Initial commit

scripts/_verify_out.py

@@ -0,0 +1,23 @@
+import sqlite3, json, sys
+db = "out/cli_direct.db"
+conn = sqlite3.connect(db)
+cur = conn.cursor()
+cur.execute("SELECT count(*) FROM daily_file_records")
+print("sqlite rows:", cur.fetchone()[0])
+cur.execute("SELECT count(DISTINCT record_type), group_concat(record_type) FROM (SELECT DISTINCT record_type FROM daily_file_records)")
+print("record types:", cur.fetchone())
+cur.execute("SELECT count(*) FROM daily_file_records WHERE committee IS NOT NULL AND committee != ''")
+print("with committee ref:", cur.fetchone()[0])
+cur.execute("SELECT count(*) FROM daily_file_records WHERE record_type='hearing'")
+print("hearings:", cur.fetchone()[0])
+cur.execute("SELECT count(*) FROM daily_file_records WHERE description IS NOT NULL AND description != ''")
+print("non-empty descriptions:", cur.fetchone()[0])
+conn.close()
+
+# Show one hearing record and one committee-bill from the jsonl
+with open("out/cli_direct.jsonl") as fh:
+    hearing = next((json.loads(l) for l in fh if json.loads(l).get("record_type")=="hearing"), None)
+with open("out/cli_direct.jsonl") as fh:
+    cbill = next((json.loads(l) for l in fh if json.loads(l).get("committee") and json.loads(l).get("record_type")=="bill"), None)
+print("sample hearing record:", json.dumps(hearing, sort_keys=True) if hearing else "NONE")
+print("sample committee bill record:", json.dumps(cbill, sort_keys=True) if cbill else "NONE")