prelawsql

prelawsql validates and indexes Philippine legal-source material. It is built for two kinds of readers:

• lawyers and legal researchers who need to trace a database record back to the exact source material; and
• developers who need deterministic parsing, reviewable failures, and safe rebuilds.

The package does not silently repair the source corpus. Raw source bytes are kept as evidence, normalized fields are stored separately, and records that cannot be interpreted safely are placed in quarantine for review.

Quick start

# Validate statute routes and atomically publish data/stats.db.
pre build-statute-index --replace

# Create data/cases.db with the decision and justice schemas.
pre init-decision-index --replace

# Ingest Supreme Court monthly decision listings for calendar year 2024.
# --end is exclusive.
pre ingest-decision-listings --start 2024 --end 2025

These commands create derived SQLite databases. They do not edit the decision, statute, or codification repositories.

Important classification rule

The word Opinion is not classified without knowing the document's role:

• a ponencia headed Opinion is a Decision;
• a separate opinion headed Opinion is a Separate Opinion; and
• an unqualified Opinion with no reliable role remains unclassified.

Explicit Concurring Opinion, Dissenting Opinion, and Separate Opinion labels are always treated as separate opinions. Kapasyahan is treated as a resolution, while the observed defective label Ecision is treated as a decision.

Command outcomes

--allow-quarantine changes only code 3 to 0. It cannot hide a failed or incomplete run.

Start with the documentation overview. The evidence and review guide explains what the database can establish—and what it cannot establish—before the data is relied upon in legal work. See Releases for the version-by-version change boundary and migration examples.