supervertaler 1.9.132__tar.gz → 1.9.147__tar.gz

{supervertaler-1.9.132 → supervertaler-1.9.147}/CHANGELOG.md

@@ -2,7 +2,332 @@
 
 All notable changes to Supervertaler are documented in this file.
 
-**Current Version:** v1.9.132 (January 20, 2026)
+**Current Version:** v1.9.147 (January 21, 2026)
+
+## 📁 Persistent User Data Location (v1.9.147) - January 21, 2026
+
+**Major Enhancement:** User data (API keys, TMs, glossaries, prompts, settings) now persists across pip upgrades!
+
+**The Problem:**
+- When users ran `pip install --upgrade supervertaler`, their data was wiped
+- This happened because user_data/ was stored inside the pip package directory
+- pip replaces the entire package directory on upgrade, deleting all user files
+- Users reported losing API keys, TMs, glossaries, and prompts after every update
+
+**The Solution:**
+User data is now stored in a **platform-specific persistent location** that lives OUTSIDE the pip package:
+
+| Platform | Location |
+|----------|----------|
+| **Windows** | `%LOCALAPPDATA%\MichaelBeijer\Supervertaler\` (e.g., `C:\Users\John\AppData\Local\MichaelBeijer\Supervertaler\`) |
+| **macOS** | `~/Library/Application Support/Supervertaler/` |
+| **Linux** | `~/.local/share/Supervertaler/` (follows XDG spec) |
+| **Windows EXE** | `user_data/` folder next to the executable (portable mode) |
+| **Development** | `user_data/` or `user_data_private/` next to the script |
+
+**Automatic Migration:**
+- On first run after upgrade, existing data is automatically migrated from the old location
+- A marker file is left in the old location explaining where data went
+- No user action required - just upgrade and everything moves automatically
+
+**What This Means for Users:**
+- ✅ API keys persist across pip upgrades
+- ✅ Translation Memories are preserved
+- ✅ Glossaries (termbases) are preserved
+- ✅ Custom prompts are preserved
+- ✅ All settings are preserved
+- ✅ Windows EXE users unaffected (already portable)
+
+**Technical Details:**
+- Uses `platformdirs` library for cross-platform path handling
+- Added `migrate_user_data_if_needed()` function for one-time migration
+- Migration is safe: copies (not moves) and only runs if old location has content
+
+---
+
+## 🔑 Gemini/Google API Key Alias Fix (v1.9.146) - January 21, 2026
+
+**Bug Fix:** Fixed "Gemini API Key Missing" error when users had `google=...` instead of `gemini=...` in their api_keys.txt file.
+
+**The Problem:**
+- Users could use either `google=YOUR_KEY` or `gemini=YOUR_KEY` in api_keys.txt
+- Some code paths only checked for `gemini`, others only for `google`
+- This caused confusing "API Key Missing" dialogs even when keys were properly configured
+
+**The Solution:**
+- Added **automatic normalization** in `load_api_keys()` - if either `google` or `gemini` is set, both keys are now populated
+- Users can use either name interchangeably
+- Fixed 6+ locations where the alias wasn't being handled correctly:
+  - Single segment translation (Ctrl+T)
+  - Batch translation
+  - QuickMenu prompts
+  - Async LLM fetch for Translation Results panel
+  - Proofreading (also fixed `anthropic` → `claude` provider name bug)
+
+**For Users:**
+- Both `google=YOUR_KEY` and `gemini=YOUR_KEY` now work identically
+- No action needed - existing api_keys.txt files will continue to work
+
+---
+
+## ✨ memoQ-Style Track Changes in Compare Panel (v1.9.145) - January 20, 2026
+
+**Improvement:** Changed Compare Panel diff highlighting to match memoQ's "Track changes view" style.
+
+**New Styling (memoQ-style):**
+- **Deletions**: Red text + strikethrough (~~deleted~~)
+- **Insertions**: Red text + underline (<u>inserted</u>)
+- **Unchanged**: Normal text
+
+This is cleaner and more familiar to translators who use Word's track changes or memoQ.
+
+**Example:**
+```
+Current: In een derde aspect betreft de huidige uitvinding het gebruik van een dichtingskit...
+TM:      In een tweede aspect betreft de huidige uitvinding een dichtingskit...
+
+Compare Panel shows:
+In een ~~tweede~~ derde aspect betreft de huidige uitvinding het gebruik van een dichtingskit...
+        ^^^^^^^^  ^^^^^^                                      ^^^^^^^^^^^^^^^^
+        red+strike red+under                                  red+underline
+```
+
+---
+
+## ✨ Enhanced Compare Panel Diff Highlighting (v1.9.144) - January 20, 2026
+
+**Improvement:** The Compare Panel now shows a complete diff with additions, deletions, and changes - matching what professional CAT tools display.
+
+**Before:**
+- Only showed text that was **different** (in red)
+- Didn't show what was **added** in the current segment
+- Hard to understand what changed between TM source and current source
+
+**After:**
+- **Red with strikethrough**: Text in TM that was changed/deleted in current segment
+- **Green background**: Text in current segment that was added (not in TM)
