@struktur/sdk 2.1.1 → 2.2.0

package/dist/parsers.js

@@ -0,0 +1,492 @@
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __esm = (fn, res) => function __init() {
+  return fn && (res = (0, fn[__getOwnPropNames(fn)[0]])(fn = 0)), res;
+};
+
+// src/parsers/collect.ts
+async function collectStream(stream) {
+  const reader = stream.getReader();
+  const chunks = [];
+  try {
+    while (true) {
+      const { done, value } = await reader.read();
+      if (done) {
+        break;
+      }
+      chunks.push(value);
+    }
+  } finally {
+    reader.releaseLock();
+  }
+  const totalLength = chunks.reduce((sum, chunk) => sum + chunk.length, 0);
+  const result = new Uint8Array(totalLength);
+  let offset = 0;
+  for (const chunk of chunks) {
+    result.set(chunk, offset);
+    offset += chunk.length;
+  }
+  return Buffer.from(result);
+}
+var init_collect = __esm({
+  "src/parsers/collect.ts"() {
+    "use strict";
+  }
+});
+
+// src/parsers/pdf.ts
+async function parsePdf(input, options) {
+  const buffer = Buffer.isBuffer(input) ? input : await collectStream(input);
+  const { PDFParse } = await import("pdf-parse");
+  const parser = new PDFParse({ data: buffer });
+  const textResult = await parser.getText();
+  const pageTextMap = /* @__PURE__ */ new Map();
+  if (textResult.pages.length > 0) {
+    for (const page of textResult.pages) {
+      if (page.text && page.text.trim().length > 0) {
+        pageTextMap.set(page.num, page.text);
+      }
+    }
+  }
+  let imageResult;
+  if (options?.includeImages !== false) {
+    try {
+      imageResult = await parser.getImage({ imageBuffer: false, imageDataUrl: true });
+    } catch {
+    }
+  }
+  let screenshotResult;
+  if (options?.screenshots === true) {
+    try {
+      const screenshotParams = { imageBuffer: false, imageDataUrl: true };
+      if (options.screenshotWidth !== void 0) {
+        screenshotParams.desiredWidth = options.screenshotWidth;
+      } else {
+        screenshotParams.scale = options.screenshotScale ?? 1.5;
+      }
+      screenshotResult = await parser.getScreenshot(screenshotParams);
+    } catch {
+    }
+  }
+  const pageImageMap = /* @__PURE__ */ new Map();
+  if (imageResult) {
+    for (const pageImages of imageResult.pages) {
+      const artifactImages = pageImages.images.filter((img) => img.dataUrl).map((img) => {
+        const base64 = img.dataUrl.replace(/^data:[^;]+;base64,/, "");
+        const artifactImage = {
+          type: "image",
+          base64,
+          width: img.width,
+          height: img.height,
+          imageType: "embedded"
+        };
+        return artifactImage;
+      });
+      if (artifactImages.length > 0) {
+        pageImageMap.set(pageImages.pageNumber, artifactImages);
+      }
+    }
+  }
+  if (screenshotResult) {
+    for (const screenshot of screenshotResult.pages) {
+      if (screenshot.dataUrl) {
+        const base64 = screenshot.dataUrl.replace(/^data:[^;]+;base64,/, "");
+        const artifactImage = {
+          type: "image",
+          base64,
+          width: screenshot.width,
+          height: screenshot.height,
+          imageType: "screenshot"
+        };
+        const existing = pageImageMap.get(screenshot.pageNumber) ?? [];
+        pageImageMap.set(screenshot.pageNumber, [...existing, artifactImage]);
+      }
+    }
+  }
+  let contents;
+  if (textResult.pages.length > 0) {
+    const allPageNums = /* @__PURE__ */ new Set([
+      ...pageTextMap.keys(),
+      ...pageImageMap.keys()
+    ]);
+    contents = Array.from(allPageNums).sort((a, b) => a - b).map((pageNum) => {
+      const entry = { page: pageNum };
+      const text = pageTextMap.get(pageNum);
+      if (text) entry.text = text;
+      const media = pageImageMap.get(pageNum);
+      if (media) entry.media = media;
+      return entry;
+    });
+  } else {
+    const entry = { text: textResult.text };
+    const firstPageImages = pageImageMap.size > 0 ? pageImageMap.values().next().value : void 0;
+    if (firstPageImages) entry.media = firstPageImages;
+    contents = [entry];
+  }
+  if (contents.length === 0) {
+    contents = [{ text: "" }];
+  }
+  let infoResult;
+  try {
+    infoResult = await parser.getInfo();
+  } catch {
+  }
+  await parser.destroy();
+  return {
+    id: `artifact-${crypto.randomUUID()}`,
+    type: "pdf",
+    raw: async () => buffer,
+    contents,
+    metadata: infoResult ? {
+      numpages: textResult.total,
+      info: infoResult
+    } : { numpages: textResult.total }
+  };
+}
+var init_pdf = __esm({
+  "src/parsers/pdf.ts"() {
+    "use strict";
+    init_collect();
+  }
+});
+
+// src/parsers/runner.ts
+import os from "os";
+import path2 from "path";
+import { rm, writeFile, readFile as readFile2 } from "fs/promises";
+import { exec } from "child_process";
+import { promisify } from "util";
+
+// src/validation/validator.ts
+import Ajv from "ajv";
+import addFormats from "ajv-formats";
+var SchemaValidationError = class extends Error {
+  errors;
+  constructor(message, errors) {
+    super(message);
+    this.name = "SchemaValidationError";
+    this.errors = errors;
+  }
+};
+var ARTIFACT_ID_PATTERN = /^artifact:[^/]+\/images\/image\d+\.\w+$/;
+var createAjv = () => {
+  const ajv = new Ajv({
+    allErrors: true,
+    strict: false,
+    allowUnionTypes: true
+  });
+  addFormats(ajv);
+  ajv.addFormat("artifact-id", {
+    type: "string",
+    validate: (data) => ARTIFACT_ID_PATTERN.test(data)
+  });
+  return ajv;
+};
+var validateOrThrow = (ajv, schema, data) => {
+  const validate = ajv.compile(schema);
+  const valid = validate(data);
+  if (!valid) {
+    const errors = validate.errors ?? [];
+    const message = "Schema validation failed";
+    throw new SchemaValidationError(message, errors);
+  }
+  return data;
+};
+
+// src/artifacts/input.ts
+import { readFile } from "fs/promises";
+
+// src/parsers/mime.ts
+import path from "path";
+var MAGIC_BYTES = [
+  // PDF: %PDF
+  { mimeType: "application/pdf", bytes: [37, 80, 68, 70] },
+  // PNG: 89 50 4E 47
+  { mimeType: "image/png", bytes: [137, 80, 78, 71] },
+  // JPEG: FF D8 FF
+  { mimeType: "image/jpeg", bytes: [255, 216, 255] },
+  // GIF: GIF8
+  { mimeType: "image/gif", bytes: [71, 73, 70, 56] },
+  // ZIP / Office Open XML (DOCX/XLSX/PPTX all start with PK\x03\x04)
+  {
+    mimeType: "application/zip",
+    bytes: [80, 75, 3, 4]
+  }
+];
+var isWebP = (header) => {
+  if (header.length < 12) return false;
+  const riff = header[0] === 82 && header[1] === 73 && header[2] === 70 && header[3] === 70;
+  const webp = header[8] === 87 && header[9] === 69 && header[10] === 66 && header[11] === 80;
+  return riff && webp;
+};
+var matchesMagicBytes = (header, bytes, offset = 0) => {
+  if (header.length < offset + bytes.length) return false;
+  return bytes.every((b, i) => header[offset + i] === b);
+};
+var detectFromMagicBytes = (header) => {
+  if (isWebP(header)) return "image/webp";
+  for (const { mimeType, bytes, offset } of MAGIC_BYTES) {
