afpp 1.2.0 → 1.4.0

package/README.md

@@ -1,5 +1,12 @@
 # afpp
 
+![Version](https://img.shields.io/github/v/release/l2ysho/afpp)
+[![codecov](https://codecov.io/github/l2ysho/afpp/graph/badge.svg?token=2PE32I4M9K)](https://codecov.io/github/l2ysho/afpp)
+![Node](https://img.shields.io/badge/node-%3E%3D%2018.x-brightgreen.svg)
+![npm Downloads](https://img.shields.io/npm/dt/afpp.svg)
+![Repo Size](https://img.shields.io/github/repo-size/l2ysho/afpp)
+![Last Commit](https://img.shields.io/github/last-commit/l2ysho/afpp.svg)
+
 Another f\*cking pdf parser. (alpha)
 
 ## Why?

package/dist/index.js

@@ -1,68 +1,10 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.pdf2string = void 0;
-const promises_1 = require("node:fs/promises");
-const parsePdfFileBuffer = async (options) => import('pdfjs-dist/legacy/build/pdf.mjs').then(async (pdfjsLib) => {
-    const loadingTask = pdfjsLib.getDocument({
-        ...options,
-        verbosity: 0, // TODO enable for debug
-    });
-    const pdfDocument = await loadingTask.promise;
-    const { numPages } = pdfDocument;
-    const pageContents = new Array(numPages).fill('');
-    const pagePromises = [];
-    for (let pageNum = 1; pageNum <= numPages; pageNum += 1) {
-        pagePromises.push(pdfDocument.getPage(pageNum).then(async (page) => {
-            const textContent = await page.getTextContent({
-                includeMarkedContent: false,
-            });
-            // ? Type assertion of items to TextItem[] should be safe because {includeMarkedContent: false}
-            const items = textContent.items;
-            if (items.length === 0) {
-                pageContents[pageNum - 1] = '';
-            }
-            else {
-                const pageText = items.map((item) => item.str || '').join(' ');
-                pageContents[pageNum - 1] = pageText;
-            }
-        }));
-    }
-    await Promise.all(pagePromises);
-    return pageContents;
-});
-/**
- * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to a string.
- *
- * @async
- * @function pdf2string
- *
- * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
- * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
- * @param {string} [options.password] - The password for encrypted PDF files, if required.
- *
- * @since — v1.0.0
- *
- * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
- *
- * @throws {Error} Throws an error if the input type is invalid.
- */
-const pdf2string = async (input, options) => {
-    if (typeof input === 'string') {
-        const fileBuffer = await (0, promises_1.readFile)(input, {});
-        const data = new Uint8Array(fileBuffer);
-        return parsePdfFileBuffer({ data, ...options });
-    }
-    if (Buffer.isBuffer(input)) {
-        const data = new Uint8Array(input);
-        return parsePdfFileBuffer({ data, ...options });
-    }
-    if (input instanceof Uint8Array) {
-        return parsePdfFileBuffer({ data: input, ...options });
-    }
-    if (input instanceof URL) {
-        return parsePdfFileBuffer({ url: input, ...options });
-    }
-    throw new Error(`Invalid source type: ${typeof input}`);
-};
-exports.pdf2string = pdf2string;
+exports.pdf2string = exports.pdf2image = exports.parsePdf = void 0;
+var parsePdf_1 = require("#afpp/src/parsePdf");
+Object.defineProperty(exports, "parsePdf", { enumerable: true, get: function () { return parsePdf_1.parsePdf; } });
+var pdf2image_1 = require("#afpp/src/pdf2image");
+Object.defineProperty(exports, "pdf2image", { enumerable: true, get: function () { return pdf2image_1.pdf2image; } });
+var pdf2string_1 = require("#afpp/src/pdf2string");
+Object.defineProperty(exports, "pdf2string", { enumerable: true, get: function () { return pdf2string_1.pdf2string; } });
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":";;;AAAA,+CAA4C;AAQ5C,MAAM,kBAAkB,GAAG,KAAK,EAAE,OAA+B,EAAE,EAAE,CACnE,MAAM,CAAC,iCAAiC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE;IAChE,MAAM,WAAW,GAAG,QAAQ,CAAC,WAAW,CAAC;QACvC,GAAG,OAAO;QACV,SAAS,EAAE,CAAC,EAAE,wBAAwB;KACvC,CAAC,CAAC;IAEH,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;IAE9C,MAAM,EAAE,QAAQ,EAAE,GAAG,WAAW,CAAC;IACjC,MAAM,YAAY,GAAa,IAAI,KAAK,CAAS,QAAQ,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACpE,MAAM,YAAY,GAAmC,EAAE,CAAC;IAExD,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;QACxD,YAAY,CAAC,IAAI,CACf,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;YAC/C,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC;gBAC5C,oBAAoB,EAAE,KAAK;aAC5B,CAAC,CAAC;YACH,+FAA+F;YAC/F,MAAM,KAAK,GAAG,WAAW,CAAC,KAAmB,CAAC;YAC9C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACvB,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC;YACjC,CAAC;iBAAM,CAAC;gBACN,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAC/D,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,QAAQ,CAAC;YACvC,CAAC;QACH,CAAC,CAAC,CACH,CAAC;IACJ,CAAC;IACD,MAAM,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IAChC,OAAO,YAAY,CAAC;AACtB,CAAC,CAAC,CAAC;AAML;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,GAAG,KAAK,EACtB,KAAyC,EACzC,OAAsB,EACtB,EAAE;IACF,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,MAAM,UAAU,GAAG,MAAM,IAAA,mBAAQ,EAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC7C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,CAAC;QACxC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,KAAK,YAAY,UAAU,EAAE,CAAC;QAChC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,IAAI,KAAK,YAAY,GAAG,EAAE,CAAC;QACzB,OAAO,kBAAkB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACxD,CAAC;IACD,MAAM,IAAI,KAAK,CAAC,wBAAwB,OAAO,KAAK,EAAE,CAAC,CAAC;AAC1D,CAAC,CAAC;AAEO,gCAAU"}
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":";;;AAAA,+CAA8C;AAArC,oGAAA,QAAQ,OAAA;AACjB,iDAAgD;AAAvC,sGAAA,SAAS,OAAA;AAClB,mDAAkD;AAAzC,wGAAA,UAAU,OAAA"}

