@statewalker/content-extractors 0.2.2

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Mikhail Kotelnikov
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,26 @@
+# @statewalker/content-extractors
+
+Content extractors: turn PDF, DOCX, XLSX, Markdown, and HTML into markdown **text** via a mime-aware registry.
+
+## Installation
+
+```sh
+pnpm add @statewalker/content-extractors
+```
+
+## Usage
+
+```ts
+import { createDefaultRegistry } from "@statewalker/content-extractors";
+
+const registry = createDefaultRegistry();
+const extractor = registry.get("report.pdf");
+const markdown = await extractor?.(pdfBytes); // string
+```
+
+## API
+
+- `ContentExtractor` — the core contract: `bytes → Promise<string>`.
+- `ExtractorRegistry` / `createDefaultRegistry` — resolve an extractor by path or MIME type.
+- Format handlers under `./extractors` (PDF, DOCX, XLSX, Markdown, HTML).
+- `html-to-markdown`, `markdown-to-html`, `mime-utils`, `collect-bytes` helpers.

package/dist/collect-bytes.d.ts

@@ -0,0 +1,16 @@
+//#region src/collect-bytes.d.ts
+/**
+ * Bridges the streaming world (chunked iterables from fetch, file reads,
+ * etc.) with libraries that require a single contiguous buffer (pdfjs-dist,
+ * mammoth). Returns the original chunk when only one arrives, avoiding
+ * an extra copy in the common single-read case.
+ */
+export declare function collectBytes(content: AsyncIterable<Uint8Array> | Iterable<Uint8Array>): Promise<Uint8Array>;
+/**
+ * Shorthand for text-based extractors (plain text, markdown, HTML) that
+ * need the full content as a string. Avoids duplicating the
+ * collect-then-decode pattern across every text extractor.
+ */
+export declare function collectText(content: AsyncIterable<Uint8Array> | Iterable<Uint8Array>): Promise<string>;
+//#endregion
+//# sourceMappingURL=collect-bytes.d.ts.map

package/dist/collect-bytes.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"collect-bytes.d.ts","names":[],"sources":["../src/collect-bytes.ts"],"mappings":";;;;;;;wBAMsB,aACpB,SAAS,cAAc,cAAc,SAAS,cAC7C,QAAQ;;;;;;wBAyBW,YACpB,SAAS,cAAc,cAAc,SAAS,cAC7C"}

package/dist/collect-bytes.js

@@ -0,0 +1,37 @@
+//#region src/collect-bytes.ts
+/**
+* Bridges the streaming world (chunked iterables from fetch, file reads,
+* etc.) with libraries that require a single contiguous buffer (pdfjs-dist,
+* mammoth). Returns the original chunk when only one arrives, avoiding
+* an extra copy in the common single-read case.
+*/
+async function collectBytes(content) {
+	const chunks = [];
+	let totalLength = 0;
+	for await (const chunk of content) {
+		chunks.push(chunk);
+		totalLength += chunk.length;
+	}
+	const first = chunks[0];
+	if (chunks.length === 1 && first) return first;
+	const result = new Uint8Array(totalLength);
+	let offset = 0;
+	for (const chunk of chunks) {
+		result.set(chunk, offset);
+		offset += chunk.length;
+	}
+	return result;
+}
+/**
+* Shorthand for text-based extractors (plain text, markdown, HTML) that
+* need the full content as a string. Avoids duplicating the
+* collect-then-decode pattern across every text extractor.
+*/
+async function collectText(content) {
+	const bytes = await collectBytes(content);
+	return new TextDecoder().decode(bytes);
+}
+//#endregion
+export { collectBytes, collectText };
+
+//# sourceMappingURL=collect-bytes.js.map

package/dist/collect-bytes.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"collect-bytes.js","names":[],"sources":["../src/collect-bytes.ts"],"sourcesContent":["/**\n * Bridges the streaming world (chunked iterables from fetch, file reads,\n * etc.) with libraries that require a single contiguous buffer (pdfjs-dist,\n * mammoth). Returns the original chunk when only one arrives, avoiding\n * an extra copy in the common single-read case.\n */\nexport async function collectBytes(\n  content: AsyncIterable<Uint8Array> | Iterable<Uint8Array>,\n): Promise<Uint8Array> {\n  const chunks: Uint8Array[] = [];\n  let totalLength = 0;\n  for await (const chunk of content) {\n    chunks.push(chunk);\n    totalLength += chunk.length;\n  }\n  const first = chunks[0];\n  if (chunks.length === 1 && first) {\n    return first;\n  }\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  return result;\n}\n\n/**\n * Shorthand for text-based extractors (plain text, markdown, HTML) that\n * need the full content as a string. Avoids duplicating the\n * collect-then-decode pattern across every text extractor.\n */\nexport async function collectText(\n  content: AsyncIterable<Uint8Array> | Iterable<Uint8Array>,\n): Promise<string> {\n  const bytes = await collectBytes(content);\n  return new TextDecoder().decode(bytes);\n}\n"],"mappings":";;;;;;;AAMA,eAAsB,aACpB,SACqB;CACrB,MAAM,SAAuB,CAAC;CAC9B,IAAI,cAAc;CAClB,WAAW,MAAM,SAAS,SAAS;EACjC,OAAO,KAAK,KAAK;EACjB,eAAe,MAAM;CACvB;CACA,MAAM,QAAQ,OAAO;CACrB,IAAI,OAAO,WAAW,KAAK,OACzB,OAAO;CAET,MAAM,SAAS,IAAI,WAAW,WAAW;CACzC,IAAI,SAAS;CACb,KAAK,MAAM,SAAS,QAAQ;EAC1B,OAAO,IAAI,OAAO,MAAM;EACxB,UAAU,MAAM;CAClB;CACA,OAAO;AACT;;;;;;AAOA,eAAsB,YACpB,SACiB;CACjB,MAAM,QAAQ,MAAM,aAAa,OAAO;CACxC,OAAO,IAAI,YAAY,CAAC,CAAC,OAAO,KAAK;AACvC"}

