modern-pdf-lib 0.14.1 → 0.15.1

package/README.md

@@ -15,7 +15,7 @@ Create, parse, fill, merge, sign, and manipulate PDF documents<br />in Node, Den
 
 [![npm version](https://img.shields.io/npm/v/modern-pdf-lib?style=flat-square&color=cb3837)](https://www.npmjs.com/package/modern-pdf-lib)
 [![bundle size](https://img.shields.io/badge/gzip-36kb_core-blue?style=flat-square)](https://bundlephobia.com/package/modern-pdf-lib)
-[![tests](https://img.shields.io/badge/tests-2%2C243_passing-brightgreen?style=flat-square)](#)
+[![tests](https://img.shields.io/badge/tests-2%2C323_passing-brightgreen?style=flat-square)](#)
 [![TypeScript](https://img.shields.io/badge/TypeScript-6.0-3178c6?style=flat-square&logo=typescript&logoColor=white)](#)
 [![License: MIT](https://img.shields.io/badge/license-MIT-yellow?style=flat-square)](LICENSE)
 
@@ -68,6 +68,7 @@ const blob = await doc.saveAsBlob();     // Blob (browsers)
 - TrueType & OpenType font embedding
 - Automatic font subsetting
 - JPEG / PNG image embedding
+- Image optimization (JPEG recompression, dedup, grayscale)
 - RGB, CMYK, grayscale colors
 - Linear & radial gradients, tiling patterns
 - Text layout (multiline, combed, auto-size)
@@ -107,6 +108,7 @@ const blob = await doc.saveAsBlob();     // Blob (browsers)
 - Linearization (fast web view)
 - 60+ low-level PDF operators
 - Custom appearance providers
+- CLI: `npx modern-pdf optimize`
 
 </td>
 </tr>
@@ -195,8 +197,12 @@ const blob = await doc.saveAsBlob();     // Blob (browsers)
 <td align="center">Yes</td>
 <td align="center">No</td></tr>
 
+<tr><td><strong>Image optimization</strong></td>
+<td align="center">JPEG recompress, dedup, grayscale</td>
+<td align="center">No</td></tr>
+
 <tr><td><strong>WASM acceleration</strong></td>
-<td align="center">Optional (compression, PNG, fonts, JBIG2)</td>
+<td align="center">Optional (compression, PNG, fonts, JBIG2, JPEG)</td>
 <td align="center">No</td></tr>
 
 <tr><td><strong>Dependencies</strong></td>
@@ -352,6 +358,38 @@ for (const item of items) {
 ```
 </details>
 
+<details>
+<summary><strong>Image Optimization</strong> &mdash; batch compress, deduplicate, CLI</summary>
+
+```ts
+import { loadPdf, initWasm, optimizeAllImages, deduplicateImages } from 'modern-pdf-lib';
+
+await initWasm({ jpeg: true });
+
+const doc = await loadPdf(pdfBytes);
+
+// Deduplicate identical images
+const dedupReport = deduplicateImages(doc);
+
+// Optimize all images (JPEG recompression)
+const report = await optimizeAllImages(doc, {
+  quality: 75,
+  progressive: true,
+  autoGrayscale: true,
+});
+
+console.log(`${report.optimizedImages}/${report.totalImages} images optimized`);
+console.log(`Savings: ${report.savings.toFixed(1)}%`);
+
+const optimized = await doc.save();
+```
+
+**CLI:**
+```sh
+npx modern-pdf optimize report.pdf report-opt.pdf --quality 60 --grayscale --dedup -v
+```
+</details>
+
 <details>
 <summary><strong>PDF/A & Accessibility</strong></summary>
 
@@ -403,6 +441,7 @@ await initWasm({
   deflate: true,   // Faster compression
   png: true,       // Faster PNG decoding
   fonts: true,     // Faster font subsetting
+  jpeg: true,      // JPEG encode/decode for image optimization
 });
 ```
 
