Update PDF processing functions

README.md

@@ -96,36 +96,29 @@ Export your OCR results in the format you need:
 5. View results with bounding boxes (if enabled)
 6. Copy or download the extracted text
 
-### Processing PDFs (Multi-Page Documents) - NEW!
+### Processing PDFs (Multi-Page Documents)
 
-1. Select **"PDF Processing"** mode in the toggle
+Processing a PDF is just another way to create an OCR Job — it produces plain,
+editable text exactly like single-image OCR, so it flows into the same review and
+commit workflow.
+
+1. Select **"PDF Processing"** in the file-type toggle
 2. Upload a PDF file (up to 100MB)
-3. Choose your OCR mode (same as above)
-4. Select **output format**:
-   - 📝 **Markdown** - For documentation, wikis, GitHub
-   - 🌐 **HTML** - For web publishing, styled viewing
-   - 📄 **DOCX** - For Word editing, professional documents
-   - 📊 **JSON** - For programmatic access, data extraction
-5. Click **"Process PDF"**
-6. Watch the progress bar as pages are processed
-7. Your file downloads automatically when complete!
+3. Pick a **Model** and **Mode** (same as image OCR)
+4. Click **"Process PDF"**
+5. All pages are OCR'd and concatenated into one plain-text result
+6. Edit the text, fill in metadata, and **Commit Job** — the PDF is stored as the
+   job's source document (shown alongside the text in Browse Jobs)
+
+> The `/api/process-pdf` endpoint still supports Markdown / HTML / DOCX / JSON
+> exports for programmatic use (see the API reference below); the UI uses the
+> plain-text (JSON) path to drive the job workflow.
 
 ### Tips for Best Results
 
 - **For scanned documents**: Use higher DPI (144-300) in advanced settings
 - **For tables**: The model excels at extracting structured data
 - **For formulas**: Mathematical notation is preserved in output
-- **For images in PDFs**: Enable "Extract Images" to include them in output
-- **For large PDFs**: JSON format is fastest, DOCX takes longer due to formatting
-
-### Output Format Comparison
-
-| Format | Best For | Features | File Size |
-|--------|----------|----------|-----------|
-| **Markdown** | Documentation, GitHub, wikis | Clean text, tables, code blocks | Smallest |
-| **HTML** | Web viewing, sharing | Styled output, embedded images, tables | Medium |
-| **DOCX** | Editing, professional docs | Full formatting, images, tables | Largest |
-| **JSON** | Data processing, APIs | Structured data, metadata, page info | Small |
 
 ## Features