OAAuB,YAAY,KAAa;CACxE,OAAO,MAAM,KAAK,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,SAAS;AAClD"}

package/dist/worker/index.d.ts

@@ -0,0 +1,105 @@
+import { d as StageDescriptor, l as Stage, s as Row } from "../pipeline-DeLO-OCE.js";
+import { n as ScaleDpConfig, t as ModelProgress } from "../config-g6IrKlDC.js";
+//#region src/worker/protocol.d.ts
+type WorkerRequest = {
+  type: 'configure';
+  requestId: number;
+  config: Partial<ScaleDpConfig>;
+} | {
+  type: 'transform';
+  requestId: number;
+  stages: StageDescriptor[];
+  rows: Row[];
+} | {
+  type: 'dispose';
+  requestId: number;
+};
+type WorkerResponse = {
+  type: 'progress';
+  requestId: number;
+  progress: ModelProgress;
+} | {
+  type: 'stage';
+  requestId: number;
+  name: string;
+  ms: number;
+  rows: number;
+} | {
+  type: 'result';
+  requestId: number;
+  rows: Row[];
+} | {
+  type: 'configured';
+  requestId: number;
+} | {
+  type: 'disposed';
+  requestId: number;
+} | {
+  type: 'error';
+  requestId: number;
+  message: string;
+};
+/**
+ * Config minus its function-valued fields.
+ *
+ * `auth` and `onProgress` are callbacks and cannot cross a postMessage
+ * boundary; the client re-implements both on the worker side.
+ */
+type TransferableConfig = Omit<Partial<ScaleDpConfig>, 'auth' | 'onProgress'>;
+//#endregion
+//#region src/worker/client.d.ts
+interface ScaleDpWorkerOptions {
+  /**
+   * The worker running `startScaleDpWorker()`. Constructed by the caller,
+   * because only the consuming app's bundler can resolve a worker entry URL:
+   *
+   *   new Worker(new URL('./my-worker.ts', import.meta.url), { type: 'module' })
+   */
+  worker: Worker;
+  onProgress?: (progress: ModelProgress) => void;
+  onStage?: (name: string, ms: number, rows: number) => void;
+}
+declare class ScaleDpWorkerClient {
+  private readonly options;
+  private nextRequestId;
+  private readonly pending;
+  private queue;
+  /**
+   * Set once the worker itself fails. Requests are posted from inside the
+   * queue, so a request enqueued before the failure would otherwise sit
+   * unposted and never settle.
+   */
+  private failure;
+  constructor(options: ScaleDpWorkerOptions);
+  private send;
+  /**
+   * Push configuration into the worker.
+   *
+   * `auth` and `onProgress` are dropped: functions cannot be cloned across a
+   * postMessage boundary. Progress is forwarded back through `onProgress` on
+   * the client instead.
+   */
+  configure(config: TransferableConfig): Promise<void>;
+  /** Run a pipeline in the worker. Stages are named, not passed by reference. */
+  transform(stages: StageDescriptor[], rows: Row[]): Promise<Row[]>;
+  dispose(): Promise<void>;
+}
+/** Convenience wrapper around the client constructor. */
+declare function createScaleDpWorker(options: ScaleDpWorkerOptions): ScaleDpWorkerClient;
+//#endregion
+//#region src/worker/host.d.ts
+/** Builds a stage from its descriptor. Consumers register the stages they use. */
+type StageFactory = (descriptor: StageDescriptor) => Stage | undefined;
+/**
+ * Register stage constructors the worker may build.
+ *
+ * Explicit registration keeps the worker bundle to the engines actually used --
+ * importing every stage would pull pdf.js, ORT, PaddleOCR and Tesseract into
+ * every worker regardless.
