feat(taxonomy): add rust sidecar compact surface pipeline

2026-03-12 15:23:10 -04:00
parent f2c25fb9c6
commit 58061af006
84 changed files with 19350 additions and 265 deletions
--- a/drizzle/0011_remove_legacy_xbrl_defaults.sql
+++ b/drizzle/0011_remove_legacy_xbrl_defaults.sql
@@ -0,0 +1,100 @@
+PRAGMA foreign_keys=OFF;
+--> statement-breakpoint
+CREATE TABLE `__new_filing_taxonomy_snapshot` (
+  `id` integer PRIMARY KEY AUTOINCREMENT NOT NULL,
+  `filing_id` integer NOT NULL,
+  `ticker` text NOT NULL,
+  `filing_date` text NOT NULL,
+  `filing_type` text NOT NULL,
+  `parse_status` text NOT NULL,
+  `parse_error` text,
+  `source` text NOT NULL,
+  `parser_engine` text DEFAULT 'fiscal-xbrl' NOT NULL,
+  `parser_version` text DEFAULT 'unknown' NOT NULL,
+  `taxonomy_regime` text DEFAULT 'unknown' NOT NULL,
+  `fiscal_pack` text,
+  `periods` text,
+  `faithful_rows` text,
+  `statement_rows` text,
+  `surface_rows` text,
+  `detail_rows` text,
+  `kpi_rows` text,
+  `derived_metrics` text,
+  `validation_result` text,
+  `normalization_summary` text,
+  `facts_count` integer DEFAULT 0 NOT NULL,
+  `concepts_count` integer DEFAULT 0 NOT NULL,
+  `dimensions_count` integer DEFAULT 0 NOT NULL,
+  `created_at` text NOT NULL,
+  `updated_at` text NOT NULL,
+  FOREIGN KEY (`filing_id`) REFERENCES `filing`(`id`) ON UPDATE no action ON DELETE cascade
+);
+--> statement-breakpoint
+INSERT INTO `__new_filing_taxonomy_snapshot` (
+  `id`,
+  `filing_id`,
+  `ticker`,
+  `filing_date`,
+  `filing_type`,
+  `parse_status`,
+  `parse_error`,
+  `source`,
+  `parser_engine`,
+  `parser_version`,
+  `taxonomy_regime`,
+  `fiscal_pack`,
+  `periods`,
+  `faithful_rows`,
+  `statement_rows`,
+  `surface_rows`,
+  `detail_rows`,
+  `kpi_rows`,
+  `derived_metrics`,
+  `validation_result`,
+  `normalization_summary`,
+  `facts_count`,
+  `concepts_count`,
+  `dimensions_count`,
+  `created_at`,
+  `updated_at`
+)
+SELECT
+  `id`,
+  `filing_id`,
+  `ticker`,
+  `filing_date`,
+  `filing_type`,
+  `parse_status`,
+  `parse_error`,
+  `source`,
+  `parser_engine`,
+  `parser_version`,
+  `taxonomy_regime`,
+  `fiscal_pack`,
+  `periods`,
+  `faithful_rows`,
+  `statement_rows`,
+  `surface_rows`,
+  `detail_rows`,
+  `kpi_rows`,
+  `derived_metrics`,
+  `validation_result`,
+  `normalization_summary`,
+  `facts_count`,
+  `concepts_count`,
+  `dimensions_count`,
+  `created_at`,
+  `updated_at`
+FROM `filing_taxonomy_snapshot`;
+--> statement-breakpoint
+DROP TABLE `filing_taxonomy_snapshot`;
+--> statement-breakpoint
+ALTER TABLE `__new_filing_taxonomy_snapshot` RENAME TO `filing_taxonomy_snapshot`;
+--> statement-breakpoint
+CREATE UNIQUE INDEX `filing_taxonomy_snapshot_filing_uidx` ON `filing_taxonomy_snapshot` (`filing_id`);
+--> statement-breakpoint
+CREATE INDEX `filing_taxonomy_snapshot_ticker_date_idx` ON `filing_taxonomy_snapshot` (`ticker`,`filing_date`);
+--> statement-breakpoint
+CREATE INDEX `filing_taxonomy_snapshot_status_idx` ON `filing_taxonomy_snapshot` (`parse_status`);
+--> statement-breakpoint
+PRAGMA foreign_keys=ON;