@yolk-sdk/extractors 0.1.0-canary.98

package/dist/node/xlsx-text.mjs

@@ -0,0 +1,181 @@
+import { FileExtractionError } from "../errors.mjs";
+import { cellCount, encodeCellAddress, parseCellRange } from "./xlsx-range.mjs";
+import { makeHyperlinkLookup } from "./xlsx-hyperlinks.mjs";
+import { Predicate } from "effect";
+//#region src/node/xlsx-text.ts
+const outputLimitReason = "output limit";
+const hyperlinkLimitReason = "hyperlink limit";
+const omittedHyperlinksMarker = (reasons) => `\n\n[Some hyperlinks omitted: ${reasons.join(" and ")}]`;
+/**
+* Space reserved for the marker whenever a workbook has hyperlinks, so it always fits.
+* `maxXlsxTextCharacters` must exceed it (`minimumXlsxTextCharacters`).
+*/
+const omittedHyperlinksMarkerReserve = omittedHyperlinksMarker([outputLimitReason, hyperlinkLimitReason]).length;
+const invalidRange = () => new FileExtractionError({
+	format: "xlsx",
+	message: "Invalid XLSX worksheet range."
+});
+const tooLarge = () => new FileExtractionError({
+	format: "xlsx",
+	message: "XLSX exceeds the worksheet or cell-visit limit."
+});
+const outputTooLarge = () => new FileExtractionError({
+	format: "xlsx",
+	message: "XLSX extracted text exceeds the output limit."
+});
+/** Read an own property without walking the prototype chain (or trusting a polluted one). */
+const ownProperty = (target, key) => {
+	const descriptor = Object.getOwnPropertyDescriptor(target, key);
+	if (descriptor === void 0) return void 0;
+	return "value" in descriptor ? descriptor.value : descriptor.get?.call(target);
+};
+const cellText = (cell) => {
+	if (ownProperty(cell, "t") === "z") return "";
+	const value = ownProperty(cell, "v");
+	if (value === void 0 || value === null) return "";
+	const display = ownProperty(cell, "w");
+	if (Predicate.isString(display)) return display;
+	if (Predicate.isString(value)) return value;
+	if (Predicate.isBoolean(value)) return value ? "TRUE" : "FALSE";
+	if (value instanceof Date && Number.isFinite(value.getTime())) return value.toISOString();
+	if (Predicate.isNumber(value) && Number.isFinite(value)) return String(value);
+	throw new FileExtractionError({
+		format: "xlsx",
+		message: "Invalid XLSX cell value."
+	});
+};
+/**
+* CSV length of `text` (quotes doubled, wrapped when needed). The scan stops once the length
+* exceeds `limit`; the returned length is then only known to be above it.
+*/
+const csvField = (text, limit = Number.POSITIVE_INFINITY) => {
+	let quoteCount = 0;
+	let quote = text === "ID";
+	for (let index = 0; index < text.length; index += 1) {
+		const character = text.charCodeAt(index);
+		if (character === 34) quoteCount += 1;
+		if (character === 34 || character === 44 || character === 10 || character === 13) {
+			quote = true;
+			if (text.length + quoteCount + 2 > limit) break;
+		}
+	}
+	return {
+		quote,
+		length: text.length + quoteCount + (quote ? 2 : 0)
+	};
+};
+/** Write bounded CSV for every sheet; throws once `budget` characters would be exceeded. */
+const renderSheets = (sheets, budget, decorate) => {
+	let characters = 0;
+	const output = [];
+	const append = (text) => {
+		if (text.length > budget - characters) throw outputTooLarge();
+		characters += text.length;
+		output.push(text);
+	};
+	const appendCell = (text) => {
+		if (text.length > budget - characters) throw outputTooLarge();
+		const field = csvField(text, budget - characters);
+		if (field.length > budget - characters) throw outputTooLarge();
+		append(field.quote ? `"${text.replaceAll("\"", "\"\"")}"` : text);
+	};
+	for (const visited of sheets) {
+		const { name, sheet, range } = visited;
+		if (sheet === void 0) continue;
+		if (output.length > 0) append("\n\n");
+		append("# ");
+		append(name);
+		append("\n");
+		if (range === void 0) continue;
+		for (let row = range.start.r; row <= range.end.r; row += 1) {
+			if (row > range.start.r) append("\n");
+			for (let column = range.start.c; column <= range.end.c; column += 1) {
+				if (column > range.start.c) append(",");
+				const cell = ownProperty(sheet, encodeCellAddress({
+					r: row,
+					c: column
+				}));
+				if (!Predicate.isObject(cell)) {
+					appendCell("");
+					continue;
+				}
+				const text = cellText(cell);
+				appendCell(decorate?.({
+					sheet: visited,
+					row,
+					column,
+					text
+				}) ?? text);
+			}
+		}
+	}
+	return output.join("");
+};
+/**
+* Validate every sheet before touching any cell, then generate bounded CSV incrementally, never
+* with `sheet_to_csv` (which can walk billions of absent cells or allocate an unbounded quoted
+* string).
+*
+* Existing cells inside an external hyperlink are written as `text <url>`. Plain text is
+* rendered first, so links only use budget the plain text leaves over: in document order, an
+* annotation that no longer fits leaves its cell plain and one marker ends the output.
+*/
+const extractBoundedXlsxText = (workbook, limits, options = {}) => {
+	if (workbook.SheetNames.length > limits.maxXlsxSheets) throw tooLarge();
+	let visits = 0;
+	const sheets = workbook.SheetNames.map((name) => {
+		const candidate = ownProperty(workbook.Sheets, name);
+		const sheet = Predicate.isObject(candidate) ? candidate : void 0;
+		const ref = sheet === void 0 ? void 0 : ownProperty(sheet, "!ref");
+		if (ref !== void 0 && !Predicate.isString(ref)) throw invalidRange();
+		const range = ref === void 0 ? void 0 : parseCellRange(ref);
+		if (ref !== void 0 && range === void 0) throw invalidRange();
+		visits += range === void 0 ? 0 : cellCount(range);
+		if (!Number.isSafeInteger(visits) || visits > limits.maxXlsxCellVisits) throw tooLarge();
+		return {
+			name,
+			sheet,
+			range
+		};
+	});
+	const links = options.hyperlinks ?? /* @__PURE__ */ new Map();
+	const truncated = options.hyperlinksTruncated === true;
+	if (links.size === 0 && !truncated) return renderSheets(sheets, limits.maxXlsxTextCharacters);
+	const budget = limits.maxXlsxTextCharacters - omittedHyperlinksMarkerReserve;
+	let spare = budget - renderSheets(sheets, budget).length;
+	let dropped = false;
+	const lookups = /* @__PURE__ */ new Map();
+	const lookupFor = (sheet) => {
+		const sheetLinks = links.get(sheet.name);
+		if (sheetLinks === void 0 || sheet.range === void 0) return void 0;
+		const existing = lookups.get(sheet.name);
+		if (existing !== void 0) return existing;
+		const lookup = makeHyperlinkLookup(sheetLinks, sheet.range);
+		lookups.set(sheet.name, lookup);
+		return lookup;
+	};
+	const output = renderSheets(sheets, budget, ({ sheet, row, column, text }) => {
+		const link = lookupFor(sheet)?.at(row, column);
+		if (link === void 0 || link.target === text) return void 0;
+		const label = text.length > 0 ? text : link.display ?? "";
+		const plainLength = csvField(text).length;
+		if (label.length + (label.length > 0 ? 3 : 2) + link.target.length - plainLength > spare) {
+			dropped = true;
+			return;
+		}
+		const annotated = label.length > 0 ? `${label} <${link.target}>` : `<${link.target}>`;
+		const extra = csvField(annotated, plainLength + spare).length - plainLength;
+		if (extra > spare) {
+			dropped = true;
+			return;
+		}
+		spare -= extra;
+		return annotated;
+	});
+	const reasons = [...dropped ? [outputLimitReason] : [], ...truncated ? [hyperlinkLimitReason] : []];
+	return reasons.length === 0 ? output : output + omittedHyperlinksMarker(reasons);
+};
+//#endregion
+export { extractBoundedXlsxText, omittedHyperlinksMarkerReserve };
+
+//# sourceMappingURL=xlsx-text.mjs.map