+ */
+declare function registerStages(stages: Record<string, new (options?: never) => Stage>): void;
+/** Start listening for pipeline requests. Call once, from a worker entry. */
+declare function startScaleDpWorker(scope?: DedicatedWorkerGlobalScope): void;
+//#endregion
+export { ScaleDpWorkerClient, type ScaleDpWorkerOptions, type StageDescriptor, type StageFactory, type TransferableConfig, type WorkerRequest, type WorkerResponse, createScaleDpWorker, registerStages, startScaleDpWorker };
+//# sourceMappingURL=index.d.ts.map

package/dist/worker/index.js

@@ -0,0 +1,180 @@
+import { f as formatException, n as Pipeline, p as configure } from "../pipeline-DACqGkpN.js";
+//#region src/worker/client.ts
+var ScaleDpWorkerClient = class {
+	options;
+	nextRequestId = 1;
+	pending = /* @__PURE__ */ new Map();
+	queue = Promise.resolve();
+	/**
+	* Set once the worker itself fails. Requests are posted from inside the
+	* queue, so a request enqueued before the failure would otherwise sit
+	* unposted and never settle.
+	*/
+	failure = null;
+	constructor(options) {
+		this.options = options;
+		options.worker.onmessage = (event) => {
+			const message = event.data;
+			if (message.type === "progress") {
+				options.onProgress?.(message.progress);
+				return;
+			}
+			if (message.type === "stage") {
+				options.onStage?.(message.name, message.ms, message.rows);
+				return;
+			}
+			const entry = this.pending.get(message.requestId);
+			if (!entry) return;
+			this.pending.delete(message.requestId);
+			if (message.type === "error") entry.reject(new Error(message.message));
+			else if (message.type === "result") entry.resolve(message.rows);
+			else entry.resolve(void 0);
+		};
+		options.worker.onerror = (event) => {
+			this.failure = /* @__PURE__ */ new Error(`Worker failed: ${event.message}`);
+			for (const [, entry] of this.pending) entry.reject(this.failure);
+			this.pending.clear();
+		};
+	}
+	send(request) {
+		const run = async () => {
+			if (this.failure) throw this.failure;
+			const requestId = this.nextRequestId++;
+			return new Promise((resolve, reject) => {
+				this.pending.set(requestId, {
+					resolve,
+					reject
+				});
+				this.options.worker.postMessage({
+					...request,
+					requestId
+				});
+			});
+		};
+		const result = this.queue.then(run, run);
+		this.queue = result.catch(() => void 0);
+		return result;
+	}
+	/**
+	* Push configuration into the worker.
+	*
+	* `auth` and `onProgress` are dropped: functions cannot be cloned across a
+	* postMessage boundary. Progress is forwarded back through `onProgress` on
+	* the client instead.
+	*/
+	configure(config) {
+		return this.send({
+			type: "configure",
+			config
+		});
+	}
+	/** Run a pipeline in the worker. Stages are named, not passed by reference. */
+	transform(stages, rows) {
+		return this.send({
+			type: "transform",
+			stages,
+			rows
+		});
+	}
+	async dispose() {
+		await this.send({ type: "dispose" }).catch(() => void 0);
+		this.options.worker.terminate();
+	}
+};
+/** Convenience wrapper around the client constructor. */
+function createScaleDpWorker(options) {
+	return new ScaleDpWorkerClient(options);
+}
+//#endregion
+//#region src/worker/host.ts
+/**
+* Worker-side pipeline host.
+*
+* Import this from your own worker entry point:
+*
+*   import { startScaleDpWorker } from '@stabrise/scaledp/worker'
+*   startScaleDpWorker()
+*
+* A worker entry has to be a real module in the consuming app so the bundler
+* can resolve it; a library cannot ship a URL that its consumer's bundler will
+* honour.
+*/
+const registry = /* @__PURE__ */ new Map();
+/**
+* Register stage constructors the worker may build.
+*
+* Explicit registration keeps the worker bundle to the engines actually used --
+* importing every stage would pull pdf.js, ORT, PaddleOCR and Tesseract into
+* every worker regardless.
