Settings
Client profile, extraction patterns, Azure endpoints & keys, and prompts. Stored in the OS keyring — shared with the desktop app.
Client profile
Everything client-specific lives here: who can never be the vendor, the client's own labels, and the formats values must follow. Consolidate runs under the active profile and records it in every result.
One name per line. Matched as whole words, any case: "Karrow" does not catch "Karrowmere".
One label per line. Added to the built-in readers ("Vendor Name", "Purchase Order No.", "P.O. No."), never replacing them.
Only where this client differs from Generic: extra labels for an attribute, or a different value format.
▶ Show exactly what will be saved
▶
Extraction patterns
click to expand
Extraction patterns
Regex patterns that pre-fill metadata fields before GPT runs. Edit inline, toggle individual rows on/off, or test a regex against any text.
| On/Off | Attribute | Regex | Type | Color | Description | Test |
|---|
📋 Tag register coverage
Upload your FLOC / tag register (CSV, Excel, or TXT — one tag per row, or a column labelled Tag, Tag Number, or FLOC). The checker tests every register tag against your enabled patterns and shows which entries aren't covered yet — clustered by shape with an auto-generated regex for each cluster. Edit if you want, then publish with one click.
🔬 Regex refinement
Build a regex from example tags. Add more examples to generalise. Insist a non-matching tag is valid → the regex refines to include it. When happy, publish to the patterns list above.
Endpoints, keys & storage
Azure Document Intelligence + Azure OpenAI credentials, and the cache database location.
Credentials live in the OS keyring; the DB path lives in ~/.equipment_extractor/app_state.json.