package/dist/parsePdf.js

@@ -0,0 +1,79 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.parsePdf = void 0;
+/* eslint-disable no-underscore-dangle */
+const promises_1 = require("node:fs/promises");
+const canvas_1 = require("canvas");
+const defaultParsePdfCallback = (content) => content;
+const parsePdfFileBuffer = async (options, callback = defaultParsePdfCallback) => import('pdfjs-dist/legacy/build/pdf.mjs').then(async (pdfjsLib) => {
+    const loadingTask = pdfjsLib.getDocument({
+        ...options,
+        verbosity: 0,
+    });
+    const pdfDocument = await loadingTask.promise;
+    const { numPages } = pdfDocument;
+    const pageContents = Array.from({ length: numPages }, () => null);
+    const pagePromises = [];
+    for (let pageNum = 1; pageNum <= numPages; pageNum += 1) {
+        pagePromises.push(pdfDocument.getPage(pageNum).then(async (page) => {
+            const textContent = await page.getTextContent({
+                includeMarkedContent: false,
+            });
+            const items = textContent.items;
+            if (items.length === 0) {
+                const viewport = page.getViewport({ scale: 1.0 });
+                const canvas = (0, canvas_1.createCanvas)(viewport.width, viewport.height);
+                const context = canvas.getContext('2d');
+                await page.render({ canvasContext: context, viewport }).promise;
+                const imageBuffer = canvas.toBuffer();
+                pageContents[pageNum - 1] = callback(imageBuffer);
+            }
+            else {
+                const pageText = items.map((item) => item.str || '').join(' ');
+                pageContents[pageNum - 1] = callback(pageText);
+            }
+        }));
+    }
+    await Promise.all(pagePromises);
+    return pageContents;
+});
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL). Pages are returned in mixed array of strings (text content) and buffers (image content) with in callback function.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ * @param {function} callback - callback function to add another layer of processing, default callback returns content of page withouth any added processing.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+const parsePdf = async (input, options, callback = defaultParsePdfCallback) => {
+    if (typeof callback !== 'function') {
+        throw new Error(`Invalid callback type: ${typeof callback}`);
+    }
+    if (typeof input === 'string') {
+        const fileBuffer = await (0, promises_1.readFile)(input, {});
+        const data = new Uint8Array(fileBuffer);
+        return parsePdfFileBuffer({ data, ...options }, callback);
+    }
+    if (Buffer.isBuffer(input)) {
+        const data = new Uint8Array(input);
+        return parsePdfFileBuffer({ data, ...options }, callback);
+    }
+    if (input instanceof Uint8Array) {
+        return parsePdfFileBuffer({ data: input, ...options }, callback);
+    }
+    if (input instanceof URL) {
+        return parsePdfFileBuffer({ url: input, ...options }, callback);
+    }
+    throw new Error(`Invalid source type: ${typeof input}`);
+};
+exports.parsePdf = parsePdf;
+//# sourceMappingURL=parsePdf.js.map