package/dist/extractor-registry.d.ts

@@ -0,0 +1,38 @@
+import { ContentExtractor } from "./types.js";
+//#region src/extractor-registry.d.ts
+/**
+ * Central routing table that decouples file-format knowledge from the
+ * extraction pipeline. New formats can be added without touching pipeline
+ * code. Longest-suffix matching ensures compound extensions (e.g.
+ * `*.tar.gz`) win over shorter ones, and MIME-type fallback handles
+ * files that arrive without a recognizable extension.
+ */
+export declare class ExtractorRegistry {
+  private patternMap;
+  private mimeMap;
+  /**
+   * Associates a suffix pattern with an extractor so the registry can
+   * route files to the right handler based on their filename.
+   */
+  registerByPattern(pattern: string, extractor: ContentExtractor): void;
+  /**
+   * Provides a fallback route for files that lack a recognizable extension
+   * (e.g. temp files or HTTP responses identified only by Content-Type).
+   */
+  registerByMime(mimeType: string, extractor: ContentExtractor): void;
+  /**
+   * Resolves by filename suffix, preferring the most specific match so
+   * compound extensions like `*.test.html` beat `*.html` when both exist.
+   */
+  getByPath(path: string): ContentExtractor | undefined;
+  /** Direct MIME lookup -- used internally by `get` as the fallback path. */
+  getByMime(mimeType: string): ContentExtractor | undefined;
+  /**
+   * Single entry point for the pipeline -- tries the most reliable signal
+   * (file extension) first, then falls back to MIME type so callers never
+   * need to worry about the resolution strategy.
+   */
+  get(path: string, mimeType?: string): ContentExtractor | undefined;
+}
+//#endregion
+//# sourceMappingURL=extractor-registry.d.ts.map

package/dist/extractor-registry.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extractor-registry.d.ts","names":[],"sources":["../src/extractor-registry.ts"],"mappings":";;;;;;;;;qBASa;UACH;UACA;;;;;EAMR,kBAAkB,iBAAiB,WAAW;;;;;EAQ9C,eAAe,kBAAkB,WAAW;;;;;EAQ5C,UAAU,eAAe;;EAoBzB,UAAU,mBAAmB;;;;;;EAS7B,IAAI,cAAc,oBAAoB"}

package/dist/extractor-registry.js

@@ -0,0 +1,61 @@
+//#region src/extractor-registry.ts
+/**
+* Central routing table that decouples file-format knowledge from the
+* extraction pipeline. New formats can be added without touching pipeline
+* code. Longest-suffix matching ensures compound extensions (e.g.
+* `*.tar.gz`) win over shorter ones, and MIME-type fallback handles
+* files that arrive without a recognizable extension.
+*/
+var ExtractorRegistry = class {
+	patternMap = /* @__PURE__ */ new Map();
+	mimeMap = /* @__PURE__ */ new Map();
+	/**
+	* Associates a suffix pattern with an extractor so the registry can
+	* route files to the right handler based on their filename.
+	*/
+	registerByPattern(pattern, extractor) {
+		this.patternMap.set(pattern, extractor);
+	}
+	/**
+	* Provides a fallback route for files that lack a recognizable extension
+	* (e.g. temp files or HTTP responses identified only by Content-Type).
+	*/
+	registerByMime(mimeType, extractor) {
+		this.mimeMap.set(mimeType, extractor);
+	}
+	/**
+	* Resolves by filename suffix, preferring the most specific match so
+	* compound extensions like `*.test.html` beat `*.html` when both exist.
+	*/
+	getByPath(path) {
+		const lowerFilename = (path.includes("/") ? path.slice(path.lastIndexOf("/") + 1) : path).toLowerCase();
+		let bestMatch;
+		let bestLength = 0;
+		for (const [pattern, extractor] of this.patternMap) {
+			const suffix = pattern.startsWith("*") ? pattern.slice(1) : pattern;
+			if (lowerFilename.endsWith(suffix) && suffix.length > bestLength) {
+				bestMatch = extractor;
+				bestLength = suffix.length;
+			}
+		}
+		return bestMatch;
+	}
+	/** Direct MIME lookup -- used internally by `get` as the fallback path. */
+	getByMime(mimeType) {
+		return this.mimeMap.get(mimeType);
+	}
+	/**
+	* Single entry point for the pipeline -- tries the most reliable signal
+	* (file extension) first, then falls back to MIME type so callers never
+	* need to worry about the resolution strategy.
+	*/
+	get(path, mimeType) {
+		const byPath = this.getByPath(path);
+		if (byPath) return byPath;
+		if (mimeType) return this.getByMime(mimeType);
+	}
+};
+//#endregion
+export { ExtractorRegistry };
+
+//# sourceMappingURL=extractor-registry.js.map

