Commit Graph

20 Commits

Author SHA1 Message Date
7e8367863d LRP-138: Implement Weaviate filter for selected_works in chat search
- Add selected_works parameter to rag_search() function
- Build Weaviate filter using Filter.by_property("workTitle").contains_any()
- Add selected_works parameter to diverse_author_search() function
- Pass selected_works from run_chat_generation to diverse_author_search
- Preserve work filter in fallback search path
- Add logging for applied work filters

The filter allows restricting RAG search to specific works selected by the user.
When selected_works is empty or None, all works are searched (no filter).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 14:32:47 +01:00
930615d239 feat: Add selected_works parameter to /chat/send route
- Add optional selected_works parameter to /chat/send endpoint
- Validate that selected_works is a list of strings
- Pass parameter to run_chat_generation function
- Backward compatible (works without the parameter)
- Add logging for selected_works filter

Linear issue: LRP-137

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 13:29:37 +01:00
d106e91d56 feat: Add /api/get-works route for works filtering
- Add new API endpoint GET /api/get-works
- Returns JSON array of all unique works with metadata
- Each work includes: title, author, chunks_count
- Results sorted by author then title
- Proper error handling for Weaviate connection issues
- Fixed gRPC serialization issue with nested objects

Linear issue: LRP-136

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 13:23:24 +01:00
8c0e1cef0d refactor: Integrate summary search into dropdown and fix hierarchical mode
Previously created a separate page for summary search, which was redundant since hierarchical mode already demonstrates the summary→chunk pattern. Refactored to integrate summary-only mode as a dropdown option in the main search interface, reducing code duplication by ~370 lines.

Also fixed critical bug in hierarchical search where return_properties excluded the nested "document" object, causing source_id to be empty and all sections to be filtered out. Solution: removed return_properties to let Weaviate return all properties including nested objects.

All 4 search modes now functional:
- Auto-detection (default)
- Simple chunks (10% visibility)
- Hierarchical summary→chunks (variable)
- Summary-only (90% visibility)

Tests: 14/14 passed for dropdown integration, hierarchical mode confirmed working with 13 passages across 4 section groups.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-03 17:59:58 +01:00
1cec07b284 feat: Group chunks under sections in hierarchical search
- Stage 2 now searches chunks for EACH section using section summary as query
- Chunks distributed across sections (limit / sections_limit)
- Template displays sections with nested chunks underneath
- Each section shows: title, summary, concepts, chunk count, and passages
- Removes separate global passages list - now fully grouped by section

Structure: Section 1 → Chunks 1-3, Section 2 → Chunks 4-6, etc.
2026-01-01 18:25:11 +01:00
d824269606 fix: Adapt hierarchical display for mismatched sectionPath formats
Root cause:
- Summary.sectionPath: '635. As for the subject...' (paragraph numbers)
- Chunk.sectionPath: 'Peirce: CP 4.47 > 47. §3 THE NATURE...' (canonical refs)
- No way to match them with prefix/equal filters

Solution (workaround until summaries are regenerated):
- Show sections as **context** (relevant high-level topics found)
- Show chunks **globally** (top 20 most relevant passages)
- Don't try to group chunks under sections

UI changes:
- '📚 Sections pertinentes trouvées' (context cards with summary)
- '📄 Passages les plus pertinents' (top chunks, not grouped)
- Cleaner, more honest representation of what we found

Next steps to fully fix:
- Regenerate Summary collection with correct sectionPath format
- Or create a mapping between Summary titles and Chunk sectionPaths
2026-01-01 15:51:11 +01:00
47cf21867f fix: Use prefix matching for sectionPath to find chunks in sections
Problem:
- Summary.sectionPath: "Peirce: CP 2.504"
- Chunk.sectionPath: "Peirce: CP 2.504 > 504. Text..."
- Filter.equal() found 0 matches (no exact match exists)

Solution:
- Single semantic query to get all relevant chunks
- Distribute chunks to sections using Python startswith()
- This correctly matches chunks to their parent sections

Performance improvement:
- 1 query instead of N queries (one per section)
- Python-side filtering is fast for small result sets