package/dist/parsePdf.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parsePdf.js","sourceRoot":"","sources":["../src/parsePdf.ts"],"names":[],"mappings":";;;AAAA,yCAAyC;AACzC,+CAA4C;AAE5C,mCAAsC;AAStC,MAAM,uBAAuB,GAAsC,CAAC,OAAO,EAAE,EAAE,CAC7E,OAAO,CAAC;AAEV,MAAM,kBAAkB,GAAG,KAAK,EAC9B,OAA+B,EAC/B,WAAgC,uBAA8C,EAC9E,EAAE,CACF,MAAM,CAAC,iCAAiC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE;IAChE,MAAM,WAAW,GAAG,QAAQ,CAAC,WAAW,CAAC;QACvC,GAAG,OAAO;QACV,SAAS,EAAE,CAAC;KACb,CAAC,CAAC;IAEH,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;IAE9C,MAAM,EAAE,QAAQ,EAAE,GAAG,WAAW,CAAC;IACjC,MAAM,YAAY,GAAQ,KAAK,CAAC,IAAI,CAClC,EAAE,MAAM,EAAE,QAAQ,EAAE,EACpB,GAAG,EAAE,CAAC,IAAoB,CAC3B,CAAC;IACF,MAAM,YAAY,GAAmC,EAAE,CAAC;IAExD,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;QACxD,YAAY,CAAC,IAAI,CACf,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;YAC/C,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC;gBAC5C,oBAAoB,EAAE,KAAK;aAC5B,CAAC,CAAC;YACH,MAAM,KAAK,GAAG,WAAW,CAAC,KAAmB,CAAC;YAC9C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACvB,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,CAAC,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC;gBAClD,MAAM,MAAM,GAAG,IAAA,qBAAY,EAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;gBAC7D,MAAM,OAAO,GAAG,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;gBAExC,MAAM,IAAI,CAAC,MAAM,CAAC,EAAE,aAAa,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC;gBAChE,MAAM,WAAW,GAAG,MAAM,CAAC,QAAQ,EAAE,CAAC;gBACtC,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,QAAQ,CAAC,WAAW,CAAC,CAAC;YACpD,CAAC;iBAAM,CAAC;gBACN,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAC/D,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,CAAC;YACjD,CAAC;QACH,CAAC,CAAC,CACH,CAAC;IACJ,CAAC;IACD,MAAM,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IAChC,OAAO,YAAY,CAAC;AACtB,CAAC,CAAC,CAAC;AAML;;;;;;;;;;;;;;;;GAgBG;AAEI,MAAM,QAAQ,GAAG,KAAK,EAC3B,KAAyC,EACzC,OAAsB,EACtB,WAAgC,uBAA8C,EAC9E,EAAE;IACF,IAAI,OAAO,QAAQ,KAAK,UAAU,EAAE,CAAC;QACnC,MAAM,IAAI,KAAK,CAAC,0BAA0B,OAAO,QAAQ,EAAE,CAAC,CAAC;IAC/D,CAAC;IACD,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,MAAM,UAAU,GAAG,MAAM,IAAA,mBAAQ,EAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC7C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,CAAC;QACxC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,EAAE,QAAQ,CAAC,CAAC;IAC5D,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,EAAE,QAAQ,CAAC,CAAC;IAC5D,CAAC;IACD,IAAI,KAAK,YAAY,UAAU,EAAE,CAAC;QAChC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,EAAE,QAAQ,CAAC,CAAC;IACnE,CAAC;IACD,IAAI,KAAK,YAAY,GAAG,EAAE,CAAC;QACzB,OAAO,kBAAkB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,EAAE,QAAQ,CAAC,CAAC;IAClE,CAAC;IACD,MAAM,IAAI,KAAK,CAAC,wBAAwB,OAAO,KAAK,EAAE,CAAC,CAAC;AAC1D,CAAC,CAAC;AAxBW,QAAA,QAAQ,YAwBnB"}

