PyGlossary Files

What's Changed

Compatibility changes

• XDXF: switch to XSL transform mode by default to fix whitespace issues (#562)
• On Windows and macOS, the default GUI is now tk_wizard (use --ui=tk for the classic Tkinter UI)
• On Linux, auto-launching a GUI is disabled when neither PyGObject (gi) nor Tkinter is available (#713)

Bug fixes (security)

• DataEntry.save: refuse absolute paths and ../ segments to prevent malicious dictionary content from writing files outside the chosen output directory. Thanks to @gistrec for the private report.

Bug fixes (core)

• Avoid loading all plugin modules at startup despite reading from index.json, which could register plugins twice (#712)

Bug fixes (plugins)

• Aard2 Slob writer: fix compatibility with Python 3.14: option file_size_approx misbehaving
• Aard2 Slob writer: fix image link replacement (missing slash in replacement) (#723)
• MOBI writer: fix leading double-quote in CSS charset declaration (#715)

Bug fixes (user interface)

• Interactive CLI: fix text color on light terminal backgrounds
• CLI: exit early when inputFormat is invalid
• Gtk4 on macOS: fix AttributeError for display.get_primary_monitor (#724)
• Gtk3 / Gtk4: fix exception when convertOptions=None
• Tkinter UI: fix not applying infoOverride from flags like --name

Features

New read/write format functionalities

• Babylon BGL writer: added for testing purposes for now; wraps HTML src / href resource links into BGL binary link format when bundled data entries are present
• TMX reader (#716) — Translation Memory eXchange XML (.tmx), the standard interchange format for translation-memory tools. Use case: import source/target translation units from CAT tools or TM archives as glossary entries for lookup or conversion to StarDict, Tabfile, etc.
• XLIFF reader (#716) — XML Localization Interchange File Format (.xlf, .xliff) used in software and document localization pipelines. Use case: turn bilingual XLIFF segments into term/definition pairs for dictionaries or further processing alongside TMX workflows.
• FreeDict Markdown reader — reads FreeDict TEI (.tei) like the existing FreeDict plugin, but stores definitions as Markdown instead of HTML. Use case: feed FreeDict dictionaries into Markdown-friendly targets (e.g. Yomitan, view-glossary, or text-oriented writers) without HTML cleanup.
• WordFrequency.info reader — tab-separated COCA lemma frequency lists (.wordfrequency, e.g. lemmas_60k.txt from wordfrequency.info (www.wordfrequency.info)), with part-of-speech tags and per-genre counts. Use case: build frequency-ranked vocabulary glossaries for learners, flashcard apps, or filtered word lists.
• AppleDict Source reader — Apple Dictionary XML source bundles (.apple directories produced by Dictionary Development Kit). Use case: read and convert existing Apple dictionary source projects on any platform, edit entries outside macOS, or round-trip with the existing AppleDict writer.

New command-line features

Conversion flags and standalone CLI tools (pyglossary, pyglossary-view / view-glossary, pyglossary-diff / diff-glossary):

• --skip-term-regex=REGEX — skip entries whose headword or alternate fully matches a regex (skip_term_regex config key; documented in doc/config.rst and doc/entry-filters.md)
• pyglossary-view (view-glossary): render Markdown definitions (needs rich module)
• Expand built-in help text: entry filters, sorting, read/write options, --interactive / --no-interactive, Gtk4 launcher, --log-time, and SQLite mode notes

New user interface features

• Add ui_tk_wizard: step-by-step Tkinter wizard (--ui=tk_wizard, --tkw); default on Windows and macOS
• Add ui_qt6: Qt 6 wizard-based interface (--ui=qt6, --ui=qt, --qt6, --qt); install with pip install pyglossary[qt6]
• Gtk3 / Gtk4: add Options → Info / Metadata dialog — set Glossary Name, Source Language, and Target Language before conversion (same fields as the Tk/Qt dialogs and CLI flags --name, --source-lang, --target-lang); values are passed as glossary info to the output writer (e.g. StarDict .ifo, EPUB metadata)
• About dialogs / tabs in Gtk3, Gtk4, Tk, Qt6 UIs: show Python and GUI toolkit versions

Improvements

• Yomitan / Yomichan: support deinflection entries (#718)
• WordNet reader: read copyright message and year from comment block (#706)

Other changes

• Update README.md format table with new readers
• Refactor large modules into packages: Babylon BGL reader, Slob, Tkinter UI, interactive CLI
• Replace scripts/gen bash script with a faster Python implementation
• Add ARCHITECTURE.md, CONTRIBUTING.md, and doc/tabfile-spec.md
• Ship some test data files with the repository
• CI: add workflow to prevent large git objects; check that plugin names contain no spaces
• Linting and typing: ruff/mypy cleanup across the codebase

New Contributors

• @daxida made their first contribution in #718

Full Changelog: https://github.com/ilius/pyglossary/compare/5.3.0...5.4.0