Show HN: ChatExport Structurer – parse ChatGPT/Claude exports into queryable SQL (github.com)

Show HN: ChatExport Structurer is an open-source, local-first parser that converts exported chat logs from ChatGPT, Anthropic Claude, and X.AI Grok into a clean, queryable SQLite archive—positioning itself as a foundational “memory” layer for AI apps. The project (part of the upcoming MyChatArchive platform, Q1 2025) emphasizes data ownership and privacy: no cloud, no API keys, everything stays local. It’s aimed at developers who want to stitch conversation context into RAG systems, agents, semantic search, or analytics without losing insights buried in raw JSON exports. Technically, the tool is a Python 3.8+ CLI that uses streaming JSON parsing (ijson) and a streaming parser capable of handling multi-GB files, with SHA1-based deduplication and a preview (--test) mode. Data lands in a normalized messages table (message_id, canonical_thread_id, platform, account_id, ts, role, text, title, source_id) and a virtual FTS5 table plus a mapping table for fast full-text search and joins. Examples and queries are included (LIKE and FTS MATCH queries, counts, exports), and the parser architecture is modular: new platforms are added by authoring a simple parser yielding normalized messages. MIT-licensed and production-ready, it’s designed to be the local core for later additions like vector search, AI synthesis, a web UI, and optional cloud sync.