package/dist/pdf2image.js

@@ -0,0 +1,64 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.pdf2image = void 0;
+/* eslint-disable no-underscore-dangle */
+const promises_1 = require("node:fs/promises");
+const canvas_1 = require("canvas");
+const parsePdfFileBuffer = async (options) => import('pdfjs-dist/legacy/build/pdf.mjs').then(async (pdfjsLib) => {
+    const loadingTask = pdfjsLib.getDocument({
+        ...options,
+        verbosity: 0, // TODO enable for debug
+    });
+    const pdfDocument = await loadingTask.promise;
+    const { numPages } = pdfDocument;
+    const pageContents = new Array(numPages).fill(Buffer.from(''));
+    const pagePromises = [];
+    for (let pageNum = 1; pageNum <= numPages; pageNum += 1) {
+        pagePromises.push(pdfDocument.getPage(pageNum).then(async (page) => {
+            const viewport = page.getViewport({ scale: 1.0 });
+            const canvas = (0, canvas_1.createCanvas)(viewport.width, viewport.height);
+            const context = canvas.getContext('2d');
+            await page.render({ canvasContext: context, viewport }).promise;
+            const imageBuffer = canvas.toBuffer();
+            pageContents[pageNum - 1] = imageBuffer;
+        }));
+    }
+    await Promise.all(pagePromises);
+    return pageContents;
+});
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to an array of image buffers.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+const pdf2image = async (input, options) => {
+    if (typeof input === 'string') {
+        const fileBuffer = await (0, promises_1.readFile)(input, {});
+        const data = new Uint8Array(fileBuffer);
+        return parsePdfFileBuffer({ data, ...options });
+    }
+    if (Buffer.isBuffer(input)) {
+        const data = new Uint8Array(input);
+        return parsePdfFileBuffer({ data, ...options });
+    }
+    if (input instanceof Uint8Array) {
+        return parsePdfFileBuffer({ data: input, ...options });
+    }
+    if (input instanceof URL) {
+        return parsePdfFileBuffer({ url: input, ...options });
+    }
+    throw new Error(`Invalid source type: ${typeof input}`);
+};
+exports.pdf2image = pdf2image;
+//# sourceMappingURL=pdf2image.js.map

