Dev

.github/workflows/CI.yml

@@ -4,7 +4,7 @@ env:
   APP_NAME: pdfdown
   MACOSX_DEPLOYMENT_TARGET: '10.13'
   CARGO_INCREMENTAL: '1'
-'on':
+on:
   push:
     branches:
       - main
@@ -23,6 +23,7 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 jobs:
+
   lint:
     name: Lint
     runs-on: ubuntu-latest
@@ -203,6 +204,32 @@ jobs:
           brew update
           brew install tesseract
           tesseract --version
+      - name: Download PDFium (Linux)
+        if: runner.os == 'Linux'
+        run: |
+          ARCH=$(uname -m)
+          if [ "$ARCH" = "x86_64" ]; then
+            PDFIUM_ARCH="linux-x64"
+          elif [ "$ARCH" = "aarch64" ]; then
+            PDFIUM_ARCH="linux-arm64"
+          fi
+          curl -L "https://github.com/nicksrandall/pdfium-lib/releases/latest/download/pdfium-${PDFIUM_ARCH}.tgz" -o pdfium.tgz
+          mkdir -p /tmp/pdfium
+          tar xzf pdfium.tgz -C /tmp/pdfium
+          echo "PDFIUM_LIBRARY_PATH=/tmp/pdfium/lib/libpdfium.so" >> $GITHUB_ENV
+      - name: Download PDFium (macOS)
+        if: runner.os == 'macOS'
+        run: |
+          ARCH=$(uname -m)
+          if [ "$ARCH" = "x86_64" ]; then
+            PDFIUM_ARCH="mac-x64"
+          elif [ "$ARCH" = "arm64" ]; then
+            PDFIUM_ARCH="mac-arm64"
+          fi
+          curl -L "https://github.com/nicksrandall/pdfium-lib/releases/latest/download/pdfium-${PDFIUM_ARCH}.tgz" -o pdfium.tgz
+          mkdir -p /tmp/pdfium
+          tar xzf pdfium.tgz -C /tmp/pdfium
+          echo "PDFIUM_LIBRARY_PATH=/tmp/pdfium/lib/libpdfium.dylib" >> $GITHUB_ENV
       - name: Install dependencies
         run: yarn install
       - name: Build OCR
@@ -447,6 +474,32 @@ jobs:
           TESS_PREFIX=$(tesseract --list-langs 2>&1 | head -1 | sed 's/.*"\(.*\)".*/\1/')
           echo "TESSDATA_PREFIX=${TESS_PREFIX}" >> $GITHUB_ENV
           echo "Tesseract data path: ${TESS_PREFIX}"
+      - name: Download PDFium (Linux)
+        if: runner.os == 'Linux'
+        run: |
+          ARCH=$(uname -m)
+          if [ "$ARCH" = "x86_64" ]; then
+            PDFIUM_ARCH="linux-x64"
+          elif [ "$ARCH" = "aarch64" ]; then
+            PDFIUM_ARCH="linux-arm64"
+          fi
+          curl -L "https://github.com/nicksrandall/pdfium-lib/releases/latest/download/pdfium-${PDFIUM_ARCH}.tgz" -o pdfium.tgz
+          mkdir -p /tmp/pdfium
+          tar xzf pdfium.tgz -C /tmp/pdfium
+          echo "PDFIUM_LIBRARY_PATH=/tmp/pdfium/lib/libpdfium.so" >> $GITHUB_ENV
+      - name: Download PDFium (macOS)
+        if: runner.os == 'macOS'
+        run: |
+          ARCH=$(uname -m)
+          if [ "$ARCH" = "x86_64" ]; then
+            PDFIUM_ARCH="mac-x64"
+          elif [ "$ARCH" = "arm64" ]; then
+            PDFIUM_ARCH="mac-arm64"
+          fi
+          curl -L "https://github.com/nicksrandall/pdfium-lib/releases/latest/download/pdfium-${PDFIUM_ARCH}.tgz" -o pdfium.tgz
+          mkdir -p /tmp/pdfium
+          tar xzf pdfium.tgz -C /tmp/pdfium
+          echo "PDFIUM_LIBRARY_PATH=/tmp/pdfium/lib/libpdfium.dylib" >> $GITHUB_ENV
       - name: Install dependencies
         run: yarn install
       - name: Download OCR artifact