package/dist/node/xlsx-text.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-text.mjs","names":[],"sources":["../../src/node/xlsx-text.ts"],"sourcesContent":["import { Predicate } from 'effect'\nimport { FileExtractionError } from '../errors.ts'\nimport type { FileExtractorLimits } from '../limits.ts'\nimport { makeHyperlinkLookup } from './xlsx-hyperlinks.ts'\nimport type { HyperlinkLookup, XlsxHyperlinks } from './xlsx-hyperlinks.ts'\nimport type { CellRange } from './xlsx-range.ts'\nimport { cellCount, encodeCellAddress, parseCellRange } from './xlsx-range.ts'\n\nexport type XlsxTextLimits = Pick<\n  FileExtractorLimits,\n  'maxXlsxCellVisits' | 'maxXlsxSheets' | 'maxXlsxTextCharacters'\n>\n\n/** The parsed workbook as SheetJS returns it; sheets and cells are read as own properties. */\nexport type XlsxWorkbook = {\n  readonly SheetNames: ReadonlyArray<string>\n  readonly Sheets: object\n}\n\nexport type XlsxTextOptions = {\n  readonly hyperlinks?: XlsxHyperlinks\n  /** Some hyperlinks were never read because the workbook exceeded `maxXlsxHyperlinks`. */\n  readonly hyperlinksTruncated?: boolean\n}\n\nconst outputLimitReason = 'output limit'\n\nconst hyperlinkLimitReason = 'hyperlink limit'\n\nconst omittedHyperlinksMarker = (reasons: ReadonlyArray<string>) =>\n  `\\n\\n[Some hyperlinks omitted: ${reasons.join(' and ')}]`\n\n/**\n * Space reserved for the marker whenever a workbook has hyperlinks, so it always fits.\n * `maxXlsxTextCharacters` must exceed it (`minimumXlsxTextCharacters`).\n */\nexport const omittedHyperlinksMarkerReserve = omittedHyperlinksMarker([\n  outputLimitReason,\n  hyperlinkLimitReason\n]).length\n\nconst invalidRange = () =>\n  new FileExtractionError({ format: 'xlsx', message: 'Invalid XLSX worksheet range.' })\n\nconst tooLarge = () =>\n  new FileExtractionError({\n    format: 'xlsx',\n    message: 'XLSX exceeds the worksheet or cell-visit limit.'\n  })\n\nconst outputTooLarge = () =>\n  new FileExtractionError({\n    format: 'xlsx',\n    message: 'XLSX extracted text exceeds the output limit.'\n  })\n\n/** Read an own property without walking the prototype chain (or trusting a polluted one). */\nconst ownProperty = (target: object, key: string): unknown => {\n  const descriptor = Object.getOwnPropertyDescriptor(target, key)\n\n  if (descriptor === undefined) return undefined\n\n  return 'value' in descriptor ? descriptor.value : descriptor.get?.call(target)\n}\n\nconst cellText = (cell: object): string => {\n  if (ownProperty(cell, 't') === 'z') return ''\n\n  const value = ownProperty(cell, 'v')\n\n  // SheetJS runs with `cellFormula: false`: only cached values exist, formula-only cells are empty.\n  if (value === undefined || value === null) return ''\n\n  // Prefer parser-provided display text. Do not run an untrusted format template here.\n  const display = ownProperty(cell, 'w')\n\n  if (Predicate.isString(display)) return display\n\n  if (Predicate.isString(value)) return value\n\n  if (Predicate.isBoolean(value)) return value ? 'TRUE' : 'FALSE'\n\n  if (value instanceof Date && Number.isFinite(value.getTime())) return value.toISOString()\n\n  if (Predicate.isNumber(value) && Number.isFinite(value)) return String(value)\n\n  throw new FileExtractionError({ format: 'xlsx', message: 'Invalid XLSX cell value.' })\n}\n\n/**\n * CSV length of `text` (quotes doubled, wrapped when needed). The scan stops once the length\n * exceeds `limit`; the returned length is then only known to be above it.\n */\nconst csvField = (text: string, limit = Number.POSITIVE_INFINITY) => {\n  let quoteCount = 0\n  let quote = text === 'ID'\n\n  for (let index = 0; index < text.length; index += 1) {\n    const character = text.charCodeAt(index)\n\n    // '\"', ',', '\\n', '\\r'\n    if (character === 34) quoteCount += 1\n\n    if (character === 34 || character === 44 || character === 10 || character === 13) {\n      quote = true\n\n      if (text.length + quoteCount + 2 > limit) break\n    }\n  }\n\n  return { quote, length: text.length + quoteCount + (quote ? 2 : 0) }\n}\n\ntype VisitedSheet = {\n  readonly name: string\n  readonly sheet: object | undefined\n  readonly range: CellRange | undefined\n}\n\n/** Annotated text for a cell, or `undefined` to write it plain. */\ntype CellDecorator = (input: {\n  readonly sheet: VisitedSheet\n  readonly row: number\n  readonly column: number\n  readonly text: string\n}) => string | undefined\n\n/** Write bounded CSV for every sheet; throws once `budget` characters would be exceeded. */\nconst renderSheets = (\n  sheets: ReadonlyArray<VisitedSheet>,\n  budget: number,\n  decorate?: CellDecorator\n) => {\n  let characters = 0\n  const output: Array<string> = []\n\n  const append = (text: string) => {\n    if (text.length > budget - characters) throw outputTooLarge()\n\n    characters += text.length\n    output.push(text)\n  }\n\n  const appendCell = (text: string) => {\n    if (text.length > budget - characters) throw outputTooLarge()\n\n    const field = csvField(text, budget - characters)\n\n    if (field.length > budget - characters) throw outputTooLarge()\n\n    append(field.quote ? `\"${text.replaceAll('\"', '\"\"')}\"` : text)\n  }\n\n  for (const visited of sheets) {\n    const { name, sheet, range } = visited\n\n    if (sheet === undefined) continue\n\n    if (output.length > 0) append('\\n\\n')\n\n    append('# ')\n    append(name)\n    append('\\n')\n\n    if (range === undefined) continue\n\n    for (let row = range.start.r; row <= range.end.r; row += 1) {\n      if (row > range.start.r) append('\\n')\n\n      for (let column = range.start.c; column <= range.end.c; column += 1) {\n        if (column > range.start.c) append(',')\n\n        const cell = ownProperty(sheet, encodeCellAddress({ r: row, c: column }))\n\n        if (!Predicate.isObject(cell)) {\n          appendCell('')\n          continue\n        }\n\n        const text = cellText(cell)\n\n        appendCell(decorate?.({ sheet: visited, row, column, text }) ?? text)\n      }\n    }\n  }\n\n  return output.join('')\n}\n\n/**\n * Validate every sheet before touching any cell, then generate bounded CSV incrementally, never\n * with `sheet_to_csv` (which can walk billions of absent cells or allocate an unbounded quoted\n * string).\n *\n * Existing cells inside an external hyperlink are written as `text <url>`. Plain text is\n * rendered first, so links only use budget the plain text leaves over: in document order, an\n * annotation that no longer fits leaves its cell plain and one marker ends the output.\n */\nexport const extractBoundedXlsxText = (\n  workbook: XlsxWorkbook,\n  limits: XlsxTextLimits,\n  options: XlsxTextOptions = {}\n) => {\n  if (workbook.SheetNames.length > limits.maxXlsxSheets) throw tooLarge()\n\n  let visits = 0\n\n  const sheets = workbook.SheetNames.map((name): VisitedSheet => {\n    const candidate = ownProperty(workbook.Sheets, name)\n    const sheet = Predicate.isObject(candidate) ? candidate : undefined\n    const ref = sheet === undefined ? undefined : ownProperty(sheet, '!ref')\n\n    if (ref !