+*/
+function registerStages(stages) {
+	for (const [name, Ctor] of Object.entries(stages)) registry.set(name, (options) => new Ctor(options));
+}
+function buildStage(descriptor) {
+	const factory = registry.get(descriptor.type);
+	if (!factory) throw new Error(`Stage "${descriptor.type}" is not registered in this worker. Call registerStages({ ${descriptor.type} }) in the worker entry.`);
+	return factory(descriptor.options);
+}
+/** Start listening for pipeline requests. Call once, from a worker entry. */
+function startScaleDpWorker(scope = self) {
+	const post = (message) => scope.postMessage(message);
+	let queue = Promise.resolve();
+	let active = null;
+	scope.onmessage = (event) => {
+		const message = event.data;
+		queue = queue.then(async () => {
+			try {
+				switch (message.type) {
+					case "configure":
+						configure({
+							...message.config,
+							onProgress: (progress) => post({
+								type: "progress",
+								requestId: message.requestId,
+								progress
+							})
+						});
+						post({
+							type: "configured",
+							requestId: message.requestId
+						});
+						break;
+					case "transform": {
+						await active?.dispose();
+						active = new Pipeline(message.stages.map(buildStage));
+						const rows = await active.transform(message.rows, { onStage: (name, ms, count) => post({
+							type: "stage",
+							requestId: message.requestId,
+							name,
+							ms,
+							rows: count
+						}) });
+						post({
+							type: "result",
+							requestId: message.requestId,
+							rows
+						});
+						break;
+					}
+					case "dispose":
+						await active?.dispose();
+						active = null;
+						post({
+							type: "disposed",
+							requestId: message.requestId
+						});
+				}
+			} catch (error) {
+				post({
+					type: "error",
+					requestId: message.requestId,
+					message: formatException("worker", error)
+				});
+			}
+		});
+	};
+}
+//#endregion
+export { ScaleDpWorkerClient, createScaleDpWorker, registerStages, startScaleDpWorker };
+
+//# sourceMappingURL=index.js.map

package/dist/worker/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":[],"sources":["../../src/worker/client.ts","../../src/worker/host.ts"],"sourcesContent":["/**\n * Main-thread client for a pipeline worker.\n *\n * Presents the same `transform` shape as `Pipeline` while the work happens off\n * the main thread. Everything below the stages is already OffscreenCanvas-based\n * and DOM-free, so OCR, detection and NER all move across -- not just NER, as\n * in the pdftools prototype.\n */\n\nimport type { ModelProgress, ScaleDpConfig } from '../core/config.js'\nimport type { Row } from '../core/pipeline.js'\nimport type {\n    StageDescriptor,\n    TransferableConfig,\n    WorkerRequest,\n    WorkerRequestInit,\n    WorkerResponse,\n} from './protocol.js'\n\nexport interface ScaleDpWorkerOptions {\n    /**\n     * The worker running `startScaleDpWorker()`. Constructed by the caller,\n     * because only the consuming app's bundler can resolve a worker entry URL:\n     *\n     *   new Worker(new URL('./my-worker.ts', import.meta.url), { type: 'module' })\n     */\n    worker: Worker\n    onProgress?: (progress: ModelProgress) => void\n    onStage?: (name: string, ms: number, rows: number) => void\n}\n\nexport class ScaleDpWorkerClient {\n    private nextRequestId = 1\n    private readonly pending = new Map<\n        number,\n        { resolve: (value: never) => void; reject: (reason: Error) => void }\n    >()\n\n    // ORT runs one inference at a time, so requests are serialised here as well\n    // as in the host -- this keeps the *queueing* observable to callers.