Search Doc Tool

Lire en français

A tool for searching terms across Word (.docx) and PDF files. Available as a desktop app (PyQt6) and a web app (Django).

Web app

Dark mode	Light mode

Features

• Search through .docx and .pdf files (recursively or not)
• Advanced query syntax:
  • Space-separated words → OR mode (at least one term)
  • + operator → AND mode (all terms must be present)
  • "exact phrase" → literal phrase search
• Accent-insensitive search by default
• Options: case sensitivity, whole-word matching
• Context display around each match (highlighted term)
• Page number with direct link to the matching page in the PDF viewer
• Favorites sidebar for frequently used folders (per user, resizable)
• SQLite FTS5 index for near-instant searches on already-indexed folders
• Background indexing with progress bar and stop button
• Light/dark mode toggle

Web app — additional features

• Multi-user: Django authentication (login required)
• Per-folder shared index: one index per folder path, shared across all users
• DOCX → PDF conversion via LibreOffice at indexing time
• PDF served inline with highlighted search terms
• Indexing status persisted in database (survives server restarts)
• Robust background indexing: stale job detection, automatic recovery after unexpected shutdown
• Favorites organized into named groups (create, rename, delete, drag & drop between groups)
• Favorite aliases: set a custom display name per folder (path shown by default)
• Non-indexable file tracking: files that fail indexing (password-protected, read errors) are flagged separately and displayed in the badge (e.g. "392 indexed • 6 non-indexable")
• Optimized search on large folders: indexed files are searched directly from the FTS5 database (no disk I/O), unindexed files fall back to PDF reading
• Superuser tools:
  • Configure allowed root directories for the file browser (users cannot navigate outside these paths)
  • Orphaned data cleanup: remove .data/folders/ directories no longer referenced by any favorite

Requirements

• Python 3.12+
• Web app only: LibreOffice installed (DOCX → PDF conversion at indexing time)
• Web app — strongly recommended: Microsoft Word installed on the server. Word COM automation is used as the primary converter (faster and higher fidelity); LibreOffice is the fallback when Word is unavailable or fails.

Setup & Running

Desktop (PyQt6)

cd desktop
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
python -m search_tool

Web (Django) — development

cd web2
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
cd search_tool_project
python manage.py migrate
python manage.py createsuperuser
python manage.py runserver

Then open http://127.0.0.1:8000 in your browser.

Web (Django) — production (Waitress)

cd web2/search_tool_project
python manage.py collectstatic
DJANGO_DEBUG=false python run.py --host 0.0.0.0 --port 8000

Usage

1. Select a folder from the favorites sidebar or type/browse a path
2. Click Indexer to build the FTS5 index (first time or after adding files)
3. Enter one or more terms and click Lancer
4. Click a result to open the PDF at the matching page, with the term highlighted

Index & data storage

• Per-folder data stored in .data/folders/<folder_name>_<hash>/:
  • index.db — FTS5 SQLite index (one row per page)
  • pdf_cache/ — converted PDFs (content-addressed, invalidated on file change)
  • docx_copy/ — local DOCX copies (network drive workaround)
• Indexing status (start time, counts, errors) persisted in the Django database

Configuration

Last folder and recurse setting saved automatically to ~/.search_tool_config.json. Favorites are stored per user in the Django database.