src.dackar.RCA.ner.equipment_ID_extractor

Functions

extract_equipment_ids(text, *[, patterns, normalize, ...])

Extract likely nuclear plant equipment / tag identifiers from free text.

Module Contents

src.dackar.RCA.ner.equipment_ID_extractor.extract_equipment_ids(text, *, patterns=None, normalize=True, unique=True, max_ids=500)[source]

Extract likely nuclear plant equipment / tag identifiers from free text.

This helper is intended to populate NERSeed.equipment_ids and/or chunk metadata for filtering and retrieval (e.g., in Chroma).

Inputs

text:

Arbitrary text (string). Can be SOP/CR/WO/ECA chunk text.

patterns:

Optional list of regex patterns (strings). If omitted, a reasonable default set is used. Patterns should contain a single capturing group for the tag itself OR match the tag as the full match.

normalize:
If True, normalize extracted tags:

  • uppercase

  • collapse whitespace

  • convert underscores/spaces to hyphens where appropriate

  • strip trailing punctuation

unique:

If True, return unique tags in first-seen order.

max_ids:

Safety limit; stops collecting after this many matches.

Output

List[str]:

Extracted equipment IDs/tags as strings (e.g., “P-101”, “MOV-204A”, “PT-1102”).

Notes

  • Tag naming conventions vary by plant/site. You should tune patterns to your org’s conventions once you see real data.

  • This function is conservative by default and tries to avoid grabbing years or generic hyphenated numbers.

  • Recommended usage:

    eq_ids = extract_equipment_ids(chunk_text) seed = NERSeed(…, equipment_ids=eq_ids, …)

Parameters:

  • text (str)

  • patterns (Optional[List[str]])

  • normalize (bool)

  • unique (bool)

  • max_ids (int)

Return type:

List[str]