+- **Normal text**: Identical text in both
+
+**Example:**
+```
+Current: In een derde aspect betreft de huidige uitvinding het gebruik van een dichtingskit...
+TM:      In een tweede aspect betreft de huidige uitvinding een dichtingskit...
+
+Compare Panel now shows:
+In een [tweede→derde] aspect betreft de huidige uitvinding [+het gebruik van] een dichtingskit...
+```
+
+**Technical:**
+- Changed from character-level to word-level diffing for better readability
+- Additions shown inline in green where they would appear
+- Deletions shown with strikethrough so you can see what was removed
+
+---
+
+## 🐛 Fix: Deleting Glossary Entry from Termview (v1.9.143) - January 20, 2026
+
+**Bug Fix:** Deleting glossary entries via right-click in Termview failed with an error.
+
+**Error in Logs:**
+```
+✗ Error deleting glossary entry from database: 'DatabaseManager' object has no attribute 'get_connection'
+```
+
+**Root Cause:**
+- The delete handler was trying to call `db_manager.get_connection()` which doesn't exist
+- Was using raw SQL instead of the existing `termbase_mgr.delete_term()` method
+
+**The Fix:**
+- Changed to use `self.termbase_mgr.delete_term(term_id)` which already exists and works correctly
+
+---
+
+## 🐛 Fix: Error When Editing Glossary Entry from Termview (v1.9.142) - January 20, 2026
+
+**Bug Fix:** After editing a glossary entry via right-click → Edit in Termview, an error would occur preventing proper refresh.
+
+**Error in Logs:**
+```
+Error refreshing segment matches: SupervertalerQt.on_cell_selected() missing 2 required positional arguments: 'previous_row' and 'previous_col'
+```
+
+**Root Cause:**
+- The `_refresh_current_segment_matches()` method was calling `on_cell_selected(current_row, 2)` with only 2 arguments
+- But `on_cell_selected()` requires 4 arguments: `current_row`, `current_col`, `previous_row`, `previous_col`
+
+**The Fix:**
+- Simplified `_refresh_current_segment_matches()` to use the targeted `_refresh_termbase_display_for_current_segment()` method
+- This method already handles termbase cache clearing and display refresh correctly
+
+---
+
+## 🐛 Fix: Termview Blank After Adding Glossary Term (v1.9.141) - January 20, 2026
+
+**Bug Fix:** After adding a term to a glossary via Alt+Down (or similar shortcuts), the Termview pane would go blank and stay blank until navigating to another segment.
+
+**Error in Logs:**
+```
+Error updating termview: list index out of range
+```
+
+**Root Cause:**
+- The `_refresh_termbase_display_for_current_segment()` method (added in v1.9.140) was using incorrect dictionary keys
+- `find_termbase_matches_in_source()` returns dict with keys `source` and `translation`
+- But the refresh method was looking for keys `source_term` and `target_term`
+- This resulted in empty strings being passed to the termview, causing the index error
+
+**The Fix:**
+- Changed `match.get('source_term', '')` → `match.get('source', '')`
+- Changed `match.get('target_term', '')` → `match.get('translation', '')`
+- Applied same fix to Translation Results panel update section
+
+---
+
+## 🐛 Fix: Adding Glossary Term No Longer Triggers TM Search (v1.9.140) - January 20, 2026
+
+**Performance Fix ([#118](https://github.com/michaelbeijer/Supervertaler/issues/118)):** Adding a term to a glossary was unnecessarily triggering a full Translation Memory search, causing delays.
+
+**The Problem:**
+- When adding a term via Ctrl+E, Ctrl+Q, or Ctrl+Shift+1/2, the app cleared BOTH caches and called `on_cell_selected()`
+- This triggered a complete TM search even though adding a glossary term doesn't change the source/target text
+
+**The Fix:**
+- Created new targeted method `_refresh_termbase_display_for_current_segment()`
+- Only clears the termbase cache (not TM cache)
+- Updates TermView widget with fresh glossary matches
+- Updates Translation Results panel's Termbases section only
+- Re-highlights termbase matches in source cell
+- TM results remain untouched
+
+**Also in this release:**
+- 🎤 Renamed "Voice OFF/ON" button to "Voice Commands OFF/ON" for clarity
+- 🎤 Renamed "Dictate" button to "Dictation"
+
+---
+
+## 📐 Auto-Sizing Segment Number Column (v1.9.138) - January 20, 2026
+
+**Grid UX Improvement:** The segment number column (#) now automatically sizes itself based on:
+- The current font size (scales when you zoom in/out)
+- The number of segments in the project (fits 3-digit, 4-digit numbers, etc.)
+
+Uses Qt font metrics to calculate the exact pixel width needed. No more truncated numbers or wasted space!
+
+**Also in this release:**
+- Status column made slightly narrower (70px → 60px)
+
+---
+
+## 🏷️ Termview Punctuated Terms Fix (v1.9.138) - January 20, 2026
+
+**Fixed: Glossary terms with punctuation now appear in Termview**
+