== undefined && !Predicate.isString(ref)) throw invalidRange()\n\n    const range = ref === undefined ? undefined : parseCellRange(ref)\n\n    if (ref !== undefined && range === undefined) throw invalidRange()\n\n    visits += range === undefined ? 0 : cellCount(range)\n\n    if (!Number.isSafeInteger(visits) || visits > limits.maxXlsxCellVisits) throw tooLarge()\n\n    return { name, sheet, range }\n  })\n\n  const links = options.hyperlinks ?? new Map()\n  const truncated = options.hyperlinksTruncated === true\n\n  if (links.size === 0 && !truncated) return renderSheets(sheets, limits.maxXlsxTextCharacters)\n\n  // Reserve the marker's space so it always fits inside the character limit.\n  const budget = limits.maxXlsxTextCharacters - omittedHyperlinksMarkerReserve\n  let spare = budget - renderSheets(sheets, budget).length\n  let dropped = false\n  const lookups = new Map<string, HyperlinkLookup>()\n\n  const lookupFor = (sheet: VisitedSheet) => {\n    const sheetLinks = links.get(sheet.name)\n\n    if (sheetLinks === undefined || sheet.range === undefined) return undefined\n\n    const existing = lookups.get(sheet.name)\n\n    if (existing !== undefined) return existing\n\n    const lookup = makeHyperlinkLookup(sheetLinks, sheet.range)\n    lookups.set(sheet.name, lookup)\n\n    return lookup\n  }\n\n  const output = renderSheets(sheets, budget, ({ sheet, row, column, text }) => {\n    const link = lookupFor(sheet)?.at(row, column)\n\n    if (link === undefined || link.target === text) return undefined\n\n    const label = text.length > 0 ? text : (link.display ?? '')\n    const plainLength = csvField(text).length\n    const annotatedLength = label.length + (label.length > 0 ? 3 : 2) + link.target.length\n\n    // Constant-time bound first (CSV escaping only adds), then a scan capped at the budget.\n    if (annotatedLength - plainLength > spare) {\n      dropped = true\n\n      return undefined\n    }\n\n    const annotated = label.length > 0 ? `${label} <${link.target}>` : `<${link.target}>`\n    const extra = csvField(annotated, plainLength + spare).length - plainLength\n\n    if (extra > spare) {\n      dropped = true\n\n      return undefined\n    }\n\n    spare -= extra\n\n    return annotated\n  })\n\n  const reasons = [\n    ...(dropped ? [outputLimitReason] : []),\n    ...(truncated ? [hyperlinkLimitReason] : [])\n  ]\n\n  return reasons.length === 0 ? output : output + omittedHyperlinksMarker(reasons)\n}\n"],"mappings":";;;;;AAyBA,MAAM,oBAAoB;AAE1B,MAAM,uBAAuB;AAE7B,MAAM,2BAA2B,YAC/B,iCAAiC,QAAQ,KAAK,OAAO,EAAE;;;;;AAMzD,MAAa,iCAAiC,wBAAwB,CACpE,mBACA,oBACF,CAAC,EAAE;AAEH,MAAM,qBACJ,IAAI,oBAAoB;CAAE,QAAQ;CAAQ,SAAS;AAAgC,CAAC;AAEtF,MAAM,iBACJ,IAAI,oBAAoB;CACtB,QAAQ;CACR,SAAS;AACX,CAAC;AAEH,MAAM,uBACJ,IAAI,oBAAoB;CACtB,QAAQ;CACR,SAAS;AACX,CAAC;;AAGH,MAAM,eAAe,QAAgB,QAAyB;CAC5D,MAAM,aAAa,OAAO,yBAAyB,QAAQ,GAAG;CAE9D,IAAI,eAAe,KAAA,GAAW,OAAO,KAAA;CAErC,OAAO,WAAW,aAAa,WAAW,QAAQ,WAAW,KAAK,KAAK,MAAM;AAC/E;AAEA,MAAM,YAAY,SAAyB;CACzC,IAAI,YAAY,MAAM,GAAG,MAAM,KAAK,OAAO;CAE3C,MAAM,QAAQ,YAAY,MAAM,GAAG;CAGnC,IAAI,UAAU,KAAA,KAAa,UAAU,MAAM,OAAO;CAGlD,MAAM,UAAU,YAAY,MAAM,GAAG;CAErC,IAAI,UAAU,SAAS,OAAO,GAAG,OAAO;CAExC,IAAI,UAAU,SAAS,KAAK,GAAG,OAAO;CAEtC,IAAI,UAAU,UAAU,KAAK,GAAG,OAAO,QAAQ,SAAS;CAExD,IAAI,iBAAiB,QAAQ,OAAO,SAAS,MAAM,QAAQ,CAAC,GAAG,OAAO,MAAM,YAAY;CAExF,IAAI,UAAU,SAAS,KAAK,KAAK,OAAO,SAAS,KAAK,GAAG,OAAO,OAAO,KAAK;CAE5E,MAAM,IAAI,oBAAoB;EAAE,QAAQ;EAAQ,SAAS;CAA2B,CAAC;AACvF;;;;;AAMA,MAAM,YAAY,MAAc,QAAQ,OAAO,sBAAsB;CACnE,IAAI,aAAa;CACjB,IAAI,QAAQ,SAAS;CAErB,KAAK,IAAI,QAAQ,GAAG,QAAQ,KAAK,QAAQ,SAAS,GAAG;EACnD,MAAM,YAAY,KAAK,WAAW,KAAK;EAGvC,IAAI,cAAc,IAAI,cAAc;EAEpC,IAAI,cAAc,MAAM,cAAc,MAAM,cAAc,MAAM,cAAc,IAAI;GAChF,QAAQ;GAER,IAAI,KAAK,SAAS,aAAa,IAAI,OAAO;EAC5C;CACF;CAEA,OAAO;EAAE;EAAO,QAAQ,KAAK,SAAS,cAAc,QAAQ,IAAI;CAAG;AACrE;;AAiBA,MAAM,gBACJ,QACA,QACA,aACG;CACH,IAAI,aAAa;CACjB,MAAM,SAAwB,CAAC;CAE/B,MAAM,UAAU,SAAiB;EAC/B,IAAI,KAAK,SAAS,SAAS,YAAY,MAAM,eAAe;EAE5D,cAAc,KAAK;EACnB,OAAO,KAAK,IAAI;CAClB;CAEA,MAAM,cAAc,SAAiB;EACnC,IAAI,KAAK,SAAS,SAAS,YAAY,MAAM,eAAe;EAE5D,MAAM,QAAQ,SAAS,MAAM,SAAS,UAAU;EAEhD,IAAI,MAAM,SAAS,SAAS,YAAY,MAAM,eAAe;EAE7D,OAAO,MAAM,QAAQ,IAAI,KAAK,WAAW,MAAK,MAAI,EAAE,KAAK,IAAI;CAC/D;CAEA,KAAK,MAAM,WAAW,QAAQ;EAC5B,MAAM,EAAE,MAAM,OAAO,UAAU;EAE/B,IAAI,UAAU,KAAA,GAAW;EAEzB,IAAI,OAAO,SAAS,GAAG,OAAO,MAAM;EAEpC,OAAO,IAAI;EACX,OAAO,IAAI;EACX,OAAO,IAAI;EAEX,IAAI,UAAU,KAAA,GAAW;EAEzB,KAAK,IAAI,MAAM,MAAM,MAAM,GAAG,OAAO,MAAM,IAAI,GAAG,OAAO,GAAG;GAC1D,IAAI,MAAM,MAAM,MAAM,GAAG,OAAO,IAAI;GAEpC,KAAK,IAAI,SAAS,MAAM,MAAM,GAAG,UAAU,MAAM,IAAI,GAAG,UAAU,GAAG;IACnE,IAAI,SAAS,MAAM,MAAM,GAAG,OAAO,GAAG;IAEtC,MAAM,OAAO,YAAY,OAAO,kBAAkB;KAAE,GAAG;KAAK,GAAG;IAAO,CAAC,CAAC;IAExE,IAAI,CAAC,UAAU,SAAS,IAAI,GAAG;KAC7B,WAAW,EAAE;KACb;IACF;IAEA,MAAM,OAAO,SAAS,IAAI;IAE1B,WAAW,WAAW;KAAE,OAAO;KAAS;KAAK;KAAQ;IAAK,CAAC,KAAK,IAAI;GACtE;EACF;CACF;CAEA,OAAO,OAAO,KAAK,EAAE;AACvB;;;;;;;;;;AAWA,MAAa,0BACX,UACA,QACA,UAA2B,CAAC,MACzB;CACH,IAAI,SAAS,WAAW,SAAS,OAAO,eAAe,MAAM,SAAS;CAEtE,IAAI,SAAS;CAEb,MAAM,SAAS,SAAS,WAAW,KAAK,SAAuB;EAC7D,MAAM,YAAY,YAAY,SAAS,QAAQ,IAAI;EACnD,MAAM,QAAQ,UAAU,SAAS,SAAS,IAAI,YAAY,KAAA;EAC1D,MAAM,MAAM,UAAU,KAAA,IAAY,KAAA,IAAY,YAAY,OAAO,MAAM;EAEvE,IAAI,QAAQ,KAAA,KAAa,CAAC,UAAU,SAAS,GAAG,GAAG,MAAM,aAAa;EAEtE,MAAM,QAAQ,QAAQ,KAAA,IAAY,KAAA,IAAY,eAAe,GAAG;EAEhE,IAAI,QAAQ,KAAA,KAAa,UAAU,KAAA,GAAW,MAAM,aAAa;EAEjE,UAAU,UAAU,KAAA,IAAY,IAAI,UAAU,KAAK;EAEnD,IAAI,CAAC,OAAO,cAAc,MAAM,KAAK,SAAS,OAAO,mBAAmB,MAAM,SAAS;EAEvF,OAAO;GAAE;GAAM;GAAO;EAAM;CAC9B,CAAC;CAED,MAAM,QAAQ,QAAQ,8BAAc,IAAI,IAAI;CAC5C,MAAM,YAAY,QAAQ,wBAAwB;CAElD,IAAI,MAAM,SAAS,KAAK,CAAC,WAAW,OAAO,aAAa,QAAQ,OAAO,qBAAqB;CAG5F,MAAM,SAAS,OAAO,wBAAwB;CAC9C,IAAI,QAAQ,SAAS,aAAa,QAAQ,MAAM,EAAE;CAClD,IAAI,UAAU;CACd,MAAM,0BAAU,IAAI,IAA6B;CAEjD,MAAM,aAAa,UAAwB;EACzC,MAAM,aAAa,MAAM,IAAI,MAAM,IAAI;EAEvC,IAAI,eAAe,KAAA,KAAa,MAAM,UAAU,KAAA,GAAW,OAAO,KAAA;EAElE,MAAM,WAAW,QAAQ,IAAI,MAAM,IAAI;EAEvC,IAAI,aAAa,KAAA,GAAW,OAAO;EAEnC,MAAM,SAAS,oBAAoB,YAAY,MAAM,KAAK;EAC1D,QAAQ,IAAI,MAAM,MAAM,MAAM;EAE9B,OAAO;CACT;CAEA,MAAM,SAAS,aAAa,QAAQ,SAAS,EAAE,OAAO,KAAK,QAAQ,WAAW;EAC5E,MAAM,OAAO,UAAU,KAAK,GAAG,GAAG,KAAK,MAAM;EAE7C,IAAI,SAAS,KAAA,KAAa,KAAK,WAAW,MAAM,OAAO,KAAA;EAEvD,MAAM,QAAQ,KAAK,SAAS,IAAI,OAAQ,KAAK,WAAW;EACxD,MAAM,cAAc,SAAS,IAAI,EAAE;EAInC,IAHwB,MAAM,UAAU,MAAM,SAAS,IAAI,IAAI,KAAK,KAAK,OAAO,SAG1D,cAAc,OAAO;GACzC,UAAU;GAEV;EACF;EAEA,MAAM,YAAY,MAAM,SAAS,IAAI,GAAG,MAAM,IAAI,KAAK,OAAO,KAAK,IAAI,KAAK,OAAO;EACnF,MAAM,QAAQ,SAAS,WAAW,cAAc,KAAK,EAAE,SAAS;EAEhE,IAAI,QAAQ,OAAO;GACjB,UAAU;GAEV;EACF;EAEA,SAAS;EAET,OAAO;CACT,CAAC;CAED,MAAM,UAAU,CACd,GAAI,UAAU,CAAC,iBAAiB,IAAI,CAAC,GACrC,GAAI,YAAY,CAAC,oBAAoB,IAAI,CAAC,CAC5C;CAEA,OAAO,QAAQ,WAAW,IAAI,SAAS,SAAS,wBAAwB,OAAO;AACjF"}