Cargo.toml

@@ -3,7 +3,7 @@
 edition = "2024"
 
 name    = "pdfdown"
-version = "0.9.7"
+version = "0.9.8"
 
 [lib]
 crate-type = ["cdylib"]
@@ -20,10 +20,17 @@ napi = { version = "3.0.0", features = ["napi6"] }
 napi-derive = "3.0.0"
 rayon = "1.11"
 tesseract-rs = { version = "0.1", optional = true }
+pdfium-render = { version = "0.8", optional = true, default-features = false, features = [
+  "pdfium_latest",
+  "image_025",
+  "thread_safe",
+  "sync",
+] }
 
 [features]
 default = []
 ocr = ["tesseract-rs"]
+render = ["pdfium-render"]
 
 [build-dependencies]
 napi-build = "2"

README.md

@@ -230,7 +230,7 @@ for (const img of images) {
   console.log(`Page ${img.page} image ${img.imageIndex}: ${img.width}x${img.height} ${img.colorSpace}`)
 }
 ```
-
+ 
 #### Extract images as PNG (async)
 
 ```typescript

index.d.ts

@@ -2,30 +2,19 @@
 /* eslint-disable */
 export declare class PdfDown {
   constructor(buffer: Buffer)
-  /** Sync: extract text per page (reuses the already-parsed document) */
   textPerPage(): Array<PageText>
-  /** Sync: extract images per page (reuses the already-parsed document) */
   imagesPerPage(): Array<PageImage>
-  /** Sync: extract annotations per page (reuses the already-parsed document) */
   annotationsPerPage(): Array<PageAnnotation>
-  /** Sync: get PDF metadata (reuses the already-parsed document) */
   metadata(): PdfMeta
-  /** Async: extract text per page on the libuv thread pool (shares parsed document via Arc) */
   textPerPageAsync(): Promise<Array<PageText>>
-  /** Async: extract images per page on the libuv thread pool (shares parsed document via Arc) */
   imagesPerPageAsync(): Promise<Array<PageImage>>
-  /** Async: extract annotations per page on the libuv thread pool (shares parsed document via Arc) */
   annotationsPerPageAsync(): Promise<Array<PageAnnotation>>
-  /** Async: get PDF metadata on the libuv thread pool (shares parsed document via Arc) */
   metadataAsync(): Promise<PdfMeta>
-  /** Sync: extract everything from the PDF in one call (reuses the already-parsed document) */
   document(): PdfDocument
-  /** Async: extract everything from the PDF on the libuv thread pool (shares parsed document via Arc) */
   documentAsync(): Promise<PdfDocument>
-  /** Sync: extract structured text with header/footer detection */
   structuredText(): Array<StructuredPageText>
-  /** Async: extract structured text with header/footer detection */
   structuredTextAsync(): Promise<Array<StructuredPageText>>
+  renderPagesAsync(opts?: RenderOptions | undefined | null): Promise<Array<RenderedPage>>
 }
 
 export declare const enum BoxType {
@@ -34,6 +23,13 @@ export declare const enum BoxType {
   Unknown = 'Unknown'
 }
 
+export declare function capabilities(): Capabilities
+
+export interface Capabilities {
+  ocr: boolean
+  render: boolean
+}
+
 export declare function extractAnnotationsPerPage(buffer: Buffer): Array<PageAnnotation>
 
 export declare function extractAnnotationsPerPageAsync(buffer: Buffer): Promise<Array<PageAnnotation>>
@@ -133,6 +129,27 @@ export declare function pdfMetadata(buffer: Buffer): PdfMeta
 
 export declare function pdfMetadataAsync(buffer: Buffer): Promise<PdfMeta>
 
+export interface RenderedPage {
+  page: number
+  width: number
+  height: number
+  dpi: number
+  data: Buffer
+}
+
+export declare const enum RenderMode {
+  Auto = 'Auto',
+  Never = 'Never',
+  Always = 'Always'
+}
+
+export interface RenderOptions {
+  dpi?: number
+  mode?: RenderMode
+}
+
+export declare function renderPagesAsync(buffer: Buffer, opts?: RenderOptions | undefined | null): Promise<Array<RenderedPage>>
+
 export interface StructuredPageText {
   page: number
   header: string