package/dist/extractor-registry.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extractor-registry.js","names":[],"sources":["../src/extractor-registry.ts"],"sourcesContent":["import type { ContentExtractor } from \"./types.js\";\n\n/**\n * Central routing table that decouples file-format knowledge from the\n * extraction pipeline. New formats can be added without touching pipeline\n * code. Longest-suffix matching ensures compound extensions (e.g.\n * `*.tar.gz`) win over shorter ones, and MIME-type fallback handles\n * files that arrive without a recognizable extension.\n */\nexport class ExtractorRegistry {\n  private patternMap = new Map<string, ContentExtractor>();\n  private mimeMap = new Map<string, ContentExtractor>();\n\n  /**\n   * Associates a suffix pattern with an extractor so the registry can\n   * route files to the right handler based on their filename.\n   */\n  registerByPattern(pattern: string, extractor: ContentExtractor): void {\n    this.patternMap.set(pattern, extractor);\n  }\n\n  /**\n   * Provides a fallback route for files that lack a recognizable extension\n   * (e.g. temp files or HTTP responses identified only by Content-Type).\n   */\n  registerByMime(mimeType: string, extractor: ContentExtractor): void {\n    this.mimeMap.set(mimeType, extractor);\n  }\n\n  /**\n   * Resolves by filename suffix, preferring the most specific match so\n   * compound extensions like `*.test.html` beat `*.html` when both exist.\n   */\n  getByPath(path: string): ContentExtractor | undefined {\n    const filename = path.includes(\"/\") ? path.slice(path.lastIndexOf(\"/\") + 1) : path;\n    const lowerFilename = filename.toLowerCase();\n\n    let bestMatch: ContentExtractor | undefined;\n    let bestLength = 0;\n\n    for (const [pattern, extractor] of this.patternMap) {\n      // Pattern format: *.ext or *.compound.ext\n      const suffix = pattern.startsWith(\"*\") ? pattern.slice(1) : pattern;\n      if (lowerFilename.endsWith(suffix) && suffix.length > bestLength) {\n        bestMatch = extractor;\n        bestLength = suffix.length;\n      }\n    }\n\n    return bestMatch;\n  }\n\n  /** Direct MIME lookup -- used internally by `get` as the fallback path. */\n  getByMime(mimeType: string): ContentExtractor | undefined {\n    return this.mimeMap.get(mimeType);\n  }\n\n  /**\n   * Single entry point for the pipeline -- tries the most reliable signal\n   * (file extension) first, then falls back to MIME type so callers never\n   * need to worry about the resolution strategy.\n   */\n  get(path: string, mimeType?: string): ContentExtractor | undefined {\n    const byPath = this.getByPath(path);\n    if (byPath) return byPath;\n    if (mimeType) return this.getByMime(mimeType);\n    return undefined;\n  }\n}\n"],"mappings":";;;;;;;;AASA,IAAa,oBAAb,MAA+B;CAC7B,6BAAqB,IAAI,IAA8B;CACvD,0BAAkB,IAAI,IAA8B;;;;;CAMpD,kBAAkB,SAAiB,WAAmC;EACpE,KAAK,WAAW,IAAI,SAAS,SAAS;CACxC;;;;;CAMA,eAAe,UAAkB,WAAmC;EAClE,KAAK,QAAQ,IAAI,UAAU,SAAS;CACtC;;;;;CAMA,UAAU,MAA4C;EAEpD,MAAM,iBADW,KAAK,SAAS,GAAG,IAAI,KAAK,MAAM,KAAK,YAAY,GAAG,IAAI,CAAC,IAAI,KAAA,CAC/C,YAAY;EAE3C,IAAI;EACJ,IAAI,aAAa;EAEjB,KAAK,MAAM,CAAC,SAAS,cAAc,KAAK,YAAY;GAElD,MAAM,SAAS,QAAQ,WAAW,GAAG,IAAI,QAAQ,MAAM,CAAC,IAAI;GAC5D,IAAI,cAAc,SAAS,MAAM,KAAK,OAAO,SAAS,YAAY;IAChE,YAAY;IACZ,aAAa,OAAO;GACtB;EACF;EAEA,OAAO;CACT;;CAGA,UAAU,UAAgD;EACxD,OAAO,KAAK,QAAQ,IAAI,QAAQ;CAClC;;;;;;CAOA,IAAI,MAAc,UAAiD;EACjE,MAAM,SAAS,KAAK,UAAU,IAAI;EAClC,IAAI,QAAQ,OAAO;EACnB,IAAI,UAAU,OAAO,KAAK,UAAU,QAAQ;CAE9C;AACF"}

package/dist/extractors/docx-extractor.d.ts

@@ -0,0 +1,11 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/docx-extractor.d.ts
+/**
+ * Unlocks Word document content for text-based pipelines via a two-stage
+ * conversion (DOCX -> HTML -> markdown). Supplies both `buffer` and
+ * `arrayBuffer` to mammoth so the same code runs in Node.js and browsers
+ * without environment detection.
+ */
+export declare const docxExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=docx-extractor.d.ts.map

package/dist/extractors/docx-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"docx-extractor.d.ts","names":[],"sources":["../../src/extractors/docx-extractor.ts"],"mappings":";;;;;;;;qBAWa,eAAe"}

package/dist/extractors/docx-extractor.js

@@ -0,0 +1,23 @@
+import { collectBytes } from "../collect-bytes.js";
+import { htmlToMarkdown } from "../html-to-markdown.js";
+import mammoth from "mammoth";
+//#region src/extractors/docx-extractor.ts
+/**
+* Unlocks Word document content for text-based pipelines via a two-stage
+* conversion (DOCX -> HTML -> markdown). Supplies both `buffer` and
+* `arrayBuffer` to mammoth so the same code runs in Node.js and browsers
+* without environment detection.
+*/
+const docxExtractor = async (content) => {
+	const bytes = await collectBytes(content);
+	const arrayBuffer = bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength);
+	const result = await mammoth.convertToHtml({
+		buffer: arrayBuffer,
+		arrayBuffer
+	});
+	return htmlToMarkdown(result.value);
+};
+//#endregion
+export { docxExtractor };
+
+//# sourceMappingURL=docx-extractor.js.map