package/dist/node/xlsx-workbook.d.mts

@@ -0,0 +1,32 @@
+//#region src/node/xlsx-workbook.d.ts
+/**
+ * `xl/workbook.xml` is never handed to SheetJS as uploaded. SheetJS's workbook parser slices and
+ * decodes the whole prefix of the part at every `</definedName>` (quadratic), and walks every
+ * `<sheet>` its own tag pattern finds, resolving each through its `r:id`, so one worksheet can be
+ * parsed once per declaration. The extractor reads the sheet list itself, checks that its strict
+ * scan and SheetJS's tag grammar see the same sheets, and writes a minimal workbook: the sheets in
+ * order (name, `sheetId`, hidden state, a fresh `r:id`) and the 1904 date system.
+ */
+type WorkbookSheetDeclaration = {
+  /** The sheet name exactly as SheetJS reads it (`unescapexml(utf8read(name))`). */readonly name: string;
+  readonly sheetId: string;
+  readonly state: 'hidden' | 'veryHidden' | undefined; /** The `r:id` of the sheet's relationship in the uploaded workbook, entity-decoded. */
+  readonly relationshipId: string | undefined;
+};
+type WorkbookModel = {
+  readonly sheets: ReadonlyArray<WorkbookSheetDeclaration>;
+  readonly date1904: boolean;
+};
+declare const malformedWorkbookMessage = "XLSX workbook is malformed.";
+/**
+ * The sheets of `xl/workbook.xml` and its date system. Throws `FileExtractionError` when the
+ * workbook declares more than `maxSheets` sheets (counted by either scan), when the strict scan
+ * and SheetJS's grammar disagree on the number or names of sheets, when a name is missing, holds
+ * CDATA, or repeats another ignoring case, or when there are no sheets.
+ */
+declare const readWorkbookModel: (bytes: Uint8Array, maxSheets: number) => WorkbookModel;
+/** The generated workbook: sheet `n` has `r:id="rId<n>"`. */
+declare const workbookXml: (model: WorkbookModel) => string;
+//#endregion
+export { WorkbookModel, WorkbookSheetDeclaration, malformedWorkbookMessage, readWorkbookModel, workbookXml };
+//# sourceMappingURL=xlsx-workbook.d.mts.map

package/dist/node/xlsx-workbook.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-workbook.d.mts","names":[],"sources":["../../src/node/xlsx-workbook.ts"],"mappings":";;AAuBA;;;;;;;KAAY,wBAAA;EAMa,2FAJd,IAAA;EAAA,SACA,OAAA;EAAA,SACA,KAAA;WAEA,cAAA;AAAA;AAAA,KAGC,aAAA;EAAA,SACD,MAAA,EAAQ,aAAa,CAAC,wBAAA;EAAA,SACtB,QAAA;AAAA;AAAA,cAGE,wBAAA;AAAb;;;;AAAqC;AAsBrC;AAtBA,cAsBa,iBAAA,GAAqB,KAAA,EAAO,UAAA,EAAY,SAAA,aAAoB,aA4DxE;;cAGY,WAAA,GAAe,KAAoB,EAAb,aAAa"}

