feat: add monitoring marked articles and retention updates

server/migrations_monitoring/20260617113000_create_user_marked_cited_articles.down.sql

@@ -0,0 +1,7 @@
+DROP INDEX IF EXISTS idx_monitoring_citation_facts_content_source;
+
+ALTER TABLE monitoring_citation_facts
+    DROP COLUMN IF EXISTS content_source_id,
+    DROP COLUMN IF EXISTS content_source_type;
+
+DROP TABLE IF EXISTS monitoring_user_marked_articles;

server/migrations_monitoring/20260617113000_create_user_marked_cited_articles.up.sql

@@ -0,0 +1,41 @@
+CREATE TABLE IF NOT EXISTS monitoring_user_marked_articles (
+    id                 BIGSERIAL PRIMARY KEY,
+    tenant_id          BIGINT NOT NULL,
+    brand_id           BIGINT,
+    marked_by_user_id  BIGINT,
+    article_title      TEXT NOT NULL,
+    original_url       TEXT NOT NULL,
+    normalized_url     TEXT NOT NULL,
+    host               VARCHAR(255) NOT NULL,
+    registrable_domain VARCHAR(255) NOT NULL,
+    site_key           VARCHAR(255) NOT NULL,
+    last_path_segment  VARCHAR(255),
+    publish_platform   VARCHAR(100) NOT NULL DEFAULT '外部文章',
+    marked_at          TIMESTAMPTZ NOT NULL DEFAULT NOW(),
+    expires_at         TIMESTAMPTZ NOT NULL DEFAULT (NOW() + INTERVAL '6 months'),
+    created_at         TIMESTAMPTZ NOT NULL DEFAULT NOW(),
+    updated_at         TIMESTAMPTZ NOT NULL DEFAULT NOW()
+);
+
+CREATE UNIQUE INDEX IF NOT EXISTS uk_monitoring_user_marked_article_url
+    ON monitoring_user_marked_articles(tenant_id, normalized_url);
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_user_marked_articles_domain
+    ON monitoring_user_marked_articles(tenant_id, registrable_domain, created_at DESC, id DESC);
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_user_marked_articles_title
+    ON monitoring_user_marked_articles USING GIN (to_tsvector('simple', article_title));
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_user_marked_articles_created
+    ON monitoring_user_marked_articles(tenant_id, created_at DESC, id DESC);
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_user_marked_articles_expires
+    ON monitoring_user_marked_articles(expires_at, id);
+
+ALTER TABLE monitoring_citation_facts
+    ADD COLUMN IF NOT EXISTS content_source_type VARCHAR(30),
+    ADD COLUMN IF NOT EXISTS content_source_id BIGINT;
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_citation_facts_content_source
+    ON monitoring_citation_facts(tenant_id, content_source_type, content_source_id)
+    WHERE content_source_type IS NOT NULL AND content_source_id IS NOT NULL;

server/migrations_monitoring/20260617124500_optimize_monitoring_content_citations.down.sql

@@ -0,0 +1,5 @@
+DROP INDEX IF EXISTS idx_article_url_alias_article_lookup;
+DROP INDEX IF EXISTS idx_citation_facts_content_window;
+DROP INDEX IF EXISTS idx_citation_facts_normalized_url;
+DROP INDEX IF EXISTS idx_citation_facts_tenant_run;
+DROP INDEX IF EXISTS idx_monitor_runs_citation_summary;

server/migrations_monitoring/20260617124500_optimize_monitoring_content_citations.up.sql

@@ -0,0 +1,63 @@
+CREATE INDEX IF NOT EXISTS idx_monitor_runs_citation_summary
+    ON question_monitor_runs (
+        tenant_id,
+        brand_id,
+        collector_type,
+        status,
+        business_date,
+        question_id,
+        ai_platform_id,
+        id
+    );
+
+CREATE INDEX IF NOT EXISTS idx_citation_facts_tenant_run
+    ON monitoring_citation_facts(tenant_id, run_id);
+
+CREATE INDEX IF NOT EXISTS idx_citation_facts_normalized_url
+    ON monitoring_citation_facts(tenant_id, normalized_url);
+
+CREATE INDEX IF NOT EXISTS idx_citation_facts_content_window
+    ON monitoring_citation_facts (
+        tenant_id,
+        brand_id,
+        business_date,
+        ai_platform_id,
+        content_source_type,
+        content_source_id,
+        run_id
+    )
+    WHERE content_source_type IS NOT NULL AND content_source_id IS NOT NULL;
+
+CREATE INDEX IF NOT EXISTS idx_article_url_alias_article_lookup
+    ON monitoring_article_url_aliases(tenant_id, article_id, updated_at DESC, id DESC)
+    WHERE confidence = 'high';
+
+UPDATE monitoring_citation_facts cf
+SET content_source_type = 'published_article',
+    content_source_id = alias.article_id
+FROM monitoring_article_url_aliases alias
+WHERE alias.tenant_id = cf.tenant_id
+  AND alias.normalized_url = cf.normalized_url
+  AND alias.confidence = 'high'
+  AND alias.article_id IS NOT NULL
+  AND (
+    cf.content_source_type IS DISTINCT FROM 'published_article'
+    OR cf.content_source_id IS DISTINCT FROM alias.article_id
+  );
+
+UPDATE monitoring_citation_facts
+SET content_source_type = 'published_article',
+    content_source_id = article_id
+WHERE content_source_type IS NULL
+  AND content_source_id IS NULL
+  AND article_id IS NOT NULL;
+
+UPDATE monitoring_citation_facts cf
+SET content_source_type = 'manual_mark',
+    content_source_id = marked.id
+FROM monitoring_user_marked_articles marked
+WHERE marked.tenant_id = cf.tenant_id
+  AND marked.normalized_url = cf.normalized_url
+  AND marked.expires_at > NOW()
+  AND cf.content_source_type IS NULL
+  AND cf.content_source_id IS NULL;