\n    private queue: Promise<unknown> = Promise.resolve()\n\n    /**\n     * Set once the worker itself fails. Requests are posted from inside the\n     * queue, so a request enqueued before the failure would otherwise sit\n     * unposted and never settle.\n     */\n    private failure: Error | null = null\n\n    constructor(private readonly options: ScaleDpWorkerOptions) {\n        options.worker.onmessage = (event: MessageEvent<WorkerResponse>) => {\n            const message = event.data\n            if (message.type === 'progress') {\n                options.onProgress?.(message.progress)\n                return\n            }\n            if (message.type === 'stage') {\n                options.onStage?.(message.name, message.ms, message.rows)\n                return\n            }\n\n            const entry = this.pending.get(message.requestId)\n            if (!entry) return\n            this.pending.delete(message.requestId)\n\n            if (message.type === 'error') entry.reject(new Error(message.message))\n            else if (message.type === 'result') entry.resolve(message.rows as never)\n            else entry.resolve(undefined as never)\n        }\n\n        options.worker.onerror = (event) => {\n            this.failure = new Error(`Worker failed: ${event.message}`)\n            for (const [, entry] of this.pending) entry.reject(this.failure)\n            this.pending.clear()\n        }\n    }\n\n    private send<T>(request: WorkerRequestInit): Promise<T> {\n        const run = async (): Promise<T> => {\n            if (this.failure) throw this.failure\n            const requestId = this.nextRequestId++\n            return new Promise<T>((resolve, reject) => {\n                this.pending.set(requestId, {\n                    resolve: resolve as (value: never) => void,\n                    reject,\n                })\n                this.options.worker.postMessage({ ...request, requestId } as WorkerRequest)\n            })\n        }\n\n        // Chain onto the queue but never let a rejection break it, otherwise one\n        // failed request poisons every request that follows.\n        const result = this.queue.then(run, run)\n        this.queue = result.catch(() => undefined)\n        return result\n    }\n\n    /**\n     * Push configuration into the worker.\n     *\n     * `auth` and `onProgress` are dropped: functions cannot be cloned across a\n     * postMessage boundary. Progress is forwarded back through `onProgress` on\n     * the client instead.\n     */\n    configure(config: TransferableConfig): Promise<void> {\n        return this.send({ type: 'configure', config: config as Partial<ScaleDpConfig> })\n    }\n\n    /** Run a pipeline in the worker. Stages are named, not passed by reference. */\n    transform(stages: StageDescriptor[], rows: Row[]): Promise<Row[]> {\n        return this.send<Row[]>({ type: 'transform', stages, rows })\n    }\n\n    async dispose(): Promise<void> {\n        await this.send({ type: 'dispose' }).catch(() => undefined)\n        this.options.worker.terminate()\n    }\n}\n\n/** Convenience wrapper around the client constructor. */\nexport function createScaleDpWorker(options: ScaleDpWorkerOptions): ScaleDpWorkerClient {\n    return new ScaleDpWorkerClient(options)\n}\n","/**\n * Worker-side pipeline host.