Citation Utilities

citation-utils extracts and normalizes Philippine Supreme Court citations that include a docket, a report citation, or both. It composes citation-date and citation-report for the LawSQL data pipeline.

The package is corpus-driven. It preserves the reported citation components that are present in the source rather than inferring missing case metadata. It helps locate and normalize references; it does not authenticate a document or prove the correctness of an OCR reading. Retain source text and location for legal review.

Quick Example

from citation_utils import Citation

raw = "G.R. No. 147033, April 30, 2003, 374 Phil. 1"
citation = Citation.extract_citation(raw)

assert citation is not None
assert citation.model_dump() == {
    "cat": "gr",
    "num": "147033",
    "date": "2003-04-30",
    "phil": "374 phil. 1",
    "scra": None,
    "offg": None,
}

Documentation

See the documentation for source-review limits, extraction and counting behavior, supported docket types, SQLite-oriented output, and local development checks. The 1.0 release history compares the stable release with the previously pushed 0.5.0 package.

Boundaries

• citation-date owns date grammar and date normalization.
• citation-report owns reporter grammar and report normalization.
• citation-utils owns docket grammar, joins dockets to reports, and provides database-ready citation records.

Administrative Matter and Bar Matter references can name rules rather than decisions. The default extraction flow excludes only known statutory serials using an exact canonical match; pass exclude_docket_rules=False only when the calling workflow needs those rule references as evidence.