package/dist/extractors/docx-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"docx-extractor.js","names":[],"sources":["../../src/extractors/docx-extractor.ts"],"sourcesContent":["import mammoth from \"mammoth\";\nimport { collectBytes } from \"../collect-bytes.js\";\nimport { htmlToMarkdown } from \"../html-to-markdown.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Unlocks Word document content for text-based pipelines via a two-stage\n * conversion (DOCX -> HTML -> markdown). Supplies both `buffer` and\n * `arrayBuffer` to mammoth so the same code runs in Node.js and browsers\n * without environment detection.\n */\nexport const docxExtractor: ContentExtractor = async (content) => {\n  const bytes = await collectBytes(content);\n  const arrayBuffer = bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength);\n  // mammoth uses \"buffer\" in Node.js and \"arrayBuffer\" in browsers\n  const result = await mammoth.convertToHtml({\n    buffer: arrayBuffer,\n    arrayBuffer,\n  } as Parameters<typeof mammoth.convertToHtml>[0]);\n  return htmlToMarkdown(result.value);\n};\n"],"mappings":";;;;;;;;;;AAWA,MAAa,gBAAkC,OAAO,YAAY;CAChE,MAAM,QAAQ,MAAM,aAAa,OAAO;CACxC,MAAM,cAAc,MAAM,OAAO,MAAM,MAAM,YAAY,MAAM,aAAa,MAAM,UAAU;CAE5F,MAAM,SAAS,MAAM,QAAQ,cAAc;EACzC,QAAQ;EACR;CACF,CAAgD;CAChD,OAAO,eAAe,OAAO,KAAK;AACpC"}

package/dist/extractors/html-extractor.d.ts

@@ -0,0 +1,10 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/html-extractor.d.ts
+/**
+ * Converts HTML files to markdown so the pipeline outputs a uniform
+ * text format. Delegates to `htmlToMarkdown` for the actual conversion,
+ * keeping this extractor focused on the bytes-to-string bridging step.
+ */
+export declare const htmlExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=html-extractor.d.ts.map

package/dist/extractors/html-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-extractor.d.ts","names":[],"sources":["../../src/extractors/html-extractor.ts"],"mappings":";;;;;;;qBASa,eAAe"}

package/dist/extractors/html-extractor.js

@@ -0,0 +1,16 @@
+import { collectText } from "../collect-bytes.js";
+import { htmlToMarkdown } from "../html-to-markdown.js";
+//#region src/extractors/html-extractor.ts
+/**
+* Converts HTML files to markdown so the pipeline outputs a uniform
+* text format. Delegates to `htmlToMarkdown` for the actual conversion,
+* keeping this extractor focused on the bytes-to-string bridging step.
+*/
+const htmlExtractor = async (content) => {
+	const html = await collectText(content);
+	return htmlToMarkdown(html);
+};
+//#endregion
+export { htmlExtractor };
+
+//# sourceMappingURL=html-extractor.js.map

package/dist/extractors/html-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-extractor.js","names":[],"sources":["../../src/extractors/html-extractor.ts"],"sourcesContent":["import { collectText } from \"../collect-bytes.js\";\nimport { htmlToMarkdown } from \"../html-to-markdown.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Converts HTML files to markdown so the pipeline outputs a uniform\n * text format. Delegates to `htmlToMarkdown` for the actual conversion,\n * keeping this extractor focused on the bytes-to-string bridging step.\n */\nexport const htmlExtractor: ContentExtractor = async (content) => {\n  const html = await collectText(content);\n  return htmlToMarkdown(html);\n};\n"],"mappings":";;;;;;;;AASA,MAAa,gBAAkC,OAAO,YAAY;CAChE,MAAM,OAAO,MAAM,YAAY,OAAO;CACtC,OAAO,eAAe,IAAI;AAC5B"}

package/dist/extractors/index.d.ts

@@ -0,0 +1,18 @@
+import { docxExtractor } from "./docx-extractor.js";
+import { htmlExtractor } from "./html-extractor.js";
+import { markdownExtractor } from "./markdown-extractor.js";
+import { pdfExtractor } from "./pdf-extractor.js";
+import { textExtractor } from "./text-extractor.js";
+import { xlsxExtractor } from "./xlsx-extractor.js";
+import { ExtractorRegistry } from "../extractor-registry.js";
+//#region src/extractors/index.d.ts
+/**
+ * Provides a batteries-included registry so most consumers can start
+ * extracting immediately without manual format registration. Custom
+ * registries are still supported for apps that need to override or
+ * limit the built-in set.
+ */
+export declare function createDefaultRegistry(): ExtractorRegistry;
+//#endregion
+export { docxExtractor, htmlExtractor, markdownExtractor, pdfExtractor, textExtractor, xlsxExtractor };
+//# sourceMappingURL=index.d.ts.map

package/dist/extractors/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/extractors/index.ts"],"mappings":";;;;;;;;;;;;;;wBAqBgB,yBAAyB"}

