Add to_searchable_pdf: overlay invisible OCR text layer onto image-based PDFs

[Copilot]

Image-only PDF pages have no selectable/searchable text. This adds RapidOCRPDF.to_searchable_pdf() which runs OCR on such pages and writes the results back as an invisible (render_mode=3) text layer, making the PDF fully text-searchable without altering its visual appearance.

Implementation

• _CJK_FONT_RANGES — module-level constant mapping Unicode ranges to PyMuPDF built-in font names (china-s, japan, korea)
• _select_font(text) — static helper; picks the appropriate built-in font for the text's script, falling back to helv for Latin
• _insert_ocr_text(page, box, txt, scale) — static helper that converts pixel-space OCR bounding boxes to PDF points (scale = 72 / dpi) and calls page.insert_text(..., render_mode=3)
• to_searchable_pdf(content, output_path, force_ocr, page_num_list) → bytes — main new API; reuses extract_texts to identify image-only pages, OCRs them, overlays invisible text, and optionally writes the result to disk
• CLI: --output_pdf <path> flag added to parse_args; main() routes to to_searchable_pdf when it is set

Usage

from rapidocr_pdf import RapidOCRPDF

engine = RapidOCRPDF()
engine.to_searchable_pdf("scan.pdf", output_path="searchable.pdf")

rapidocr_pdf scan.pdf --output_pdf searchable.pdf

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.