+    if (matchesMagicBytes(header, bytes, offset ?? 0)) {
+      return mimeType;
+    }
+  }
+  return null;
+};
+var EXTENSION_MIME_MAP = {
+  ".txt": "text/plain",
+  ".md": "text/markdown",
+  ".markdown": "text/markdown",
+  ".html": "text/html",
+  ".htm": "text/html",
+  ".json": "application/json",
+  ".pdf": "application/pdf",
+  ".png": "image/png",
+  ".jpg": "image/jpeg",
+  ".jpeg": "image/jpeg",
+  ".gif": "image/gif",
+  ".webp": "image/webp",
+  ".csv": "text/csv",
+  ".xml": "application/xml",
+  ".yaml": "application/yaml",
+  ".yml": "application/yaml",
+  ".docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+  ".xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+  ".pptx": "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+  ".mp4": "video/mp4",
+  ".mp3": "audio/mpeg",
+  ".wav": "audio/wav",
+  ".ogg": "audio/ogg",
+  ".svg": "image/svg+xml",
+  ".ts": "text/plain",
+  ".tsx": "text/plain",
+  ".js": "text/javascript",
+  ".jsx": "text/javascript",
+  ".css": "text/css",
+  ".toml": "application/toml"
+};
+async function detectMimeType(options) {
+  const { buffer, filePath, mimeOverride, npmParsers } = options;
+  if (mimeOverride) {
+    return mimeOverride;
+  }
+  if (buffer && buffer.length > 0) {
+    const header = buffer.subarray(0, 512);
+    const magicMime = detectFromMagicBytes(header);
+    if (magicMime) {
+      return magicMime;
+    }
+    if (npmParsers && npmParsers.length > 0) {
+      for (const entry of npmParsers) {
+        try {
+          const mod = await import(entry.def.package);
+          if (typeof mod.detectFileType === "function" && mod.detectFileType(header)) {
+            return entry.mimeType;
+          }
+        } catch {
+        }
+      }
+    }
+  }
+  if (filePath) {
+    const ext = path.extname(filePath).toLowerCase();
+    if (ext && ext in EXTENSION_MIME_MAP) {
+      return EXTENSION_MIME_MAP[ext] ?? null;
+    }
+  }
+  return null;
+}
+
+// src/artifacts/input.ts
+var serializedArtifactImageSchema = {
+  type: "object",
+  required: ["type"],
+  properties: {
+    type: { const: "image" },
+    url: { type: "string", minLength: 1 },
+    base64: { type: "string", minLength: 1 },
+    text: { type: "string" },
+    x: { type: "number" },
+    y: { type: "number" },
+    width: { type: "number" },
+    height: { type: "number" },
+    imageType: { enum: ["embedded", "screenshot"] }
+  },
+  additionalProperties: false,
+  anyOf: [{ required: ["url"] }, { required: ["base64"] }]
+};
+var serializedArtifactContentSchema = {
+  type: "object",
+  properties: {
+    page: { type: "number" },
+    text: { type: "string" },
+    media: { type: "array", items: serializedArtifactImageSchema }
+  },
+  additionalProperties: false,
+  anyOf: [{ required: ["text"] }, { required: ["media"] }]
+};
+var serializedArtifactSchema = {
+  type: "object",
+  required: ["id", "type", "contents"],
+  properties: {
+    id: { type: "string", minLength: 1 },
+    type: { enum: ["text", "image", "pdf", "file"] },
+    contents: { type: "array", items: serializedArtifactContentSchema },
+    metadata: { type: "object", additionalProperties: true },
+    tokens: { type: "number" }
+  },
+  additionalProperties: false
+};
+var serializedArtifactsSchema = {
+  anyOf: [
+    serializedArtifactSchema,
+    { type: "array", items: serializedArtifactSchema }
+  ]
+};
+var validateSerializedArtifacts = (data) => {
+  const ajv = createAjv();
+  const parsed = validateOrThrow(
+    ajv,
+    serializedArtifactsSchema,
+    data
+  );
+  return Array.isArray(parsed) ? parsed : [parsed];
+};
+var hydrateSerializedArtifacts = (items) => {
+  return items.map((item) => ({
+    ...item,
+    raw: async () => Buffer.from(JSON.stringify(item.contents ?? []))
+  }));
+};
+
+// src/parsers/runner.ts
+var execAsync = promisify(exec);
+var parseCommandOutput = (stdout) => {
+  let parsed;
+  try {
+    parsed = JSON.parse(stdout);
+  } catch (error) {
+    const message = error instanceof Error ? error.message : String(error);
+    throw new Error(`Parser command produced invalid JSON: ${message}
+Output: ${stdout.slice(0, 200)}`);
+  }
+  const serialized = validateSerializedArtifacts(parsed);
+  return hydrateSerializedArtifacts(serialized);
+};
+var spawnAndCapture = async (command, stdinBuffer) => {
+  if (!command.trim()) {
+    throw new Error(`Empty command: ${command}`);
+  }
+  try {
+    const options = stdinBuffer ? { input: stdinBuffer.toString(), maxBuffer: 50 * 1024 * 1024 } : { maxBuffer: 50 * 1024 * 1024 };
+    const { stdout } = await execAsync(command, options);
+    return stdout;
+  } catch (error) {
+    if (error instanceof Error && "stderr" in error) {
+      const stderr = error.stderr;
+      throw new Error(
+        `Parser command failed: ${command}
+Stderr: ${stderr?.slice(0, 500) ?? ""}`
+      );
+    }
+    throw error;
+  }
+};
+var runNpmParser = async (pkg, input, mimeType) => {
+  const mod = await import(pkg);
+  const hasParseFile = typeof mod.parseFile === "function";
+  const hasParseStream = typeof mod.parseStream === "function";
+  if (!hasParseFile && !hasParseStream) {
+    throw new Error(
+      `npm parser package "${pkg}" exports neither parseFile nor parseStream`
+    );
+  }
+  if (input.kind === "file") {
+    if (hasParseFile) {
+      return mod.parseFile(input.path, mimeType);
+    }
+    const { createReadStream } = await import("fs");
+    const { Readable } = await import("stream");
+    const nodeStream = createReadStream(input.path);
+    const stream = Readable.toWeb(nodeStream);
+    return mod.parseStream(stream, mimeType);
+  }
+  if (hasParseStream) {
+    const stream = new ReadableStream({
+      start(controller) {
+        controller.enqueue(input.buffer);
+        controller.close();
+      }
+    });
+    return mod.parseStream(stream, mimeType);
+  }
+  const tmpFile = path2.join(os.tmpdir(), `struktur-parse-${crypto.randomUUID()}`);
+  try {
+    await writeFile(tmpFile, input.buffer);
+    return await mod.parseFile(tmpFile, mimeType);
+  } finally {
+    await rm(tmpFile, { force: true });
+  }
+};
+var runCommandFileParser = async (command, input) => {
+  let filePath;
+  let tempFile = null;
+  if (input.kind === "file") {
+    filePath = input.path;
+  } else {
+    tempFile = path2.join(os.tmpdir(), `struktur-parse-${crypto.randomUUID()}`);
+    await writeFile(tempFile, input.buffer);
+    filePath = tempFile;
+  }
+  try {
+    const interpolated = command.replace(/FILE_PATH/g, filePath);
+    const stdout = await spawnAndCapture(interpolated);
+    return parseCommandOutput(stdout);
+  } finally {
+    if (tempFile) {
+      await rm(tempFile, { force: true });
+    }
+  }
+};
+var runCommandStdinParser = async (command, input) => {
+  let buffer;
+  if (input.kind === "file") {
+    buffer = await readFile2(input.path);
+  } else {
+    buffer = input.buffer;
+  }
+  const stdout = await spawnAndCapture(command, buffer);