package/dist/extractors/index.js

@@ -0,0 +1,31 @@
+import { docxExtractor } from "./docx-extractor.js";
+import { htmlExtractor } from "./html-extractor.js";
+import { markdownExtractor } from "./markdown-extractor.js";
+import { pdfExtractor } from "./pdf-extractor.js";
+import { textExtractor } from "./text-extractor.js";
+import { xlsxExtractor } from "./xlsx-extractor.js";
+import { ExtractorRegistry } from "../extractor-registry.js";
+//#region src/extractors/index.ts
+/**
+* Provides a batteries-included registry so most consumers can start
+* extracting immediately without manual format registration. Custom
+* registries are still supported for apps that need to override or
+* limit the built-in set.
+*/
+function createDefaultRegistry() {
+	const registry = new ExtractorRegistry();
+	registry.registerByPattern("*.md", markdownExtractor);
+	registry.registerByPattern("*.markdown", markdownExtractor);
+	registry.registerByPattern("*.txt", textExtractor);
+	registry.registerByPattern("*.html", htmlExtractor);
+	registry.registerByPattern("*.htm", htmlExtractor);
+	registry.registerByPattern("*.pdf", pdfExtractor);
+	registry.registerByPattern("*.docx", docxExtractor);
+	registry.registerByPattern("*.docm", docxExtractor);
+	registry.registerByPattern("*.xlsx", xlsxExtractor);
+	return registry;
+}
+//#endregion
+export { createDefaultRegistry, docxExtractor, htmlExtractor, markdownExtractor, pdfExtractor, textExtractor, xlsxExtractor };
+
+//# sourceMappingURL=index.js.map

package/dist/extractors/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":[],"sources":["../../src/extractors/index.ts"],"sourcesContent":["export { docxExtractor } from \"./docx-extractor.js\";\nexport { htmlExtractor } from \"./html-extractor.js\";\nexport { markdownExtractor } from \"./markdown-extractor.js\";\nexport { pdfExtractor } from \"./pdf-extractor.js\";\nexport { textExtractor } from \"./text-extractor.js\";\nexport { xlsxExtractor } from \"./xlsx-extractor.js\";\n\nimport { ExtractorRegistry } from \"../extractor-registry.js\";\nimport { docxExtractor } from \"./docx-extractor.js\";\nimport { htmlExtractor } from \"./html-extractor.js\";\nimport { markdownExtractor } from \"./markdown-extractor.js\";\nimport { pdfExtractor } from \"./pdf-extractor.js\";\nimport { textExtractor } from \"./text-extractor.js\";\nimport { xlsxExtractor } from \"./xlsx-extractor.js\";\n\n/**\n * Provides a batteries-included registry so most consumers can start\n * extracting immediately without manual format registration. Custom\n * registries are still supported for apps that need to override or\n * limit the built-in set.\n */\nexport function createDefaultRegistry(): ExtractorRegistry {\n  const registry = new ExtractorRegistry();\n  registry.registerByPattern(\"*.md\", markdownExtractor);\n  registry.registerByPattern(\"*.markdown\", markdownExtractor);\n  registry.registerByPattern(\"*.txt\", textExtractor);\n  registry.registerByPattern(\"*.html\", htmlExtractor);\n  registry.registerByPattern(\"*.htm\", htmlExtractor);\n  registry.registerByPattern(\"*.pdf\", pdfExtractor);\n  registry.registerByPattern(\"*.docx\", docxExtractor);\n  registry.registerByPattern(\"*.docm\", docxExtractor);\n  registry.registerByPattern(\"*.xlsx\", xlsxExtractor);\n  return registry;\n}\n"],"mappings":";;;;;;;;;;;;;;AAqBA,SAAgB,wBAA2C;CACzD,MAAM,WAAW,IAAI,kBAAkB;CACvC,SAAS,kBAAkB,QAAQ,iBAAiB;CACpD,SAAS,kBAAkB,cAAc,iBAAiB;CAC1D,SAAS,kBAAkB,SAAS,aAAa;CACjD,SAAS,kBAAkB,UAAU,aAAa;CAClD,SAAS,kBAAkB,SAAS,aAAa;CACjD,SAAS,kBAAkB,SAAS,YAAY;CAChD,SAAS,kBAAkB,UAAU,aAAa;CAClD,SAAS,kBAAkB,UAAU,aAAa;CAClD,SAAS,kBAAkB,UAAU,aAAa;CAClD,OAAO;AACT"}

package/dist/extractors/markdown-extractor.d.ts

@@ -0,0 +1,11 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/markdown-extractor.d.ts
+/**
+ * Passthrough extractor for markdown files. Functionally identical to
+ * `textExtractor` but registered under its own extension patterns so
+ * the registry can distinguish `.md` from `.txt` -- allowing future
+ * markdown-specific processing without breaking existing registrations.
+ */
+export declare const markdownExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=markdown-extractor.d.ts.map

package/dist/extractors/markdown-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"markdown-extractor.d.ts","names":[],"sources":["../../src/extractors/markdown-extractor.ts"],"mappings":";;;;;;;;qBASa,mBAAmB"}

package/dist/extractors/markdown-extractor.js

@@ -0,0 +1,15 @@
+import { collectText } from "../collect-bytes.js";
+//#region src/extractors/markdown-extractor.ts
+/**
+* Passthrough extractor for markdown files. Functionally identical to
+* `textExtractor` but registered under its own extension patterns so
+* the registry can distinguish `.md` from `.txt` -- allowing future
+* markdown-specific processing without breaking existing registrations.
+*/
+const markdownExtractor = async (content) => {
+	return collectText(content);
+};
+//#endregion
+export { markdownExtractor };
+
+//# sourceMappingURL=markdown-extractor.js.map

package/dist/extractors/markdown-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"markdown-extractor.js","names":[],"sources":["../../src/extractors/markdown-extractor.ts"],"sourcesContent":["import { collectText } from \"../collect-bytes.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Passthrough extractor for markdown files. Functionally identical to\n * `textExtractor` but registered under its own extension patterns so\n * the registry can distinguish `.md` from `.txt` -- allowing future\n * markdown-specific processing without breaking existing registrations.\n */\nexport const markdownExtractor: ContentExtractor = async (content) => {\n  return collectText(content);\n};\n"],"mappings":";;;;;;;;AASA,MAAa,oBAAsC,OAAO,YAAY;CACpE,OAAO,YAAY,OAAO;AAC5B"}

