docfind 1.1.1

A cross-platform document indexing and search tool

DocFind

A powerful cross-platform document indexing and search tool with both CLI and GUI interfaces.

Features

• Smart Auto Mode 🎯: Automatically chooses between full-text and metadata-only indexing based on folder size
• Full-text search using SQLite FTS5 for blazing-fast queries
• Multi-format support: PDF, DOCX, XLSX, PPTX, HTML, XML, and plain text
• Unknown format handling: Hex extraction for files with unrecognized formats
• Ripgrep integration: Lightning-fast search using ripgrep for large files
• CLI and GUI: Professional command-line and PyQt5 desktop interfaces
• Cross-platform: Works on Windows, macOS, and Linux
• Thread-safe indexing: Efficient multi-threaded document processing with stop capability
• Large file support: Handles files up to 10 GB with intelligent truncation
• Dark theme: Modern, accessible dark UI with customizable accent colors
• Real-time progress: Live progress tracking with detailed logging

Installation

Prerequisites

• Python 3.8 or higher
• (Optional) ripgrep for enhanced search

Install from PyPI

pip install docfind

Verify Installation

# Check CLI is available
docfind --help

# Launch GUI
docfind-gui

Quick Start

CLI Usage

Index documents

# Index a directory
docfind index /path/to/documents

# Index with progress display
docfind index /path/to/documents --progress

# Reindex existing documents
docfind index /path/to/documents --reindex

# Use multiple threads (default: 4)
docfind index /path/to/documents --threads 8

# Set maximum file size (in bytes)
docfind index /path/to/documents --max-size 52428800  # 50MB

Search documents

# Basic search
docfind search "python programming"

# Case-sensitive search
docfind search "Python" --case-sensitive

# Regex search
docfind search "func.*\(" --regex

# Whole word search
docfind search "test" --whole-word

# Use ripgrep for searching
docfind search "error" --use-ripgrep

# JSON output (JSONL format)
docfind search "data" --json

# Limit results
docfind search "query" --limit 50

# Filter by root path
docfind search "term" --root /path/to/documents

List indexed paths

# Show all indexed paths
docfind list

# JSON output
docfind list --json

Show statistics

# Display database statistics
docfind stats

# JSON output
docfind stats --json

Explain queries or documents

# Explain how a query would be executed
docfind explain --query "search term"

# Explain a specific document
docfind explain --path /path/to/file.pdf

# Show extracted text preview
docfind explain --path /path/to/file.pdf --show-text

Remove indexed data

# Remove specific root path
docfind remove --path /path/to/documents

# Remove all indexed data
docfind remove --all --force

Optimize database

# Optimize FTS index and vacuum database
docfind optimize

System check

# Check system configuration and dependencies
docfind doctor

GUI Usage

Launch the GUI application:

docfind-gui

GUI Features

Main Window Layout:

┌─────────────────────────────────────────────────────────────┐
│ File  Tools  Help                                           │
├───────────┬─────────────────────────────────┬───────────────┤
│           │ [Search...] [Options] [Actions] │               │
│ Projects  ├─────────────────────────────────┤ File Details  │
│           │                                 │               │
│ • /docs/  │      Results Table              │ Path: ...     │
│   (1234)  │                                 │ Type: pdf     │
│           │  Path | Type | Line | Snippet  │ Size: 2.3 MB  │
│ • /work/  │  ─────┼──────┼──────┼────────  │               │
│   (567)   │  ...  │ pdf  │  42  │ text..  │ [Actions]     │
│           │                                 │               │
│ [Add]     │                                 │ • Open Folder │
│ [Remove]  │      Preview / Text             │ • Copy Path   │
│           │                                 │ • Export      │
│           │  Extracted text with            │               │
│           │  highlighted matches...         │               │
│           │                                 │               │
├───────────┴─────────────────────────────────┴───────────────┤
│ [Progress Bar]                                              │
│ Log Console:                                                │
│ [12:34:56] [INFO] Indexing started...                       │
└─────────────────────────────────────────────────────────────┘

Keyboard Shortcuts:

• Ctrl+F - Focus search box
• Ctrl+I - Add folder to index
• Ctrl+E - Export results
• Ctrl+, - Open settings
• Ctrl+Q - Quit

Workflow:

1. Add a folder: Click "Add Folder" → Select directory → Index starts automatically
2. Search: Type in search box → Results appear in real-time (debounced)
   • For metadata-only indexed documents, ripgrep is auto-enabled
   • If ripgrep is not installed, a helpful dialog appears with download instructions
3. View results: Click result → Preview automatically loads and switches to Preview tab
   • See file details (path, type, size, extractor)
   • View file content with highlighted search matches
   • Works for both database-indexed and ripgrep-only results
4. Export: Click "Export Results" → Choose format (CSV or JSONL) → Save
   • CSV format: Excel-compatible with headers
   • JSONL format: One JSON object per line for programmatic processing

