monocr 0.1.3 → 0.1.4

package/README.md

@@ -1,6 +1,6 @@
-# monocr (JavaScript/Node.js)
+# monocr
 
-Mon language OCR using ONNX Runtime for Node.js applications.
+Mon language (mnw) OCR for Node.js.
 
 ## Installation
 
@@ -13,107 +13,65 @@ npm install monocr
 ```javascript
 const { read_image } = require("monocr");
 
-// Auto-downloads model on first run
-const text = await read_image("path/to/image.jpg");
+// Automatically downloads model on first run
+const text = await read_image("image.jpg");
 console.log(text);
 ```
 
 ## API
 
-### read_image(imagePath, [modelPath], [charsetPath])
+### `read_image(imagePath, [modelPath], [charsetPath])`
 
-Recognize text from an image file.
+Recognizes text from an image file.
 
-**Parameters:**
+- `imagePath` (string): Path to image file.
+- `modelPath` (string, optional): Path to ONNX model. Defaults to `~/.monocr/models/monocr.onnx`.
+- `charsetPath` (string, optional): Path to charset file. Defaults to bundled charset.
 
-- `imagePath` - Path to image file (jpg, png)
-- `modelPath` - (Optional) Path to ONNX model. Defaults to auto-downloaded model.
-- `charsetPath` - (Optional) Path to charset file. Defaults to auto-downloaded file.
+Returns: `Promise<string>`
 
-**Returns:** `Promise<string>` - Recognized text
+### `read_pdf(pdfPath, [modelPath], [charsetPath])`
 
-### read_pdf(pdfPath, modelPath, charsetPath)
+Recognizes text from a PDF file.
 
-Recognize text from a PDF file.
+- `pdfPath` (string): Path to PDF file.
+- `modelPath` (string, optional): As above.
+- `charsetPath` (string, optional): As above.
 
-**Parameters:**
+Returns: `Promise<string[]>` (Array of text per page)
 
-- `pdfPath` - Path to PDF file
-- `modelPath` - Path to ONNX model (optional)
-- `charsetPath` - Path to charset file (optional)
+### `read_image_with_accuracy(imagePath, groundTruth, [modelPath], [charsetPath])`
 
-**Returns:** `Promise<string[]>` - Array of text per page
+Recognizes text and calculates accuracy against ground truth.
 
-### read_image_with_accuracy(imagePath, groundTruth, modelPath, charsetPath)
+- `imagePath` (string): Path to image file.
+- `groundTruth` (string): Expected text.
 
-Recognize text with accuracy measurement.
+Returns: `Promise<{text: string, accuracy: number}>`
 
-**Parameters:**
+## CLI Usage
 
-- `imagePath` - Path to image file
-- `groundTruth` - Expected text for accuracy calculation
-- `modelPath` - Path to ONNX model (optional)
-- `charsetPath` - Path to charset file (optional)
-
-**Returns:** `Promise<{text: string, accuracy: number}>` - Text and accuracy percentage
-
-### MonOCR Class
-
-For advanced usage, use the `MonOCR` class directly:
-
-```javascript
-const { MonOCR } = require("monocr");
-
-const ocr = new MonOCR("model.onnx", "charset.txt");
-await ocr.init();
-
-// Single line
-const text = await ocr.predictLine(imageSource);
-
-// Full page (with line segmentation)
-const results = await ocr.predictPage(imagePath);
-```
-
-## CLI
-
-The package includes a command-line tool:
+The package includes a `monocr` command-line tool.
 
 ```bash
-# Single image
-monocr image path/to/image.jpg
-
-# PDF файл
-monocr pdf path/to/document.pdf
-
-# Batch processing
-monocr batch path/to/images/ -o results.json
-```
-
-## Examples
-
-See the `examples/` directory for detailed usage examples:
+# Download model to cache (optional, happens automatically on first use)
+monocr download
 
-- `simple.js` - Basic image OCR
-- `with-accuracy.js` - OCR with accuracy measurement
-- `batch.js` - Batch processing
+# Recognize single image
+monocr image input.jpg
 
-Run examples:
+# Recognize PDF
+monocr pdf document.pdf
 
-```bash
-node examples/simple.js
+# Batch process directory
+monocr batch ./images -o results.json
 ```
 
 ## Model Files
 
-Models are **automatically downloaded** on first use to `~/.monocr/models/`.
-
-You can also trigger a manual download:
-
-```bash
-monocr download
-```
+The ONNX model (`monocr.onnx`) is downloaded automatically to `~/.monocr/models/` on first use. The charset file is bundled with the package.
 
-Can also specify custom model paths if you prefer offline usage without the default cache.
+To use a custom model, provide the `modelPath` argument to the API functions or CLI.
 
 ## License
 

package/bin/monocr.js

@@ -14,7 +14,7 @@ program
     .command('image <path>')
     .description('Recognize text from an image file')
     .option('-m, --model <path>', 'Path to ONNX model (optional, auto-downloads)')
-    .option('-c, --charset <path>', 'Path to charset file (optional, auto-downloads)')
+    .option('-c, --charset <path>', 'Path to charset file (optional)')
     .action(async (imagePath, options) => {
         try {
             const text = await read_image(imagePath, options.model, options.charset);
@@ -29,7 +29,7 @@ program
     .command('pdf <path>')
     .description('Recognize text from a PDF file')
     .option('-m, --model <path>', 'Path to ONNX model (optional, auto-downloads)')
-    .option('-c, --charset <path>', 'Path to charset file (optional, auto-downloads)')
+    .option('-c, --charset <path>', 'Path to charset file (optional)')
     .action(async (pdfPath, options) => {
         try {
             const pages = await read_pdf(pdfPath, options.model, options.charset);
@@ -48,7 +48,7 @@ program
     .command('batch <directory>')
     .description('Process all images in a directory')
     .option('-m, --model <path>', 'Path to ONNX model (optional, auto-downloads)')
-    .option('-c, --charset <path>', 'Path to charset file (optional, auto-downloads)')
+    .option('-c, --charset <path>', 'Path to charset file (optional)')
     .option('-o, --output <path>', 'Output file for results (optional)')
     .action(async (directory, options) => {
         try {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "monocr",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "Cross-platform Mon (mnw) language OCR using ONNX Runtime. Supports Node.js.",
   "main": "src/index.js",
   "bin": {