Skip to content

court_opinions

Court opinions (index, no text)

Every CourtListener opinion's decision, type and author, without its text. It is what connects court_citation_map and court_parentheticals, which name opinions, to court_opinion_clusters, which holds CourtListener's decision records (one decision can have several): join opinion_id from either, then cluster_id to court_opinion_clusters.cluster_id. local_path and download_url locate the original document. Rebuilt whole from each quarterly export, locally rather than on a schedule, because the source export is 54.6 GB. All columns are VARCHAR.

Coverage. Not a range. The complete 2026-06-30 CourtListener opinions export without its eight text fields: every opinion_id unique and every cluster_id set. All but 21 name a decision in court_opinion_clusters; those 21 are RECAP trial-court opinions created between the two exports, which opinion search does not index, and resolve with the next export (receipt join-gaps-2026-09-26/i/). Opinion text is not included: the fork links out to CourtListener instead. (measured 2026-09-22)

Data quality. Author strings are the publisher's text and are not resolved to people; author_id is set only where the publisher linked a judge. opinion_type is the publisher's code, such as 010combined or 040dissent.

  • Parquet file: court_opinions.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
opinion_id VARCHAR CourtListener opinion ID. Primary key; joins court_citation_map and court_parentheticals opinion ids.
cluster_id VARCHAR CourtListener decision-cluster ID the opinion belongs to; joins court_opinion_clusters.cluster_id.
opinion_type VARCHAR Publisher opinion type code, such as 010combined, 020lead, 030concurrence or 040dissent.
author_id VARCHAR CourtListener judge ID of the author, when the publisher linked one.
author_str VARCHAR Author as the publisher's source text names them; not resolved to a person.
per_curiam VARCHAR Publisher per-curiam flag (t or f).
joined_by_str VARCHAR Judges joining the opinion, as the publisher's source text names them.
page_count VARCHAR Page count of the original document when known, as text.
sha1 VARCHAR Publisher's SHA-1 of the original document; not a SpicyRegs capture digest.
download_url VARCHAR Where the publisher obtained the original document, when recorded; no current accessibility claim. For court-website opinions the file name usually carries the lead docket number (a consolidated case lists two), the fallback key to court_dockets.docket_number.
local_path VARCHAR Path of the original document in CourtListener's storage, not a SpicyRegs file.
extracted_by_ocr VARCHAR Publisher flag for text extracted by OCR (t or f).
date_created VARCHAR Timestamp the publisher created the opinion record.
date_modified VARCHAR Timestamp the publisher last modified the opinion record.
dump_date VARCHAR The CourtListener bulk export the row was copied from (ISO date).