Agent-System/deploy/postgres/goose/00060_2026-09-21-document-canonical-task-key.sql

-- +goose Up
-- 正式台账 doc_proj* 上，旧任务键与发布键（intake 与 bd_intake）不能再并存。
-- 只归档「同一项目里已经有发布键行」的旧键行；孤立旧键行保留，避免晋升时丢掉唯一数据。
-- 唯一索引不覆盖 doc_test_db_*：dev 库项目 1 有大量同键测试残留，全表唯一会误伤。
-- document_snapshots / document_content_blobs 有不可变触发器，不删除；正文副本写入归档 payload。

-- +goose StatementBegin
CREATE OR REPLACE FUNCTION canonical_document_task_key(task_key text)
RETURNS text
LANGUAGE sql
IMMUTABLE
PARALLEL SAFE
AS $$
  SELECT CASE btrim(COALESCE(task_key, ''))
    WHEN 'intake' THEN 'bd_intake'
    WHEN 'research' THEN 'bd_research'
    WHEN 'feasibility_dev' THEN 'bd_feasibility_dev'
    WHEN 'feasibility_reg' THEN 'bd_feasibility_reg'
    WHEN 'user_requirements' THEN 'bd_user_requirements'
    WHEN 'prelim_design' THEN 'bd_prelim_design'
    WHEN 'cdo_quote' THEN 'bd_cdo_quote'
    WHEN 'system_reqs' THEN 'pm_system_reqs'
    WHEN 'test_manager' THEN 'tm_system_verification'
    WHEN 'software_charter' THEN 'sw_software_charter'
    WHEN 'source_inventory' THEN 'sw_source_inventory'
    WHEN 'current_state_audit' THEN 'sw_current_state_audit'
    WHEN 'architecture_baseline' THEN 'sw_architecture_baseline'
    WHEN 'software_risk_register' THEN 'sw_software_risk_register'
    WHEN 'product_requirements' THEN 'sw_product_requirements'
    WHEN 'scope_backlog' THEN 'sw_scope_backlog'
    WHEN 'target_architecture' THEN 'sw_target_architecture'
    WHEN 'decision_log' THEN 'sw_decision_log'
    WHEN 'bootstrap_roadmap' THEN 'sw_bootstrap_roadmap'
    WHEN 'verification_strategy' THEN 'sw_verification_strategy'
    WHEN 'operations_baseline' THEN 'sw_operations_baseline'
    WHEN 'bootstrap_readiness' THEN 'sw_bootstrap_readiness'
    WHEN 'change_brief' THEN 'sw_change_brief'
    WHEN 'implementation_plan' THEN 'sw_implementation_plan'
    WHEN 'change_manifest' THEN 'sw_change_manifest'
    WHEN 'adr_delta' THEN 'sw_adr_delta'
    WHEN 'test_report' THEN 'sw_test_report'
    WHEN 'security_review' THEN 'sw_security_review'
    WHEN 'release_plan' THEN 'sw_release_plan'
    WHEN 'release_notes' THEN 'sw_release_notes'
    WHEN 'retrospective' THEN 'sw_retrospective'
    WHEN 'backlog_delta' THEN 'sw_backlog_delta'
    ELSE btrim(COALESCE(task_key, ''))
  END;
$$;
-- +goose StatementEnd

CREATE TABLE document_records_legacy_key_archive (LIKE document_records INCLUDING DEFAULTS);
ALTER TABLE document_records_legacy_key_archive ALTER COLUMN id DROP DEFAULT;
ALTER TABLE document_records_legacy_key_archive
    ADD COLUMN archive_id BIGSERIAL PRIMARY KEY,
    ADD COLUMN archived_at TIMESTAMPTZ NOT NULL DEFAULT now(),
    ADD COLUMN archive_reason TEXT NOT NULL DEFAULT '',
    ADD COLUMN related_payload JSONB NOT NULL DEFAULT '{}'::jsonb;

CREATE TABLE document_content_projections_legacy_key_archive (LIKE document_content_projections INCLUDING DEFAULTS);
ALTER TABLE document_content_projections_legacy_key_archive
    ADD COLUMN archive_id BIGSERIAL PRIMARY KEY,
    ADD COLUMN archived_at TIMESTAMPTZ NOT NULL DEFAULT now(),
    ADD COLUMN archive_reason TEXT NOT NULL DEFAULT '';