package/dist/extractors/pdf-extractor.d.ts

@@ -0,0 +1,10 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/pdf-extractor.d.ts
+/**
+ * Extracts PDF content page-by-page, formatting each page as a markdown
+ * section with a `## Page N` header. This preserves page boundaries in
+ * the indexed content, making search results more precise.
+ */
+export declare const pdfExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=pdf-extractor.d.ts.map

package/dist/extractors/pdf-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-extractor.d.ts","names":[],"sources":["../../src/extractors/pdf-extractor.ts"],"mappings":";;;;;;;qBASa,cAAc"}

package/dist/extractors/pdf-extractor.js

@@ -0,0 +1,22 @@
+import { collectBytes } from "../collect-bytes.js";
+import { extractText } from "unpdf";
+//#region src/extractors/pdf-extractor.ts
+/**
+* Extracts PDF content page-by-page, formatting each page as a markdown
+* section with a `## Page N` header. This preserves page boundaries in
+* the indexed content, making search results more precise.
+*/
+const pdfExtractor = async (content) => {
+	const bytes = await collectBytes(content);
+	const { text: pages } = await extractText(bytes);
+	const sections = [];
+	for (let i = 0; i < pages.length; i++) {
+		const pageText = pages[i]?.trim();
+		if (pageText) sections.push(`## Page ${i + 1}\n\n${pageText}`);
+	}
+	return sections.join("\n\n");
+};
+//#endregion
+export { pdfExtractor };
+
+//# sourceMappingURL=pdf-extractor.js.map

package/dist/extractors/pdf-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-extractor.js","names":[],"sources":["../../src/extractors/pdf-extractor.ts"],"sourcesContent":["import { extractText } from \"unpdf\";\nimport { collectBytes } from \"../collect-bytes.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Extracts PDF content page-by-page, formatting each page as a markdown\n * section with a `## Page N` header. This preserves page boundaries in\n * the indexed content, making search results more precise.\n */\nexport const pdfExtractor: ContentExtractor = async (content) => {\n  const bytes = await collectBytes(content);\n  const { text: pages } = await extractText(bytes);\n\n  const sections: string[] = [];\n  for (let i = 0; i < pages.length; i++) {\n    const pageText = pages[i]?.trim();\n    if (pageText) {\n      sections.push(`## Page ${i + 1}\\n\\n${pageText}`);\n    }\n  }\n\n  return sections.join(\"\\n\\n\");\n};\n"],"mappings":";;;;;;;;AASA,MAAa,eAAiC,OAAO,YAAY;CAC/D,MAAM,QAAQ,MAAM,aAAa,OAAO;CACxC,MAAM,EAAE,MAAM,UAAU,MAAM,YAAY,KAAK;CAE/C,MAAM,WAAqB,CAAC;CAC5B,KAAK,IAAI,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;EACrC,MAAM,WAAW,MAAM,EAAE,EAAE,KAAK;EAChC,IAAI,UACF,SAAS,KAAK,WAAW,IAAI,EAAE,MAAM,UAAU;CAEnD;CAEA,OAAO,SAAS,KAAK,MAAM;AAC7B"}

package/dist/extractors/text-extractor.d.ts

@@ -0,0 +1,10 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/text-extractor.d.ts
+/**
+ * Identity extractor for plain text -- no conversion needed, just
+ * decode bytes to a string. Exists as a distinct entry so `.txt` files
+ * route through the same pipeline as every other format.
+ */
+export declare const textExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=text-extractor.d.ts.map

package/dist/extractors/text-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"text-extractor.d.ts","names":[],"sources":["../../src/extractors/text-extractor.ts"],"mappings":";;;;;;;qBAQa,eAAe"}

package/dist/extractors/text-extractor.js

@@ -0,0 +1,14 @@
+import { collectText } from "../collect-bytes.js";
+//#region src/extractors/text-extractor.ts
+/**
+* Identity extractor for plain text -- no conversion needed, just
+* decode bytes to a string. Exists as a distinct entry so `.txt` files
+* route through the same pipeline as every other format.
+*/
+const textExtractor = async (content) => {
+	return collectText(content);
+};
+//#endregion
+export { textExtractor };
+
+//# sourceMappingURL=text-extractor.js.map

package/dist/extractors/text-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"text-extractor.js","names":[],"sources":["../../src/extractors/text-extractor.ts"],"sourcesContent":["import { collectText } from \"../collect-bytes.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Identity extractor for plain text -- no conversion needed, just\n * decode bytes to a string. Exists as a distinct entry so `.txt` files\n * route through the same pipeline as every other format.\n */\nexport const textExtractor: ContentExtractor = async (content) => {\n  return collectText(content);\n};\n"],"mappings":";;;;;;;AAQA,MAAa,gBAAkC,OAAO,YAAY;CAChE,OAAO,YAAY,OAAO;AAC5B"}

package/dist/extractors/xlsx-extractor.d.ts

@@ -0,0 +1,11 @@
+import { ContentExtractor } from "../types.js";
+//#region src/extractors/xlsx-extractor.d.ts
+/**
+ * Converts Excel spreadsheets into markdown tables so the content
+ * is searchable and consumable by text-based pipelines. Each worksheet
+ * becomes a separate section with a level-2 heading. Empty sheets are
+ * skipped to avoid noise.
+ */
+export declare const xlsxExtractor: ContentExtractor;
+//#endregion
+//# sourceMappingURL=xlsx-extractor.d.ts.map