@@ -413,6 +452,7 @@ await initWasm({
 | ttf | Font parsing & subsetting | ~3x |
 | shaping | Complex script layout | ~10x |
 | jbig2 | JBIG2 bilevel image decoding | ~3x |
+| jpeg | JPEG encode/decode for image optimization | Required |
 
 <br />
 
@@ -434,8 +474,9 @@ modern-pdf-lib/
     layers/         Optional content groups (OCG)
     outline/        Bookmarks / document outline
     metadata/       XMP metadata, viewer preferences
-    wasm/           Rust crate sources (5 modules)
-  tests/            2,243 tests across 103 suites
+    wasm/           Rust crate sources (6 modules)
+    cli/            CLI tool (modern-pdf optimize)
+  tests/            2,323 tests across 110 suites
   docs/             VitePress documentation
 ```
 
@@ -447,7 +488,7 @@ modern-pdf-lib/
 git clone https://github.com/ABCrimson/modern-pdf-lib.git
 cd modern-pdf-lib
 npm install
-npm test          # 2,243 tests
+npm test          # 2,323 tests
 npm run typecheck # TypeScript 6.0 strict
 npm run build     # ESM + CJS + declarations
 ```

package/dist/batchOptimize-ClXizv19.mjs

@@ -0,0 +1,306 @@
+import { t as __exportAll } from "./rolldown-runtime-95iHPtFO.mjs";
+import { i as PdfName } from "./pdfObjects-DZZ2GPRW.mjs";
+import { n as extractImages, t as decodeImageStream } from "./imageExtract-vjyQyFcT.mjs";
+
+//#region src/assets/image/grayscaleDetect.ts
+/**
+* @module assets/image/grayscaleDetect
+*
+* Grayscale detection and conversion for image optimization.
+*
+* Detects RGB images where all pixels are effectively grayscale
+* (R ≈ G ≈ B) and converts them to single-channel grayscale,
+* reducing data size by ~66%.
+*
+* No Buffer — uses Uint8Array exclusively.
+*/
+/**
+* Check whether an RGB/RGBA image is effectively grayscale.
+*
+* Scans all pixels and checks if R, G, and B channels are within
+* `tolerance` of each other. If ≥99% of pixels pass, the image
+* is considered grayscale.
+*
+* @param pixels    - Raw pixel data (row-major, channel-interleaved).
+* @param width     - Image width in pixels.
+* @param height    - Image height in pixels.
+* @param channels  - Number of channels: 3 (RGB) or 4 (RGBA).
+* @param tolerance - Maximum allowed difference between R, G, and B
+*                    values for a pixel to be considered gray.
+*                    Default: `2`.
+* @returns `true` if the image is effectively grayscale.
+*
+* @example
+* ```ts
+* import { isGrayscaleImage, convertToGrayscale } from 'modern-pdf-lib';
+*
+* if (isGrayscaleImage(pixels, width, height, 3)) {
+*   const grayPixels = convertToGrayscale(pixels, width, height, 3);
+*   // grayPixels has 1 byte per pixel instead of 3
+* }
+* ```
+*/
+function isGrayscaleImage(pixels, width, height, channels, tolerance = 2) {
+	const pixelCount = width * height;
+	const maxNonGray = Math.floor(pixelCount * .01);
+	let nonGrayCount = 0;
+	for (let i = 0; i < pixelCount; i++) {
+		const r = pixels[i * channels];
+		const g = pixels[i * channels + 1];
+		const b = pixels[i * channels + 2];
+		if (Math.max(r, g, b) - Math.min(r, g, b) > tolerance) {
+			nonGrayCount++;
+			if (nonGrayCount > maxNonGray) return false;
+		}
+	}
+	return true;
+}
+/**
+* Convert an RGB/RGBA image to single-channel grayscale.
+*
+* Uses the ITU-R BT.601 luma formula:
+* ```
+* gray = 0.299 × R + 0.587 × G + 0.114 × B
+* ```
+*
+* The alpha channel (if present) is discarded.
+*
+* @param pixels   - Raw pixel data (row-major, channel-interleaved).
+* @param width    - Image width in pixels.