package/dist/node/xlsx-workbook.mjs

@@ -0,0 +1,70 @@
+import { FileExtractionError } from "../errors.mjs";
+import { officeDocumentRelationshipsNamespace, sheetJsAttributeEscape, sheetJsAttributeText, sheetJsTags, sheetJsXmlHeader, spreadsheetMainNamespace, stripSheetJsNamespace } from "./sheetjs-xml.mjs";
+import { decodeXmlEntities, prefixedAttribute, rawXmlAttributes } from "./xml-text.mjs";
+import { Buffer } from "node:buffer";
+//#region src/node/xlsx-workbook.ts
+const malformedWorkbookMessage = "XLSX workbook is malformed.";
+const malformed = () => new FileExtractionError({
+	format: "xlsx",
+	message: malformedWorkbookMessage
+});
+const tooManySheets = () => new FileExtractionError({
+	format: "xlsx",
+	message: "XLSX exceeds the worksheet or cell-visit limit."
+});
+const strictSheetTag = /<(?:[\w.-]+:)?sheet(?=[\s/>])[^<>]*>/g;
+const plainSheetId = /^[1-9]\d{0,8}$/;
+/**
+* The sheets of `xl/workbook.xml` and its date system. Throws `FileExtractionError` when the
+* workbook declares more than `maxSheets` sheets (counted by either scan), when the strict scan
+* and SheetJS's grammar disagree on the number or names of sheets, when a name is missing, holds
+* CDATA, or repeats another ignoring case, or when there are no sheets.
+*/
+const readWorkbookModel = (bytes, maxSheets) => {
+	const text = Buffer.from(bytes.buffer, bytes.byteOffset, bytes.byteLength).toString("latin1");
+	const sheetJsSheets = [];
+	let date1904 = false;
+	for (const tag of sheetJsTags(text)) {
+		const head = stripSheetJsNamespace(tag.head);
+		if (head === "<sheet") {
+			sheetJsSheets.push(tag);
+			if (sheetJsSheets.length > maxSheets) throw tooManySheets();
+		} else if (head === "<workbookPr" || head === "<workbookPr/>") {
+			const value = tag.attributes.get("date1904");
+			if (value !== void 0) date1904 = value === "1" || value === "true";
+		}
+	}
+	const strictSheets = [];
+	for (const [tag] of text.matchAll(strictSheetTag)) {
+		strictSheets.push(rawXmlAttributes(tag));
+		if (strictSheets.length > maxSheets) throw tooManySheets();
+	}
+	if (sheetJsSheets.length === 0 || sheetJsSheets.length !== strictSheets.length) throw malformed();
+	const seen = /* @__PURE__ */ new Set();
+	return {
+		sheets: sheetJsSheets.map((tag, index) => {
+			const strict = strictSheets[index];
+			const rawName = tag.attributes.get("name");
+			if (strict === void 0 || rawName === void 0 || strict.get("name") !== rawName) throw malformed();
+			const name = sheetJsAttributeText(rawName);
+			if (name === void 0 || seen.has(name.toLowerCase())) throw malformed();
+			seen.add(name.toLowerCase());
+			const state = tag.attributes.get("state");
+			const sheetId = tag.attributes.get("sheetId");
+			const relationshipId = prefixedAttribute(strict, "id");
+			return {
+				name,
+				sheetId: sheetId !== void 0 && plainSheetId.test(sheetId) ? sheetId : `${index + 1}`,
+				state: state === "hidden" || state === "veryHidden" ? state : void 0,
+				relationshipId: relationshipId === void 0 ? void 0 : decodeXmlEntities(relationshipId)
+			};
+		}),
+		date1904
+	};
+};
+/** The generated workbook: sheet `n` has `r:id="rId<n>"`. */
+const workbookXml = (model) => `${sheetJsXmlHeader}<workbook xmlns="${spreadsheetMainNamespace}" xmlns:r="${officeDocumentRelationshipsNamespace}">${model.date1904 ? "<workbookPr date1904=\"1\"/>" : ""}<sheets>${model.sheets.map((sheet, index) => `<sheet name="${sheetJsAttributeEscape(sheet.name)}" sheetId="${sheet.sheetId}"${sheet.state === void 0 ? "" : ` state="${sheet.state}"`} r:id="rId${index + 1}"/>`).join("")}</sheets></workbook>`;
+//#endregion
+export { malformedWorkbookMessage, readWorkbookModel, workbookXml };
+
+//# sourceMappingURL=xlsx-workbook.mjs.map