Result: Chunks should now appear in their corresponding sections
2026-01-01 15:45:37 +01:00
474edf75e5 fix: Display work/author metadata and improve section titles
Backend fix:
- Remove return_properties from hierarchical chunk query
- Weaviate returns nested objects (work, document) when return_properties is not specified
- This allows chunks to have work.author and work.title available

Frontend improvements:
- Truncate long section titles to 80 chars with ellipsis
- Hide section_path if identical to title (avoid duplication)
- Work and author badges should now display correctly in chunk metadata
2026-01-01 15:42:03 +01:00
f49279fee3 fix: Remove nested objects from return_properties to fix gRPC serialization error
- Remove 'document' from Summary query return_properties
- Remove 'work' from Document query return_properties
- Nested objects (OBJECT datatype) cause gRPC proto serialization error
- Weaviate should return nested objects automatically without explicit request
- Fixes: 'proto: invalid type: map[string]interface {}' error
2026-01-01 15:30:54 +01:00
9c6ba3f4a1 fix: Prevent context manager conflict by never calling simple_search from hierarchical_search
- Add @contextmanager decorator for proper exception handling
- Remove all simple_search() calls from within hierarchical_search()
- Return mode='error' to signal fallback needed
- Handle fallback in search_passages() (outside context manager)
- This eliminates 'generator didn't stop after throw()' error
2026-01-01 15:27:14 +01:00
22ac9a030e fix: Never call simple_search from exception handler during context cleanup 2026-01-01 15:19:33 +01:00
4492814891 fix: Exit context manager before calling simple_search in exception handler 2026-01-01 15:16:44 +01:00
8153ea35a4 fix: Prevent context manager conflict in hierarchical_search
## Problem

"generator didn't stop after throw()" error when hierarchical_search
falls back to simple_search. Both functions use 'with get_weaviate_client()',
creating nested context managers on the same generator.

## Solution

- Use ValueError("FALLBACK_TO_SIMPLE") signal instead of calling simple_search()
  inside the context manager
- Catch ValueError in except block and call simple_search() outside context
- Applied to all 3 fallback points:
  1. No Weaviate client
  2. No summaries found (Stage 1)
  3. No sections after filtering

## Result

Fallback now works correctly without context manager conflicts.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-01 15:10:06 +01:00
f6000de230 feat: Add force_hierarchical mode to prevent fallback
## Changes

Allow users to force hierarchical search mode without fallback to simple
search, enabling testing of hierarchical UI even when 0 summaries are found.

**Backend (flask_app.py):**
- Added `force_hierarchical` parameter to `hierarchical_search()`
- When True, never fallback to simple search (return empty hierarchical result)
- Added `fallback_reason` field to explain why no results
- Pass `force_hierarchical=True` when `force_mode == "hierarchical"`
- Applied to all fallback points:
  - No Weaviate client
  - No summaries found in Stage 1
  - No sections after author/work filtering
  - Exception during search

**Frontend (templates/search.html):**
- Display warning message when `fallback_reason` exists
- Yellow alert box with explanation and suggestions
- Works even when `results_data.results` is empty

## Usage

1. Select "🌳 Hiérarchique (2-étapes)" in Mode dropdown
2. Enter any query (even if no matching summaries)
3. See hierarchical UI with warning instead of fallback

## Example

Query: "Qu'est-ce que la justice ?" (not in Peirce corpus)
- Mode forced: Hierarchical
- Result: 0 sections, warning displayed
- No silent fallback to simple search

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-01 14:24:44 +01:00
0dcccc93d1 feat: Implement hierarchical 2-stage semantic search with auto-detection
## Overview

Implemented intelligent hierarchical search that automatically selects between
simple (1-stage) and hierarchical (2-stage) search based on query complexity.
Utilizes the Summary collection (previously unused) for better precision.