+* @param height   - Image height in pixels.
+* @param channels - Number of channels: 3 (RGB) or 4 (RGBA).
+* @returns Grayscale pixel data (1 byte per pixel).
+*/
+function convertToGrayscale(pixels, width, height, channels) {
+	const pixelCount = width * height;
+	const gray = new Uint8Array(pixelCount);
+	for (let i = 0; i < pixelCount; i++) {
+		const r = pixels[i * channels];
+		const g = pixels[i * channels + 1];
+		const b = pixels[i * channels + 2];
+		gray[i] = Math.round(.299 * r + .587 * g + .114 * b);
+	}
+	return gray;
+}
+
+//#endregion
+//#region src/assets/image/batchOptimize.ts
+var batchOptimize_exports = /* @__PURE__ */ __exportAll({ optimizeAllImages: () => optimizeAllImages });
+/** Minimum image size to bother optimizing (10 KB). */
+const SMALL_IMAGE_THRESHOLD = 10240;
+/**
+* Optimize all images in a PDF document by recompressing them as JPEG.
+*
+* Walks every image XObject in the document, decodes its pixel data,
+* recompresses it as JPEG using the WASM encoder (if available), and
+* replaces the stream data in-place when the result is smaller.
+*
+* **Requires the JPEG WASM module to be initialized** via
+* `initJpegWasm()` or `initWasm({ jpeg: true })`.  Without it,
+* no images will be optimized (all will be skipped).
+*
+* @param doc     - A parsed `PdfDocument` (from `loadPdf()`).
+* @param options - Optimization settings.
+* @returns A report summarizing the optimization results.
+*
+* @example
+* ```ts
+* import { loadPdf, initWasm, optimizeAllImages } from 'modern-pdf-lib';
+*
+* await initWasm({ jpeg: true });
+*
+* const doc = await loadPdf(pdfBytes);
+* const report = await optimizeAllImages(doc);
+*
+* console.log(`Optimized ${report.optimizedImages} of ${report.totalImages} images`);
+* console.log(`Savings: ${report.savings.toFixed(1)}%`);
+*
+* const optimizedBytes = await doc.save();
+* ```
+*/
+async function optimizeAllImages(doc, options = {}) {
+	const quality = options.quality ?? 80;
+	const minSavingsPercent = options.minSavingsPercent ?? 10;
+	const skipSmall = options.skipSmallImages ?? false;
+	const progressive = options.progressive ?? false;
+	const chromaSubsampling = options.chromaSubsampling ?? "4:2:0";
+	const { encodeJpegWasm, isJpegWasmReady } = await import("./bridge-DpzMOnHd.mjs").then((n) => n.t);
+	const { decodeJpegWasm } = await import("./bridge-DpzMOnHd.mjs").then((n) => n.t);
+	const images = extractImages(doc);
+	const perImage = [];
+	let totalOriginal = 0;
+	let totalNew = 0;
+	let optimizedCount = 0;
+	for (const img of images) {
+		totalOriginal += img.compressedSize;
+		if (!isJpegWasmReady()) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: "JPEG WASM encoder not initialized"
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		if (skipSmall && img.compressedSize < SMALL_IMAGE_THRESHOLD) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: `Below size threshold (${SMALL_IMAGE_THRESHOLD} bytes)`
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		if (img.bitsPerComponent !== 8) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: `Unsupported bits per component: ${img.bitsPerComponent}`
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		if (img.colorSpace === "Indexed") {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: "Indexed color space not suitable for JPEG"
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		let pixels;
+		let channels = img.channels;
+		try {
+			if (img.filters[0] === "DCTDecode") {
+				const decoded = decodeJpegWasm(img.stream.data);