package/dist/node/xlsx-workbook.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-workbook.mjs","names":[],"sources":["../../src/node/xlsx-workbook.ts"],"sourcesContent":["import { Buffer } from 'node:buffer'\nimport { FileExtractionError } from '../errors.ts'\nimport {\n  sheetJsAttributeEscape,\n  sheetJsAttributeText,\n  sheetJsTags,\n  stripSheetJsNamespace,\n  sheetJsXmlHeader,\n  spreadsheetMainNamespace,\n  officeDocumentRelationshipsNamespace\n} from './sheetjs-xml.ts'\nimport type { SheetJsTag } from './sheetjs-xml.ts'\nimport { decodeXmlEntities, prefixedAttribute, rawXmlAttributes } from './xml-text.ts'\n\n/**\n * `xl/workbook.xml` is never handed to SheetJS as uploaded. SheetJS's workbook parser slices and\n * decodes the whole prefix of the part at every `</definedName>` (quadratic), and walks every\n * `<sheet>` its own tag pattern finds, resolving each through its `r:id`, so one worksheet can be\n * parsed once per declaration. The extractor reads the sheet list itself, checks that its strict\n * scan and SheetJS's tag grammar see the same sheets, and writes a minimal workbook: the sheets in\n * order (name, `sheetId`, hidden state, a fresh `r:id`) and the 1904 date system.\n */\n\nexport type WorkbookSheetDeclaration = {\n  /** The sheet name exactly as SheetJS reads it (`unescapexml(utf8read(name))`). */\n  readonly name: string\n  readonly sheetId: string\n  readonly state: 'hidden' | 'veryHidden' | undefined\n  /** The `r:id` of the sheet's relationship in the uploaded workbook, entity-decoded. */\n  readonly relationshipId: string | undefined\n}\n\nexport type WorkbookModel = {\n  readonly sheets: ReadonlyArray<WorkbookSheetDeclaration>\n  readonly date1904: boolean\n}\n\nexport const malformedWorkbookMessage = 'XLSX workbook is malformed.'\n\nconst malformed = () =>\n  new FileExtractionError({ format: 'xlsx', message: malformedWorkbookMessage })\n\nconst tooManySheets = () =>\n  new FileExtractionError({\n    format: 'xlsx',\n    message: 'XLSX exceeds the worksheet or cell-visit limit.'\n  })\n\n// `[^<>]` keeps each candidate inside one tag, so the strict scan stays linear.\nconst strictSheetTag = /<(?:[\\w.-]+:)?sheet(?=[\\s/>])[^<>]*>/g\n\nconst plainSheetId = /^[1-9]\\d{0,8}$/\n\n/**\n * The sheets of `xl/workbook.xml` and its date system. Throws `FileExtractionError` when the\n * workbook declares more than `maxSheets` sheets (counted by either scan), when the strict scan\n * and SheetJS's grammar disagree on the number or names of sheets, when a name is missing, holds\n * CDATA, or repeats another ignoring case, or when there are no sheets.\n */\nexport const readWorkbookModel = (bytes: Uint8Array, maxSheets: number): WorkbookModel => {\n  // SheetJS parses the Latin-1 (\"binary\") view and decodes attribute text as UTF-8 afterwards.\n  const text = Buffer.from(bytes.buffer, bytes.byteOffset, bytes.byteLength).toString('latin1')\n  const sheetJsSheets: Array<SheetJsTag> = []\n  let date1904 = false\n\n  for (const tag of sheetJsTags(text)) {\n    const head = stripSheetJsNamespace(tag.head)\n\n    if (head === '<sheet') {\n      sheetJsSheets.push(tag)\n\n      if (sheetJsSheets.length > maxSheets) throw tooManySheets()\n    } else if (head === '<workbookPr' || head === '<workbookPr/>') {\n      // Each `workbookPr` that carries the attribute overrides the last (SheetJS `parsexmlbool`).\n      const value = tag.attributes.get('date1904')\n\n      if (value !== undefined) date1904 = value === '1' || value === 'true'\n    }\n  }\n\n  const strictSheets: Array<ReadonlyMap<string, string>> = []\n\n  for (const [tag] of text.matchAll(strictSheetTag)) {\n    strictSheets.push(rawXmlAttributes(tag))\n\n    if (strictSheets.length > maxSheets) throw tooManySheets()\n  }\n\n  if (sheetJsSheets.length === 0 || sheetJsSheets.length !== strictSheets.length) throw malformed()\n\n  const seen = new Set<string>()\n\n  const sheets = sheetJsSheets.map((tag, index): WorkbookSheetDeclaration => {\n    const strict = strictSheets[index]\n    const rawName = tag.attributes.get('name')\n\n    if (strict === undefined || rawName === undefined || strict.get('name') !== rawName)\n      throw malformed()\n\n    const name = sheetJsAttributeText(rawName)\n\n    // Excel compares sheet names ignoring case.\n    if (name === undefined || seen.has(name.toLowerCase())) throw malformed()\n\n    seen.add(name.toLowerCase())\n\n    const state = tag.attributes.get('state')\n    const sheetId = tag.attributes.get('sheetId')\n    const relationshipId = prefixedAttribute(strict, 'id')\n\n    return {\n      name,\n      sheetId: sheetId !== undefined && plainSheetId.test(sheetId) ? sheetId : `${index + 1}`,\n      state: state === 'hidden' || state === 'veryHidden' ? state : undefined,\n      relationshipId: relationshipId === undefined ? undefined : decodeXmlEntities(relationshipId)\n    }\n  })\n\n  return { sheets, date1904 }\n}\n\n/** The generated workbook: sheet `n` has `r:id=\"rId<n>\"`. */\nexport const workbookXml = (model: WorkbookModel) =>\n  `${sheetJsXmlHeader}<workbook xmlns=\"${spreadsheetMainNamespace}\" xmlns:r=\"${officeDocumentRelationshipsNamespace}\">${\n    model.date1904 ? '<workbookPr date1904=\"1\"/>' : ''\n  }<sheets>${model.sheets\n    .map(\n      (sheet, index) =>\n        `<sheet name=\"${sheetJsAttributeEscape(sheet.name)}\" sheetId=\"${sheet.sheetId}\"${\n          sheet.state === undefined ? '' : ` state=\"${sheet.state}\"`\n        } r:id=\"rId${index + 1}\"/>`\n    )\n    .join('')}</sheets></workbook>`\n"],"mappings":";;;;;AAqCA,MAAa,2BAA2B;AAExC,MAAM,kBACJ,IAAI,oBAAoB;CAAE,QAAQ;CAAQ,SAAS;AAAyB,CAAC;AAE/E,MAAM,sBACJ,IAAI,oBAAoB;CACtB,QAAQ;CACR,SAAS;AACX,CAAC;AAGH,MAAM,iBAAiB;AAEvB,MAAM,eAAe;;;;;;;AAQrB,MAAa,qBAAqB,OAAmB,cAAqC;CAExF,MAAM,OAAO,OAAO,KAAK,MAAM,QAAQ,MAAM,YAAY,MAAM,UAAU,EAAE,SAAS,QAAQ;CAC5F,MAAM,gBAAmC,CAAC;CAC1C,IAAI,WAAW;CAEf,KAAK,MAAM,OAAO,YAAY,IAAI,GAAG;EACnC,MAAM,OAAO,sBAAsB,IAAI,IAAI;EAE3C,IAAI,SAAS,UAAU;GACrB,cAAc,KAAK,GAAG;GAEtB,IAAI,cAAc,SAAS,WAAW,MAAM,cAAc;EAC5D,OAAO,IAAI,SAAS,iBAAiB,SAAS,iBAAiB;GAE7D,MAAM,QAAQ,IAAI,WAAW,IAAI,UAAU;GAE3C,IAAI,UAAU,KAAA,GAAW,WAAW,UAAU,OAAO,UAAU;EACjE;CACF;CAEA,MAAM,eAAmD,CAAC;CAE1D,KAAK,MAAM,CAAC,QAAQ,KAAK,SAAS,cAAc,GAAG;EACjD,aAAa,KAAK,iBAAiB,GAAG,CAAC;EAEvC,IAAI,aAAa,SAAS,WAAW,MAAM,cAAc;CAC3D;CAEA,IAAI,cAAc,WAAW,KAAK,cAAc,WAAW,aAAa,QAAQ,MAAM,UAAU;CAEhG,MAAM,uBAAO,IAAI,IAAY;CA4B7B,OAAO;EAAE,QA1BM,cAAc,KAAK,KAAK,UAAoC;GACzE,MAAM,SAAS,aAAa;GAC5B,MAAM,UAAU,IAAI,WAAW,IAAI,MAAM;GAEzC,IAAI,WAAW,KAAA,KAAa,YAAY,KAAA,KAAa,OAAO,IAAI,MAAM,MAAM,SAC1E,MAAM,UAAU;GAElB,MAAM,OAAO,qBAAqB,OAAO;GAGzC,IAAI,SAAS,KAAA,KAAa,KAAK,IAAI,KAAK,YAAY,CAAC,GAAG,MAAM,UAAU;GAExE,KAAK,IAAI,KAAK,YAAY,CAAC;GAE3B,MAAM,QAAQ,IAAI,WAAW,IAAI,OAAO;GACxC,MAAM,UAAU,IAAI,WAAW,IAAI,SAAS;GAC5C,MAAM,iBAAiB,kBAAkB,QAAQ,IAAI;GAErD,OAAO;IACL;IACA,SAAS,YAAY,KAAA,KAAa,aAAa,KAAK,OAAO,IAAI,UAAU,GAAG,QAAQ;IACpF,OAAO,UAAU,YAAY,UAAU,eAAe,QAAQ,KAAA;IAC9D,gBAAgB,mBAAmB,KAAA,IAAY,KAAA,IAAY,kBAAkB,cAAc;GAC7F;EACF,CAEc;EAAG;CAAS;AAC5B;;AAGA,MAAa,eAAe,UAC1B,GAAG,iBAAiB,mBAAmB,yBAAyB,aAAa,qCAAqC,IAChH,MAAM,WAAW,iCAA+B,GACjD,UAAU,MAAM,OACd,KACE,OAAO,UACN,gBAAgB,uBAAuB,MAAM,IAAI,EAAE,aAAa,MAAM,QAAQ,GAC5E,MAAM,UAAU,KAAA,IAAY,KAAK,WAAW,MAAM,MAAM,GACzD,YAAY,QAAQ,EAAE,IAC3B,EACC,KAAK,EAAE,EAAE"}

package/dist/node/xml-text.d.mts

@@ -0,0 +1,12 @@
+//#region src/node/xml-text.d.ts
+/** Decode the predefined XML entities and numeric character references; keep unknown ones. */
+declare const decodeXmlEntities: (text: string) => string;
+/** Attributes of one start tag as written (not decoded), keyed by qualified name; first wins. */
+declare const rawXmlAttributes: (tag: string) => ReadonlyMap<string, string>;
+/** Attributes of one start tag, entity-decoded, keyed by their qualified name. */
+declare const xmlAttributes: (tag: string) => ReadonlyMap<string, string>;
+/** The value of a namespace-prefixed attribute such as `r:id`, whatever the prefix. */
+declare const prefixedAttribute: (attributes: ReadonlyMap<string, string>, localName: string) => string | undefined;
+//#endregion
+export { decodeXmlEntities, prefixedAttribute, rawXmlAttributes, xmlAttributes };
+//# sourceMappingURL=xml-text.d.mts.map

package/dist/node/xml-text.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xml-text.d.mts","names":[],"sources":["../../src/node/xml-text.ts"],"mappings":";;cAuCa,iBAAA,GAAqB,IAAY;;cAQjC,gBAAA,GAAoB,GAAA,aAAc,WAAW;;cAe7C,aAAA,GAAiB,GAAA,aAAc,WAAW;AAfvD;AAAA,cAqBa,iBAAA,GAAqB,UAAA,EAAY,WAAW,kBAAkB,SAAA"}

package/dist/node/xml-text.mjs

@@ -0,0 +1,51 @@
+//#region src/node/xml-text.ts
+const xmlEntity = /&([^;&\s]{1,16});/g;
+const hexEntity = /^#x([0-9a-fA-F]+)$/;
+const decimalEntity = /^#(\d+)$/;
+const namedEntities = new Map([
+	["amp", "&"],
+	["lt", "<"],
+	["gt", ">"],
+	["quot", "\""],
+	["apos", "'"]
+]);
+const decodeCodePoint = (raw, codePointText, radix) => {
+	const codePoint = Number.parseInt(codePointText, radix);
+	if (!Number.isInteger(codePoint) || codePoint < 0 || codePoint > 1114111) return raw;
+	return String.fromCodePoint(codePoint);
+};
+const decodeXmlEntity = (raw, entity) => {
+	const named = namedEntities.get(entity);
+	if (named !== void 0) return named;
+	const hex = hexEntity.exec(entity)?.[1];
+	if (hex !== void 0) return decodeCodePoint(raw, hex, 16);
+	const decimal = decimalEntity.exec(entity)?.[1];
+	if (decimal !== void 0) return decodeCodePoint(raw, decimal, 10);
+	return raw;
+};
+/** Decode the predefined XML entities and numeric character references; keep unknown ones. */
+const decodeXmlEntities = (text) => text.replace(xmlEntity, (raw, entity) => decodeXmlEntity(raw, entity));
+const attributePattern = /(?<![\w.:-])([\w.:-]+)\s*=\s*(?:"([^"]*)"|'([^']*)')/g;
+/** Attributes of one start tag as written (not decoded), keyed by qualified name; first wins. */
+const rawXmlAttributes = (tag) => {
+	const attributes = /* @__PURE__ */ new Map();
+	for (const match of tag.matchAll(attributePattern)) {
+		const name = match[1];
+		const value = match[2] ?? match[3];
+		if (name !== void 0 && value !== void 0 && !attributes.has(name)) attributes.set(name, value);
+	}
+	return attributes;
+};
+/** Attributes of one start tag, entity-decoded, keyed by their qualified name. */
+const xmlAttributes = (tag) => new Map(Array.from(rawXmlAttributes(tag), ([name, value]) => [name, decodeXmlEntities(value)]));
+/** The value of a namespace-prefixed attribute such as `r:id`, whatever the prefix. */
+const prefixedAttribute = (attributes, localName) => {
+	for (const [name, value] of attributes) {
+		const separator = name.indexOf(":");
+		if (separator > 0 && name.slice(separator + 1) === localName) return value;
+	}
+};
+//#endregion
+export { decodeXmlEntities, prefixedAttribute, rawXmlAttributes, xmlAttributes };
+
+//# sourceMappingURL=xml-text.mjs.map