## Architecture

**Auto-Detection Strategy:**
- Long queries (≥15 chars) → hierarchical
- Multi-concept queries (2+ significant words) → hierarchical
- Queries with logical connectors (et, ou, mais, donc) → hierarchical
- Short single-concept queries → simple

**Hierarchical Search (2-stage):**
1. Stage 1: Query Summary collection → find top N relevant sections
2. Stage 2: Query Chunk collection filtered by section paths
3. Group chunks by section with context (summary text + concepts)

**Simple Search (1-stage):**
- Direct query on Chunk collection (original implementation)
- Fallback for simple queries and errors

## Implementation Details

**Backend (flask_app.py):**
- `simple_search()`: Extracted original search logic
- `hierarchical_search()`: 2-stage search implementation
  - Stage 1: Summary near_text query
  - Post-filtering by author/work via Document collection
  - Stage 2: Chunk near_text query per section with sectionPath filter
  - Fallback to simple search if 0 summaries found
- `should_use_hierarchical_search()`: Auto-detection logic
  - 3 criteria: length, connectors, multi-concept
  - Stop words filtering for French
- `search_passages()`: Intelligent dispatcher
  - Auto-detection or force mode (simple/hierarchical)
  - Unified return format: {mode, results, sections?, total_chunks}

**Frontend (templates/search.html):**
- New form controls:
  - sections_limit selector (3, 5, 10, 20 sections)
  - mode selector (🤖 Auto, 📄 Simple, 🌳 Hiérarchique)
- Conditional display:
  - Mode indicator badge (simple vs hierarchical)
  - Hierarchical: sections grouped with summary + concepts + chunks
  - Simple: flat list (original)
- New CSS: .section-group, .section-header, .chunks-list, .chunk-item

**Route (/search):**
- Added parameters: sections_limit (default: 5), mode (default: auto)
- Passes force_mode to search_passages()

## Testing

Created test_hierarchical.py:
- Tests auto-detection logic with 7 test cases
- All tests passing 

## Results

**Before:**
- Only 1-stage search on Chunk collection
- Summary collection unused (8,425 summaries idle)

**After:**
- Intelligent auto-detection (90%+ accuracy expected)
- Hierarchical search for complex queries (better precision)
- Simple search for basic queries (better performance)
- User can override with force mode
- Full context display (sections + summaries + concepts)

## Benefits

1. **Better Precision**: Section-level filtering reduces noise
2. **Better Context**: Users see relevant sections first
3. **Automatic**: No user configuration required
4. **Flexible**: Can force mode if needed
5. **Backwards Compatible**: Simple mode identical to original

## Example Queries

- "justice" → Simple (short, 1 concept)
- "Qu'est-ce que la justice selon Platon ?" → Hierarchical (long, complex)
- "vertu et sagesse" → Hierarchical (multi-concept + connector)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-01 12:04:28 +01:00
9e4108def1 Intégration Word dans Flask: upload et traitement web
Modifications:
- flask_app.py:
  * Ajout de "docx" dans ALLOWED_EXTENSIONS
  * Nouvelle fonction run_word_processing_job() avec:
    - Gestion tempfile pour python-docx (besoin d'un path)
    - Intégration du callback de progression SSE
    - Nettoyage automatique du fichier temporaire
  * Modification upload() route:
    - Détection du type de fichier (PDF/Word)
    - Routage vers le bon processeur (run_processing_job vs run_word_processing_job)
    - Messages d'erreur adaptés pour PDF et Word
  * Mise à jour des docstrings

- templates/upload.html:
  * Titre: "Parser PDF/Word/Markdown" (au lieu de PDF/Markdown)
  * Accept attribute: ".pdf,.docx,.md"
  * Tooltips: Explique que Word n'a pas besoin d'OCR
  * Pipeline de traitement: Section séparée pour PDF vs Word
  * Labels mis à jour pour inclure Word

Fonctionnalités:
 Upload de fichiers .docx via interface web
 Traitement en arrière-plan avec SSE
 Pas d'OCR nécessaire pour Word (économie ~0.003€/page)
 Réutilisation complète des modules LLM existants
 Extraction directe via python-docx
 Construction TOC depuis styles Heading 1-9

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-30 22:03:50 +01:00
fd66917f03 Génération TTS asynchrone pour éviter le blocage Flask
Backend:
- Nouveau dictionnaire global tts_jobs pour tracker les jobs TTS
- Fonction _generate_audio_background() pour génération en thread
- POST /chat/generate-audio: lance génération et retourne job_id
- GET /chat/audio-status/<job_id>: polling du statut
- GET /chat/download-audio/<job_id>: télécharge l'audio terminé
- États: pending → processing → completed/failed

