bill_family_archives

Bill-family archive listings

One row per BILLSTATUS bulk folder a bill-family run read, holding the listing entry it read; the next run compares against it before downloading that folder's zip only when the generation's metadata lists the folder as complete (every member parsed and every rebuilt bill shaped), so a row records what was read, never that the folder may be skipped. Printing bodies are read apart from status, by each printing's own state; the same metadata names the BILLS bulk printings a zip refused, with the zip entry they were read at, so an unchanged zip is not read again for them. It also records, per bill, the reader that last read its status (this rollup's status rule and the SpicyDocs code): an unchanged bill is skipped only while that reader is the running one. And it records each status read that settled nothing, with the zip it read and its reader: a document the reader refused, a bill whose rows leave its status in doubt, and a bill its folder's zip no longer holds; none of them reopens its folder until the zip moves or another reader runs. A folder with such a record is complete. All columns are stored as VARCHAR.

Coverage. Not a range. The bill family's processing state: one row per GovInfo BILLSTATUS folder (Congress and bill type) it has listed, which the next run compares with the live listing to decide whether a zip needs downloading. (measured 2026-09-28)

Data quality. Only the four fields the skip compares are retained — name and link, which must name the folder's own zip or the comparison is refused outright, and modified_at and size, which must match for the zip to be skipped. The other six the publisher's listing states are deliberately absent rather than filled with guesses. All columns are stored as VARCHAR.

• Parquet file: bill_family_archives.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
name	VARCHAR	The zip's own file name in the folder listing, as the publisher spells it.
link	VARCHAR	The publisher's path to that zip; it must rebuild this folder's own locator or the skip is refused.
formatted_last_modified_time	VARCHAR	The publisher's own DD-Mon-YYYY HH:MM stamp string, kept as the evidence modified_at was read from.
modified_at	VARCHAR	That stamp as a UTC instant, which is one of the two values the unchanged comparison tests.
size	VARCHAR	The zip's byte size as the listing states it; the other value the comparison tests.
congress	VARCHAR	The numbered Congress whose folder this entry describes.
bill_type	VARCHAR	The publisher's lowercase bill-type folder name (hr, s, hjres, sjres, hconres, sconres, hres, sres).
observed_at	VARCHAR	When this run captured the listing the entry was read from.