package/dist/pdf2image.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf2image.js","sourceRoot":"","sources":["../src/pdf2image.ts"],"names":[],"mappings":";;;AAAA,yCAAyC;AACzC,+CAA4C;AAE5C,mCAAsC;AAItC,MAAM,kBAAkB,GAAG,KAAK,EAAE,OAA+B,EAAE,EAAE,CACnE,MAAM,CAAC,iCAAiC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE;IAChE,MAAM,WAAW,GAAG,QAAQ,CAAC,WAAW,CAAC;QACvC,GAAG,OAAO;QACV,SAAS,EAAE,CAAC,EAAE,wBAAwB;KACvC,CAAC,CAAC;IAEH,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;IAE9C,MAAM,EAAE,QAAQ,EAAE,GAAG,WAAW,CAAC;IACjC,MAAM,YAAY,GAAa,IAAI,KAAK,CAAS,QAAQ,CAAC,CAAC,IAAI,CAC7D,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAChB,CAAC;IACF,MAAM,YAAY,GAAmC,EAAE,CAAC;IAExD,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;QACxD,YAAY,CAAC,IAAI,CACf,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;YAC/C,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,CAAC,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC;YAClD,MAAM,MAAM,GAAG,IAAA,qBAAY,EAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;YAC7D,MAAM,OAAO,GAAG,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;YAExC,MAAM,IAAI,CAAC,MAAM,CAAC,EAAE,aAAa,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,CAAC,OAAO,CAAC;YAChE,MAAM,WAAW,GAAG,MAAM,CAAC,QAAQ,EAAE,CAAC;YACtC,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,WAAW,CAAC;QAC1C,CAAC,CAAC,CACH,CAAC;IACJ,CAAC;IACD,MAAM,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IAChC,OAAO,YAAY,CAAC;AACtB,CAAC,CAAC,CAAC;AAML;;;;;;;;;;;;;;;GAeG;AACI,MAAM,SAAS,GAAG,KAAK,EAC5B,KAAyC,EACzC,OAAsB,EACtB,EAAE;IACF,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,MAAM,UAAU,GAAG,MAAM,IAAA,mBAAQ,EAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC7C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,CAAC;QACxC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,KAAK,YAAY,UAAU,EAAE,CAAC;QAChC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,IAAI,KAAK,YAAY,GAAG,EAAE,CAAC;QACzB,OAAO,kBAAkB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACxD,CAAC;IACD,MAAM,IAAI,KAAK,CAAC,wBAAwB,OAAO,KAAK,EAAE,CAAC,CAAC;AAC1D,CAAC,CAAC;AApBW,QAAA,SAAS,aAoBpB"}

package/dist/pdf2string.js

@@ -0,0 +1,68 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.pdf2string = void 0;
+const promises_1 = require("node:fs/promises");
+const parsePdfFileBuffer = async (options) => import('pdfjs-dist/legacy/build/pdf.mjs').then(async (pdfjsLib) => {
+    const loadingTask = pdfjsLib.getDocument({
+        ...options,
+        verbosity: 0, // TODO enable for debug
+    });
+    const pdfDocument = await loadingTask.promise;
+    const { numPages } = pdfDocument;
+    const pageContents = new Array(numPages).fill('');
+    const pagePromises = [];
+    for (let pageNum = 1; pageNum <= numPages; pageNum += 1) {
+        pagePromises.push(pdfDocument.getPage(pageNum).then(async (page) => {
+            const textContent = await page.getTextContent({
+                includeMarkedContent: false,
+            });
+            // ? Type assertion of items to TextItem[] should be safe because {includeMarkedContent: false}
+            const items = textContent.items;
+            if (items.length === 0) {
+                pageContents[pageNum - 1] = '';
+            }
+            else {
+                const pageText = items.map((item) => item.str || '').join(' ');
+                pageContents[pageNum - 1] = pageText;
+            }
+        }));
+    }
+    await Promise.all(pagePromises);
+    return pageContents;
+});
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to a string.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+const pdf2string = async (input, options) => {
+    if (typeof input === 'string') {
+        const fileBuffer = await (0, promises_1.readFile)(input, {});
+        const data = new Uint8Array(fileBuffer);
+        return parsePdfFileBuffer({ data, ...options });
+    }
+    if (Buffer.isBuffer(input)) {
+        const data = new Uint8Array(input);
+        return parsePdfFileBuffer({ data, ...options });
+    }
+    if (input instanceof Uint8Array) {
+        return parsePdfFileBuffer({ data: input, ...options });
+    }
+    if (input instanceof URL) {
+        return parsePdfFileBuffer({ url: input, ...options });
+    }
+    throw new Error(`Invalid source type: ${typeof input}`);
+};
+exports.pdf2string = pdf2string;
+//# sourceMappingURL=pdf2string.js.map