Settings:

Access via File → Settings (or press Ctrl+,):

• Threads: Number of parallel indexing threads (1-32)
  • Lower values (1-2) recommended for very large files to avoid database locking
  • Higher values (4+) optimal for many small files
• Max file size: Maximum file size to process (up to 10 GB)
  • Files larger than 500 MB will have text truncated for database storage
• Index mode: Choose indexing strategy ⭐
  • Auto (Recommended): Automatically selects best mode based on folder size
    • Uses Full Text for folders ≤400 MB (configurable threshold)
    • Uses Metadata Only for folders >400 MB
  • Metadata Only (Fast): Only indexes file metadata, uses ripgrep for search
    • Lightning fast indexing (seconds instead of hours)
    • Perfect for very large files or massive document collections
    • Requires ripgrep for searching
  • Full Text (Slow): Extracts and indexes all text content
    • Best for small to medium collections
    • Enables FTS5 search without external dependencies
• Auto mode threshold: Configure the size threshold for Auto mode (default: 400 MB)
• Trust external tools: Enable external conversion tools
• Ripgrep path: Path to ripgrep executable
• UI accent color: Customize interface color

Supported File Formats

Native Support

Format	Extensions	Extractor
PDF	.pdf	pdfminer.six
Word	.docx	python-docx
Excel	.xlsx	openpyxl
PowerPoint	.pptx	python-pptx
HTML	.html, .htm	beautifulsoup4
XML	.xml	beautifulsoup4
Text	.txt, .md, .rst, .log	Native
Source Code	.py, .js, .java, .c, .cpp, .h, .cs, .go, .rs, .rb, .php, .sh, .bat, .ps1	Native
Data	.json, .csv	Native

Fallback Support

For unknown file formats, DocFind uses hex extraction to extract readable ASCII/UTF-16 text strings from binary files.

Legacy Formats

• .doc, .xls, .ppt - Extracted via hex extractor (native support requires external tools)

Architecture

Core Components

DocFind consists of several key modules:

• CLI Interface (docfind): Command-line tool for indexing and searching
• GUI Application (docfind-gui): PyQt5 desktop application with dark theme
• Database Layer: SQLite with FTS5 full-text search engine
• Document Indexer: Multi-threaded extraction and indexing engine
• Search Engine: Supports both FTS5 and optional ripgrep integration
• Format Extractors: PDF, Office, HTML, text, and hex-based fallback

Database Schema

documents table:

• Stores file metadata (path, type, size, hash, mtime, status)
• Tracks indexing status and errors

documents_fts (FTS5 virtual table):

• Full-text search index with Porter stemming
• Unicode tokenization for international text
• BM25 ranking for relevance scoring

extracted_text table:

• Stores complete extracted text for preview
• Linked to documents via foreign key

Threading Model

GUI Application:

• Main Thread: UI updates and user interaction
• IndexWorker Thread: Background document indexing with progress signals
• SearchWorker Thread: Async search operations
• Database: Thread-local connections with WAL mode for concurrent access

CLI Application:

• Main Thread: User interface and coordination
• ThreadPoolExecutor: Parallel document processing (configurable thread count)
• Database: Thread-safe with connection pooling

Configuration

Configuration is stored in platform-specific locations:

• Windows: %APPDATA%\docfind\config.json
• macOS: ~/Library/Application Support/docfind/config.json
• Linux: ~/.config/docfind/config.json

Default Configuration

{
  "max_file_size": 419430400,
  "threads": 2,
  "ignore_globs": [
    "*.pyc",
    "__pycache__",
    ".git",
    ".svn",
    "node_modules",
    ".venv",
    "venv",
    "*.log"
  ],
  "trust_external_tools": false,
  "ripgrep_path": "rg",
  "theme": "dark",
  "accent_color": "#3a7bd5",
  "db_path": "<platform-specific-data-dir>/docfind.db",
  "index_mode": "auto",
  "auto_mode_threshold": 419430400
}

Advanced Usage

Custom Configuration

You can customize DocFind behavior by editing the configuration file:

# Linux/macOS
~/.config/docfind/config.json

# Windows
%APPDATA%\docfind\config.json

Database Location

By default, the database is stored in:

# Linux
~/.local/share/docfind/docfind.db

# macOS
~/Library/Application Support/docfind/docfind.db

# Windows
%LOCALAPPDATA%\docfind\docfind.db

You can back up this single file to preserve your entire index.

