feat: Add comprehensive autocorrection system with batch import and m…

[pzauner]

…ulti-language support

This PR introduces a complete autocorrection system with automatic loading of accent/umlaut replacement rules and a user-friendly batch import interface.

🎯 Core Features

1. Automatic Autocorrection Loading (51,537 rules across 6 languages) MOST IMPORTANT: Autocorrection rules are now automatically loaded from assets/common/autocorrect/ at app startup - no user action required!

Generated rules are immediately active after installation:

• German (DE): 7,911 rules (ä→ae, ö→oe, ü→ue, ß→ss)
• French (FR): 15,449 rules (é→e, è→e, à→a, ç→c, etc.)
• Spanish (ES): 10,683 rules (á→a, ñ→n, etc.)
• English (EN): 121 rules (preserves 51 existing manual rules)
• Italian (IT): 1,816 rules (preserves 18 existing manual rules)
• Polish (PL): 15,472 rules

How it works:

App Launch → AutoCorrector.loadCorrections()
  → Loads assets/common/autocorrect/auto_corrections_{lang}.json
  → User types "ueber" → automatically becomes "über" ✨

2. Batch Import UI (AutoCorrectionImportActivity) User-friendly interface for importing custom autocorrection rules:

Features:

• File picker integration (JSON import)
• Language selector: Choose target language before import
• Real-time preview: filename, rule count, language code/name
• Validation with detailed error messages
• Progress indicator and success/error feedback
• Automatic activation: Imported rules are enabled immediately

Access: Settings → Auto-Correction → "Regeln Batch-Import"

JSON Format:

{
  "language": "de",
  "name": "Deutsch",
  "rules": {
    "ueber": "über",
    "fuer": "für"
  }
}

3. Delete All Feature (AutoCorrectEditScreen)

Safely delete all autocorrection rules for a specific language:

Features:

• Delete button (🗑️ icon) next to Add button in top bar
• Only visible when rules exist
• Confirmation dialog showing rule count and language
• Warning: "This action cannot be undone"
• Immediate reload after deletion

Use Cases:

• Undo incorrect batch imports
• Switch between different rule sets
• Testing and development

4. Performance Optimizations

Problem: UI freeze when displaying 7,000+ rules Solution: Replaced Column + forEach with LazyColumn + items()

Results:

• Smooth scrolling for 7,000+ rules
• Instant load time
• Virtualized rendering (only visible items rendered)

Changed Files:

• AutoCorrectEditScreen.kt: LazyColumn implementation

5. Universal Autocorrection Generator (generate_autocorrections.py) Python script for generating autocorrection rules from base dictionaries.

Features:

• Multi-language support: de, fr, es, en, it, pl
• Language-specific transformations:
  • German: ä→ae, ö→oe, ü→ue, ß→ss
  • Others: Generic accent removal (NFD normalization)
• Preserves manually defined rules by default
• Outputs directly to assets/common/autocorrect/

Usage:

# Generate for all supported languages
python3 scripts/generate_autocorrections.py

# Generate for specific languages only
python3 scripts/generate_autocorrections.py de fr

# Overwrite existing rules (don't preserve manual edits)
python3 scripts/generate_autocorrections.py --no-preserve

Output:

DE:   7,911 rules → auto_corrections_de.json (254 KB)
FR:  15,449 rules → auto_corrections_fr.json (439 KB)
ES:  10,683 rules → auto_corrections_es.json (299 KB)
EN:    121 rules → auto_corrections_en.json (3 KB)
IT:   1,816 rules → auto_corrections_it.json (48 KB)
PL:  15,472 rules → auto_corrections_pl.json (440 KB)

📋 Technical Changes

New Files

• app/src/main/java/.../AutoCorrectionImportActivity.kt (469 lines)
  • Batch import UI with file picker, validation, language selector
• scripts/generate_autocorrections.py (228 lines)
  • Universal generator for multi-language autocorrection rules
• scripts/convert_dictionaries.py (moved from root)
  • Dictionary format converter (organization cleanup)

Modified Files

• app/src/main/AndroidManifest.xml

  • Registered AutoCorrectionImportActivity

• app/src/main/java/.../AutoCorrectionCategoryScreen.kt

  • Added "Regeln Batch-Import" button with cloud upload icon
  • Links to new import activity

• app/src/main/java/.../AutoCorrectEditScreen.kt

  • Performance: Replaced Column + verticalScroll + forEach with LazyColumn + items()
  • Feature: Added "Delete All" button with confirmation dialog
  • Import: DeleteSweep icon in error color