package/dist/node/xml-text.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"xml-text.mjs","names":[],"sources":["../../src/node/xml-text.ts"],"sourcesContent":["const xmlEntity = /&([^;&\\s]{1,16});/g\n\nconst hexEntity = /^#x([0-9a-fA-F]+)$/\n\nconst decimalEntity = /^#(\\d+)$/\n\nconst namedEntities: ReadonlyMap<string, string> = new Map([\n  ['amp', '&'],\n  ['lt', '<'],\n  ['gt', '>'],\n  ['quot', '\"'],\n  ['apos', \"'\"]\n])\n\nconst decodeCodePoint = (raw: string, codePointText: string, radix: number) => {\n  const codePoint = Number.parseInt(codePointText, radix)\n\n  if (!Number.isInteger(codePoint) || codePoint < 0 || codePoint > 0x10ffff) return raw\n\n  return String.fromCodePoint(codePoint)\n}\n\nconst decodeXmlEntity = (raw: string, entity: string) => {\n  const named = namedEntities.get(entity)\n\n  if (named !== undefined) return named\n\n  const hex = hexEntity.exec(entity)?.[1]\n\n  if (hex !== undefined) return decodeCodePoint(raw, hex, 16)\n\n  const decimal = decimalEntity.exec(entity)?.[1]\n\n  if (decimal !== undefined) return decodeCodePoint(raw, decimal, 10)\n\n  return raw\n}\n\n/** Decode the predefined XML entities and numeric character references; keep unknown ones. */\nexport const decodeXmlEntities = (text: string) =>\n  text.replace(xmlEntity, (raw, entity: string) => decodeXmlEntity(raw, entity))\n\n// A name may only start after a non-name character, so a long run of name characters is scanned\n// once rather than once per starting position.\nconst attributePattern = /(?<![\\w.:-])([\\w.:-]+)\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)')/g\n\n/** Attributes of one start tag as written (not decoded), keyed by qualified name; first wins. */\nexport const rawXmlAttributes = (tag: string): ReadonlyMap<string, string> => {\n  const attributes = new Map<string, string>()\n\n  for (const match of tag.matchAll(attributePattern)) {\n    const name = match[1]\n    const value = match[2] ?? match[3]\n\n    if (name !== undefined && value !== undefined && !attributes.has(name))\n      attributes.set(name, value)\n  }\n\n  return attributes\n}\n\n/** Attributes of one start tag, entity-decoded, keyed by their qualified name. */\nexport const xmlAttributes = (tag: string): ReadonlyMap<string, string> =>\n  new Map(\n    Array.from(rawXmlAttributes(tag), ([name, value]) => [name, decodeXmlEntities(value)] as const)\n  )\n\n/** The value of a namespace-prefixed attribute such as `r:id`, whatever the prefix. */\nexport const prefixedAttribute = (attributes: ReadonlyMap<string, string>, localName: string) => {\n  for (const [name, value] of attributes) {\n    const separator = name.indexOf(':')\n\n    if (separator > 0 && name.slice(separator + 1) === localName) return value\n  }\n\n  return undefined\n}\n"],"mappings":";AAAA,MAAM,YAAY;AAElB,MAAM,YAAY;AAElB,MAAM,gBAAgB;AAEtB,MAAM,gBAA6C,IAAI,IAAI;CACzD,CAAC,OAAO,GAAG;CACX,CAAC,MAAM,GAAG;CACV,CAAC,MAAM,GAAG;CACV,CAAC,QAAQ,IAAG;CACZ,CAAC,QAAQ,GAAG;AACd,CAAC;AAED,MAAM,mBAAmB,KAAa,eAAuB,UAAkB;CAC7E,MAAM,YAAY,OAAO,SAAS,eAAe,KAAK;CAEtD,IAAI,CAAC,OAAO,UAAU,SAAS,KAAK,YAAY,KAAK,YAAY,SAAU,OAAO;CAElF,OAAO,OAAO,cAAc,SAAS;AACvC;AAEA,MAAM,mBAAmB,KAAa,WAAmB;CACvD,MAAM,QAAQ,cAAc,IAAI,MAAM;CAEtC,IAAI,UAAU,KAAA,GAAW,OAAO;CAEhC,MAAM,MAAM,UAAU,KAAK,MAAM,IAAI;CAErC,IAAI,QAAQ,KAAA,GAAW,OAAO,gBAAgB,KAAK,KAAK,EAAE;CAE1D,MAAM,UAAU,cAAc,KAAK,MAAM,IAAI;CAE7C,IAAI,YAAY,KAAA,GAAW,OAAO,gBAAgB,KAAK,SAAS,EAAE;CAElE,OAAO;AACT;;AAGA,MAAa,qBAAqB,SAChC,KAAK,QAAQ,YAAY,KAAK,WAAmB,gBAAgB,KAAK,MAAM,CAAC;AAI/E,MAAM,mBAAmB;;AAGzB,MAAa,oBAAoB,QAA6C;CAC5E,MAAM,6BAAa,IAAI,IAAoB;CAE3C,KAAK,MAAM,SAAS,IAAI,SAAS,gBAAgB,GAAG;EAClD,MAAM,OAAO,MAAM;EACnB,MAAM,QAAQ,MAAM,MAAM,MAAM;EAEhC,IAAI,SAAS,KAAA,KAAa,UAAU,KAAA,KAAa,CAAC,WAAW,IAAI,IAAI,GACnE,WAAW,IAAI,MAAM,KAAK;CAC9B;CAEA,OAAO;AACT;;AAGA,MAAa,iBAAiB,QAC5B,IAAI,IACF,MAAM,KAAK,iBAAiB,GAAG,IAAI,CAAC,MAAM,WAAW,CAAC,MAAM,kBAAkB,KAAK,CAAC,CAAU,CAChG;;AAGF,MAAa,qBAAqB,YAAyC,cAAsB;CAC/F,KAAK,MAAM,CAAC,MAAM,UAAU,YAAY;EACtC,MAAM,YAAY,KAAK,QAAQ,GAAG;EAElC,IAAI,YAAY,KAAK,KAAK,MAAM,YAAY,CAAC,MAAM,WAAW,OAAO;CACvE;AAGF"}

package/dist/sanitize.d.mts

@@ -0,0 +1,6 @@
+//#region src/sanitize.d.ts
+/** Normalize line endings, drop control characters, and collapse layout noise. */
+declare const sanitizeExtractedText: (text: string) => string;
+//#endregion
+export { sanitizeExtractedText };
+//# sourceMappingURL=sanitize.d.mts.map

package/dist/sanitize.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sanitize.d.mts","names":[],"sources":["../src/sanitize.ts"],"mappings":";;cASa,qBAAA,GAAyB,IAAY"}