+  return parseCommandOutput(stdout);
+};
+var runParser = async (def, input, mimeType) => {
+  switch (def.type) {
+    case "npm":
+      return runNpmParser(def.package, input, mimeType);
+    case "command-file":
+      return runCommandFileParser(def.command, input);
+    case "command-stdin":
+      return runCommandStdinParser(def.command, input);
+    case "inline": {
+      let buffer;
+      if (input.kind === "file") {
+        buffer = await readFile2(input.path);
+      } else {
+        buffer = input.buffer;
+      }
+      return [await def.handler(buffer)];
+    }
+    default: {
+      const _exhaustive = def;
+      throw new Error(`Unknown parser type: ${_exhaustive.type}`);
+    }
+  }
+};
+
+// src/parsers/index.ts
+init_collect();
+init_pdf();
+export {
+  collectStream,
+  detectMimeType,
+  parsePdf,
+  runParser
+};
+//# sourceMappingURL=parsers.js.map

package/dist/parsers.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/parsers/collect.ts","../src/parsers/pdf.ts","../src/parsers/runner.ts","../src/validation/validator.ts","../src/artifacts/input.ts","../src/parsers/mime.ts","../src/parsers/index.ts"],"sourcesContent":["/**\n * Collects a ReadableStream<Uint8Array> into a Buffer.\n * Uses Web Streams API — compatible with Bun and Node 18+.\n * Exported as a public utility for npm parser authors.\n */\nexport async function collectStream(stream: ReadableStream<Uint8Array>): Promise<Buffer> {\n  const reader = stream.getReader();\n  const chunks: Uint8Array[] = [];\n\n  try {\n    while (true) {\n      const { done, value } = await reader.read();\n      if (done) {\n        break;\n      }\n      chunks.push(value);\n    }\n  } finally {\n    reader.releaseLock();\n  }\n\n  const totalLength = chunks.reduce((sum, chunk) => sum + chunk.length, 0);\n  const result = new Uint8Array(totalLength);\n  let offset = 0;\n  for (const chunk of chunks) {\n    result.set(chunk, offset);\n    offset += chunk.length;\n  }\n\n  return Buffer.from(result);\n}\n","import type { Artifact, ArtifactContent, ArtifactImage } from \"../types\";\nimport { collectStream } from \"./collect\";\n\nexport type ParsePdfOptions = {\n  /**\n   * Whether to extract embedded images from each page and include them as\n   * base64-encoded ArtifactImage entries in the media field.\n   * Defaults to true. Pass false to skip image extraction entirely.\n   */\n  includeImages?: boolean;\n  /**\n   * Whether to render page screenshots and include them as ArtifactImage entries.\n   * When true, each page is rendered to a PNG image and added to the media field.\n   * Defaults to false.\n   */\n  screenshots?: boolean;\n  /**\n   * Scale factor for screenshots. Higher values produce larger, higher-quality images.\n   * Defaults to 1.5.\n   */\n  screenshotScale?: number;\n  /**\n   * Target width in pixels for screenshots. If specified, takes precedence over screenshotScale.\n   * Height is calculated to maintain aspect ratio.\n   */\n  screenshotWidth?: number;\n};\n\n/**\n * Built-in PDF parser using pdf-parse.\n *\n * Accepts a Buffer or ReadableStream<Uint8Array> and extracts text per-page\n * into ArtifactContent[] with page numbers set. Embedded images on each page\n * are extracted and included as base64-encoded ArtifactImage entries in the\n * media field of the corresponding content block (unless includeImages is\n * false). Returns an Artifact with type: \"pdf\".\n */\nexport async function parsePdf(\n  input: Buffer | ReadableStream<Uint8Array>,\n  options?: ParsePdfOptions,\n): Promise<Artifact> {\n  const buffer = Buffer.isBuffer(input) ? input : await collectStream(input);\n\n  // Dynamic import to avoid bundling issues\n  const { PDFParse } = await import(\"pdf-parse\");\n\n  const parser = new PDFParse({ data: buffer });\n  const textResult = await parser.getText();\n\n  // Build a page-number → text map from per-page results\n  const pageTextMap = new Map<number, string>();\n  if (textResult.pages.length > 0) {\n    for (const page of textResult.pages) {\n      if (page.text && page.text.trim().length > 0) {\n        pageTextMap.set(page.num, page.text);\n      }\n    }\n  }\n\n  // Extract embedded images unless the caller opted out.\n  // imageBuffer=false saves memory (we only need the data URL).\n  let imageResult;\n  if (options?.includeImages !== false) {\n    try {\n      imageResult = await parser.getImage({ imageBuffer: false, imageDataUrl: true });\n    } catch {\n      // Image extraction is optional — continue without images if it fails\n    }\n  }\n\n  // Render page screenshots if requested\n  let screenshotResult;\n  if (options?.screenshots === true) {\n    try {\n      const screenshotParams: {\n        imageBuffer: boolean;\n        imageDataUrl: boolean;\n        scale?: number;\n        desiredWidth?: number;\n      } = { imageBuffer: false, imageDataUrl: true };\n      \n      if (options.screenshotWidth !== undefined) {\n        screenshotParams.desiredWidth = options.screenshotWidth;\n      } else {\n        screenshotParams.scale = options.screenshotScale ?? 1.5;\n      }\n      \n      screenshotResult = await parser.getScreenshot(screenshotParams);\n    } catch {\n      // Screenshot rendering is optional — continue without screenshots if it fails\n    }\n  }\n\n  // Build a page-number → ArtifactImage[] map from extracted images\n  const pageImageMap = new Map<number, ArtifactImage[]>();\n  if (imageResult) {\n    for (const pageImages of imageResult.pages) {\n      const artifactImages: ArtifactImage[] = pageImages.images\n        .filter((img) => img.dataUrl)\n        .map((img) => {\n          // Strip the \"data:<mime>;base64,\" prefix to get the raw base64 string\n          const base64 = img.dataUrl.replace(/^data:[^;]+;base64,/, \"\");\n          const artifactImage: ArtifactImage = {\n            type: \"image\",\n            base64,\n            width: img.width,\n            height: img.height,\n            imageType: \"embedded\",\n          };\n          return artifactImage;\n        });\n      if (artifactImages.length > 0) {\n        pageImageMap.set(pageImages.pageNumber, artifactImages);\n      }\n    }\n  }\n\n  // Add screenshots to the pageImageMap\n  if (screenshotResult) {\n    for (const screenshot of screenshotResult.pages) {\n      if (screenshot.dataUrl) {\n        // Strip the \"data:<mime>;base64,\" prefix to get the raw base64 string\n        const base64 = screenshot.dataUrl.replace(/^data:[^;]+;base64,/, \"\");\n        const artifactImage: ArtifactImage = {\n          type: \"image\",\n          base64,\n          width: screenshot.width,\n          height: screenshot.height,\n          imageType: \"screenshot\",\n        };\n        // Append to existing images for this page, or create new entry\n        const existing = pageImageMap.get(screenshot.pageNumber) ?? [];\n        pageImageMap.set(screenshot.pageNumber, [...existing, artifactImage]);\n      }\n    }\n  }\n\n  let contents: ArtifactContent[];\n\n  if (textResult.pages.length > 0) {\n    // Collect all page numbers that have text or images\n    const allPageNums = new Set<number>([\n      ...pageTextMap.keys(),\n      ...pageImageMap.keys(),\n    ]);\n\n    contents = Array.from(allPageNums)\n      .sort((a, b) => a - b)\n      .map((pageNum) => {\n        const entry: ArtifactContent = { page: pageNum };\n        const text = pageTextMap.get(pageNum);\n        if (text) entry.text = text;\n        const media = pageImageMap.get(pageNum);\n        if (media) entry.media = media;\n        return entry;\n      });\n  } else {\n    // Fallback: no per-page info — use full concatenated text\n    const entry: ArtifactContent = { text: textResult.text };\n    // Attach any images from the first page if available\n    const firstPageImages = pageImageMap.size > 0\n      ? pageImageMap.values().next().value\n      : undefined;\n    if (firstPageImages) entry.media = firstPageImages;\n    contents = [entry];\n  }\n\n  // Ensure we have at least one content entry\n  if (contents.length === 0) {\n    contents = [{ text: \"\" }];\n  }\n\n  let infoResult;\n  try {\n    infoResult = await parser.getInfo();\n  } catch {\n    // Info extraction is optional\n  }\n\n  await parser.destroy();\n\n  return {\n    id: `artifact-${crypto.randomUUID()}`,\n    type: \"pdf\",\n    raw: async () => buffer,\n    contents,\n    metadata: infoResult\n      ? {\n          numpages: textResult.total,\n          info: infoResult,\n        }\n      : { numpages: textResult.total },\n  };\n}\n","import os from \"node:os\";\nimport path from \"node:path\";\nimport { rm, writeFile, readFile } from \"node:fs/promises\";\nimport { exec } from \"node:child_process\";\nimport { promisify } from \"node:util\";\nimport type { Artifact } from \"../types\";\nimport type { ParserDef, ParserInput } from \"./types\";\nimport type { NpmParserModule } from \"./npm\";\nimport {\n  hydrateSerializedArtifacts,\n  validateSerializedArtifacts,\n} from \"../artifacts/input\";\n\nconst execAsync = promisify(exec);\n\nconst parseCommandOutput = (stdout: string): Artifact[] => {\n  let parsed: unknown;\n  try {\n    parsed = JSON.parse(stdout);\n  } catch (error) {\n    const message = error instanceof Error ? error.message : String(error);\n    throw new Error(`Parser command produced invalid JSON: ${message}\\nOutput: ${stdout.slice(0, 200)}`);\n  }\n  const serialized = validateSerializedArtifacts(parsed);\n  return hydrateSerializedArtifacts(serialized);\n};\n\nconst spawnAndCapture = async (command: string, stdinBuffer?: Buffer): Promise<string> => {\n  if (!command.trim()) {\n    throw new Error(`Empty command: ${command}`);\n  }\n\n  try {\n    const options = stdinBuffer\n      ? { input: stdinBuffer.toString(), maxBuffer: 50 * 1024 * 1024 }\n      : { maxBuffer: 50 * 1024 * 1024 };\n    const { stdout } = await execAsync(command, options);\n    return stdout;\n  } catch (error) {\n    if (error instanceof Error && \"stderr\" in error) {\n      const stderr = (error as { stderr: string }).stderr;\n      throw new Error(\n        `Parser command failed: ${command}\\nStderr: ${stderr?.slice(0, 500) ?? \"\"}`\n      );\n    }\n    throw error;\n  }\n};\n\nconst runNpmParser = async (\n  pkg: string,\n  input: ParserInput,\n  mimeType: string,\n): Promise<Artifact[]> => {\n  const mod = (await import(pkg)) as NpmParserModule;\n\n  const hasParseFile = typeof mod.parseFile === \"function\";\n  const hasParseStream = typeof mod.parseStream === \"function\";\n\n  if (!hasParseFile && !hasParseStream) {\n    throw new Error(\n      `npm parser package \"${pkg}\" exports neither parseFile nor parseStream`\n    );\n  }\n\n  if (input.kind === \"file\") {\n    // Prefer parseFile for zero-copy\n    if (hasParseFile) {\n      return mod.parseFile!(input.path, mimeType);\n    }\n    // Fallback: open file as stream\n    const { createReadStream } = await import(\"node:fs\");\n    const { Readable } = await import(\"node:stream\");\n    const nodeStream = createReadStream(input.path);\n    const stream = Readable.toWeb(nodeStream) as ReadableStream<Uint8Array>;\n    return mod.parseStream!(stream, mimeType);\n  }\n\n  // input.kind === \"buffer\"\n  if (hasParseStream) {\n    // Prefer parseStream for buffers\n    const stream = new ReadableStream<Uint8Array>({\n      start(controller) {\n        controller.enqueue(input.buffer);\n        controller.close();\n      },\n    });\n    return mod.parseStream!(stream, mimeType);\n  }\n\n  // Fallback: write buffer to temp file, call parseFile, clean up\n  const tmpFile = path.join(os.tmpdir(), `struktur-parse-${crypto.randomUUID()}`);\n  try {\n    await writeFile(tmpFile, input.buffer);\n    return await mod.parseFile!(tmpFile, mimeType);\n  } finally {\n    await rm(tmpFile, { force: true });\n  }\n};\n\nconst runCommandFileParser = async (\n  command: string,\n  input: ParserInput,\n): Promise<Artifact[]> => {\n  let filePath: string;\n  let tempFile: string | null = null;\n\n  if (input.kind === \"file\") {\n    filePath = input.path;\n  } else {\n    // Write buffer to temp file\n    tempFile = path.join(os.tmpdir(), `struktur-parse-${crypto.randomUUID()}`);\n    await writeFile(tempFile, input.buffer);\n    filePath = tempFile;\n  }\n\n  try {\n    const interpolated = command.replace(/FILE_PATH/g, filePath);\n    const stdout = await spawnAndCapture(interpolated);\n    return parseCommandOutput(stdout);\n  } finally {\n    if (tempFile) {\n      await rm(tempFile, { force: true });\n    }\n  }\n};\n\nconst runCommandStdinParser = async (\n  command: string,\n  input: ParserInput,\n): Promise<Artifact[]> => {\n  let buffer: Buffer;\n\n  if (input.kind === \"file\") {\n    buffer = await readFile(input.path);\n  } else {\n    buffer = input.buffer;\n  }\n\n  const stdout = await spawnAndCapture(command, buffer);\n  return parseCommandOutput(stdout);\n};\n\nexport const runParser = async (\n  def: ParserDef,\n  input: ParserInput,\n  mimeType: string,\n): Promise<Artifact[]> => {\n  switch (def.type) {\n    case \"npm\":\n      return runNpmParser(def.package, input, mimeType);\n    case \"command-file\":\n      return runCommandFileParser(def.command, input);\n    case \"command-stdin\":\n      return runCommandStdinParser(def.command, input);\n    case \"inline\": {\n      let buffer: Buffer;\n      if (input.kind === \"file\") {\n        buffer = await readFile(input.path);\n      } else {\n        buffer = input.buffer;\n      }\n      return [await def.handler(buffer)];\n    }\n    default: {\n      const _exhaustive: never = def;\n      throw new Error(`Unknown parser type: ${(_exhaustive as { type: string }).type}`);\n    }\n  }\n};\n","import Ajv, { type AnySchema, type ErrorObject, type JSONSchemaType } from \"ajv\";\nimport addFormats from \"ajv-formats\";\n\nexport type ValidationErrors = ErrorObject[];\n\nexport type ValidationMode = 'strict' | 'lenient';\n\nexport type ValidationResult<T> = \n  | { valid: true; data: T }\n  | { valid: false; errors: ErrorObject[] };\n\nexport class SchemaValidationError extends Error {\n  public readonly errors: ValidationErrors;\n\n  constructor(message: string, errors: ValidationErrors) {\n    super(message);\n    this.name = \"SchemaValidationError\";\n    this.errors = errors;\n  }\n}\n\nconst ARTIFACT_ID_PATTERN = /^artifact:[^/]+\\/images\\/image\\d+\\.