• app/src/main/assets/common/autocorrect/auto_corrections_*.json (6 files)

  • Populated with generated rules (51,537 total)
  • Preserved existing manual rules where applicable

• scripts/README.md

  • Added documentation for generate_autocorrections.py
  • Added documentation for convert_dictionaries.py
  • Organized into sections: Main Scripts, Legacy Scripts

• .gitignore

  • Added .idea/deploymentTargetSelector.xml
  • Added app/build.properties
  • Prevents IDE-specific files from being committed

🚀 How to Use (For Developers)

Generate Autocorrection Rules

# Generate for all languages (recommended after dictionary updates)
cd /path/to/project
python3 scripts/generate_autocorrections.py

# Or for specific languages only
python3 scripts/generate_autocorrections.py de fr

Regenerate After Dictionary Changes

# When base dictionaries are updated:
python3 scripts/generate_autocorrections.py --no-preserve

This overwrites existing rules. Use with caution if manual rules exist.

📱 How to Use (For End Users)

Option 1: Automatic (Default)

No action required! Autocorrection rules are automatically active:

1. Install/update Pastiera
2. Start typing: "ueber" → "über", "cafe" → "café"
3. Works immediately for all 6 supported languages

Option 2: Batch Import (Custom Rules)

1. Open Pastiera Settings
2. Navigate to: Settings → Auto-Correction
3. Tap: "Regeln Batch-Import"
4. Select JSON file from device
5. (Optional) Change target language
6. Tap: "Alle Regeln importieren"
7. Done! Custom rules override defaults

JSON Format Example:

{
  "language": "de",
  "name": "Meine Regeln",
  "rules": {
    "hallo": "Hallo",
    "danke": "Danke!"
  }
}

Option 3: Delete All Rules (Per Language)

1. Settings → Auto-Correction → Select language (e.g., Deutsch)
2. Tap 🗑️ icon in top-right (next to + button)
3. Confirm deletion in dialog
4. All rules for that language are removed

🔄 System Architecture

Loading Priority

1. Custom Rules (from Batch Import)
   ↓ (if exists, skip step 2)
2. Asset Rules (automatic, built-in)
   ↓ (loaded from assets/common/autocorrect/)
3. Runtime Application

Key Point: Custom imports override asset files. This allows users to:

• Customize built-in rules
• Add new languages
• Test different rule sets

File Locations

Built-in Rules (Automatic):
└─ app/src/main/assets/common/autocorrect/
   ├─ auto_corrections_de.json
   ├─ auto_corrections_fr.json
   └─ ...

Custom Rules (User Imports):
└─ SharedPreferences
   └─ "auto_correct_custom_{language}"

🧪 Testing

Manual Testing Checklist

• Install APK
• Type "ueber" in any text field → Should become "über"
• Type "cafe" → Should become "café"
• Settings → Auto-Correction → Batch Import
• Import test JSON file
• Verify rules are applied immediately
• Delete all rules for a language
• Confirm rules are removed

Test JSON File

Create test_rules.json:

{
  "language": "de",
  "name": "Test",
  "rules": {
    "test": "TEST",
    "hallo": "HALLO"
  }
}

📊 Statistics

• 14 files changed
• 52,401 insertions, 218 deletions
• 51,537 autocorrection rules generated
• 469 lines of new UI code (AutoCorrectionImportActivity)
• 228 lines of Python generation code
• 6 languages supported out of the box

🎯 Benefits

For Users

✅ Faster typing: No need to access special characters ✅ Multi-language: Works for German umlauts, French accents, etc. ✅ Automatic: No setup required, works immediately ✅ Customizable: Import custom rules via JSON
✅ Safe: Delete all feature with confirmation

For Developers

✅ Maintainable: Single script generates all languages ✅ Extensible: Easy to add new languages
✅ Preserved: Manual rules are kept by default
✅ Documented: Complete README in scripts/
✅ Organized: All scripts in scripts/ folder

🔍 Breaking Changes

None. This is a new feature with backward compatibility:

• Existing custom rules are preserved
• App works without autocorrection files (fallback)
• No changes to existing autocorrection behavior

Notes

• Asset files are loaded first at app startup (see AutoCorrector.loadCorrections())
• Custom imports take precedence over asset files
• Generated rules preserve existing manual edits by default
• All autocorrection files use simple JSON: {"from": "to"}
• Language codes follow standard: de, en, fr, es, it, pl

Acknowledgments

• Preserves existing manual rules in FR (40), EN (51), IT (18)
• Generator script respects frequency data for collision resolution
• UI follows Material Design 3 guidelines