package/dist/pdf2string.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf2string.js","sourceRoot":"","sources":["../src/pdf2string.ts"],"names":[],"mappings":";;;AAAA,+CAA4C;AAQ5C,MAAM,kBAAkB,GAAG,KAAK,EAAE,OAA+B,EAAE,EAAE,CACnE,MAAM,CAAC,iCAAiC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE;IAChE,MAAM,WAAW,GAAG,QAAQ,CAAC,WAAW,CAAC;QACvC,GAAG,OAAO;QACV,SAAS,EAAE,CAAC,EAAE,wBAAwB;KACvC,CAAC,CAAC;IACH,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;IAE9C,MAAM,EAAE,QAAQ,EAAE,GAAG,WAAW,CAAC;IACjC,MAAM,YAAY,GAAa,IAAI,KAAK,CAAS,QAAQ,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACpE,MAAM,YAAY,GAAmC,EAAE,CAAC;IAExD,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;QACxD,YAAY,CAAC,IAAI,CACf,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;YAC/C,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC;gBAC5C,oBAAoB,EAAE,KAAK;aAC5B,CAAC,CAAC;YACH,+FAA+F;YAC/F,MAAM,KAAK,GAAG,WAAW,CAAC,KAAmB,CAAC;YAC9C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACvB,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC;YACjC,CAAC;iBAAM,CAAC;gBACN,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAC/D,YAAY,CAAC,OAAO,GAAG,CAAC,CAAC,GAAG,QAAQ,CAAC;YACvC,CAAC;QACH,CAAC,CAAC,CACH,CAAC;IACJ,CAAC;IACD,MAAM,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IAChC,OAAO,YAAY,CAAC;AACtB,CAAC,CAAC,CAAC;AAML;;;;;;;;;;;;;;;GAeG;AACI,MAAM,UAAU,GAAG,KAAK,EAC7B,KAAyC,EACzC,OAAsB,EACtB,EAAE;IACF,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,MAAM,UAAU,GAAG,MAAM,IAAA,mBAAQ,EAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC7C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,CAAC;QACxC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3B,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IAClD,CAAC;IACD,IAAI,KAAK,YAAY,UAAU,EAAE,CAAC;QAChC,OAAO,kBAAkB,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACzD,CAAC;IACD,IAAI,KAAK,YAAY,GAAG,EAAE,CAAC;QACzB,OAAO,kBAAkB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACxD,CAAC;IACD,MAAM,IAAI,KAAK,CAAC,wBAAwB,OAAO,KAAK,EAAE,CAAC,CAAC;AAC1D,CAAC,CAAC;AApBW,QAAA,UAAU,cAoBrB"}

package/dist/types/index.d.ts

@@ -1,21 +1,3 @@
-type ParseOptions = {
-    password?: string;
-};
-/**
- * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to a string.
- *
- * @async
- * @function pdf2string
- *
- * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
- * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
- * @param {string} [options.password] - The password for encrypted PDF files, if required.
- *
- * @since — v1.0.0
- *
- * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
- *
- * @throws {Error} Throws an error if the input type is invalid.
- */
-declare const pdf2string: (input: Buffer | URL | Uint8Array | string, options?: ParseOptions) => Promise<string[]>;
-export { pdf2string };
+export { parsePdf } from '#afpp/src/parsePdf';
+export { pdf2image } from '#afpp/src/pdf2image';
+export { pdf2string } from '#afpp/src/pdf2string';