\\w+$/;\n\nexport const createAjv = () => {\n  const ajv = new Ajv({\n    allErrors: true,\n    strict: false,\n    allowUnionTypes: true,\n  });\n  addFormats(ajv);\n  \n  ajv.addFormat(\"artifact-id\", {\n    type: \"string\",\n    validate: (data: string) => ARTIFACT_ID_PATTERN.test(data),\n  });\n  \n  return ajv;\n};\n\nexport type SchemaInput<T> = JSONSchemaType<T> | AnySchema;\n\nexport const validateOrThrow = <T>(\n  ajv: Ajv,\n  schema: SchemaInput<T>,\n  data: unknown\n): T => {\n  const validate = ajv.compile<T>(schema);\n  const valid = validate(data);\n\n  if (!valid) {\n    const errors = validate.errors ?? [];\n    const message = \"Schema validation failed\";\n    throw new SchemaValidationError(message, errors);\n  }\n\n  return data as T;\n};\n\nexport const isRequiredError = (error: ErrorObject): boolean => {\n  return error.keyword === \"required\";\n};\n\nexport const validateAllowingMissingRequired = <T>(\n  ajv: Ajv,\n  schema: SchemaInput<T>,\n  data: unknown,\n  isFinalAttempt: boolean = true\n): ValidationResult<T> => {\n  const validate = ajv.compile<T>(schema);\n  const valid = validate(data);\n\n  if (valid) {\n    return { valid: true, data: data as T };\n  }\n\n  const errors = validate.errors ?? [];\n  const nonRequiredErrors = errors.filter((error) => !isRequiredError(error));\n\n  if (nonRequiredErrors.length === 0) {\n    // Only required field errors\n    // On final attempt, accept partial data\n    // On non-final attempts, return invalid to trigger retry\n    if (isFinalAttempt) {\n      return { valid: true, data: data as T };\n    }\n    return { valid: false, errors };\n  }\n\n  return { valid: false, errors: nonRequiredErrors };\n};\n","import type { Artifact, ArtifactContent, ArtifactImage, ArtifactType } from \"../types\";\nimport { createAjv, validateOrThrow } from \"../validation/validator\";\nimport { defaultArtifactProviders, type ArtifactProviders } from \"./providers\";\nimport type { ParsersConfig } from \"../parsers/types\";\nimport { runParser } from \"../parsers/runner\";\nimport type { ParsePdfOptions } from \"../parsers/pdf\";\nimport { readFile } from \"node:fs/promises\";\nimport { detectMimeType } from \"../parsers/mime\";\n\nexport type SerializedArtifactImage = Omit<ArtifactImage, \"contents\"> & {\n  contents?: never;\n};\n\nexport type SerializedArtifactContent = Omit<ArtifactContent, \"media\"> & {\n  media?: SerializedArtifactImage[];\n};\n\nexport type SerializedArtifact = Omit<Artifact, \"raw\" | \"contents\"> & {\n  contents: SerializedArtifactContent[];\n  raw?: never;\n};\n\nexport type SerializedArtifacts = SerializedArtifact | SerializedArtifact[];\n\nexport type ArtifactInput =\n  | { kind: \"artifact-json\"; data: unknown }\n  | { kind: \"text\"; text: string; id?: string }\n  | { kind: \"file\"; path: string; mimeType?: string; id?: string }\n  | { kind: \"buffer\"; buffer: Buffer; mimeType: string; id?: string };\n\nexport type ArtifactInputParser = {\n  name: string;\n  canParse: (input: ArtifactInput) => boolean;\n  parse: (\n    input: ArtifactInput,\n    options?: { \n      providers?: ArtifactProviders; \n      parsers?: ParsersConfig; \n      includeImages?: boolean;\n      screenshots?: boolean;\n      screenshotScale?: number;\n      screenshotWidth?: number;\n    }\n  ) => Promise<Artifact[]>;\n};\n\nconst serializedArtifactImageSchema = {\n  type: \"object\",\n  required: [\"type\"],\n  properties: {\n    type: { const: \"image\" },\n    url: { type: \"string\", minLength: 1 },\n    base64: { type: \"string\", minLength: 1 },\n    text: { type: \"string\" },\n    x: { type: \"number\" },\n    y: { type: \"number\" },\n    width: { type: \"number\" },\n    height: { type: \"number\" },\n    imageType: { enum: [\"embedded\", \"screenshot\"] },\n  },\n  additionalProperties: false,\n  anyOf: [{ required: [\"url\"] }, { required: [\"base64\"] }],\n};\n\nconst serializedArtifactContentSchema = {\n  type: \"object\",\n  properties: {\n    page: { type: \"number\" },\n    text: { type: \"string\" },\n    media: { type: \"array\", items: serializedArtifactImageSchema },\n  },\n  additionalProperties: false,\n  anyOf: [{ required: [\"text\"] }, { required: [\"media\"] }],\n};\n\nconst serializedArtifactSchema = {\n  type: \"object\",\n  required: [\"id\", \"type\", \"contents\"],\n  properties: {\n    id: { type: \"string\", minLength: 1 },\n    type: { enum: [\"text\", \"image\", \"pdf\", \"file\"] as ArtifactType[] },\n    contents: { type: \"array\", items: serializedArtifactContentSchema },\n    metadata: { type: \"object\", additionalProperties: true },\n    tokens: { type: \"number\" },\n  },\n  additionalProperties: false,\n};\n\nconst serializedArtifactsSchema = {\n  anyOf: [\n    serializedArtifactSchema,\n    { type: \"array\", items: serializedArtifactSchema },\n  ],\n};\n\nconst inputParsers: ArtifactInputParser[] = [];\n\nexport const registerArtifactInputParser = (parser: ArtifactInputParser) => {\n  inputParsers.push(parser);\n};\n\nexport const clearArtifactInputParsers = () => {\n  inputParsers.length = 0;\n};\n\nexport const validateSerializedArtifacts = (data: unknown): SerializedArtifact[] => {\n  const ajv = createAjv();\n  const parsed = validateOrThrow<SerializedArtifacts>(\n    ajv,\n    serializedArtifactsSchema,\n    data\n  );\n  return Array.isArray(parsed) ? parsed : [parsed];\n};\n\nexport const hydrateSerializedArtifacts = (items: SerializedArtifact[]): Artifact[] => {\n  return items.map((item) => ({\n    ...item,\n    raw: async () => Buffer.from(JSON.stringify(item.contents ?? [])),\n  }));\n};\n\nexport const parseSerializedArtifacts = (text: string): SerializedArtifact[] => {\n  const parsed = JSON.parse(text) as unknown;\n  return validateSerializedArtifacts(parsed);\n};\n\nexport const splitTextIntoContents = (text: string): ArtifactContent[] => {\n  const blocks = text\n    .split(/\\n\\s*\\n/g)\n    .map((block) => block.trim())\n    .filter((block) => block.length > 0);\n\n  if (blocks.length === 0) {\n    return [{ text }];\n  }\n\n  return blocks.map((block) => ({ text: block }));\n};\n\nconst bufferToTextArtifact = (buffer: Buffer, id?: string): Artifact => {\n  const text = buffer.toString();\n  return {\n    id: id ?? `artifact-${crypto.randomUUID()}`,\n    type: \"text\",\n    raw: async () => buffer,\n    contents: splitTextIntoContents(text),\n  };\n};\n\nconst bufferToImageArtifact = (buffer: Buffer, id?: string): Artifact => {\n  return {\n    id: id ?? `artifact-${crypto.randomUUID()}`,\n    type: \"image\",\n    raw: async () => buffer,\n    contents: [\n      {\n        