natural-pdf 0.1.24__tar.gz → 0.1.27__tar.gz

{natural_pdf-0.1.24/natural_pdf.egg-info → natural_pdf-0.1.27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: natural-pdf
-Version: 0.1.24
+Version: 0.1.27
 Summary: A more intuitive interface for working with PDFs
 Author-email: Jonathan Soma <jonathan.soma@gmail.com>
 License-Expression: MIT

{natural_pdf-0.1.24 → natural_pdf-0.1.27}/docs/ocr/index.md

@@ -6,12 +6,12 @@ Got a PDF that's actually just a bunch of scanned images? Or maybe a PDF where t
 
 Natural PDF supports multiple OCR engines, each with different strengths:
 
-| Engine              | Best For | Speed | Memory | Notes |
-|---------------------|----------|-------|--------|-------|
-| **EasyOCR**         | General documents, handwritten text | Moderate | Higher | Good all-around choice |
-| **PaddleOCR**       | Asian languages, when speed matters | Fast | Efficient | Great for Chinese, Japanese, Korean |
-| **Surya OCR**       | Highest accuracy needed | Moderate | Higher (GPU helps) | Best quality results |
-| **Gemini**          | Complex layouts (via API) | Depends on API | N/A | Requires API key |
+- [EasyOCR](https://github.com/JaidedAI/EasyOCR)
+- [PaddleOCR](https://paddlepaddle.github.io/PaddleOCR/latest/en/index.html)
+- [Surya](https://github.com/datalab-to/surya)
+- [DocTR](https://github.com/mindee/doctr)
+
+What are those strengths??? It honestly doesn't even matter, *it's so easy to try each of them you can just see what works best for you*.
 
 If you try to use an engine that isn't installed, Natural PDF will tell you exactly what to install.
 
@@ -27,7 +27,7 @@ pdf = PDF("https://github.com/jsoma/natural-pdf/raw/refs/heads/main/pdfs/needs-o
 page = pdf.pages[0]
 
 # Apply OCR using the default engine
-ocr_elements = page.apply_ocr(languages=['en'])
+ocr_elements = page.apply_ocr()
 
 # Extract the text (uses OCR results automatically)
 text = page.extract_text()
@@ -68,29 +68,16 @@ easy_opts = EasyOCROptions(
     batch_size=8         # Process multiple regions at once
 )
 ocr_elements = page.apply_ocr(engine='easyocr', options=easy_opts)
-
-# Configure Surya for high-accuracy line detection
-surya_opts = SuryaOCROptions(
-    languages=['en', 'de'],
-    min_confidence=0.4   # Minimum confidence for results
-)
-ocr_elements = page.apply_ocr(engine='surya', options=surya_opts)
 ```
 
-## How OCR Actually Works
+## OCRing regions
 
-When you run `page.apply_ocr()`, here's what happens:
+Don't want to apply OCR to an entire page? You don't need to!
 
 ```python
-# Apply OCR to a page - this adds text elements to the page
-ocr_elements = page.apply_ocr(engine='easyocr')
-print(f"Found {len(ocr_elements)} text elements via OCR")
-
-# You can also OCR just a specific region
-title = page.find('text:contains("Title")')
-if title:
-    content_region = title.below(height=300)
-    region_ocr_elements = content_region.apply_ocr(engine='paddle', languages=['en'])
+# Grab the top half of the page
+region = page.region(0, 0, height=page.height/2, width=page.width)
+region.apply_ocr(engine='paddle')
 ```
 
 *Note: Running OCR again on the same area will replace the previous OCR results.*
@@ -195,7 +182,10 @@ Natural PDF includes a web app for reviewing and correcting OCR results:
    create_correction_task_package(pdf, "correction_package.zip", overwrite=True)
    ```
 
-2. **Start the web app:**
+2. **Visit [the live OCR tool](https://jsoma.github.io/natural-pdf/ocr-tool)** and upload your zip file.
+
+If you're a crazy person, alternatively you can do it locally like this:
+
    ```bash
    # Find where Natural PDF is installed
    NATURAL_PDF_PATH=$(python -c "import site; print(site.getsitepackages()[0])")/natural_pdf
@@ -203,10 +193,9 @@ Natural PDF includes a web app for reviewing and correcting OCR results:
    # Start the web server
    cd $NATURAL_PDF_PATH/templates/spa
    python -m http.server 8000
-   ```
 
-3. **Use the app:**
-   Open `http://localhost:8000` in your browser and drag in your `correction_package.zip` file to review and edit the OCR results.
+   # Open http://localhost:8000 in your browser
+   ```
 
 ## Next Steps
 

{natural_pdf-0.1.24/natural_pdf/templates/spa → natural_pdf-0.1.27/docs/ocr-tool}/css/style.css

