LibreOffice plugin to pipe whole Writer documents through Google Translate, that ought to keep most of the page formatting.

βŒˆβŒ‹ βŽ‡ branch:  PageTranslate


Update of "PageTranslate"

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview

Artifact ID: d20405d267ea78bcad23b23a675940ff3f4f86ef
Page Name:PageTranslate
Date: 2020-06-07 13:23:57
Original User: mario
Mimetype:text/x-markdown
Parent: 066c965e0596b86c830fbfdb5a2711724aaeb057 (diff)
Next 3d17fefd92bd3c640cfaa6acc0bb852261fbb10b
Content

OpenOffice/LibreOffice extension that translates whole documents, whilst retaining some formatting.

  • Iterates over paragraphs / text segments / and tables (only linebreaks kept)
  • Utilizes Google translate (or DeepL, Microsoft Translate, MyMemory).
  • And is therefore rather slow.
    (Writer freezes during process for long documents!)
  • Still allows for text-selection mode translation (from original extension).

Test/Install

Download the package file, and import it in LibreOffice via Tools β†’ Extension Manager β†’ Add.

Location Features Link
Source repository Newest package, Linux-only package, no python dependencies pagetranslate.oxt
LibreOffice Extensions Older/stable versions, bundles python requests library https://extensions.libreoffice.org/extensions/pagetranslate
OpenOffice Extensions Full releases, AOO/LO/Windows compatible (bundles translate+requests) https://extensions.openoffice.org/en/project/pagetranslate

Usage

  • Hit the Tβ†’πŸ‡¬πŸ‡§ toolbar button to get entire document translated to English. Have some trust and patience for longer documents.
  • The second button 🏴 translates to the local/system language.
  • Or select a paragraph, to get a partial translation.
  • Alternatively pick a menu entry from Extras > PageTranslate.
  • The β†’System lang entry refers to your OpenOffice/Desktop setting.
  • Whereas β†’ Paragraph locale respects the text language as set in the Writer status bar.
  • Configure options in Toolsβ†’Optionsβ†’Language Settingsβ†’PageTranslate

Caveats

  • Does not yet support anything but Writer documents. (Structure is entirely distinct for Draw/Impress).
  • The plugin is also likely to crash (won't affect Writer) when encountering complex documents (embedded drawings, Calc sheets, OLE elements, etc). It's only been tested with flow text and text tables.
  • Generally there's a log file in /tmp.
  • Not sure if the toolbar icons work at all, or if that's a bug in the Linux version of LibreOffice.
  • OpenOffice support is somewhat preliminary still (albeit it's mostly Python 2.7 fallbacks and XCU variations).
  • For direct content.xml translations, there's an inconvenient tool in off/. (Each office document .odt/.odc is a zip file. Extract the content.xml, apply the script, then repackage the content.xml into the .odc)

ToDo

  • Finalize DeepL web translation mode (for testing really). The API mode is untested, but likely faster because longer paragraphs can be translated at once. And reportedly it yields smoother/more natural translations. (No XML block translation yet.)
  • Unfortunately not sure yet how to bundle requests.
  • Nor how to craft an options/settings page.
  • Support for TextFrames
  • Figure out if getText() provides for some XML serialization, or how to traverse Paragraphs piece by piece (super slow mode: retain more formatting).