media: [{ type: \"image\", contents: buffer }],\n      },\n    ],\n  };\n};\n\nconst parseBufferInput = async (\n  buffer: Buffer,\n  mimeType: string,\n  id?: string,\n  providers?: ArtifactProviders,\n  parsers?: ParsersConfig,\n  includeImages?: boolean,\n  screenshots?: boolean,\n  screenshotScale?: number,\n  screenshotWidth?: number,\n): Promise<Artifact[]> => {\n  // Resolution order:\n  // 1. parsers config (custom ParserDef) — if MIME type has a configured parser, use it\n  if (parsers) {\n    const parserDef = parsers[mimeType];\n    if (parserDef) {\n      return runParser(parserDef, { kind: \"buffer\", buffer }, mimeType);\n    }\n  }\n\n  // 2. providers registry (user-registered ArtifactProvider functions)\n  const registry = providers ?? defaultArtifactProviders;\n  const provider = registry[mimeType];\n  if (provider) {\n    return [await provider(buffer)];\n  }\n\n  // JSON auto-detection: if MIME is application/json, try to parse as SerializedArtifact[]\n  if (mimeType === \"application/json\") {\n    try {\n      const parsed = JSON.parse(buffer.toString()) as unknown;\n      const serialized = validateSerializedArtifacts(parsed);\n      return hydrateSerializedArtifacts(serialized);\n    } catch {\n      // If no custom parser is configured for application/json, throw clear error\n      throw new Error(\n        \"Input is JSON but not in SerializedArtifact format. To parse arbitrary JSON files, configure a parser: struktur config parsers add --mime application/json ...\"\n      );\n    }\n  }\n\n  // 3. Built-in PDF → pdf artifact\n  if (mimeType === \"application/pdf\") {\n    const { parsePdf } = await import(\"../parsers/pdf\");\n    const pdfOptions: ParsePdfOptions = { \n      includeImages, \n      screenshots, \n      screenshotScale, \n      screenshotWidth \n    };\n    return [await parsePdf(buffer, pdfOptions)];\n  }\n\n  // 4. Built-in text/* → text artifact\n  if (mimeType.startsWith(\"text/\")) {\n    return [bufferToTextArtifact(buffer, id)];\n  }\n\n  // 5. Built-in image/* → image artifact\n  if (mimeType.startsWith(\"image/\")) {\n    return [bufferToImageArtifact(buffer, id)];\n  }\n\n  throw new Error(`Unsupported MIME type: ${mimeType}`);\n};\n\nconst artifactJsonParser: ArtifactInputParser = {\n  name: \"artifact-json\",\n  canParse: (input) => input.kind === \"artifact-json\",\n  parse: async (input) => {\n    if (input.kind !== \"artifact-json\") {\n      return [];\n    }\n    const serialized = validateSerializedArtifacts(input.data);\n    return hydrateSerializedArtifacts(serialized);\n  },\n};\n\nconst textParser: ArtifactInputParser = {\n  name: \"text\",\n  canParse: (input) => input.kind === \"text\",\n  parse: async (input) => {\n    if (input.kind !== \"text\") {\n      return [];\n    }\n    const buffer = Buffer.from(input.text);\n    return [bufferToTextArtifact(buffer, input.id)];\n  },\n};\n\nconst fileParser: ArtifactInputParser = {\n  name: \"file\",\n  canParse: (input) => input.kind === \"file\",\n  parse: async (input, options) => {\n    if (input.kind !== \"file\") {\n      return [];\n    }\n    const mimeType = input.mimeType ?? (await detectMimeType({ filePath: input.path })) ?? \"application/octet-stream\";\n\n    // JSON auto-detection: if MIME type is application/json, first try to validate as SerializedArtifact[]\n    if (mimeType === \"application/json\") {\n      const text = await readFile(input.path, \"utf-8\");\n      try {\n        const parsed = JSON.parse(text) as unknown;\n        const serialized = validateSerializedArtifacts(parsed);\n        return hydrateSerializedArtifacts(serialized);\n      } catch {\n        // Not valid artifact JSON — try custom parser or throw\n        if (options?.parsers) {\n          const parserDef = options.parsers[mimeType];\n          if (parserDef) {\n            return runParser(parserDef, { kind: \"file\", path: input.path }, mimeType);\n          }\n        }\n        throw new Error(\n          `File \"${input.path}\" is JSON but not in SerializedArtifact format. To parse arbitrary JSON files, configure a parser: struktur config parsers add --mime application/json ...`\n        );\n      }\n    }\n\n    const buffer = await readFile(input.path);\n    return parseBufferInput(\n      buffer, \n      mimeType, \n      input.id, \n      options?.providers, \n      options?.parsers, \n      options?.includeImages,\n      options?.screenshots,\n      options?.screenshotScale,\n      options?.screenshotWidth,\n    );\n  },\n};\n\nconst bufferParser: ArtifactInputParser = {\n  name: \"buffer\",\n  canParse: (input) => input.kind === \"buffer\",\n  parse: async (input, options) => {\n    if (input.kind !== \"buffer\") {\n      return [];\n    }\n    return parseBufferInput(\n      input.buffer,\n      input.mimeType,\n      input.id,\n      options?.providers,\n      options?.parsers,\n      options?.includeImages,\n      options?.screenshots,\n      options?.screenshotScale,\n      options?.screenshotWidth,\n    );\n  },\n};\n\nexport const parse = async (\n  input: ArtifactInput,\n  options?: { \n    parsers?: ArtifactInputParser[]; \n    providers?: ArtifactProviders; \n    parserConfig?: ParsersConfig; \n    includeImages?: boolean;\n    screenshots?: boolean;\n    screenshotScale?: number;\n    screenshotWidth?: number;\n  }\n): Promise<Artifact[]> => {\n  const parsers =\n    options?.parsers ??\n    [\n      ...inputParsers,\n      artifactJsonParser,\n      textParser,\n      fileParser,\n      bufferParser,\n    ];\n  const parser = parsers.find((candidate) => candidate.canParse(input));\n\n  if (!parser) {\n    throw new Error(`No artifact input parser available for ${input.kind}`);\n  }\n\n  return parser.parse(input, { \n    providers: options?.providers, \n    parsers: options?.parserConfig, \n    includeImages: options?.includeImages,\n    screenshots: options?.screenshots,\n    screenshotScale: options?.screenshotScale,\n    screenshotWidth: options?.screenshotWidth,\n  });\n};\n","import path from \"node:path\";\nimport type { NpmParserDef } from \"./types\";\n\n// Magic byte signatures for common file types\nconst MAGIC_BYTES: Array<{ mimeType: string; bytes: number[]; offset?: number }> = [\n  // PDF: %PDF\n  { mimeType: \"application/pdf\", bytes: [0x25, 0x50, 0x44, 0x46] },\n  // PNG: 89 50 4E 47\n  { mimeType: \"image/png\", bytes: [0x89, 0x50, 0x4e, 0x47] },\n  // JPEG: FF D8 FF\n  { mimeType: \"image/jpeg\", bytes: [0xff, 0xd8, 0xff] },\n  // GIF: GIF8\n  { mimeType: \"image/gif\", bytes: [0x47, 0x49, 0x46, 0x38] },\n  // ZIP / Office Open XML (DOCX/XLSX/PPTX all start with PK\\x03\\x04)\n  {\n    mimeType: \"application/zip\",\n    bytes: [0x50, 0x4b, 0x03, 0x04],\n  },\n];\n\n// WebP has RIFF at offset 0 and WEBP at offset 8\nconst isWebP = (header: Uint8Array): boolean => {\n  if (header.length < 12) return false;\n  const riff =\n    header[0] === 0x52 && header[1] === 0x49 && header[2] === 0x46 && header[3] === 0x46;\n  const webp =\n    header[8] === 0x57 && header[9] === 0x45 && header[10] === 0x42 && header[11] === 0x50;\n  return riff && webp;\n};\n\nconst matchesMagicBytes = (header: Uint8Array, bytes: number[], offset = 0): boolean => {\n  if (header.length < offset + bytes.length) return false;\n  return bytes.every((b, i) => header[offset + i] === b);\n};\n\nconst detectFromMagicBytes = (header: Uint8Array): string | null => {\n  if (isWebP(header)) return \"image/webp\";\n\n  for (const { mimeType, bytes, offset } of MAGIC_BYTES) {\n    if (matchesMagicBytes(header, bytes, offset ?? 