phase-10.5: screener enhancements

server/domains/shared/db/queries.constant.ts

@@ -163,6 +163,68 @@ export const UNIVERSE_QUERIES = {
     WHERE type != 'crypto'
     ORDER BY ticker
   `,
+
+  // Every ticker screened recently (snapshot ledger) — part of the news universe
+  DISTINCT_SNAPSHOT_TICKERS_SINCE: `
+    SELECT DISTINCT ticker FROM signal_snapshots
+    WHERE snapshot_date >= ?
+    ORDER BY ticker
+  `,
+};
+
+// ── News Queries (FREE-DATA-STACK §2–5 — free-tier news pipeline) ───────────
+
+export const NEWS_QUERIES = {
+  // INSERT OR IGNORE — url_hash PK is the first dedupe line (returns 0 changes on dup)
+  INSERT_ARTICLE: `
+    INSERT OR IGNORE INTO news_articles
+      (url_hash, title_hash, ticker_list, headline, body, source, catalyst, url, published_at, created_at)
+    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `,
+
+  // Second dedupe line: same (normalized) title seen recently → syndicated copy
+  TITLE_SEEN_SINCE: `
+    SELECT 1 FROM news_articles
+    WHERE title_hash = ? AND published_at >= ?
+    LIMIT 1
+  `,
+
+  INSERT_CATALYST_LINK: `
+    INSERT OR IGNORE INTO ticker_catalysts (ticker, day, url_hash)
+    VALUES (?, ?, ?)
+  `,
+
+  // Per-ticker daily cap check (FREE-DATA-STACK §4.4)
+  COUNT_TICKER_DAY: `
+    SELECT COUNT(*) AS n FROM ticker_catalysts
+    WHERE ticker = ? AND day = ?
+  `,
+
+  // Stories for one ticker since a given day — what the UI reads (never Yahoo live)
+  SELECT_TICKER_NEWS: `
+    SELECT a.* FROM ticker_catalysts c
+    JOIN news_articles a ON a.url_hash = c.url_hash
+    WHERE c.ticker = ? AND c.day >= ?
+    ORDER BY a.published_at DESC
+  `,
+
+  SELECT_RECENT: `
+    SELECT * FROM news_articles
+    ORDER BY published_at DESC
+    LIMIT ?
+  `,
+
+  // Retention (FREE-DATA-STACK §5): purge bodies after 90d, drop unreferenced after 18mo
+  PURGE_BODIES_BEFORE: `
+    UPDATE news_articles SET body = NULL
+    WHERE body IS NOT NULL AND published_at < ?
+  `,
+
+  DELETE_UNREFERENCED_BEFORE: `
+    DELETE FROM news_articles
+    WHERE published_at < ?
+      AND url_hash NOT IN (SELECT url_hash FROM ticker_catalysts)
+  `,
 };
 
 // ── Signal Snapshot Queries (P0.1 — signal track record) ────────────────────
@@ -287,6 +349,31 @@ export const DDL = `
 
   CREATE INDEX IF NOT EXISTS idx_snapshots_date   ON signal_snapshots(snapshot_date);
   CREATE INDEX IF NOT EXISTS idx_snapshots_signal ON signal_snapshots(signal, snapshot_date);
+
+  CREATE TABLE IF NOT EXISTS news_articles (
+    url_hash     TEXT PRIMARY KEY,          -- sha256(url)
+    title_hash   TEXT NOT NULL,             -- sha256(normalized headline) — syndication dedupe
+    ticker_list  TEXT NOT NULL,             -- JSON array of matched universe tickers
+    headline     TEXT NOT NULL,
+    body         TEXT,                      -- nullable; purged after 90 days (retention job)
+    source       TEXT NOT NULL,             -- 'edgar' | 'prwire' | 'yahoo'
+    catalyst     TEXT,                      -- 'earnings'|'ma'|'guidance'|'regulatory'|'macro'|NULL
+    url          TEXT NOT NULL,
+    published_at TEXT NOT NULL,             -- ISO timestamp
+    created_at   TEXT NOT NULL
+  );
+
+  CREATE INDEX IF NOT EXISTS idx_news_published ON news_articles(published_at DESC);
+  CREATE INDEX IF NOT EXISTS idx_news_title     ON news_articles(title_hash, published_at);
+
+  CREATE TABLE IF NOT EXISTS ticker_catalysts (
+    ticker   TEXT NOT NULL,
+    day      TEXT NOT NULL,                 -- YYYY-MM-DD (published date)
+    url_hash TEXT NOT NULL REFERENCES news_articles(url_hash),
+    PRIMARY KEY (ticker, day, url_hash)
+  );
+
+  CREATE INDEX IF NOT EXISTS idx_catalysts_ticker ON ticker_catalysts(ticker, day DESC);
 `;
 
 // ── Runtime migrations (ALTER TABLE for existing DBs) ────────────────────────