`bill_sections`

Bill sections

One row per content-bearing node of one bill version, in document order. All columns are stored as VARCHAR. Published as one file per Congress; the server and the publication readers present them as one table, but a raw download holds one file for each Congress.

Coverage. Sampled: the sections parsed from printings whose bodies the bill family acquired (source govinfo in bill_versions). (measured 2026-09-28)

Data quality. A printing with no parsed section tree is a counted refusal, never a row with invented values. On the measured cold-start run of the 119th (receipt d1-measured-run-2026-09-19/) the family reported 21,466 such refusals against 6,301 published rows — one per printing whose body the 600-per-run fetch cap did not reach, plus those offered only as PDF. The refusal count is therefore a measure of how far the cap is from the corpus, not of anything wrong with the printings; it falls as successive runs spend the cap on ground earlier ones did not reach. A section is keyed on its seq within the printing since spicy-docs 0.35.0; the older key (match_path, body_index) repeats inside a printing. All columns are stored as VARCHAR.

Parquet file: bill_sections.parquet
MCP query_sql support: Configured; requires an available artifact.
Publication status: Not established by this schema page or its measurement date.
Row count: Not stated here; the MCP describe_table reply gives the live count under publication.

Column	Type	Description
`bill_id`	`VARCHAR`	The bill this section belongs to.
`version_code`	`VARCHAR`	The printing this section was read from.
`source`	`VARCHAR`	Which acquisition path supplied the printing; part of the parent version's key.
`match_path`	`VARCHAR`	The normalized, division-free cross-version key, unit-separator joined; it can repeat.
`match_path_json`	`VARCHAR`	The same path as a JSON array, so a consumer need not split on a separator.
`display_path_json`	`VARCHAR`	The human-facing path the engine composes, as a JSON array.
`element_id`	`VARCHAR`	The publisher's own id attribute on the element this node came from.
`section_number`	`VARCHAR`	The section's own number as displayed (Sec. 1), where it has one.
`heading`	`VARCHAR`	The section's header text, empty where the node carries none.
`body`	`VARCHAR`	The section's body text, at the grain people actually query.
`display_text`	`VARCHAR`	The engine's display rendering of the body, kept apart from the body itself.
`division_label`	`VARCHAR`	The division this section sits under, as the engine composes the label.
`division_key`	`VARCHAR`	The normalized division key, which is what a cross-version match ignores.
`body_index`	`VARCHAR`	Which body element this node came from; a reported bill carries two.
`seq`	`VARCHAR`	Position in document order; the section's key within its printing, which the match path cannot be.
`body_chars`	`VARCHAR`	Character length of body.
`body_sha256`	`VARCHAR`	Digest of body's UTF-8 bytes, so an unchanged section is recognisable without a join.
`version_date`	`VARCHAR`	The parent printing's date, carried so this table versions with its parent.
`congress`	`VARCHAR`	The Congress of the bill, the prefix of bill_id (119 for 119-hr-1), so a host can store the table one file per Congress.