DTLarchive

It turns the conversations*.json files from a ChatGPT export into a persistent SQLite index. You can then retrieve a word, phrase, or combination of terms, restore the context of matches, and reuse the findings in other knowledge-management tools.

Why DTLarchive?

A ChatGPT export can contain thousands of conversations spread across several JSON files. DTLarchive turns these archives into a persistent local index so you can quickly retrieve a subject, restore its context, and reuse the results.

Main features

• local, persistent SQLite full-text index;
• incremental import of new or modified archives;
• searches by word, phrase, combination, exclusion, or word prefix;
• date filtering and searches across questions, answers, or both;
• French and English interface, contextual help, and reports;
• relevance ranking, immediate contexts, conversation pages, and JSON output.

Search

The syntax accepts AND/ET to require several terms, OR/OU, a comma, or a semicolon for alternatives, -word to exclude, and print* to search for a word prefix.

asylum AND John Doe, pension
backup, -network
print*

Results

The HTML report presents statistics, term distribution, the most frequent matching conversation titles, relevance scores, and contexts. A button opens the complete conversation at the first matching message.

mining_results.json contains the same findings in a reusable format. This structured output allows DTLarchive to act as a first stage for knowledge extraction, comparison, or knowledge-base enrichment tools.

Index and command line

Later searches reuse DTLarchive-index.sqlite. The --reindex option forces a complete rebuild. Both the executable and Python script accept files or folders, dates, a scope, an output directory, and a language.

.\DTLarchive.exe "D:\Archives\ChatGPT" `
  --mots-cles "asylum AND John Doe" `
  --role user --lang en

Building with PyInstaller

The Windows build uses DTLarchive.spec. After installing PyInstaller, the following command produces dist\DTLarchive.exe.

python -m PyInstaller --clean --noconfirm .\DTLarchive.spec

Diagnostics

Each run appends to a daily HTML log in logs\DTLarchive_YYYYMMDD.html. It records startup, index updates, search settings, and any errors.

Privacy