server/migrations_monitoring/20260617132000_tune_external_article_citation_indexes.down.sql

@@ -0,0 +1,7 @@
+DROP INDEX IF EXISTS idx_marked_articles_title_trgm;
+DROP INDEX IF EXISTS idx_marked_articles_domain_trgm;
+DROP INDEX IF EXISTS idx_citation_facts_day_content_run;
+DROP INDEX IF EXISTS idx_citation_facts_run_content;
+
+CREATE INDEX IF NOT EXISTS idx_monitoring_user_marked_articles_title
+    ON monitoring_user_marked_articles USING GIN (to_tsvector('simple', article_title));

server/migrations_monitoring/20260617132000_tune_external_article_citation_indexes.up.sql

@@ -0,0 +1,44 @@
+CREATE EXTENSION IF NOT EXISTS pg_trgm;
+
+CREATE INDEX IF NOT EXISTS idx_citation_facts_run_content
+    ON monitoring_citation_facts (
+        tenant_id,
+        run_id,
+        content_source_type,
+        content_source_id
+    )
+    WHERE content_source_type IS NOT NULL AND content_source_id IS NOT NULL;
+
+CREATE INDEX IF NOT EXISTS idx_citation_facts_day_content_run
+    ON monitoring_citation_facts (
+        tenant_id,
+        brand_id,
+        business_date,
+        ai_platform_id,
+        run_id,
+        content_source_type,
+        content_source_id
+    )
+    WHERE content_source_type IS NOT NULL AND content_source_id IS NOT NULL;
+
+CREATE INDEX IF NOT EXISTS idx_marked_articles_domain_trgm
+    ON monitoring_user_marked_articles
+    USING GIN (
+        lower(registrable_domain) gin_trgm_ops,
+        lower(host) gin_trgm_ops,
+        lower(site_key) gin_trgm_ops
+    );
+
+DROP INDEX IF EXISTS idx_monitoring_user_marked_articles_title;
+
+CREATE INDEX IF NOT EXISTS idx_marked_articles_title_trgm
+    ON monitoring_user_marked_articles
+    USING GIN (lower(article_title) gin_trgm_ops);
+
+ALTER TABLE monitoring_user_marked_articles
+    ALTER COLUMN publish_platform SET DEFAULT '外部文章';
+
+UPDATE monitoring_user_marked_articles
+SET publish_platform = '外部文章',
+    updated_at = NOW()
+WHERE publish_platform IN ('自有文章', 'Own Article', 'Owned Article', 'Manual Article');

server/migrations_monitoring/20260617133500_tune_marked_article_domain_lookup.down.sql

@@ -0,0 +1,10 @@
+DROP INDEX IF EXISTS idx_marked_articles_site_key_lookup;
+DROP INDEX IF EXISTS idx_marked_articles_host_lookup;
+
+CREATE INDEX IF NOT EXISTS idx_marked_articles_domain_trgm
+    ON monitoring_user_marked_articles
+    USING GIN (
+        lower(registrable_domain) gin_trgm_ops,
+        lower(host) gin_trgm_ops,
+        lower(site_key) gin_trgm_ops
+    );

server/migrations_monitoring/20260617133500_tune_marked_article_domain_lookup.up.sql

@@ -0,0 +1,7 @@
+DROP INDEX IF EXISTS idx_marked_articles_domain_trgm;
+
+CREATE INDEX IF NOT EXISTS idx_marked_articles_host_lookup
+    ON monitoring_user_marked_articles(tenant_id, host, created_at DESC, id DESC);
+
+CREATE INDEX IF NOT EXISTS idx_marked_articles_site_key_lookup
+    ON monitoring_user_marked_articles(tenant_id, site_key, created_at DESC, id DESC);