@arcships/light-ocr 0.5.3 → 0.5.4

package/README.md

@@ -1,10 +1,36 @@
 # @arcships/light-ocr
 
-The default PP-OCRv6 Small entry for `light-ocr`. It exact-pins one compatible model-free runtime and one Small model package, and remains the only owner of the `light-ocr` command.
+The stable PP-OCRv6 Small entry for local image OCR on Node.js 22 and 24.
 
 ```bash
 npm install @arcships/light-ocr
-light-ocr info --version
+light-ocr image.png --format text
+light-ocr doctor --json
 ```
 
-`0.4.0` is the N2 topology cutover: the facade contains only Small model configuration and delegates the API, native loading, EXIF handling, and CLI implementation to the shared runtime. Tiny and Medium use separate packages and commands, so installing this package still installs only Small.
+The package exact-pins one model-free runtime, the Small model, and the native
+component for the current platform. It has no install script and its complete
+release closure is tested with npm offline and scripts disabled.
+
+```js
+const { createEngine } = require('@arcships/light-ocr');
+
+const engine = await createEngine();
+try {
+  const result = await engine.recognizeEncoded(imageBytes);
+  console.log(result.lines);
+} finally {
+  await engine.close();
+}
+```
+
+PDF and multi-page processing are intentionally separate:
+
+```bash
+npm install @arcships/light-ocr-document@next
+light-ocr-document report.pdf --format jsonl
+```
+
+The Document package is preview software and runs its pinned PDF renderer's
+prebuild installer. Keeping that dependency explicit prevents PDF tooling from
+changing the stable image package's installation contract.

package/package.json

@@ -7,7 +7,7 @@
   },
   "dependencies": {
     "@arcships/light-ocr-model-ppocrv6-small": "0.3.4",
-    "@arcships/light-ocr-runtime": "0.1.3"
+    "@arcships/light-ocr-runtime": "0.1.4"
   },
   "description": "Offline PP-OCRv6 Small OCR for Node.js — the stable default tier",
   "engines": {
@@ -31,9 +31,6 @@
   "main": "./src/index.cjs",
   "module": "./src/index.mjs",
   "name": "@arcships/light-ocr",
-  "optionalDependencies": {
-    "pdfium-native": "0.6.1"
-  },
   "publishConfig": {
     "access": "public",
     "provenance": true
@@ -44,5 +41,5 @@
   },
   "type": "commonjs",
   "types": "./src/index.d.ts",
-  "version": "0.5.3"
+  "version": "0.5.4"
 }

package/src/index.cjs

@@ -1,20 +1,8 @@
 'use strict';
 
-const path = require('node:path');
+const { createModelFacade } = require('@arcships/light-ocr-runtime/facade');
 
