Parse MobilityDB's vendored pgtypes headers into the catalog

parser/parser.py

@@ -109,12 +109,15 @@ def merge_meta(idl: dict, meta_path: Path) -> dict:
 )
 
 
-def parse_meos(entry: Path, include_dir: Path) -> dict:
+def parse_meos(entry: Path, include_dir: Path,
+               extra_headers: tuple[Path, ...] = ()) -> dict:
     index = clang.cindex.Index.create()
+    extra_dirs = sorted({h.parent.resolve() for h in extra_headers})
     tu = index.parse(str(entry), args=[
         "-x", "c",
         "-std=c11",
         f"-I{include_dir}",
+        *(f"-I{d}" for d in extra_dirs),
         "-DMEOS",
         # Define the MEOS ``UNUSED`` attribute macro on the command line so it is
         # always in scope: the amalgamated entry point may parse a header that
@@ -131,6 +134,7 @@ def parse_meos(entry: Path, include_dir: Path) -> dict:
 
     # Collect all .h files belonging to the project
     own_files = {str(p.resolve()) for p in include_dir.glob("**/*.h")}
+    own_files.update(str(h.resolve()) for h in extra_headers)
 
     # First pass: build a mapping "anonymous struct location -> typedef name"
     typedef_map: dict[str, str] = {}
@@ -199,21 +203,25 @@ def _dedup(items: list) -> list:
     return resolve_idl_types(idl, mappings_path)
 
 
-def build_entry_point(headers_dir: Path) -> str:
+def build_entry_point(headers_dir: Path,
+                      extra_headers: tuple[Path, ...] = ()) -> str:
     lines = []
     for h in sorted(headers_dir.glob("**/*.h")):
         lines.append(f'#include "{h.resolve()}"')
+    for h in extra_headers:
+        lines.append(f'#include "{h.resolve()}"')
     return "\n".join(lines)
 
 
-def parse_all_headers(headers_dir: Path) -> dict:
-    entry_src = build_entry_point(headers_dir)
+def parse_all_headers(headers_dir: Path,
+                      extra_headers: tuple[Path, ...] = ()) -> dict:
+    entry_src = build_entry_point(headers_dir, extra_headers)
 
     with tempfile.NamedTemporaryFile(suffix=".h", mode="w", delete=False) as f:
         f.write(entry_src)
         tmp_path = f.name
 
     try:
-        return parse_meos(Path(tmp_path), headers_dir)
+        return parse_meos(Path(tmp_path), headers_dir, extra_headers)
     finally:
         os.unlink(tmp_path)

run.py

@@ -60,12 +60,39 @@ def _source_commit():
     return None
 
 
+def _public_pgtypes_headers() -> tuple[Path, ...]:
+    """MobilityDB's vendored PostgreSQL base-type headers, when they sit outside HEADERS_DIR.
+
+    MobilityDB keeps the PostgreSQL 18 base types in a `pgtypes/` library at the repo root and
+    installs `pgtypes.h` and the `pg_*.h` set into the include prefix, so they are as public as
+    `meos.h`. Reading an INSTALLED prefix therefore already covers them and this adds nothing;
+    reading the source tree points at `meos/include`, which does not contain them, and without
+    this the catalog silently loses the whole base-type surface (`interval_make`, the base I/O
+    functions, …) — the same ref then yields two different catalogs depending on the header root.
+    """
+    root = Path(os.environ.get("MDB_SRC_ROOT", "./_mobilitydb")) / "pgtypes"
+    if not root.is_dir():
+        return ()
+    seen = {p.name for p in HEADERS_DIR.glob("**/*.h")}
+    candidates = sorted(root.glob("pg_*.h")) + [root / "pgtypes.h"]
+    headers = [h for h in candidates
+               # pg_config*.h carry PostgreSQL's build configuration, not API; the
+               # install set leaves them behind and so does the catalog.
+               if h.is_file() and not h.name.startswith("pg_config")
+               and h.name not in seen]
+    return tuple(headers)
+
+
 def main():
     OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
 
     # 1. Parse C headers
     print(f"[1/4] Parsing {HEADERS_DIR}...", file=sys.stderr)
-    idl = parse_all_headers(HEADERS_DIR)
+    pgtypes = _public_pgtypes_headers()
+    if pgtypes:
+        print(f"      + {len(pgtypes)} vendored pgtypes headers from "
+              f"{pgtypes[0].parent}", file=sys.stderr)
+    idl = parse_all_headers(HEADERS_DIR, pgtypes)
 
     # 1b. Recover PG-vendored C types the preprocessor collapsed to int
     #     (bool / int64 / Timestamp(Tz) / H3Index) from the header text.