@@ -195,6 +195,10 @@ button.secondary:hover {
     /* display: inline-block; /* Remove this if using flex parent */
 }
 
+.image-clip-canvas {
+    min-width: 70%;
+}
+
 .editing-content {
     font-size: 18px;
     text-align: center;

natural_pdf-0.1.27/docs/ocr-tool/index.html

@@ -0,0 +1,31 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>OCR Correction Tool</title>
+    <link rel="stylesheet" href="css/style.css">
+</head>
+<body>
+    <header>
+        <h1>OCR Correction Tool</h1>
+    </header>
+
+    <main id="app">
+        <p>Loading application...</p>
+    </main>
+
+    <footer>
+        <p>Generated by natural-pdf</p>
+    </footer>
+
+    <script src="js/vendor/react.development.js"></script>
+    <script src="js/vendor/react-dom.development.js"></script>
+    <script src="js/vendor/babel.min.js"></script>
+    <script src="js/vendor/jszip.min.js"></script>
+    <script src="js/vendor/FileSaver.min.js"></script>
+
+    <script type="text/babel" src="js/app.js"></script>
+
+</body>
+</html> 

natural_pdf-0.1.27/docs/ocr-tool/js/vendor/FileSaver.min.js

@@ -0,0 +1,3 @@
+(function(a,b){if("function"==typeof define&&define.amd)define([],b);else if("undefined"!=typeof exports)b();else{b(),a.FileSaver={exports:{}}.exports}})(this,function(){"use strict";function b(a,b){return"undefined"==typeof b?b={autoBom:!1}:"object"!=typeof b&&(console.warn("Deprecated: Expected third argument to be a object"),b={autoBom:!b}),b.autoBom&&/^\s*(?:text\/\S*|application\/xml|\S*\/\S*\+xml)\s*;.*charset\s*=\s*utf-8/i.test(a.type)?new Blob(["\uFEFF",a],{type:a.type}):a}function c(a,b,c){var d=new XMLHttpRequest;d.open("GET",a),d.responseType="blob",d.onload=function(){g(d.response,b,c)},d.onerror=function(){console.error("could not download file")},d.send()}function d(a){var b=new XMLHttpRequest;b.open("HEAD",a,!1);try{b.send()}catch(a){}return 200<=b.status&&299>=b.status}function e(a){try{a.dispatchEvent(new MouseEvent("click"))}catch(c){var b=document.createEvent("MouseEvents");b.initMouseEvent("click",!0,!0,window,0,0,0,80,20,!1,!1,!1,!1,0,null),a.dispatchEvent(b)}}var f="object"==typeof window&&window.window===window?window:"object"==typeof self&&self.self===self?self:"object"==typeof global&&global.global===global?global:void 0,a=f.navigator&&/Macintosh/.test(navigator.userAgent)&&/AppleWebKit/.test(navigator.userAgent)&&!/Safari/.test(navigator.userAgent),g=f.saveAs||("object"!=typeof window||window!==f?function(){}:"download"in HTMLAnchorElement.prototype&&!a?function(b,g,h){var i=f.URL||f.webkitURL,j=document.createElement("a");g=g||b.name||"download",j.download=g,j.rel="noopener","string"==typeof b?(j.href=b,j.origin===location.origin?e(j):d(j.href)?c(b,g,h):e(j,j.target="_blank")):(j.href=i.createObjectURL(b),setTimeout(function(){i.revokeObjectURL(j.href)},4E4),setTimeout(function(){e(j)},0))}:"msSaveOrOpenBlob"in navigator?function(f,g,h){if(g=g||f.name||"download","string"!=typeof f)navigator.msSaveOrOpenBlob(b(f,h),g);else if(d(f))c(f,g,h);else{var i=document.createElement("a");i.href=f,i.target="_blank",setTimeout(function(){e(i)})}}:function(b,d,e,g){if(g=g||open("","_blank"),g&&(g.document.title=g.document.body.innerText="downloading..."),"string"==typeof b)return c(b,d,e);var h="application/octet-stream"===b.type,i=/constructor/i.test(f.HTMLElement)||f.safari,j=/CriOS\/[\d]+/.test(navigator.userAgent);if((j||h&&i||a)&&"undefined"!=typeof FileReader){var k=new FileReader;k.onloadend=function(){var a=k.result;a=j?a:a.replace(/^data:[^;]*;/,"data:attachment/file;"),g?g.location.href=a:location=a,g=null},k.readAsDataURL(b)}else{var l=f.URL||f.webkitURL,m=l.createObjectURL(b);g?g.location=m:location.href=m,g=null,setTimeout(function(){l.revokeObjectURL(m)},4E4)}});f.saveAs=g.saveAs=g,"undefined"!=typeof module&&(module.exports=g)});
+
+//# sourceMappingURL=FileSaver.min.js.map