BHI layer v1: docs, schema, Phase A ingestion stubs

jobs/ingestion/cms_hospital_compare.py

@@ -0,0 +1,77 @@
+#!/usr/bin/env python3
+# READY TO DEPLOY — requires base Brain Postgres schema + run schemas/bhi_tables.sql
+"""
+CMS Hospital General Information (Care Compare) — used to cross-reference
+which acute hospitals host behavioral health units and to capture CCN-level
+facility metadata.
+
+Source: https://data.cms.gov/provider-data/api/1/datastore/query/xubh-q36u/0
+"""
+import logging
+import sys
+from _common import RateLimitedSession, bulk_insert, job_run
+
+LOG = logging.getLogger("bhi.cms_hospital_compare")
+BASE = "https://data.cms.gov/provider-data/api/1/datastore/query/xubh-q36u/0"
+PAGE = 500
+
+
+def test_endpoint():
+    s = RateLimitedSession()
+    r = s.get(BASE, params={"limit": 2}).json()
+    rows = r.get("results", [])
+    print(f"OK: {len(rows)} rows, sample:", rows[0].get("facility_name") if rows else None)
+    return bool(rows)
+
+
+def fetch_rows():
+    s = RateLimitedSession(min_interval=0.25)
+    offset, out = 0, []
+    while True:
+        b = s.get(BASE, params={"limit": PAGE, "offset": offset}).json().get("results", [])
+        if not b:
+            break
+        out.extend(b)
+        if len(b) < PAGE:
+            break
+        offset += PAGE
+    LOG.info("fetched %d hospitals", len(out))
+    return out
+
+
+def write_rows(conn, raw):
+    cols = [
+        "ccn","npi","name","address","city","state","zip","county_fips",
+        "lat","lon","facility_type","ownership","bed_count","psych_bed_count",
+        "pediatric_psych_bed_count","adolescent_unit","young_adult_unit",
+        "services_offered","populations_served","payment_accepted",
+        "medicaid_accepted","accreditation","opened_date","closed_date",
+        "last_verified","source","source_raw_id",
+    ]
+    rows = []
+    for r in raw:
+        rows.append((
+            r.get("facility_id"), None,
+            r.get("facility_name"), r.get("address"),
+            r.get("citytown"), r.get("state"), r.get("zip_code"), None,
+            None, None,
+            (r.get("hospital_type") or "hospital"),
+            r.get("hospital_ownership"),
+            None, None, None, None, None,
+            [], [], [], None, None, None, None, None,
+            "cms_hospital_compare", None,
+        ))
+    bulk_insert(conn, "bhi_facilities", cols, rows)
+    return len(rows)
+
+
+def main():
+    with job_run("bhi_cms_hospital_compare") as (conn, _):
+        n = write_rows(conn, fetch_rows())
+        LOG.info("inserted %d", n)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1 and sys.argv[1] == "test":
+        sys.exit(0 if test_endpoint() else 1)
+    main()