create master people db

src/ossql/people_to_sqlite.py

@@ -19,11 +19,16 @@ def load_people_yaml(dir_path: pathlib.Path):
     # ensure all tables exist
     db.create_tables([Person, PersonLink, PersonSource, PersonRole, PersonOffice])
 
+    # output stats
+    unused = set()
+    created = 0
+
     files = list(dir_path.glob("*.yml"))
-    print(f"preparing to load {len(files)} files")
+    print(f"preparing to load {len(files)} files from {dir_path}")
     for file in files:
         pdata = yaml.safe_load(file.read_text())
 
+        created += 1
         person = Person.create(
             id=pdata.pop("id"),
             name=pdata.pop("name"),
@@ -32,23 +37,23 @@ def load_people_yaml(dir_path: pathlib.Path):
             birth_date=pdata.pop("birth_date", None),
             gender=pdata.pop("gender"),
             email=pdata.pop("email", ""),
-            image=pdata.pop("image"),
+            image=pdata.pop("image", ""),
             party=pdata.pop("party"),
             extras=pdata.pop("extras", {}),
         )
-        to_links(person, pdata.pop("links"), PersonLink)
+        to_links(person, pdata.pop("links", []), PersonLink)
-        to_links(person, pdata.pop("sources"), PersonSource)
+        to_links(person, pdata.pop("sources", []), PersonSource)
 
         for role in pdata.pop("roles"):
             PersonRole.create(
                 person=person,
                 jurisdiction=role.pop("jurisdiction"),
-                district=role.pop("district"),
+                district=role.pop("district", ""),
                 type=role.pop('type'),
                 start_date=role.pop('start_date', None),
                 end_date=role.pop('end_date', None),
             )
-        for office in pdata.pop("offices"):
+        for office in pdata.pop("offices", []):
             PersonOffice.create(
                 person=person,
                 classification=office.pop("classification"),
@@ -59,10 +64,25 @@ def load_people_yaml(dir_path: pathlib.Path):
 
         # currently not using other_names, other_identifiers
         if pdata.keys():
-            print(pdata.keys(), "left unused")
+            unused.update(pdata.keys())
+
+    if unused:
+        print(unused, "left unused")
+
+    return created
 
 
 
 if __name__ == "__main__":
     path = pathlib.Path(sys.argv[1])
-    load_people_yaml(path)
+    n_people = 0
+    if path.name == "data":
+        # load all states
+        states = list(path.glob("??"))
+        for state in sorted(states):
+            n_people += load_people_yaml(state / "legislature")
+    else:
+        # exact path, one state
+        n_people += load_people_yaml(path)
+
+    print(f"Created {n_people} people")

src/ossql/schemas/common.py

@@ -2,7 +2,7 @@ from peewee import SqliteDatabase, Model
 from playhouse.sqlite_ext import SqliteExtDatabase
 
 db = SqliteExtDatabase('openstates.db', pragmas=(
-    ('cache_size', -1024 * 64),  # 64MB page-cache.
+    ('cache_size', 1024 * 64),  # 64MB page-cache.
     ('journal_mode', 'wal'),  # Use WAL-mode (you should always use this!).
     ('foreign_keys', 1))
 )