+				if (!decoded) {
+					perImage.push({
+						name: img.name,
+						pageIndex: img.pageIndex,
+						originalSize: img.compressedSize,
+						newSize: img.compressedSize,
+						skipped: true,
+						reason: "Failed to decode existing JPEG"
+					});
+					totalNew += img.compressedSize;
+					continue;
+				}
+				pixels = decoded.pixels;
+				channels = decoded.channels;
+			} else pixels = decodeImageStream(img);
+		} catch {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: "Failed to decode image stream"
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		const expectedLen = img.width * img.height * channels;
+		if (pixels.length !== expectedLen) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: `Pixel data length mismatch: got ${pixels.length}, expected ${expectedLen}`
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		if (channels === 4 && img.colorSpace === "DeviceCMYK") {
+			const rgb = new Uint8Array(img.width * img.height * 3);
+			for (let i = 0; i < img.width * img.height; i++) {
+				const c = pixels[i * 4] / 255;
+				const m = pixels[i * 4 + 1] / 255;
+				const y = pixels[i * 4 + 2] / 255;
+				const k = pixels[i * 4 + 3] / 255;
+				rgb[i * 3] = Math.round(255 * (1 - c) * (1 - k));
+				rgb[i * 3 + 1] = Math.round(255 * (1 - m) * (1 - k));
+				rgb[i * 3 + 2] = Math.round(255 * (1 - y) * (1 - k));
+			}
+			pixels = rgb;
+			channels = 3;
+		}
+		if (options.autoGrayscale && (channels === 3 || channels === 4)) {
+			if (isGrayscaleImage(pixels, img.width, img.height, channels)) {
+				pixels = convertToGrayscale(pixels, img.width, img.height, channels);
+				channels = 1;
+			}
+		}
+		const jpegBytes = encodeJpegWasm(pixels, img.width, img.height, channels, quality, progressive, chromaSubsampling);
+		if (!jpegBytes) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: "JPEG encoding failed"
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		const savingsPercent = (img.compressedSize - jpegBytes.length) / img.compressedSize * 100;
+		if (savingsPercent < minSavingsPercent) {
+			perImage.push({
+				name: img.name,
+				pageIndex: img.pageIndex,
+				originalSize: img.compressedSize,
+				newSize: img.compressedSize,
+				skipped: true,
+				reason: `Savings ${savingsPercent.toFixed(1)}% below threshold ${minSavingsPercent}%`
+			});
+			totalNew += img.compressedSize;
+			continue;
+		}
+		img.stream.data = jpegBytes;
+		img.stream.syncLength();
+		const dict = img.stream.dict;
+		dict.set("/Filter", PdfName.of("/DCTDecode"));
+		if (img.colorSpace === "DeviceCMYK" && channels === 3) dict.set("/ColorSpace", PdfName.of("/DeviceRGB"));
+		if (channels === 1) dict.set("/ColorSpace", PdfName.of("/DeviceGray"));
+		dict.delete("/DecodeParms");
+		if (img.colorSpace === "DeviceCMYK") dict.delete("/Decode");
+		optimizedCount++;
+		perImage.push({
+			name: img.name,
+			pageIndex: img.pageIndex,
+			originalSize: img.compressedSize,
+			newSize: jpegBytes.length,
+			skipped: false
+		});
+		totalNew += jpegBytes.length;
+	}
+	const overallSavings = totalOriginal > 0 ? (totalOriginal - totalNew) / totalOriginal * 100 : 0;
+	return {
+		totalImages: images.length,
+		optimizedImages: optimizedCount,
+		originalTotalBytes: totalOriginal,
+		optimizedTotalBytes: totalNew,
+		savings: overallSavings,
+		perImage
+	};
+}
+
+//#endregion
+export { isGrayscaleImage as i, optimizeAllImages as n, convertToGrayscale as r, batchOptimize_exports as t };
+//# sourceMappingURL=batchOptimize-ClXizv19.mjs.map