Environment Variables

• DOCFIND_DB_PATH: Override default database location
• DOCFIND_CONFIG_PATH: Override default config location
• DOCFIND_LOG_LEVEL: Set logging level (DEBUG, INFO, WARNING, ERROR)

Performance Tips

Indexing

• Index Mode: Choose the right mode for your use case:
  • Auto (Recommended): Automatically picks best mode based on folder size
  • Metadata Only: Lightning-fast indexing (10-20 seconds for hundreds of GB)
  • Full Text: Slower but enables FTS5 search without ripgrep
• Threads: Default is 2 to avoid database locking. Use --threads 1 for very large files (>500 MB)
• File size: Limit with --max-size to skip very large files (default: 10 GB)
• Ignore patterns: Configure patterns for files/folders to skip
• Reindex: Only use --reindex when necessary (slower)

Searching

• Automatic mode detection: GUI auto-enables ripgrep for metadata-only indexed documents
• FTS5: Fast for most queries, supports phrase search (full-text mode)
• Ripgrep: Lightning-fast for simple string matches, regex support (metadata-only mode)
• Rich results: File path, line numbers, and text snippets displayed in GUI
• Pagination: Use --limit and --offset for large result sets
• Filters: Use --root to narrow search scope

Database

• Optimize: Run docfind optimize periodically to compact database
• Backup: Database is a single .db file - easy to backup
• Location: Store on SSD for better performance

Troubleshooting

"Database locked" errors

• Reduce number of indexing threads in settings (try 1 thread for very large files)
• Close other DocFind instances accessing the same database
• Use metadata-only mode for large file collections (skips heavy database writes)
• Database timeout is 60 seconds with automatic retry (3 attempts with exponential backoff)

"ripgrep not found" or "Ripgrep Required" dialog

DocFind automatically enables ripgrep when searching metadata-only indexed documents. If ripgrep is not installed:

• Easy Installation: Click "Open Download Page" in the dialog to download ripgrep
• Manual Download: Visit https://github.com/BurntSushi/ripgrep/releases
• Windows: Download the .zip file, extract rg.exe, and add it to your PATH
• Linux: sudo apt install ripgrep (Ubuntu/Debian) or use your package manager
• macOS: brew install ripgrep
• Custom path: Specify path in Settings → Ripgrep path

After installation, restart DocFind to detect ripgrep automatically.

GUI doesn't start

• Check PyQt5 installation: pip install --upgrade PyQt5
• On Linux, install: sudo apt-get install python3-pyqt5
• Check logs: ~/.local/share/docfind/docfind_gui.log (Linux)

Extraction fails for PDF/Office files

• Ensure dependencies are installed: pip install -r requirements.txt
• For legacy formats (.doc, .xls, .ppt), use hex extraction (automatic fallback)
• Check file isn't corrupted: Try opening in native application

High memory usage

• Use metadata-only or auto mode instead of full-text mode
• Reduce max_file_size in config
• Use fewer indexing threads (1-2 instead of 4+)
• Process large directories in smaller batches

Indexing is too slow

• For large file collections (>400 MB total): Use Auto or Metadata Only mode
  • Metadata mode indexes hundreds of GB in 10-20 seconds
  • Only stores file paths and metadata (ripgrep searches actual files)
• For small collections (<400 MB total): Use Full Text mode
  • Extracts and indexes all text content
  • Enables fast FTS5 search without external tools
  • Adjust the threshold in Settings if needed
• Check Settings → Index Mode to change behavior

Security Considerations

• External tools: Disabled by default (trust_external_tools: false)
• System paths: GUI warns before indexing system directories
• Network drives: Warning displayed before indexing
• File execution: DocFind never executes indexed files
• SQL injection: Parameterized queries prevent injection

License

MIT License - see LICENSE file for details.

Credits

Built with:

• PyQt5 - GUI framework
• SQLite FTS5 - Full-text search
• pdfminer.six - PDF extraction
• python-docx - DOCX extraction
• openpyxl - XLSX extraction
• python-pptx - PPTX extraction
• BeautifulSoup - HTML/XML parsing
• ripgrep - Optional fast search

Changelog

Version 1.1.0

Major performance improvements and bug fixes for large folder search:

🚀 Performance Improvements

• Smart Auto Mode: Automatically chooses between full-text and metadata-only indexing based on folder size (configurable threshold, default 400 MB)
• Lightning-fast indexing:
  • First-time indexing: ~10 minutes for 285 GB (metadata-only mode)
  • Re-indexing unchanged files: Under 1 second (batch query + in-memory skip checks)
• Performance optimizations:
  • Skipped hash calculation in metadata-only mode (eliminates reading entire files)
  • Batch-load existing documents for skip checking (1 query vs 285 queries)
  • In-memory duplicate detection (nanoseconds vs milliseconds per file)

