adding caching to presque gratos

storage.py

@@ -0,0 +1,55 @@
+# storage.py
+from __future__ import annotations
+import sqlite3, pathlib, datetime as dt
+from typing import Optional, Iterable, Tuple
+import os
+DB_PATH = "/data/published.db"  # bind-mount ./data:/data in docker
+
+_SCHEMA = """
+PRAGMA journal_mode = WAL;
+CREATE TABLE IF NOT EXISTS published_items(
+  platform TEXT NOT NULL,            -- e.g. xgp | egs | psplus
+  key      TEXT PRIMARY KEY,         -- your dedupe key (see below)
+  first_seen_utc TEXT NOT NULL,      -- ISO-8601
+  last_post_id   TEXT                -- Ghost post id that recorded it
+);
+CREATE INDEX IF NOT EXISTS idx_platform ON published_items(platform);
+"""
+
+class Storage:
+    def __init__(self, db_path: str = DB_PATH):
+
+        
+        if not os.path.isfile(db_path):
+            db_path = os.environ.get("DB_FILE_FALLBACK", r"f:\workspace\Substack_JV\data\published.db")
+
+        print(db_path)
+        pathlib.Path(db_path).parent.mkdir(parents=True, exist_ok=True)
+        self.conn = sqlite3.connect(db_path)
+        self.conn.execute("PRAGMA foreign_keys = ON;")
+        for stmt in filter(None, _SCHEMA.split(";")):
+            if stmt.strip():
+                self.conn.execute(stmt)
+
+    def seen(self, key: str) -> bool:
+        cur = self.conn.execute("SELECT 1 FROM published_items WHERE key=?", (key,))
+        return cur.fetchone() is not None
+
+    def remember(self, platform: str, key: str, post_id: Optional[str]):
+        self.conn.execute(
+            "INSERT OR IGNORE INTO published_items(platform,key,first_seen_utc,last_post_id) VALUES(?,?,?,?)",
+            (platform, key, dt.datetime.utcnow().isoformat(), post_id),
+        )
+        if post_id:
+            self.conn.execute("UPDATE published_items SET last_post_id=? WHERE key=?", (post_id, key))
+        self.conn.commit()
+
+    def bulk_remember(self, platform: str, pairs: Iterable[Tuple[str, Optional[str]]]):
+        rows = [(platform, k, dt.datetime.utcnow().isoformat(), pid) for (k, pid) in pairs]
+        self.conn.executemany(
+            "INSERT OR IGNORE INTO published_items(platform,key,first_seen_utc,last_post_id) VALUES(?,?,?,?)",
+            rows
+        )
+        self.conn.commit()
+
+