package/dist/types/parsePdf.d.ts

@@ -0,0 +1,23 @@
+type ParsePdfCallback<T> = (content: Buffer | string) => T;
+type ParseOptions = {
+    password?: string;
+};
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL). Pages are returned in mixed array of strings (text content) and buffers (image content) with in callback function.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ * @param {function} callback - callback function to add another layer of processing, default callback returns content of page withouth any added processing.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+export declare const parsePdf: <T>(input: Buffer | URL | Uint8Array | string, options?: ParseOptions, callback?: ParsePdfCallback<T>) => Promise<T[]>;
+export {};

package/dist/types/pdf2image.d.ts

@@ -0,0 +1,21 @@
+type ParseOptions = {
+    password?: string;
+};
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to an array of image buffers.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+export declare const pdf2image: (input: Buffer | URL | Uint8Array | string, options?: ParseOptions) => Promise<Buffer[]>;
+export {};

package/dist/types/pdf2string.d.ts

@@ -0,0 +1,21 @@
+type ParseOptions = {
+    password?: string;
+};
+/**
+ * Converts a PDF file from various input formats (Buffer, Uint8Array, string path, or URL) to a string.
+ *
+ * @async
+ * @function pdf2string
+ *
+ * @param {Buffer|Uint8Array|string|URL} input - The PDF source, which can be a file path, URL, Buffer, or Uint8Array.
+ * @param {Object} [options] - Optional parsing options for customizing the PDF parsing process.
+ * @param {string} [options.password] - The password for encrypted PDF files, if required.
+ *
+ * @since — v1.0.0
+ *
+ * @returns {Promise<string>} - A promise that resolves to the string representation of the PDF content.
+ *
+ * @throws {Error} Throws an error if the input type is invalid.
+ */
+export declare const pdf2string: (input: Buffer | URL | Uint8Array | string, options?: ParseOptions) => Promise<string[]>;
+export {};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "afpp",
-  "version": "1.2.0",
+  "version": "1.4.0",
   "description": "another f*cking pdf parser",
   "types": "./dist/types/index.d.ts",
   "main": "./dist/index.js",
@@ -13,8 +13,9 @@
     "lint": "eslint .",
     "prebuild": "rm -rf dist",
     "prepare": "husky",
-    "test": "NODE_ENV=test npx tsx --test --test-reporter=@voxpelli/node-test-pretty-reporter test/*.test.ts",
-    "test:coverage": "NODE_ENV=test npx tsx --test --experimental-test-coverage test/*.test.ts",
+    "test": "NODE_ENV=test npx tsx --test --test-reporter @voxpelli/node-test-pretty-reporter test/*.test.ts",
+    "pretest:coverage": "rm -rf coverage",
+    "test:coverage": "c8 --reporter=lcov npm test",
     "typecheck": "tsc -p tsconfig.json --noEmit"
   },
   "repository": {
@@ -38,10 +39,6 @@
     "url": "https://github.com/l2ysho/afpp/issues"
   },
   "homepage": "https://github.com/l2ysho/afpp#readme",
-  "dependencies": {
-    "pdfjs-dist": "4.6.82",
-    "typescript": "5.6.2"
-  },
   "devDependencies": {
     "@commitlint/cli": "19.5.0",
     "@commitlint/config-conventional": "19.5.0",
@@ -49,6 +46,7 @@
     "@typescript-eslint/eslint-plugin": "7.18.0",
     "@typescript-eslint/parser": "7.18.0",
     "@voxpelli/node-test-pretty-reporter": "1.1.2",
+    "c8": "10.1.2",
     "commitizen": "4.3.0",
     "cz-conventional-changelog": "3.3.0",
     "eslint": "8.56.0",
@@ -64,6 +62,11 @@
     "husky": "9.1.6",
     "lint-staged": "15.2.10",
     "semantic-release": "24.1.1",
-    "tsx": "4.19.1"
+    "tsx": "4.19.1",
+    "typescript": "5.6.2"
+  },
+  "dependencies": {
+    "canvas": "2.11.2",
+    "pdfjs-dist": "4.6.82"
   }
 }