+Glossary entries like "ca." (with period) or "typisch" (when appearing in parentheses as "(typisch)") were found by the termbase search (grid showed green highlighting) but did NOT appear in the Termview pane.
+
+**Root Cause - Key Mismatch:**
+- `matches_dict` keys were built using the raw source term: `"ca."` (with period)
+- But lookup stripped punctuation from tokens: `"ca."` → `"ca"`
+- Result: `matches_dict.get("ca")` returned nothing because the key was `"ca."`
+
+**The Fix:**
+- When building `matches_dict`, keys are now normalized by stripping punctuation
+- Both the dictionary keys and lookup keys now use the same normalization
+- Added brackets `()[]` to punctuation chars so "(typisch)" matches "typisch"
+
+**Files Modified:**
+- `modules/termview_widget.py` - Normalize punctuation in `matches_dict` keys and lookup
+
+---
+
+## 🔧 Termview Race Condition Fix (v1.9.137) - January 20, 2026
+
+**Fixed: Glossary terms now appear in Termview pane**
+
+When navigating to a segment, the Termview pane showed "No glossary matches" even though:
+- The grid highlighted terms in green (glossary highlighting worked)
+- Force Refresh (F5) would then show the terms correctly
+
+**Root Cause - Timing/Race Condition:**
+- `update_tab_segment_editor()` was called EARLY in the cell selection process
+- It tried to read from `termbase_cache[segment_id]` to update Termview
+- But the termbase search (`find_termbase_matches_in_source()`) hadn't run yet!
+- Result: Termview called with empty list before matches were found
+
+**The Fix:**
+1. Removed premature Termview update from `update_tab_segment_editor()` 
+2. Termview is now updated ONLY after the termbase search completes in `_on_cell_selected_full()`
+3. Also fixed: Termview now updates even when no matches found (shows "No matches" state)
+
+**Files Modified:**
+- `Supervertaler.py` - Removed premature Termview update, fixed condition to always update Termview
+- `modules/termview_widget.py` - Removed debug logging
+
+---
+
+## 📚 Glossary Matching Fix for Punctuation (v1.9.136) - January 20, 2026
+
+**Fixed: Glossary terms with trailing punctuation now match correctly**
+
+Glossary entries like "ca." (with period), "psi", and "typisch" were not being found in the Termview window when they appeared in source text like "ca. 2,2 (270 psi)".
+
+**Root Cause:**
+- When tokenizing source text, punctuation was stripped from words ("ca." → "ca")
+- The database search then looked for "ca" but the glossary had "ca." — no match
+- Short terms in parentheses like "(psi)" were also affected
+
+**The Fix:**
+1. Now searches for BOTH the stripped word AND the original word with punctuation
+2. Database query enhanced with reverse matching: finds glossary terms where the search word matches the term with trailing punctuation stripped
+3. Handles entries like "ca.", "gew.%", "psi", etc.
+
+**Files Modified:**
+- `Supervertaler.py` - Enhanced `find_termbase_matches_in_source()` to search with original punctuation
+- `modules/database_manager.py` - Enhanced `search_termbases()` with punctuation-tolerant matching
+
+---
+
+## 🔍 Filter Now Searches Entire Document (v1.9.135) - January 20, 2026
+
+**Fixed: Filter Source/Target boxes now search across ALL pages**
+
+Previously, the Filter Source and Filter Target boxes above the grid would only search within the currently visible page. If the text you were looking for was on a different page, it wouldn't be found.
+
+Now, filtering searches through the **entire document** regardless of pagination. When a filter is active, ALL matching rows are displayed (pagination is temporarily ignored). When you clear the filter, normal pagination resumes.
+
+**Files Modified:**
+- `Supervertaler.py` - Fixed `_apply_pagination_to_grid()` to show all filter matches
+
+---
+
+## 🔊 Fuzzy TM Match Sound Effect (v1.9.134) - January 20, 2026
+
+**New sound effect option: "Fuzzy TM match found"**
+
+Plays when navigating to a segment and a fuzzy TM match (50-99%) is found, but NO 100% match exists.
+
+**Access:** Settings → General → Sound Effects → "Fuzzy TM match found"
+
+Disabled by default (set to "None"). Works with the same sound options as other effects.
+
+---
+
+## 🔊 New Sound Effects (v1.9.133) - January 20, 2026
+
+**Two new sound effect options added:**
+
+1. **Segment confirmed** - Plays when you confirm a segment with Ctrl+Enter
+2. **100% TM match alert** - Plays when navigating to a segment and a 100% TM match is found and auto-inserted
+
+**Access:** Settings → General → Sound Effects
+
+Both sounds are disabled by default (set to "None"). Users can configure them to any of the available Windows system sounds:
+- OK, Asterisk, Exclamation, Hand, Question
+- Windows .wav files: Restore, Navigation Start, Speech Disambiguation, etc.
+
+**Files Modified:**
+- `Supervertaler.py` - Added sound effect options and trigger points
+
+---
 