0)) {\n      return mimeType;\n    }\n  }\n\n  return null;\n};\n\n// Extension → MIME type lookup\nconst EXTENSION_MIME_MAP: Record<string, string> = {\n  \".txt\": \"text/plain\",\n  \".md\": \"text/markdown\",\n  \".markdown\": \"text/markdown\",\n  \".html\": \"text/html\",\n  \".htm\": \"text/html\",\n  \".json\": \"application/json\",\n  \".pdf\": \"application/pdf\",\n  \".png\": \"image/png\",\n  \".jpg\": \"image/jpeg\",\n  \".jpeg\": \"image/jpeg\",\n  \".gif\": \"image/gif\",\n  \".webp\": \"image/webp\",\n  \".csv\": \"text/csv\",\n  \".xml\": \"application/xml\",\n  \".yaml\": \"application/yaml\",\n  \".yml\": \"application/yaml\",\n  \".docx\":\n    \"application/vnd.openxmlformats-officedocument.wordprocessingml.document\",\n  \".xlsx\":\n    \"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet\",\n  \".pptx\":\n    \"application/vnd.openxmlformats-officedocument.presentationml.presentation\",\n  \".mp4\": \"video/mp4\",\n  \".mp3\": \"audio/mpeg\",\n  \".wav\": \"audio/wav\",\n  \".ogg\": \"audio/ogg\",\n  \".svg\": \"image/svg+xml\",\n  \".ts\": \"text/plain\",\n  \".tsx\": \"text/plain\",\n  \".js\": \"text/javascript\",\n  \".jsx\": \"text/javascript\",\n  \".css\": \"text/css\",\n  \".toml\": \"application/toml\",\n};\n\nexport type NpmParserEntry = {\n  mimeType: string;\n  def: NpmParserDef;\n};\n\nexport async function detectMimeType(options: {\n  buffer?: Buffer;\n  filePath?: string;\n  mimeOverride?: string;\n  npmParsers?: NpmParserEntry[];\n}): Promise<string | null> {\n  const { buffer, filePath, mimeOverride, npmParsers } = options;\n\n  // --mime override takes precedence\n  if (mimeOverride) {\n    return mimeOverride;\n  }\n\n  // Layer 1: magic bytes (authoritative)\n  if (buffer && buffer.length > 0) {\n    const header = buffer.subarray(0, 512);\n    const magicMime = detectFromMagicBytes(header);\n    if (magicMime) {\n      return magicMime;\n    }\n\n    // Layer 3: npm parser detectFileType callbacks (after built-ins)\n    if (npmParsers && npmParsers.length > 0) {\n      for (const entry of npmParsers) {\n        try {\n          const mod = await import(entry.def.package) as {\n            detectFileType?: (header: Uint8Array) => boolean;\n          };\n          if (typeof mod.detectFileType === \"function\" && mod.detectFileType(header)) {\n            return entry.mimeType;\n          }\n        } catch {\n          // If the package fails to load, skip it\n        }\n      }\n    }\n  }\n\n  // Layer 2: extension database (for file inputs)\n  if (filePath) {\n    const ext = path.extname(filePath).toLowerCase();\n    if (ext && ext in EXTENSION_MIME_MAP) {\n      return EXTENSION_MIME_MAP[ext] ?? null;\n    }\n  }\n\n  return null;\n}\n","export type { ParserDef, ParsersConfig, NpmParserDef, CommandFileDef, CommandStdinDef, InlineParserDef, ParserInput } from \"./types\";\nexport { runParser } from \"./runner\";\nexport { detectMimeType } from \"./mime\";\nexport { collectStream } from \"./collect\";\nexport { parsePdf } from \"./pdf\";\nexport type { ParsePdfOptions } from \"./pdf\";\n"],"mappings":";;;;;;AAKA,eAAsB,cAAc,QAAqD;AACvF,QAAM,SAAS,OAAO,UAAU;AAChC,QAAM,SAAuB,CAAC;AAE9B,MAAI;AACF,WAAO,MAAM;AACX,YAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,UAAI,MAAM;AACR;AAAA,MACF;AACA,aAAO,KAAK,KAAK;AAAA,IACnB;AAAA,EACF,UAAE;AACA,WAAO,YAAY;AAAA,EACrB;AAEA,QAAM,cAAc,OAAO,OAAO,CAAC,KAAK,UAAU,MAAM,MAAM,QAAQ,CAAC;AACvE,QAAM,SAAS,IAAI,WAAW,WAAW;AACzC,MAAI,SAAS;AACb,aAAW,SAAS,QAAQ;AAC1B,WAAO,IAAI,OAAO,MAAM;AACxB,cAAU,MAAM;AAAA,EAClB;AAEA,SAAO,OAAO,KAAK,MAAM;AAC3B;AA9BA;AAAA;AAAA;AAAA;AAAA;;;ACqCA,eAAsB,SACpB,OACA,SACmB;AACnB,QAAM,SAAS,OAAO,SAAS,KAAK,IAAI,QAAQ,MAAM,cAAc,KAAK;AAGzE,QAAM,EAAE,SAAS,IAAI,MAAM,OAAO,WAAW;AAE7C,QAAM,SAAS,IAAI,SAAS,EAAE,MAAM,OAAO,CAAC;AAC5C,QAAM,aAAa,MAAM,OAAO,QAAQ;AAGxC,QAAM,cAAc,oBAAI,IAAoB;AAC5C,MAAI,WAAW,MAAM,SAAS,GAAG;AAC/B,eAAW,QAAQ,WAAW,OAAO;AACnC,UAAI,KAAK,QAAQ,KAAK,KAAK,KAAK,EAAE,SAAS,GAAG;AAC5C,oBAAY,IAAI,KAAK,KAAK,KAAK,IAAI;AAAA,MACrC;AAAA,IACF;AAAA,EACF;AAIA,MAAI;AACJ,MAAI,SAAS,kBAAkB,OAAO;AACpC,QAAI;AACF,oBAAc,MAAM,OAAO,SAAS,EAAE,aAAa,OAAO,cAAc,KAAK,CAAC;AAAA,IAChF,QAAQ;AAAA,IAER;AAAA,EACF;AAGA,MAAI;AACJ,MAAI,SAAS,gBAAgB,MAAM;AACjC,QAAI;AACF,YAAM,mBAKF,EAAE,aAAa,OAAO,cAAc,KAAK;AAE7C,UAAI,QAAQ,oBAAoB,QAAW;AACzC,yBAAiB,eAAe,QAAQ;AAAA,MAC1C,OAAO;AACL,yBAAiB,QAAQ,QAAQ,mBAAmB;AAAA,MACtD;AAEA,yBAAmB,MAAM,OAAO,cAAc,gBAAgB;AAAA,IAChE,QAAQ;AAAA,IAER;AAAA,EACF;AAGA,QAAM,eAAe,oBAAI,IAA6B;AACtD,MAAI,aAAa;AACf,eAAW,cAAc,YAAY,OAAO;AAC1C,YAAM,iBAAkC,WAAW,OAChD,OAAO,CAAC,QAAQ,IAAI,OAAO,EAC3B,IAAI,CAAC,QAAQ;AAEZ,cAAM,SAAS,IAAI,QAAQ,QAAQ,uBAAuB,EAAE;AAC5D,cAAM,gBAA+B;AAAA,UACnC,MAAM;AAAA,UACN;AAAA,UACA,OAAO,IAAI;AAAA,UACX,QAAQ,IAAI;AAAA,UACZ,WAAW;AAAA,QACb;AACA,eAAO;AAAA,MACT,CAAC;AACH,UAAI,eAAe,SAAS,GAAG;AAC7B,qBAAa,IAAI,WAAW,YAAY,cAAc;AAAA,MACxD;AAAA,IACF;AAAA,EACF;AAGA,MAAI,kBAAkB;AACpB,eAAW,cAAc,iBAAiB,OAAO;AAC/C,UAAI,WAAW,SAAS;AAEtB,cAAM,SAAS,WAAW,QAAQ,QAAQ,uBAAuB,EAAE;AACnE,cAAM,gBAA+B;AAAA,UACnC,MAAM;AAAA,UACN;AAAA,UACA,OAAO,WAAW;AAAA,UAClB,QAAQ,WAAW;AAAA,UACnB,WAAW;AAAA,QACb;AAEA,cAAM,WAAW,aAAa,IAAI,WAAW,UAAU,KAAK,CAAC;AAC7D,qBAAa,IAAI,WAAW,YAAY,CAAC,GAAG,UAAU,aAAa,CAAC;AAAA,MACtE;AAAA,IACF;AAAA,EACF;AAEA,MAAI;AAEJ,MAAI,WAAW,MAAM,SAAS,GAAG;AAE/B,UAAM,cAAc,oBAAI,IAAY;AAAA,MAClC,GAAG,YAAY,KAAK;AAAA,MACpB,GAAG,aAAa,KAAK;AAAA,IACvB,CAAC;AAED,eAAW,MAAM,KAAK,WAAW,EAC9B,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC,EACpB,IAAI,CAAC,YAAY;AAChB,YAAM,QAAyB,EAAE,MAAM,QAAQ;AAC/C,YAAM,OAAO,YAAY,IAAI,OAAO;AACpC,UAAI,KAAM,OAAM,OAAO;AACvB,YAAM,QAAQ,aAAa,IAAI,OAAO;AACtC,UAAI,MAAO,OAAM,QAAQ;AACzB,aAAO;AAAA,IACT,CAAC;AAAA,EACL,OAAO;AAEL,UAAM,QAAyB,EAAE,MAAM,WAAW,KAAK;AAEvD,UAAM,kBAAkB,aAAa,OAAO,IACxC,aAAa,OAAO,EAAE,KAAK,EAAE,QAC7B;AACJ,QAAI,gBAAiB,OAAM,QAAQ;AACnC,eAAW,CAAC,KAAK;AAAA,EACnB;AAGA,MAAI,SAAS,WAAW,GAAG;AACzB,eAAW,CAAC,EAAE,MAAM,GAAG,CAAC;AAAA,EAC1B;AAEA,MAAI;AACJ,MAAI;AACF,iBAAa,MAAM,OAAO,QAAQ;AAAA,EACpC,QAAQ;AAAA,EAER;AAEA,QAAM,OAAO,QAAQ;AAErB,SAAO;AAAA,IACL,IAAI,YAAY,OAAO,WAAW,CAAC;AAAA,IACnC,MAAM;AAAA,IACN,KAAK,YAAY;AAAA,IACjB;AAAA,IACA,UAAU,aACN;AAAA,MACE,UAAU,WAAW;AAAA,MACrB,MAAM;AAAA,IACR,IACA,EAAE,UAAU,WAAW,MAAM;AAAA,EACnC;AACF;AAjMA;AAAA;AAAA;AACA;AAAA;AAAA;;;ACDA,OAAO,QAAQ;AACf,OAAOA,WAAU;AACjB,SAAS,IAAI,WAAW,YAAAC,iBAAgB;AACxC,SAAS,YAAY;AACrB,SAAS,iBAAiB;;;ACJ1B,OAAO,SAAoE;AAC3E,OAAO,gBAAgB;AAUhB,IAAM,wBAAN,cAAoC,MAAM;AAAA,EAC/B;AAAA,EAEhB,YAAY,SAAiB,QAA0B;AACrD,UAAM,OAAO;AACb,SAAK,OAAO;AACZ,SAAK,SAAS;AAAA,EAChB;AACF;AAEA,IAAM,sBAAsB;AAErB,IAAM,YAAY,MAAM;AAC7B,QAAM,MAAM,IAAI,IAAI;AAAA,IAClB,WAAW;AAAA,IACX,QAAQ;AAAA,IACR,iBAAiB;AAAA,EACnB,CAAC;AACD,aAAW,GAAG;AAEd,MAAI,UAAU,eAAe;AAAA,IAC3B,MAAM;AAAA,IACN,UAAU,CAAC,SAAiB,oBAAoB,KAAK,IAAI;AAAA,EAC3D,CAAC;AAED,SAAO;AACT;AAIO,IAAM,kBAAkB,CAC7B,KACA,QACA,SACM;AACN,QAAM,WAAW,IAAI,QAAW,MAAM;AACtC,QAAM,QAAQ,SAAS,IAAI;AAE3B,MAAI,CAAC,OAAO;AACV,UAAM,SAAS,SAAS,UAAU,CAAC;AACnC,UAAM,UAAU;AAChB,UAAM,IAAI,sBAAsB,SAAS,MAAM;AAAA,EACjD;AAEA,SAAO;AACT;;;AClDA,SAAS,gBAAgB;;;ACNzB,OAAO,UAAU;AAIjB,IAAM,cAA6E;AAAA;AAAA,EAEjF,EAAE,UAAU,mBAAmB,OAAO,CAAC,IAAM,IAAM,IAAM,EAAI,EAAE;AAAA;AAAA,EAE/D,EAAE,UAAU,aAAa,OAAO,CAAC,KAAM,IAAM,IAAM,EAAI,EAAE;AAAA;AAAA,EAEzD,EAAE,UAAU,cAAc,OAAO,CAAC,KAAM,KAAM,GAAI,EAAE;AAAA;AAAA,EAEpD,EAAE,UAAU,aAAa,OAAO,CAAC,IAAM,IAAM,IAAM,EAAI,EAAE;AAAA;AAAA,EAEzD;AAAA,IACE,UAAU;AAAA,IACV,OAAO,CAAC,IAAM,IAAM,GAAM,CAAI;AAAA,EAChC;AACF;AAGA,IAAM,SAAS,CAAC,WAAgC;AAC9C,MAAI,OAAO,SAAS,GAAI,QAAO;AAC/B,QAAM,OACJ,OAAO,CAAC,MAAM,MAAQ,OAAO,CAAC,MAAM,MAAQ,OAAO,CAAC,MAAM,MAAQ,OAAO,CAAC,MAAM;AAClF,QAAM,OACJ,OAAO,CAAC,MAAM,MAAQ,OAAO,CAAC,MAAM,MAAQ,OAAO,EAAE,MAAM,MAAQ,OAAO,EAAE,MAAM;AACpF,SAAO,QAAQ;AACjB;AAEA,IAAM,oBAAoB,CAAC,QAAoB,OAAiB,SAAS,MAAe;AACtF,MAAI,OAAO,SAAS,SAAS,MAAM,OAAQ,QAAO;AAClD,SAAO,MAAM,MAAM,CAAC,GAAG,MAAM,OAAO,SAAS,CAAC,MAAM,CAAC;AACvD;AAEA,IAAM,uBAAuB,CAAC,WAAsC;AAClE,MAAI,OAAO,MAAM,EAAG,QAAO;AAE3B,aAAW,EAAE,UAAU,OAAO,OAAO,KAAK,aAAa;AACrD,QAAI,kBAAkB,QAAQ,OAAO,UAAU,CAAC,GAAG;AACjD,aAAO;AAAA,IACT;AAAA,EACF;AAEA,SAAO;AACT;AAGA,IAAM,qBAA6C;AAAA,EACjD,QAAQ;AAAA,EACR,OAAO;AAAA,EACP,aAAa;AAAA,EACb,SAAS;AAAA,EACT,QAAQ;AAAA,EACR,SAAS;AAAA,EACT,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,SAAS;AAAA,EACT,QAAQ;AAAA,EACR,SAAS;AAAA,EACT,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,SAAS;AAAA,EACT,QAAQ;AAAA,EACR,SACE;AAAA,EACF,SACE;AAAA,EACF,SACE;AAAA,EACF,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,OAAO;AAAA,EACP,QAAQ;AAAA,EACR,OAAO;AAAA,EACP,QAAQ;AAAA,EACR,QAAQ;AAAA,EACR,SAAS;AACX;AAOA,eAAsB,eAAe,SAKV;AACzB,QAAM,EAAE,QAAQ,UAAU,cAAc,WAAW,IAAI;AAGvD,MAAI,cAAc;AAChB,WAAO;AAAA,EACT;AAGA,MAAI,UAAU,OAAO,SAAS,GAAG;AAC/B,UAAM,SAAS,OAAO,SAAS,GAAG,GAAG;AACrC,UAAM,YAAY,qBAAqB,MAAM;AAC7C,QAAI,WAAW;AACb,aAAO;AAAA,IACT;AAGA,QAAI,cAAc,WAAW,SAAS,GAAG;AACvC,iBAAW,SAAS,YAAY;AAC9B,YAAI;AACF,gBAAM,MAAM,MAAM,OAAO,MAAM,IAAI;AAGnC,cAAI,OAAO,IAAI,mBAAmB,cAAc,IAAI,eAAe,MAAM,GAAG;AAC1E,mBAAO,MAAM;AAAA,UACf;AAAA,QACF,QAAQ;AAAA,QAER;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAGA,MAAI,UAAU;AACZ,UAAM,MAAM,KAAK,QAAQ,QAAQ,EAAE,YAAY;AAC/C,QAAI,OAAO,OAAO,oBAAoB;AACpC,aAAO,mBAAmB,GAAG,KAAK;AAAA,IACpC;AAAA,EACF;AAEA,SAAO;AACT;;;AD1FA,IAAM,gCAAgC;AAAA,EACpC,MAAM;AAAA,EACN,UAAU,CAAC,MAAM;AAAA,EACjB,YAAY;AAAA,IACV,MAAM,EAAE,OAAO,QAAQ;AAAA,IACvB,KAAK,EAAE,MAAM,UAAU,WAAW,EAAE;AAAA,IACpC,QAAQ,EAAE,MAAM,UAAU,WAAW,EAAE;AAAA,IACvC,MAAM,EAAE,MAAM,SAAS;AAAA,IACvB,GAAG,EAAE,MAAM,SAAS;AAAA,IACpB,GAAG,EAAE,MAAM,SAAS;AAAA,IACpB,OAAO,EAAE,MAAM,SAAS;AAAA,IACxB,QAAQ,EAAE,MAAM,SAAS;AAAA,IACzB,WAAW,EAAE,MAAM,CAAC,YAAY,YAAY,EAAE;AAAA,EAChD;AAAA,EACA,sBAAsB;AAAA,EACtB,OAAO,CAAC,EAAE,UAAU,CAAC,KAAK,EAAE,GAAG,EAAE,UAAU,CAAC,QAAQ,EAAE,CAAC;AACzD;AAEA,IAAM,kCAAkC;AAAA,EACtC,MAAM;AAAA,EACN,YAAY;AAAA,IACV,MAAM,EAAE,MAAM,SAAS;AAAA,IACvB,MAAM,EAAE,MAAM,SAAS;AAAA,IACvB,OAAO,EAAE,MAAM,SAAS,OAAO,8BAA8B;AAAA,EAC/D;AAAA,EACA,sBAAsB;AAAA,EACtB,OAAO,CAAC,EAAE,UAAU,CAAC,MAAM,EAAE,GAAG,EAAE,UAAU,CAAC,OAAO,EAAE,CAAC;AACzD;AAEA,IAAM,2BAA2B;AAAA,EAC/B,MAAM;AAAA,EACN,UAAU,CAAC,MAAM,QAAQ,UAAU;AAAA,EACnC,YAAY;AAAA,IACV,IAAI,EAAE,MAAM,UAAU,WAAW,EAAE;AAAA,IACnC,MAAM,EAAE,MAAM,CAAC,QAAQ,SAAS,OAAO,MAAM,EAAoB;AAAA,IACjE,UAAU,EAAE,MAAM,SAAS,OAAO,gCAAgC;AAAA,IAClE,UAAU,EAAE,MAAM,UAAU,sBAAsB,KAAK;AAAA,IACvD,QAAQ,EAAE,MAAM,SAAS;AAAA,EAC3B;AAAA,EACA,sBAAsB;AACxB;AAEA,IAAM,4BAA4B;AAAA,EAChC,OAAO;AAAA,IACL;AAAA,IACA,EAAE,MAAM,SAAS,OAAO,yBAAyB;AAAA,EACnD;AACF;AAYO,IAAM,8BAA8B,CAAC,SAAwC;AAClF,QAAM,MAAM,UAAU;AACtB,QAAM,SAAS;AAAA,IACb;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACA,SAAO,MAAM,QAAQ,MAAM,IAAI,SAAS,CAAC,MAAM;AACjD;AAEO,IAAM,6BAA6B,CAAC,UAA4C;AACrF,SAAO,MAAM,IAAI,CAAC,UAAU;AAAA,IAC1B,GAAG;AAAA,IACH,KAAK,YAAY,OAAO,KAAK,KAAK,UAAU,KAAK,YAAY,CAAC,CAAC,CAAC;AAAA,EAClE,EAAE;AACJ;;;AF3GA,IAAM,YAAY,UAAU,IAAI;AAEhC,IAAM,qBAAqB,CAAC,WAA+B;AACzD,MAAI;AACJ,MAAI;AACF,aAAS,KAAK,MAAM,MAAM;AAAA,EAC5B,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,IAAI,MAAM,yCAAyC,OAAO;AAAA,UAAa,OAAO,MAAM,GAAG,GAAG,CAAC,EAAE;AAAA,EACrG;AACA,QAAM,aAAa,4BAA4B,MAAM;AACrD,SAAO,2BAA2B,UAAU;AAC9C;AAEA,IAAM,kBAAkB,OAAO,SAAiB,gBAA0C;AACxF,MAAI,CAAC,QAAQ,KAAK,GAAG;AACnB,UAAM,IAAI,MAAM,kBAAkB,OAAO,EAAE;AAAA,EAC7C;AAEA,MAAI;AACF,UAAM,UAAU,cACZ,EAAE,OAAO,YAAY,SAAS,GAAG,WAAW,KAAK,OAAO,KAAK,IAC7D,EAAE,WAAW,KAAK,OAAO,KAAK;AAClC,UAAM,EAAE,OAAO,IAAI,MAAM,UAAU,SAAS,OAAO;AACnD,WAAO;AAAA,EACT,SAAS,OAAO;AACd,QAAI,iBAAiB,SAAS,YAAY,OAAO;AAC/C,YAAM,SAAU,MAA6B;AAC7C,YAAM,IAAI;AAAA,QACR,0BAA0B,OAAO;AAAA,UAAa,QAAQ,MAAM,GAAG,GAAG,KAAK,EAAE;AAAA,MAC3E;AAAA,IACF;AACA,UAAM;AAAA,EACR;AACF;AAEA,IAAM,eAAe,OACnB,KACA,OACA,aACwB;AACxB,QAAM,MAAO,MAAM,OAAO;AAE1B,QAAM,eAAe,OAAO,IAAI,cAAc;AAC9C,QAAM,iBAAiB,OAAO,IAAI,gBAAgB;AAElD,MAAI,CAAC,gBAAgB,CAAC,gBAAgB;AACpC,UAAM,IAAI;AAAA,MACR,uBAAuB,GAAG;AAAA,IAC5B;AAAA,EACF;AAEA,MAAI,MAAM,SAAS,QAAQ;AAEzB,QAAI,cAAc;AAChB,aAAO,IAAI,UAAW,MAAM,MAAM,QAAQ;AAAA,IAC5C;AAEA,UAAM,EAAE,iBAAiB,IAAI,MAAM,OAAO,IAAS;AACnD,UAAM,EAAE,SAAS,IAAI,MAAM,OAAO,QAAa;AAC/C,UAAM,aAAa,iBAAiB,MAAM,IAAI;AAC9C,UAAM,SAAS,SAAS,MAAM,UAAU;AACxC,WAAO,IAAI,YAAa,QAAQ,QAAQ;AAAA,EAC1C;AAGA,MAAI,gBAAgB;AAElB,UAAM,SAAS,IAAI,eAA2B;AAAA,MAC5C,MAAM,YAAY;AAChB,mBAAW,QAAQ,MAAM,MAAM;AAC/B,mBAAW,MAAM;AAAA,MACnB;AAAA,IACF,CAAC;AACD,WAAO,IAAI,YAAa,QAAQ,QAAQ;AAAA,EAC1C;AAGA,QAAM,UAAUC,MAAK,KAAK,GAAG,OAAO,GAAG,kBAAkB,OAAO,WAAW,CAAC,EAAE;AAC9E,MAAI;AACF,UAAM,UAAU,SAAS,MAAM,MAAM;AACrC,WAAO,MAAM,IAAI,UAAW,SAAS,QAAQ;AAAA,EAC/C,UAAE;AACA,UAAM,GAAG,SAAS,EAAE,OAAO,KAAK,CAAC;AAAA,EACnC;AACF;AAEA,IAAM,uBAAuB,OAC3B,SACA,UACwB;AACxB,MAAI;AACJ,MAAI,WAA0B;AAE9B,MAAI,MAAM,SAAS,QAAQ;AACzB,eAAW,MAAM;AAAA,EACnB,OAAO;AAEL,eAAWA,MAAK,KAAK,GAAG,OAAO,GAAG,kBAAkB,OAAO,WAAW,CAAC,EAAE;AACzE,UAAM,UAAU,UAAU,MAAM,MAAM;AACtC,eAAW;AAAA,EACb;AAEA,MAAI;AACF,UAAM,eAAe,QAAQ,QAAQ,cAAc,QAAQ;AAC3D,UAAM,SAAS,MAAM,gBAAgB,YAAY;AACjD,WAAO,mBAAmB,MAAM;AAAA,EAClC,UAAE;AACA,QAAI,UAAU;AACZ,YAAM,GAAG,UAAU,EAAE,OAAO,KAAK,CAAC;AAAA,IACpC;AAAA,EACF;AACF;AAEA,IAAM,wBAAwB,OAC5B,SACA,UACwB;AACxB,MAAI;AAEJ,MAAI,MAAM,SAAS,QAAQ;AACzB,aAAS,MAAMC,UAAS,MAAM,IAAI;AAAA,EACpC,OAAO;AACL,aAAS,MAAM;AAAA,EACjB;AAEA,QAAM,SAAS,MAAM,gBAAgB,SAAS,MAAM;AACpD,SAAO,mBAAmB,MAAM;AAClC;AAEO,IAAM,YAAY,OACvB,KACA,OACA,aACwB;AACxB,UAAQ,IAAI,MAAM;AAAA,IAChB,KAAK;AACH,aAAO,aAAa,IAAI,SAAS,OAAO,QAAQ;AAAA,IAClD,KAAK;AACH,aAAO,qBAAqB,IAAI,SAAS,KAAK;AAAA,IAChD,KAAK;AACH,aAAO,sBAAsB,IAAI,SAAS,KAAK;AAAA,IACjD,KAAK,UAAU;AACb,UAAI;AACJ,UAAI,MAAM,SAAS,QAAQ;AACzB,iBAAS,MAAMA,UAAS,MAAM,IAAI;AAAA,MACpC,OAAO;AACL,iBAAS,MAAM;AAAA,MACjB;AACA,aAAO,CAAC,MAAM,IAAI,QAAQ,MAAM,CAAC;AAAA,IACnC;AAAA,IACA,SAAS;AACP,YAAM,cAAqB;AAC3B,YAAM,IAAI,MAAM,wBAAyB,YAAiC,IAAI,EAAE;AAAA,IAClF;AAAA,EACF;AACF;;;AItKA;AACA;","names":["path","readFile","path","readFile"]}