CREATE TEMP TABLE legacy_key_docs_to_archive ON COMMIT DROP AS
SELECT legacy.doc_id
FROM document_records legacy
JOIN document_records canon
  ON canon.project_id = legacy.project_id
 AND canon.doc_id <> legacy.doc_id
 AND canon.doc_id LIKE 'doc_proj%'
 AND legacy.doc_id LIKE 'doc_proj%'
 AND canonical_document_task_key(canon.task_key) = canonical_document_task_key(legacy.task_key)
 AND canon.task_key = canonical_document_task_key(canon.task_key)
 AND legacy.task_key <> canonical_document_task_key(legacy.task_key);

INSERT INTO document_records_legacy_key_archive
SELECT d.*,
       nextval('document_records_legacy_key_archive_archive_id_seq'),
       now(),
       'coexisting_legacy_task_key',
       '{}'::jsonb
FROM document_records d
JOIN legacy_key_docs_to_archive x ON x.doc_id = d.doc_id;

UPDATE document_records_legacy_key_archive a
SET related_payload = jsonb_build_object(
    'comments', COALESCE((
        SELECT jsonb_agg(to_jsonb(c)) FROM document_comments c WHERE c.doc_id = a.doc_id
    ), '[]'::jsonb),
    'change_proposals', COALESCE((
        SELECT jsonb_agg(to_jsonb(p)) FROM document_change_proposals p WHERE p.doc_id = a.doc_id
    ), '[]'::jsonb),
    'audit_logs', COALESCE((
        SELECT jsonb_agg(to_jsonb(l)) FROM document_audit_logs l WHERE l.doc_id = a.doc_id
    ), '[]'::jsonb),
    'versions', COALESCE((
        SELECT jsonb_agg(to_jsonb(v)) FROM document_versions v WHERE v.doc_id = a.doc_id
    ), '[]'::jsonb),
    'snapshots', COALESCE((
        SELECT jsonb_agg(jsonb_build_object(
            'id', s.id,
            'content_sha256', s.content_sha256,
            'source_kind', s.source_kind,
            'created_at', s.created_at,
            'markdown', b.content_markdown
        ))
        FROM document_snapshots s
        JOIN document_content_blobs b ON b.content_sha256 = s.content_sha256
        WHERE s.doc_id = a.doc_id
    ), '[]'::jsonb)
);

INSERT INTO document_content_projections_legacy_key_archive
SELECT p.*,
       nextval('document_content_projections_legacy_key_archive_archive_id_seq'),
       now(),
       'coexisting_legacy_task_key'
FROM document_content_projections p
JOIN legacy_key_docs_to_archive x ON x.doc_id = p.doc_id;

DELETE FROM document_content_projections p
USING legacy_key_docs_to_archive x
WHERE p.doc_id = x.doc_id;

DELETE FROM document_records d
USING legacy_key_docs_to_archive x
WHERE d.doc_id = x.doc_id;

-- +goose StatementBegin
DO $$
DECLARE
    dup text;
BEGIN
    SELECT string_agg(project_id::text || ':' || canon, ', ')
    INTO dup
    FROM (
        SELECT project_id, canonical_document_task_key(task_key) AS canon
        FROM document_records
        WHERE btrim(task_key) <> '' AND doc_id LIKE 'doc_proj%'
        GROUP BY project_id, canonical_document_task_key(task_key)
        HAVING count(*) > 1
    ) s;
    IF dup IS NOT NULL THEN
        RAISE EXCEPTION 'doc_proj canonical task key still duplicated: %', dup;
    END IF;
END $$;
-- +goose StatementEnd

CREATE UNIQUE INDEX uq_document_records_doc_proj_canonical_task_key
    ON document_records (project_id, canonical_document_task_key(task_key))
    WHERE btrim(task_key) <> '' AND doc_id LIKE 'doc_proj%';

-- +goose Down
DROP INDEX IF EXISTS uq_document_records_doc_proj_canonical_task_key;

INSERT INTO document_records
SELECT (jsonb_populate_record(
    NULL::document_records,
    to_jsonb(a) - 'archive_id' - 'archived_at' - 'archive_reason' - 'related_payload'
)).*
FROM document_records_legacy_key_archive a
ON CONFLICT (doc_id) DO NOTHING;

INSERT INTO document_content_projections
SELECT (jsonb_populate_record(
    NULL::document_content_projections,
    to_jsonb(a) - 'archive_id' - 'archived_at' - 'archive_reason'
)).*
FROM document_content_projections_legacy_key_archive a
ON CONFLICT (doc_id, snapshot_id) DO NOTHING;

DROP TABLE IF EXISTS document_content_projections_legacy_key_archive;
DROP TABLE IF EXISTS document_records_legacy_key_archive;
DROP FUNCTION IF EXISTS canonical_document_task_key(text);