 ## 🐛 Ctrl+K Superlookup Shortcut Fix (v1.9.132) - January 20, 2026
 

{supervertaler-1.9.132 → supervertaler-1.9.147}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: supervertaler
-Version: 1.9.132
+Version: 1.9.147
 Summary: Professional AI-powered translation workbench with multi-LLM support, glossary system, TM, spellcheck, voice commands, and PyQt6 interface. Batteries included (core).
 Home-page: https://supervertaler.com
 Author: Michael Beijer
@@ -46,6 +46,7 @@ Requires-Dist: chardet>=5.0.0
 Requires-Dist: pyyaml>=6.0.0
 Requires-Dist: markdown>=3.4.0
 Requires-Dist: pyspellchecker>=0.7.0
+Requires-Dist: platformdirs>=4.0.0
 Requires-Dist: sounddevice>=0.4.6
 Requires-Dist: numpy>=1.24.0
 Requires-Dist: PyMuPDF>=1.23.0
@@ -71,7 +72,7 @@ Dynamic: home-page
 Dynamic: license-file
 Dynamic: requires-python
 
-# 🚀 Supervertaler v1.9.128
+# 🚀 Supervertaler v1.9.147
 
 [![PyPI version](https://badge.fury.io/py/supervertaler.svg)](https://pypi.org/project/Supervertaler/)
 [![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
@@ -79,7 +80,40 @@ Dynamic: requires-python
 
 AI-enhanced CAT tool with multi-LLM support (GPT-4, Claude, Gemini, Ollama), innovative Superlookup concordance system offering access to multiple terminology sources (TMs, glossaries, web resources, etc.), and seamless CAT tool integration (memoQ, Trados, CafeTran, Phrase).
 