🔍 Search Enhancements

• Fixed ripgrep search for large folders (>threshold):
  • Metadata-only indexed documents now properly searched with ripgrep
  • Auto-detects metadata-only mode and enables ripgrep automatically
  • Fixed auto-enable logic to count ALL metadata-only documents (not just first)
  • Auto-checks ripgrep checkbox when enabled so users can see it's active
• Enhanced debugging and logging:
  • Detailed INFO-level logging for ripgrep command execution
  • Logs ripgrep return codes, output length, and parsed results count
  • Better error messages when ripgrep returns empty results or errors
  • Debug logging for JSON parsing showing each match found
• Improved result handling:
  • Files not in database now included in results with minimal metadata
  • Better error handling with full tracebacks in debug mode
  • Fixed JSON parsing to handle edge cases
• Helpful ripgrep installation:
  • Clear dialog if ripgrep is not found (with download link and instructions)
  • One-click "Open Download Page" button for easy installation
  • Shows count of metadata-only documents when auto-enabling
• Rich search results: File path, line numbers, and text snippets displayed in GUI

💾 Database & File Handling

• Large file support: Increased max file size to 10 GB with automatic text truncation at 500 MB
• Stop functionality: Can now stop indexing operations in GUI with proper cleanup
• Database improvements:
  • Increased timeout from 30s to 60s
  • Added retry logic with exponential backoff for locked database
  • Reduced default threads from 4 to 2 to minimize locking issues
  • Batch metadata queries for 100x+ faster skip checking
• Enhanced error handling: Automatic truncation and retry for oversized text content

🎨 UI/UX Improvements

• Real-time progress: Live progress tracking with detailed logging in GUI
• Settings improvements: Added index mode selector (Auto/Metadata Only/Full Text) with helpful tooltips
• Better warnings: Alert users when no files found due to size limits or patterns
• Log console: Detailed color-coded logs showing indexing and search operations
• Auto-preview on click: Clicking a search result automatically loads and displays the file preview
  • Automatically switches to Preview tab for better UX
  • Highlights search matches in the preview
  • Works for both database-indexed and ripgrep-only results
  • Smart context loading: For ripgrep results, shows ±50 lines around the match (not entire file)
  • Matched line is highlighted with ">>>" prefix and line numbers shown
  • Chunked file reader for extremely large files (500+ GB):
    • Files >400 MB use memory-efficient line-by-line reading
    • Only reads the specific lines needed (±50 around match)
    • Prevents memory issues and crashes with massive files
    • Automatic detection and logging when chunked reader is activated
• CSV export option: Export search results as CSV (Excel-compatible) or JSONL
  • CSV includes headers and all result fields
  • JSONL maintains backward compatibility for programmatic processing

🐛 Bug Fixes

• Fixed search returning 0 results: Ripgrep now properly searches files in large folders
  • Fixed critical Unicode encoding error when ripgrep searches binary/mixed-encoding files
  • Added UTF-8 encoding with error replacement to handle all file types
  • Added --text flag to ripgrep to search binary files without errors
• Fixed preview not loading for metadata-only indexed documents:
  • Fixed empty string handling - now properly detects and falls back to file reading
  • Preview now works for all search results, whether from database or ripgrep
  • Added detailed logging to debug preview loading issues
• Fixed test failures: Added missing existing_docs parameter to index_single_file method
• Fixed auto-enable detection: Now correctly counts all metadata-only documents, not just the first one
• Fixed NoneType error: Safe handling when ripgrep stdout is None due to encoding issues

Version 1.0.2

Metadata and configuration updates:

• Updated project URLs to cmdeniz.dev homepage
• Cleaned up package metadata

Version 1.0.1

Bug fixes and improvements:

• Fixed FTS5 database schema issue causing "no such column: T.content" error
• Fixed GUI tests crashing on Linux CI environments
• Fixed reindex test timing issues on Windows
• Improved database update logic for document reindexing
• Updated README for PyPI publication

Version 1.0.0

Initial release with:

• Full-text search using SQLite FTS5 with BM25 ranking
• CLI and PyQt5 GUI interfaces
• Support for PDF, DOCX, XLSX, PPTX, HTML, XML, and text files
• Hex extraction fallback for unknown formats
• Multi-threaded indexing with progress tracking
• Optional ripgrep integration for fast regex search
• Cross-platform support (Windows, macOS, Linux)
• Dark theme GUI with customizable accents
• Thread-safe database with WAL mode
• Comprehensive test suite (30+ tests)

Support

For issues, questions, or feature requests, please visit:

• PyPI Package: https://pypi.org/project/docfind/
• GitHub Issues: https://github.com/CihanMertDeniz/docfind/issues
• Documentation: Full documentation available in this README

---

DocFind - Find anything in your documents, instantly. 🔍