package/dist/extractors/xlsx-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-extractor.d.ts","names":[],"sources":["../../src/extractors/xlsx-extractor.ts"],"mappings":";;;;;;;;qBAUa,eAAe"}

package/dist/extractors/xlsx-extractor.js

@@ -0,0 +1,49 @@
+import { collectBytes } from "../collect-bytes.js";
+import ExcelJS from "exceljs";
+//#region src/extractors/xlsx-extractor.ts
+/**
+* Converts Excel spreadsheets into markdown tables so the content
+* is searchable and consumable by text-based pipelines. Each worksheet
+* becomes a separate section with a level-2 heading. Empty sheets are
+* skipped to avoid noise.
+*/
+const xlsxExtractor = async (content) => {
+	const bytes = await collectBytes(content);
+	const workbook = new ExcelJS.Workbook();
+	const arrayBuffer = bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength);
+	await workbook.xlsx.load(arrayBuffer);
+	const sections = [];
+	workbook.eachSheet((sheet) => {
+		const rows = [];
+		sheet.eachRow((row) => {
+			const cells = [];
+			row.eachCell({ includeEmpty: true }, (cell) => {
+				try {
+					const v = cell.value;
+					cells.push(v == null ? "" : String(v));
+				} catch {
+					cells.push("");
+				}
+			});
+			rows.push(cells);
+		});
+		if (rows.length === 0) return;
+		const colCount = Math.max(...rows.map((r) => r.length));
+		const [headers, ...dataRows] = rows.map((r) => {
+			while (r.length < colCount) r.push("");
+			return r;
+		});
+		if (!headers) return;
+		const table = [
+			`| ${headers.join(" | ")} |`,
+			`| ${headers.map(() => "---").join(" | ")} |`,
+			...dataRows.map((r) => `| ${r.join(" | ")} |`)
+		].join("\n");
+		sections.push(`## ${sheet.name}\n\n${table}`);
+	});
+	return sections.join("\n\n");
+};
+//#endregion
+export { xlsxExtractor };
+
+//# sourceMappingURL=xlsx-extractor.js.map

package/dist/extractors/xlsx-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"xlsx-extractor.js","names":[],"sources":["../../src/extractors/xlsx-extractor.ts"],"sourcesContent":["import ExcelJS from \"exceljs\";\nimport { collectBytes } from \"../collect-bytes.js\";\nimport type { ContentExtractor } from \"../types.js\";\n\n/**\n * Converts Excel spreadsheets into markdown tables so the content\n * is searchable and consumable by text-based pipelines. Each worksheet\n * becomes a separate section with a level-2 heading. Empty sheets are\n * skipped to avoid noise.\n */\nexport const xlsxExtractor: ContentExtractor = async (content) => {\n  const bytes = await collectBytes(content);\n  const workbook = new ExcelJS.Workbook();\n  const arrayBuffer = bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength);\n  await workbook.xlsx.load(arrayBuffer as ArrayBuffer);\n\n  const sections: string[] = [];\n\n  workbook.eachSheet((sheet) => {\n    const rows: string[][] = [];\n\n    sheet.eachRow((row) => {\n      const cells: string[] = [];\n      row.eachCell({ includeEmpty: true }, (cell) => {\n        try {\n          const v = cell.value;\n          cells.push(v == null ? \"\" : String(v));\n        } catch {\n          cells.push(\"\");\n        }\n      });\n      rows.push(cells);\n    });\n\n    if (rows.length === 0) return;\n\n    // Normalize column count to the widest row\n    const colCount = Math.max(...rows.map((r) => r.length));\n    const padded = rows.map((r) => {\n      while (r.length < colCount) r.push(\"\");\n      return r;\n    });\n\n    // First row as headers, rest as data\n    const [headers, ...dataRows] = padded;\n    if (!headers) return;\n\n    const headerLine = `| ${headers.join(\" | \")} |`;\n    const separator = `| ${headers.map(() => \"---\").join(\" | \")} |`;\n    const bodyLines = dataRows.map((r) => `| ${r.join(\" | \")} |`);\n\n    const table = [headerLine, separator, ...bodyLines].join(\"\\n\");\n    sections.push(`## ${sheet.name}\\n\\n${table}`);\n  });\n\n  return sections.join(\"\\n\\n\");\n};\n"],"mappings":";;;;;;;;;AAUA,MAAa,gBAAkC,OAAO,YAAY;CAChE,MAAM,QAAQ,MAAM,aAAa,OAAO;CACxC,MAAM,WAAW,IAAI,QAAQ,SAAS;CACtC,MAAM,cAAc,MAAM,OAAO,MAAM,MAAM,YAAY,MAAM,aAAa,MAAM,UAAU;CAC5F,MAAM,SAAS,KAAK,KAAK,WAA0B;CAEnD,MAAM,WAAqB,CAAC;CAE5B,SAAS,WAAW,UAAU;EAC5B,MAAM,OAAmB,CAAC;EAE1B,MAAM,SAAS,QAAQ;GACrB,MAAM,QAAkB,CAAC;GACzB,IAAI,SAAS,EAAE,cAAc,KAAK,IAAI,SAAS;IAC7C,IAAI;KACF,MAAM,IAAI,KAAK;KACf,MAAM,KAAK,KAAK,OAAO,KAAK,OAAO,CAAC,CAAC;IACvC,QAAQ;KACN,MAAM,KAAK,EAAE;IACf;GACF,CAAC;GACD,KAAK,KAAK,KAAK;EACjB,CAAC;EAED,IAAI,KAAK,WAAW,GAAG;EAGvB,MAAM,WAAW,KAAK,IAAI,GAAG,KAAK,KAAK,MAAM,EAAE,MAAM,CAAC;EAOtD,MAAM,CAAC,SAAS,GAAG,YANJ,KAAK,KAAK,MAAM;GAC7B,OAAO,EAAE,SAAS,UAAU,EAAE,KAAK,EAAE;GACrC,OAAO;EACT,CAGoC;EACpC,IAAI,CAAC,SAAS;EAMd,MAAM,QAAQ;GAAC,KAJS,QAAQ,KAAK,KAAK,EAAE;GAIjB,KAHJ,QAAQ,UAAU,KAAK,CAAC,CAAC,KAAK,KAAK,EAAE;GAGtB,GAFpB,SAAS,KAAK,MAAM,KAAK,EAAE,KAAK,KAAK,EAAE,GAER;EAAC,CAAC,CAAC,KAAK,IAAI;EAC7D,SAAS,KAAK,MAAM,MAAM,KAAK,MAAM,OAAO;CAC9C,CAAC;CAED,OAAO,SAAS,KAAK,MAAM;AAC7B"}