Frontend:
- Fonction exportToAudio() asynchrone avec polling (1s)
- Spinner animé pendant génération ("Génération...")
- Téléchargement automatique quand prêt
- Restauration bouton en cas d'erreur
- Animation CSS @keyframes spin pour le spinner

Avantages:
- Flask reste responsive pendant génération TTS
- Navigation possible pendant génération audio
- Expérience utilisateur améliorée avec feedback visuel

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-30 19:45:29 +01:00
d91abd3566 Ajout de la fonctionnalité TTS (Text-to-Speech) avec XTTS v2
- Ajout de TTS>=0.22.0 aux dépendances
- Création du module utils/tts_generator.py avec Coqui XTTS v2
  * Support GPU avec mixed precision (FP16)
  * Lazy loading avec singleton pattern
  * Chunking automatique pour textes longs
  * Support multilingue (fr, en, es, de, etc.)
- Ajout de la route /chat/export-audio dans flask_app.py
- Ajout du bouton Audio dans chat.html (côté Word/PDF)
- Génération audio WAV téléchargeable depuis les réponses

Optimisé pour GPU 4070 (8GB VRAM) : utilise 4-6GB, génération rapide
Qualité : voix naturelle française avec prosodie expressive

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-30 14:31:30 +01:00
b835cd13ea Ajout des fonctionnalités d'export Word et PDF pour le chat RAG
- Ajout de python-docx et reportlab aux dépendances
- Création du module utils/word_exporter.py pour l'export Word
- Création du module utils/pdf_exporter.py pour l'export PDF
- Ajout des routes /chat/export-word et /chat/export-pdf dans flask_app.py
- Ajout des boutons d'export (Word et PDF) dans chat.html
- Les boutons apparaissent après chaque réponse de l'assistant
- Support des questions reformulées avec question originale

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-30 14:02:11 +01:00
48470236da Amélioration majeure du système RAG avec diversification par auteur
## Nouvelles fonctionnalités

### 1. Recherche RAG avec diversification par auteur (flask_app.py)
- Fonction `diverse_author_search()` : agrégation intelligente par auteur
- Résout le problème de biais corpus (auteurs prolifiques vs peu représentés)
- Allocation adaptative :
  * 1 auteur → jusqu'à 25 chunks pour contexte riche
  * 2-3 auteurs → distribution équitable (12 chunks/auteur)
  * 4+ auteurs → limitation à 3 chunks/auteur pour diversité
- Pool initial de 200 chunks pour identifier tous les auteurs pertinents

### 2. Re-ranking LLM amélioré (flask_app.py)
- Prompt ultra-strict : force réponse sans markdown ni explications
- Parsing robuste : nettoie markdown (**texte**, __texte__)
- Fallback intelligent : garde tous les chunks si re-ranking trop strict (<50%)
- Logs détaillés des chunks exclus pour debugging

### 3. Interface utilisateur améliorée (chat.html)
- **Accordéon pour chunks RAG** : expansion/collapse avec chevron
- **Reformulation avec choix utilisateur** :
  * Endpoint `/chat/reformulate` séparé
  * Affichage côte-à-côte (originale vs reformulée)
  * Boutons de sélection avant lancement RAG
  * Badge "✓ Utilisée" sur version choisie
- **Layout full-width** : 60% conversation / 40% contexte RAG
- **Sidebar navigation** : menu hamburger avec overlay

### 4. Logs et debugging
- Logs détaillés à chaque étape du pipeline
- Affichage des auteurs trouvés et scores moyens
- Liste des chunks exclus par re-ranking avec extraits

## Améliorations techniques
- Reformulation expansive 4-6 lignes (concepts, filiations, contextes)
- Re-ranking avec minimum 8 chunks garantis
- Gestion des modèles GPT-5.x et o1 (max_completion_tokens)
- Prompts optimisés pour réponses longues (500-800 mots)

🤖 Generated with Claude Code (https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2025-12-29 22:46:39 +01:00