\n *\n * Import this from your own worker entry point:\n *\n *   import { startScaleDpWorker } from '@stabrise/scaledp/worker'\n *   startScaleDpWorker()\n *\n * A worker entry has to be a real module in the consuming app so the bundler\n * can resolve it; a library cannot ship a URL that its consumer's bundler will\n * honour.\n */\n\nimport { configure } from '../core/config.js'\nimport { formatException } from '../core/errors.js'\nimport { Pipeline, type Row, type Stage } from '../core/pipeline.js'\nimport type { StageDescriptor, WorkerRequest, WorkerResponse } from './protocol.js'\n\n/** Builds a stage from its descriptor. Consumers register the stages they use. */\nexport type StageFactory = (descriptor: StageDescriptor) => Stage | undefined\n\nconst registry = new Map<string, (options?: Record<string, unknown>) => Stage>()\n\n/**\n * Register stage constructors the worker may build.\n *\n * Explicit registration keeps the worker bundle to the engines actually used --\n * importing every stage would pull pdf.js, ORT, PaddleOCR and Tesseract into\n * every worker regardless.\n */\nexport function registerStages(stages: Record<string, new (options?: never) => Stage>): void {\n    for (const [name, Ctor] of Object.entries(stages)) {\n        registry.set(name, (options) => new Ctor(options as never))\n    }\n}\n\nfunction buildStage(descriptor: StageDescriptor): Stage {\n    const factory = registry.get(descriptor.type)\n    if (!factory) {\n        throw new Error(\n            `Stage \"${descriptor.type}\" is not registered in this worker. ` +\n                `Call registerStages({ ${descriptor.type} }) in the worker entry.`\n        )\n    }\n    return factory(descriptor.options)\n}\n\n/** Start listening for pipeline requests. Call once, from a worker entry. */\nexport function startScaleDpWorker(scope: DedicatedWorkerGlobalScope = self as never): void {\n    const post = (message: WorkerResponse) => scope.postMessage(message)\n\n    // ORT sessions run one inference at a time, so requests are serialised\n    // rather than run concurrently. Without this, a second transform started\n    // mid-flight fails with \"Session already started\".\n    let queue: Promise<unknown> = Promise.resolve()\n    let active: Pipeline | null = null\n\n    scope.onmessage = (event: MessageEvent<WorkerRequest>) => {\n        const message = event.data\n        queue = queue.then(async () => {\n            try {\n                switch (message.type) {\n                    case 'configure': {\n                        configure({\n                            ...message.config,\n                            onProgress: (progress) =>\n                                post({ type: 'progress', requestId: message.requestId, progress }),\n                        })\n                        post({ type: 'configured', requestId: message.requestId })\n                        break\n                    }\n                    case 'transform': {\n                        await active?.dispose()\n                        active = new Pipeline(message.stages.map(buildStage))\n                        const rows: Row[] = await active.transform(message.rows, {\n                            onStage: (name, ms, count) =>\n                                post({\n                                    type: 'stage',\n                                    requestId: message.requestId,\n                                    name,\n                                    ms,\n                                    rows: count,\n                                }),\n                        })\n                        post({ type: 'result', requestId: message.requestId, rows })\n                        break\n                    }\n                    case 'dispose': {\n                        await active?.dispose()\n                        active = null\n                        post({ type: 'disposed', requestId: message.requestId })\n                        break\n                    }\n                }\n            } catch (error) {\n                post({\n                    type: 'error',\n                    requestId: message.requestId,\n                    message: formatException('worker', error),\n                })\n            }\n        })\n    }\n}\n"],"mappings":";;AA+BA,IAAa,sBAAb,MAAiC;CAkBA;CAjB7B,gBAAwB;CACxB,0BAA2B,IAAI,IAG7B;CAIF,QAAkC,QAAQ,QAAQ;;;;;;CAOlD,UAAgC;CAEhC,YAAY,SAAgD;EAA/B,KAAA,UAAA;EACzB,QAAQ,OAAO,aAAa,UAAwC;GAChE,MAAM,UAAU,MAAM;GACtB,IAAI,QAAQ,SAAS,YAAY;IAC7B,QAAQ,aAAa,QAAQ,QAAQ;IACrC;GACJ;GACA,IAAI,QAAQ,SAAS,SAAS;IAC1B,QAAQ,UAAU,QAAQ,MAAM,QAAQ,IAAI,QAAQ,IAAI;IACxD;GACJ;GAEA,MAAM,QAAQ,KAAK,QAAQ,IAAI,QAAQ,SAAS;GAChD,IAAI,CAAC,OAAO;GACZ,KAAK,QAAQ,OAAO,QAAQ,SAAS;GAErC,IAAI,QAAQ,SAAS,SAAS,MAAM,OAAO,IAAI,MAAM,QAAQ,OAAO,CAAC;QAChE,IAAI,QAAQ,SAAS,UAAU,MAAM,QAAQ,QAAQ,IAAa;QAClE,MAAM,QAAQ,KAAA,CAAkB;EACzC;EAEA,QAAQ,OAAO,WAAW,UAAU;GAChC,KAAK,0BAAU,IAAI,MAAM,kBAAkB,MAAM,SAAS;GAC1D,KAAK,MAAM,GAAG,UAAU,KAAK,SAAS,MAAM,OAAO,KAAK,OAAO;GAC/D,KAAK,QAAQ,MAAM;EACvB;CACJ;CAEA,KAAgB,SAAwC;EACpD,MAAM,MAAM,YAAwB;GAChC,IAAI,KAAK,SAAS,MAAM,KAAK;GAC7B,MAAM,YAAY,KAAK;GACvB,OAAO,IAAI,SAAY,SAAS,WAAW;IACvC,KAAK,QAAQ,IAAI,WAAW;KACf;KACT;IACJ,CAAC;IACD,KAAK,QAAQ,OAAO,YAAY;KAAE,GAAG;KAAS;IAAU,CAAkB;GAC9E,CAAC;EACL;EAIA,MAAM,SAAS,KAAK,MAAM,KAAK,KAAK,GAAG;EACvC,KAAK,QAAQ,OAAO,YAAY,KAAA,CAAS;EACzC,OAAO;CACX;;;;;;;;CASA,UAAU,QAA2C;EACjD,OAAO,KAAK,KAAK;GAAE,MAAM;GAAqB;EAAiC,CAAC;CACpF;;CAGA,UAAU,QAA2B,MAA6B;EAC9D,OAAO,KAAK,KAAY;GAAE,MAAM;GAAa;GAAQ;EAAK,CAAC;CAC/D;CAEA,MAAM,UAAyB;EAC3B,MAAM,KAAK,KAAK,EAAE,MAAM,UAAU,CAAC,CAAC,CAAC,YAAY,KAAA,CAAS;EAC1D,KAAK,QAAQ,OAAO,UAAU;CAClC;AACJ;;AAGA,SAAgB,oBAAoB,SAAoD;CACpF,OAAO,IAAI,oBAAoB,OAAO;AAC1C;;;;;;;;;;;;;;;ACrGA,MAAM,2BAAW,IAAI,IAA0D;;;;;;;;AAS/E,SAAgB,eAAe,QAA8D;CACzF,KAAK,MAAM,CAAC,MAAM,SAAS,OAAO,QAAQ,MAAM,GAC5C,SAAS,IAAI,OAAO,YAAY,IAAI,KAAK,OAAgB,CAAC;AAElE;AAEA,SAAS,WAAW,YAAoC;CACpD,MAAM,UAAU,SAAS,IAAI,WAAW,IAAI;CAC5C,IAAI,CAAC,SACD,MAAM,IAAI,MACN,UAAU,WAAW,KAAK,4DACG,WAAW,KAAK,yBACjD;CAEJ,OAAO,QAAQ,WAAW,OAAO;AACrC;;AAGA,SAAgB,mBAAmB,QAAoC,MAAqB;CACxF,MAAM,QAAQ,YAA4B,MAAM,YAAY,OAAO;CAKnE,IAAI,QAA0B,QAAQ,QAAQ;CAC9C,IAAI,SAA0B;CAE9B,MAAM,aAAa,UAAuC;EACtD,MAAM,UAAU,MAAM;EACtB,QAAQ,MAAM,KAAK,YAAY;GAC3B,IAAI;IACA,QAAQ,QAAQ,MAAhB;KACI,KAAK;MACD,UAAU;OACN,GAAG,QAAQ;OACX,aAAa,aACT,KAAK;QAAE,MAAM;QAAY,WAAW,QAAQ;QAAW;OAAS,CAAC;MACzE,CAAC;MACD,KAAK;OAAE,MAAM;OAAc,WAAW,QAAQ;MAAU,CAAC;MACzD;KAEJ,KAAK,aAAa;MACd,MAAM,QAAQ,QAAQ;MACtB,SAAS,IAAI,SAAS,QAAQ,OAAO,IAAI,UAAU,CAAC;MACpD,MAAM,OAAc,MAAM,OAAO,UAAU,QAAQ,MAAM,EACrD,UAAU,MAAM,IAAI,UAChB,KAAK;OACD,MAAM;OACN,WAAW,QAAQ;OACnB;OACA;OACA,MAAM;MACV,CAAC,EACT,CAAC;MACD,KAAK;OAAE,MAAM;OAAU,WAAW,QAAQ;OAAW;MAAK,CAAC;MAC3D;KACJ;KACA,KAAK;MACD,MAAM,QAAQ,QAAQ;MACtB,SAAS;MACT,KAAK;OAAE,MAAM;OAAY,WAAW,QAAQ;MAAU,CAAC;IAG/D;GACJ,SAAS,OAAO;IACZ,KAAK;KACD,MAAM;KACN,WAAW,QAAQ;KACnB,SAAS,gBAAgB,UAAU,KAAK;IAC5C,CAAC;GACL;EACJ,CAAC;CACL;AACJ"}