ChatGPT Export Viewer (SQLite Edition)

This Flask application lets you browse ChatGPT exports locally (conversations.json, chat.html, and their associated assets). SQLite persistence avoids reparsing the raw files on every request.

Key Features

• Bootstrap web interface to browse conversations, display messages and attachments, copy Markdown, and render KaTeX.
• Automatic detection of assets (images, audio, other files) with MIME type inference.
• Full-text search powered by the SQLite FTS5 extension (Whoosh is no longer required).
• Utility scripts to inspect and clean exports (Analyse_json.py, GPT_cleaner.py, and more).
• Works entirely offline - fonts and libraries are bundled locally.
• Lightweight syntax highlighting (Python, JS/C, HTML, CSS) with no external dependency.
• Local filters to highlight conversations that contain media, audio, or images.

Quick Installation

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt  # or install Flask/ijson/python-magic manually if needed

Note: the core app only relies on the Python standard library plus Flask. If python-magic is installed, MIME detection becomes more accurate.

Starting the Server

source .venv/bin/activate  # or activate your preferred environment
FLASK_APP=app python -m flask run --no-debugger --no-reload --host 0.0.0.0 --port 5001

Drop your export folders (the ones that contain conversations.json and chat.html) at the repository root; they will be detected automatically.

Importing a ZIP Export

• Use the "Import ZIP export" form in the interface to upload the archive you received from ChatGPT.
• Provide the target path - it can point to any mounted volume (external SSD, NAS, etc.).
• The application extracts the archive, creates the export folder, indexes it in SQLite, and adds it to the list automatically.
• You can leave the "Folder name" field empty to reuse the ZIP filename.

Sphinx Documentation

Developer documentation lives in docs/. To build it:

source .venv/bin/activate
pip install -r requirements-dev.txt  # includes sphinx-rtd-theme
cd docs
sphinx-build -b html . _build/html

The generated pages describe the server architecture, asset detection, and the front-end rendering logic (math, assets, search).

SQLite Persistence

• The database file is app_data.db at the repository root.
• The first load of a folder parses all raw files and stores the results (conversations, asset mapping, MIME guesses) in SQLite.
• Subsequent requests read directly from SQLite, removing the initial delay.
• An FTS5 virtual table (conversation_search) powers the search route. If FTS5 is not available in your SQLite build, the /search route returns a 501 error.

Utility Scripts

Use the CLI script to ingest (or reingest) an export manually:

source .venv/bin/activate
python scripts/refresh_export.py 06042025

Development Notes

• Python code follows PEP 8; type annotations are being added progressively.
• Server-side modules:
  • app_core/parsing.py: extraction logic (mirrors the original behaviour).
  • app_core/db.py: SQLite connection helpers.
  • app_core/schema.py: schema creation and migrations.
  • app_core/ingest.py: ingestion pipeline + search.
• Front-end logic stays in the browser (static/script.js).

Quick Checks

source .venv/bin/activate
python -m compileall app.py app_core
python -m pytest  # add tests as needed

Roadmap

• Add unit tests for app_core.parsing.
• Handle very large exports through incremental ingestion.
• Expose a UI button that triggers a manual reindex.