-// Try to use workspace dependencies, fallback to local paths
-let createModelFacade;
-try {
-  ({ createModelFacade } = require('@arcships/light-ocr-runtime/facade'));
-} catch {
-  // Fallback to local runtime
-  const facadePath = path.join(__dirname, '..', '..', 'runtime', 'src', 'facade.cjs');
-  ({ createModelFacade } = require(facadePath));
-}
-
-const fs = require('node:fs');
-
-const facade = createModelFacade({
+module.exports = createModelFacade({
   model: 'ppocrv6-small',
   modelPackage: '@arcships/light-ocr-model-ppocrv6-small',
   compatibleBundleIds: [
@@ -37,250 +25,3 @@ const facade = createModelFacade({
     maturity: 'stable',
   },
 });
-
-// Lazy load pdfium-native
-let pdfium = null;
-let pdfiumLoaded = false;
-
-function loadPdfium() {
-  if (pdfiumLoaded) return pdfium;
-  pdfiumLoaded = true;
-  try {
-    pdfium = require('pdfium-native');
-  } catch {
-    pdfium = null;
-  }
-  return pdfium;
-}
-
-function hasPdfSupport() {
-  return loadPdfium() !== null;
-}
-
-async function* processPdf(engine, pdfBuffer, options = {}) {
-  const pdfiumNative = loadPdfium();
-  if (!pdfiumNative) {
-    throw new facade.OcrError('unsupported_capability', 'PDF support not available. Install pdfium-native.');
-  }
-
-  const {
-    pageRange,
-    dpi = 150,
-    maxPages = 100,
-    maxPagePixels = 4096 * 4096,
-    maxTotalPixels = 100 * 1024 * 1024,
-    maxFileBytes = 100 * 1024 * 1024,
-    signal,
-    ocrOptions = {}
-  } = options;
-
-  // Check file size
-  if (pdfBuffer.byteLength > maxFileBytes) {
-    throw new facade.OcrError('resource_limit_exceeded', 
-      `File size ${pdfBuffer.byteLength} exceeds maxFileBytes ${maxFileBytes}`);
-  }
-
-  let totalPixels = 0;
-
-  // Open PDF document
-  const doc = await pdfiumNative.loadDocument(pdfBuffer);
-  
-  try {
-    const pageCount = doc.pageCount;
-    
-    // Apply page range
-    const start = pageRange?.start ? Math.max(1, pageRange.start) : 1;
-    const end = pageRange?.end ? Math.min(pageCount, pageRange.end) : pageCount;
-    
-    // Check page limits
-    if (end - start + 1 > maxPages) {
-      throw new facade.OcrError('resource_limit_exceeded', 
-        `Page count ${end - start + 1} exceeds maxPages ${maxPages}`);
-    }
-
-    for (let i = start; i <= end; i++) {
-      // Check abort signal
-      if (signal?.aborted) {
-        throw new facade.OcrError('internal_error', 'Operation aborted');
-      }
-
-      const page = await doc.getPage(i - 1); // 0-indexed
-      
-      // Get page dimensions
-      const { width, height } = page;
-      const pagePixels = width * height;
-      
-      // Check pixel limits
-      if (pagePixels > maxPagePixels) {
-        throw new facade.OcrError('resource_limit_exceeded', 
-          `Page ${i} pixels ${pagePixels} exceeds maxPagePixels ${maxPagePixels}`);
-      }
-      
-      totalPixels += pagePixels;
-      if (totalPixels > maxTotalPixels) {
-        throw new facade.OcrError('resource_limit_exceeded', 
-          `Total pixels ${totalPixels} exceeds maxTotalPixels ${maxTotalPixels}`);
-      }
-
-      // Render page to PNG
-      const renderStart = Date.now();
-      const scale = dpi / 72; // PDF default is 72 DPI
-      const pngBuffer = await page.render({ scale });
-      const renderTime = (Date.now() - renderStart) * 1000;
-
-      // OCR the rendered image
-      const ocrStart = Date.now();
-      const ocrResult = await engine.recognizeEncoded(pngBuffer, ocrOptions);
-      const ocrTime = (Date.now() - ocrStart) * 1000;
-
-      // Close page to free native memory
-      await page.close();
-
-      // Build page result
-      const pageResult = {
-        index: i - 1,
-        width: ocrResult.imageWidth,
-        height: ocrResult.imageHeight,
-        coordinateSpace: 'pageSpace',
-        structure: 'ocr-order',
-        lines: ocrResult.lines.map((line, idx) => ({
-          id: `L${idx}`,
-          text: line.text,
-          confidence: line.confidence,
-          box: line.box
-        })),
-        source: {
-          kind: 'pdf',
-          mediaType: 'application/pdf',
-          identity: { pageIndex: i - 1 },
-          appliedTransforms: {
-            pdf: {
-              rotation: 0,
-              mediaBox: { x: 0, y: 0, width, height },
-              cropBox: { x: 0, y: 0, width, height },
-              dpi,
-              scale: dpi / 72
-            }
-          }
-        },
-        timingUs: {
-          total: renderTime + ocrTime,
-          decode: renderTime,
-          ocr: ocrTime
-        },
-        modelBundleId: ocrResult.modelBundleId
-      };
-
-      yield pageResult;
-    }
-  } finally {
-    doc.destroy();
-  }
-}
-
-async function* processImages(engine, imageBuffers, options = {}) {
-  const { signal, ocrOptions = {} } = options;
-
-  for (let i = 0; i < imageBuffers.length; i++) {
-    if (signal?.aborted) {
-      throw new facade.OcrError('internal_error', 'Operation aborted');
-    }
-
-    const buffer = imageBuffers[i];
-    
-    const ocrStart = Date.now();
-    const ocrResult = await engine.recognizeEncoded(buffer, {
-      ...ocrOptions,
-      applyExif: true
-    });
-    const ocrTime = (Date.now() - ocrStart) * 1000;
-
-    const pageResult = {
-      index: i,
-      width: ocrResult.imageWidth,
-      height: ocrResult.imageHeight,
-      coordinateSpace: 'pageSpace',
-      structure: 'ocr-order',
-      lines: ocrResult.lines.map((line, idx) => ({
-        id: `L${idx}`,
-        text: line.text,
-        confidence: line.confidence,
-        box: line.box
-      })),
-      source: {
-        kind: 'image',
-        mediaType: 'image/png',
-        identity: { index: i },
-        appliedTransforms: {
-          exif: { orientation: 1, applied: false }
-        }
-      },
-      timingUs: {
-        total: ocrTime,
-        decode: 0,
-        ocr: ocrTime
-      },
-      modelBundleId: ocrResult.modelBundleId
-    };
-
-    yield pageResult;
-  }
-}
-
-async function* recognizeDocument(source, options = {}) {
-  // Create engine if not provided
-  let engine = options.engine;
-  let engineCreated = false;
-  
-  if (!engine) {
-    engine = await facade.createEngine();
-    engineCreated = true;
-  }
-  
-  try {
-    let buffers;
-    let isPdf = false;
-    
-    if (Array.isArray(source)) {
-      // Multiple images
-      buffers = [];
-      for (const s of source) {
-        if (typeof s === 'string') {
-          buffers.push(fs.readFileSync(s));
-        } else {
-          buffers.push(s);
-        }
-      }
-    } else if (typeof source === 'string') {
-      // File path
-      const ext = path.extname(source).toLowerCase();
-      if (ext === '.pdf') {
-        isPdf = true;
-        buffers = [fs.readFileSync(source)];
-      } else {
-        buffers = [fs.readFileSync(source)];
-      }
-    } else {
-      // Buffer
-      isPdf = source[0] === 0x25 && source[1] === 0x50 && source[2] === 0x44 && source[3] === 0x46;
-      buffers = [source];
-    }
-    
-    if (isPdf) {
-      yield* processPdf(engine, buffers[0], options);
-    } else {
-      yield* processImages(engine, buffers, options);
-    }
-  } finally {
-    if (engineCreated) {
-      await engine.close();
-    }
-  }
-}
-
-// Export facade + document capabilities
-module.exports = {
-  ...facade,
-  hasPdfSupport,
-  recognizeDocument
-};