package/dist/html-to-markdown.d.ts

@@ -0,0 +1,10 @@
+//#region src/html-to-markdown.d.ts
+/**
+ * Normalizes HTML into markdown so downstream consumers get a uniform
+ * text format regardless of whether the source was HTML, DOCX (via
+ * mammoth's HTML output), or any other HTML-producing extractor.
+ * Uses Turndown with GFM tables and strikethrough support.
+ */
+export declare function htmlToMarkdown(html: string): string;
+//#endregion
+//# sourceMappingURL=html-to-markdown.d.ts.map

package/dist/html-to-markdown.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-to-markdown.d.ts","names":[],"sources":["../src/html-to-markdown.ts"],"mappings":";;;;;;;wBAUgB,eAAe"}

package/dist/html-to-markdown.js

@@ -0,0 +1,25 @@
+import { gfm } from "@joplin/turndown-plugin-gfm";
+import TurndownService from "turndown";
+//#region src/html-to-markdown.ts
+/**
+* Normalizes HTML into markdown so downstream consumers get a uniform
+* text format regardless of whether the source was HTML, DOCX (via
+* mammoth's HTML output), or any other HTML-producing extractor.
+* Uses Turndown with GFM tables and strikethrough support.
+*/
+function htmlToMarkdown(html) {
+	const turndownService = new TurndownService({
+		headingStyle: "atx",
+		hr: "---",
+		bulletListMarker: "*",
+		codeBlockStyle: "fenced",
+		emDelimiter: "*",
+		strongDelimiter: "**"
+	});
+	turndownService.use(gfm);
+	return turndownService.turndown(html);
+}
+//#endregion
+export { htmlToMarkdown };
+
+//# sourceMappingURL=html-to-markdown.js.map

package/dist/html-to-markdown.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-to-markdown.js","names":[],"sources":["../src/html-to-markdown.ts"],"sourcesContent":["/// <reference path=\"./turndown-plugin-gfm.d.ts\" />\nimport { gfm } from \"@joplin/turndown-plugin-gfm\";\nimport TurndownService from \"turndown\";\n\n/**\n * Normalizes HTML into markdown so downstream consumers get a uniform\n * text format regardless of whether the source was HTML, DOCX (via\n * mammoth's HTML output), or any other HTML-producing extractor.\n * Uses Turndown with GFM tables and strikethrough support.\n */\nexport function htmlToMarkdown(html: string): string {\n  const turndownService = new TurndownService({\n    headingStyle: \"atx\",\n    hr: \"---\",\n    bulletListMarker: \"*\",\n    codeBlockStyle: \"fenced\",\n    emDelimiter: \"*\",\n    strongDelimiter: \"**\",\n  });\n  turndownService.use(gfm);\n  return turndownService.turndown(html);\n}\n"],"mappings":";;;;;;;;;AAUA,SAAgB,eAAe,MAAsB;CACnD,MAAM,kBAAkB,IAAI,gBAAgB;EAC1C,cAAc;EACd,IAAI;EACJ,kBAAkB;EAClB,gBAAgB;EAChB,aAAa;EACb,iBAAiB;CACnB,CAAC;CACD,gBAAgB,IAAI,GAAG;CACvB,OAAO,gBAAgB,SAAS,IAAI;AACtC"}

package/dist/index.d.ts

@@ -0,0 +1,14 @@
+import { ContentExtractor, ContentNormalizer } from "./types.js";
+import { docxExtractor } from "./extractors/docx-extractor.js";
+import { htmlExtractor } from "./extractors/html-extractor.js";
+import { markdownExtractor } from "./extractors/markdown-extractor.js";
+import { pdfExtractor } from "./extractors/pdf-extractor.js";
+import { textExtractor } from "./extractors/text-extractor.js";
+import { xlsxExtractor } from "./extractors/xlsx-extractor.js";
+import { ExtractorRegistry } from "./extractor-registry.js";
+import { createDefaultRegistry } from "./extractors/index.js";
+import { collectBytes, collectText } from "./collect-bytes.js";
+import { htmlToMarkdown } from "./html-to-markdown.js";
+import { markdownToHtml } from "./markdown-to-html.js";
+import { getMimeType } from "./mime-utils.js";
+export { ContentExtractor, ContentNormalizer, ExtractorRegistry, collectBytes, collectText, createDefaultRegistry, docxExtractor, getMimeType, htmlExtractor, htmlToMarkdown, markdownExtractor, markdownToHtml, pdfExtractor, textExtractor, xlsxExtractor };