package/dist/sanitize.mjs

@@ -0,0 +1,11 @@
+//#region src/sanitize.ts
+const nonPrintableCharacters = /[\u0000-\u0008\u000B\u000C\u000E-\u001F\u007F]/g;
+const longDotRuns = /\.{4,}/g;
+const horizontalWhitespaceRuns = /[\t ]{2,}/g;
+const blankLineRuns = /\n{3,}/g;
+/** Normalize line endings, drop control characters, and collapse layout noise. */
+const sanitizeExtractedText = (text) => text.replaceAll("\r\n", "\n").replaceAll("\r", "\n").replace(nonPrintableCharacters, "").replace(longDotRuns, "…").replace(horizontalWhitespaceRuns, " ").replace(blankLineRuns, "\n\n").trim();
+//#endregion
+export { sanitizeExtractedText };
+
+//# sourceMappingURL=sanitize.mjs.map

package/dist/sanitize.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"sanitize.mjs","names":[],"sources":["../src/sanitize.ts"],"sourcesContent":["const nonPrintableCharacters = /[\\u0000-\\u0008\\u000B\\u000C\\u000E-\\u001F\\u007F]/g\n\nconst longDotRuns = /\\.{4,}/g\n\nconst horizontalWhitespaceRuns = /[\\t ]{2,}/g\n\nconst blankLineRuns = /\\n{3,}/g\n\n/** Normalize line endings, drop control characters, and collapse layout noise. */\nexport const sanitizeExtractedText = (text: string) =>\n  text\n    .replaceAll('\\r\\n', '\\n')\n    .replaceAll('\\r', '\\n')\n    .replace(nonPrintableCharacters, '')\n    .replace(longDotRuns, '…')\n    .replace(horizontalWhitespaceRuns, ' ')\n    .replace(blankLineRuns, '\\n\\n')\n    .trim()\n"],"mappings":";AAAA,MAAM,yBAAyB;AAE/B,MAAM,cAAc;AAEpB,MAAM,2BAA2B;AAEjC,MAAM,gBAAgB;;AAGtB,MAAa,yBAAyB,SACpC,KACG,WAAW,QAAQ,IAAI,EACvB,WAAW,MAAM,IAAI,EACrB,QAAQ,wBAAwB,EAAE,EAClC,QAAQ,aAAa,GAAG,EACxB,QAAQ,0BAA0B,GAAG,EACrC,QAAQ,eAAe,MAAM,EAC7B,KAAK"}

package/dist/service.d.mts

@@ -0,0 +1,22 @@
+import { FileExtractorError } from "./errors.mjs";
+import { ExtractedFile, FileInput } from "./format.mjs";
+import { Context, Effect } from "effect";
+
+//#region src/service.d.ts
+type FileExtractorApi = {
+  /**
+   * Extract sanitized text from one file. Fails with `UnsupportedFileFormatError` for unknown
+   * formats, `FileExtractionError` for unreadable, oversized, or empty files, and
+   * `SheetJsUnavailableError` when an XLSX file arrives but SheetJS 0.20.3+ is not installed.
+   */
+  readonly extract: (input: FileInput) => Effect.Effect<ExtractedFile, FileExtractorError>;
+};
+declare const FileExtractor_base: Context.ServiceClass<FileExtractor, "@yolk-sdk/extractors/FileExtractor", FileExtractorApi>;
+/**
+ * The file extractor service. The tag is runtime-portable; the Node implementation lives in
+ * `@yolk-sdk/extractors/node`.
+ */
+declare class FileExtractor extends FileExtractor_base {}
+//#endregion
+export { FileExtractor, FileExtractorApi };
+//# sourceMappingURL=service.d.mts.map

package/dist/service.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"service.d.mts","names":[],"sources":["../src/service.ts"],"mappings":";;;;;KAKY,gBAAA;EAAA;;;;;EAAA,SAMD,OAAA,GAAU,KAAA,EAAO,SAAA,KAAc,MAAA,CAAO,MAAA,CAAO,aAAA,EAAe,kBAAA;AAAA;AAAA,cACtE,kBAAA;;;;;cAMY,aAAA,SAAsB,kBAElC"}

package/dist/service.mjs

@@ -0,0 +1,11 @@
+import { Context } from "effect";
+//#region src/service.ts
+/**
+* The file extractor service. The tag is runtime-portable; the Node implementation lives in
+* `@yolk-sdk/extractors/node`.
+*/
+var FileExtractor = class extends Context.Service()("@yolk-sdk/extractors/FileExtractor") {};
+//#endregion
+export { FileExtractor };
+
+//# sourceMappingURL=service.mjs.map

package/dist/service.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"service.mjs","names":[],"sources":["../src/service.ts"],"sourcesContent":["import { Context } from 'effect'\nimport type { Effect } from 'effect'\nimport type { FileExtractorError } from './errors.ts'\nimport type { ExtractedFile, FileInput } from './format.ts'\n\nexport type FileExtractorApi = {\n  /**\n   * Extract sanitized text from one file. Fails with `UnsupportedFileFormatError` for unknown\n   * formats, `FileExtractionError` for unreadable, oversized, or empty files, and\n   * `SheetJsUnavailableError` when an XLSX file arrives but SheetJS 0.20.3+ is not installed.\n   */\n  readonly extract: (input: FileInput) => Effect.Effect<ExtractedFile, FileExtractorError>\n}\n\n/**\n * The file extractor service. The tag is runtime-portable; the Node implementation lives in\n * `@yolk-sdk/extractors/node`.\n */\nexport class FileExtractor extends Context.Service<FileExtractor, FileExtractorApi>()(\n  '@yolk-sdk/extractors/FileExtractor'\n) {}\n"],"mappings":";;;;;;AAkBA,IAAa,gBAAb,cAAmC,QAAQ,QAAyC,EAClF,oCACF,EAAE,CAAC"}

package/package.json

@@ -0,0 +1,87 @@
+{
+  "name": "@yolk-sdk/extractors",
+  "version": "0.1.0-canary.98",
+  "description": "Bounded text extraction for PDF, DOCX, XLSX, PPTX, and text files: an Effect FileExtractor service with Office archive validation, hyperlink-safe XLSX text, and a knowledge extractor adapter.",
+  "license": "MIT",
+  "type": "module",
+  "sideEffects": false,
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/magoz/yolk-sdk.git",
+    "directory": "packages/extractors"
+  },
+  "bugs": {
+    "url": "https://github.com/magoz/yolk-sdk/issues"
+  },
+  "homepage": "https://github.com/magoz/yolk-sdk#readme",
+  "keywords": [
+    "extraction",
+    "pdf",
+    "docx",
+    "xlsx",
+    "pptx",
+    "knowledge",
+    "effect"
+  ],
+  "engines": {
+    "node": ">=22"
+  },
+  "exports": {
+    "./package.json": "./package.json",
+    ".": {
+      "types": "./dist/index.d.mts",
+      "import": "./dist/index.mjs",
+      "default": "./dist/index.mjs"
+    },
+    "./knowledge": {
+      "types": "./dist/knowledge.d.mts",
+      "import": "./dist/knowledge.mjs",
+      "default": "./dist/knowledge.mjs"
+    },
+    "./node": {
+      "types": "./dist/node/index.d.mts",
+      "import": "./dist/node/index.mjs",
+      "default": "./dist/node/index.mjs"
+    },
+    "./node/extraction-worker": {
+      "types": "./dist/node/extraction-worker.d.mts",
+      "import": "./dist/node/extraction-worker.mjs",
+      "default": "./dist/node/extraction-worker.mjs"
+    }
+  },
+  "files": [
+    "src/**/*.ts",
+    "!src/**/*.test.ts",
+    "!src/**/*.test.tsx",
+    "dist/**/*",
+    "README.md"
+  ],
+  "publishConfig": {
+    "access": "public",
+    "provenance": true
+  },
+  "dependencies": {
+    "effect": "4.0.0",
+    "fflate": "^0.8.3",
+    "mammoth": "^1.13.0",
+    "unpdf": "^1.8.1",
+    "@yolk-sdk/knowledge": "^0.1.0-canary.98"
+  },
+  "peerDependencies": {
+    "xlsx": ">=0.20.3"
+  },
+  "peerDependenciesMeta": {
+    "xlsx": {
+      "optional": true
+    }
+  },
+  "devDependencies": {
+    "xlsx": "https://cdn.sheetjs.com/xlsx-0.20.3/xlsx-0.20.3.tgz"
+  },
+  "scripts": {
+    "build": "tsdown",
+    "check": "tsc -p tsconfig.json --noEmit",
+    "test": "vitest run --passWithNoTests",
+    "test:run": "vitest run --passWithNoTests"
+  }
+}