Member Avatar for chand.

I need an opensource French to English converter, which is able to translate one pdf to another and vice versa.

Many thanks for helping me out.

  • 2 Contributors
  • 2 Replies
  • 237 Views
  • 1 Day Discussion Span
  • Latest Post Latest Post by chand.

Dani AI

Generated

Quick practical plan for : the task breaks into two cases — a “digital” PDF (contains selectable text) or a scanned/image PDF. The reliable, open‑source approach is the same in both cases: extract or OCR the source, run an offline translation engine, then rebuild a PDF (or convert via HTML to preserve layout). ’s hint to search is fair; the bullets below give concrete, usable toolchains and a tiny example so this can be run end‑to‑end.

Tools and when to use them

  • If the PDF already has text: extract with Poppler’s pdftotext (use -layout to keep column positions).
  • If it’s scanned: run Tesseract (or OCRmyPDF, which wraps Tesseract and automates cleaning/embedding) to produce searchable text or a searchable PDF. (manpages.debian.org)

Translation engines (open source)

  • Argos Translate: local/offline NMT with a CLI/Python API and a file‑translation helper (supports .pdf via the files helper).
  • OPUS‑MT (Helsinki‑NLP) offers ready-made fr→en transformer models (good if you script with Marian/HuggingFace).
  • LibreTranslate is an open API/web UI built on Argos that can also translate HTML fragments (useful for layout pipelines). (github.com)

A reliable layout‑preserving pipeline (short example)

  • Convert PDF → HTML, translate HTML nodes, then render HTML → PDF:
    pdf2htmlEX input.pdf output.html
# translate HTML (example using a local LibreTranslate instance; extract .translatedText)
curl -s -X POST 'http://localhost:5000/translate' -H 'Content-Type: application/json' \
-d '{"q":"<body>...HTML...</body>","source":"fr","target":"en","format":"html"}' \
| jq -r .translatedText > translated.html
wkhtmltopdf translated.html translated.pdf

    pdf2htmlEX preserves fonts/positions better than plain text extraction; wkhtmltopdf (or WeasyPrint) can render the translated HTML back to PDF. (github.com)

Cautions and tips

  • Machine translation will often break captions, footnotes, tables and math; expect manual fixes for complex layout.
  • Always test with a short sample, confirm OCR language packs (fra) and the translation model you installed, and do a human review for legal/technical documents.

Recommended Answers

All 2 Replies

Member Avatar for chand.

Is it bad if someone already know your query, and you just try yourself and also put it on the forum for quick reply. Is that bad...?

very funny but why you waste your time on this hangy bangy... :p

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.