-**Current Version:** v1.9.132 (January 20, 2026)
+**Current Version:** v1.9.147 (January 21, 2026)
+
+### NEW in v1.9.147 - 📁 Persistent User Data Location
+
+**No More Data Loss on Upgrade!** User data (API keys, TMs, glossaries, prompts, settings) is now stored in a persistent location that survives `pip install --upgrade`:
+
+| Platform | User Data Location |
+|----------|-------------------|
+| **Windows** | `%LOCALAPPDATA%\MichaelBeijer\Supervertaler\` |
+| **macOS** | `~/Library/Application Support/Supervertaler/` |
+| **Linux** | `~/.local/share/Supervertaler/` |
+| **Windows EXE** | `user_data\` next to executable (portable) |
+
+Existing data is automatically migrated on first run after upgrade.
+
+### FIXED in v1.9.146 - 🔑 Gemini/Google API Key Alias
+
+**Bug Fix:** Fixed "Gemini API Key Missing" error when users had `google=...` instead of `gemini=...` in their api_keys.txt. Both names now work identically thanks to automatic normalization at load time.
+
+### FIXED in v1.9.140 - 🐛 Glossary Add No Longer Triggers TM Search
+
+**Performance Fix ([#118](https://github.com/michaelbeijer/Supervertaler/issues/118)):** Adding a term to a glossary was unnecessarily triggering a full TM search. Now uses targeted refresh that only updates glossary display - TM results stay cached.
+
+**Also:** Renamed "Voice OFF" → "Voice Commands OFF" and "Dictate" → "Dictation" for clarity.
+
+### FIXED in v1.9.138 - 🏷️ Termview Punctuated Terms & Auto-Sizing Columns
+
+**Termview Fix:** Glossary terms with punctuation (like "ca." or "(typisch)") now correctly appear in the Termview pane. Previously they were found but not displayed due to a key normalization mismatch.
+
+**Grid UX:** Segment number column now auto-sizes based on font size and segment count - no more truncated numbers!
+
+### FIXED in v1.9.137 - 🔧 Termview Race Condition
+
+**Glossary terms now appear immediately:** Fixed timing bug where Termview showed "No glossary matches" until you pressed F5. Now updates correctly when navigating segments.
 
 ### ENHANCED in v1.9.128 - 📝 Placeholders Tab Layout Optimization
 

{supervertaler-1.9.132 → supervertaler-1.9.147}/README.md

@@ -1,4 +1,4 @@
-# 🚀 Supervertaler v1.9.128
+# 🚀 Supervertaler v1.9.147
 
 [![PyPI version](https://badge.fury.io/py/supervertaler.svg)](https://pypi.org/project/Supervertaler/)
 [![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
@@ -6,7 +6,40 @@
 
 AI-enhanced CAT tool with multi-LLM support (GPT-4, Claude, Gemini, Ollama), innovative Superlookup concordance system offering access to multiple terminology sources (TMs, glossaries, web resources, etc.), and seamless CAT tool integration (memoQ, Trados, CafeTran, Phrase).
 
-**Current Version:** v1.9.132 (January 20, 2026)
+**Current Version:** v1.9.147 (January 21, 2026)
+
+### NEW in v1.9.147 - 📁 Persistent User Data Location
+
+**No More Data Loss on Upgrade!** User data (API keys, TMs, glossaries, prompts, settings) is now stored in a persistent location that survives `pip install --upgrade`:
+
+| Platform | User Data Location |
+|----------|-------------------|
+| **Windows** | `%LOCALAPPDATA%\MichaelBeijer\Supervertaler\` |
+| **macOS** | `~/Library/Application Support/Supervertaler/` |
+| **Linux** | `~/.local/share/Supervertaler/` |
+| **Windows EXE** | `user_data\` next to executable (portable) |
+
+Existing data is automatically migrated on first run after upgrade.
+
+### FIXED in v1.9.146 - 🔑 Gemini/Google API Key Alias
+
+**Bug Fix:** Fixed "Gemini API Key Missing" error when users had `google=...` instead of `gemini=...` in their api_keys.txt. Both names now work identically thanks to automatic normalization at load time.
+
+### FIXED in v1.9.140 - 🐛 Glossary Add No Longer Triggers TM Search
+
+**Performance Fix ([#118](https://github.com/michaelbeijer/Supervertaler/issues/118)):** Adding a term to a glossary was unnecessarily triggering a full TM search. Now uses targeted refresh that only updates glossary display - TM results stay cached.
+
+**Also:** Renamed "Voice OFF" → "Voice Commands OFF" and "Dictate" → "Dictation" for clarity.
+
+### FIXED in v1.9.138 - 🏷️ Termview Punctuated Terms & Auto-Sizing Columns
+
+**Termview Fix:** Glossary terms with punctuation (like "ca." or "(typisch)") now correctly appear in the Termview pane. Previously they were found but not displayed due to a key normalization mismatch.
+
+**Grid UX:** Segment number column now auto-sizes based on font size and segment count - no more truncated numbers!
+
+### FIXED in v1.9.137 - 🔧 Termview Race Condition
+
+**Glossary terms now appear immediately:** Fixed timing bug where Termview showed "No glossary matches" until you pressed F5. Now updates correctly when navigating segments.
 
 ### ENHANCED in v1.9.128 - 📝 Placeholders Tab Layout Optimization