fix database schema updater

src/flickrhistory/basicflickrhistorydownloader.py

@@ -21,6 +21,7 @@
 from .cache import Cache
 from .cacheupdaterthread import CacheUpdaterThread
 from .config import Config
+from .database import Session
 from .licensedownloader import LicenseDownloader
 from .photodownloaderthread import PhotoDownloaderThread
 from .photoupdaterthread import PhotoUpdaterThread
@@ -56,6 +57,10 @@ def __init__(self):
 
     def download(self):
         """Download all georeferenced flickr posts."""
+        # create a session to initialise the database
+        _ = Session()
+
+        # download an updated list of possible licenses
         LicenseDownloader(self._api_key_manager).update_licenses()
 
         for gap in self.gaps_in_download_history:

src/flickrhistory/database/databaseschemaupdater.py

@@ -20,22 +20,55 @@
 SCHEMA_UPDATES = {
     # 0 -> 1
     1: """
-        ALTER TABLE
-            photos
-        ADD COLUMN IF NOT EXISTS
-            geo_accuracy SMALLINT;
-
-        CREATE TABLE IF NOT EXISTS
-            licenses (
-                id INTEGER,
-                name TEXT,
-                url TEXT
-            );
-
-        ALTER TABLE
-            photos
-        ADD COLUMN IF NOT EXISTS
-            license INTEGER REFERENCES licenses(id);
+        CREATE TABLE licenses (
+            id integer NOT NULL,
+            name text,
+            url text
+        );
+
+        CREATE SEQUENCE licenses_id_seq
+            AS integer
+            START WITH 1
+            INCREMENT BY 1
+            NO MINVALUE
+            NO MAXVALUE
+            CACHE 1;
+
+        ALTER SEQUENCE licenses_id_seq OWNED BY licenses.id;
+
+        ALTER TABLE ONLY photos ADD COLUMN geo_accuracy SMALLINT, ADD COLUMN license_id INTEGER;
+
+        CREATE TABLE tag_photo_associations (
+            tag_tag text NOT NULL,
+            photo_id bigint NOT NULL
+        );
+
+        CREATE TABLE tags (
+            tag text NOT NULL
+        );
+
+        ALTER TABLE ONLY licenses ALTER COLUMN id SET DEFAULT nextval('licenses_id_seq'::regclass);
+
+        ALTER TABLE ONLY licenses
+            ADD CONSTRAINT licenses_pkey PRIMARY KEY (id);
+
+        ALTER TABLE ONLY tag_photo_associations
+            ADD CONSTRAINT tag_photo_associations_pkey PRIMARY KEY (tag_tag, photo_id);
+
+        ALTER TABLE ONLY tags
+            ADD CONSTRAINT tags_pkey PRIMARY KEY (tag);
+
+        CREATE INDEX ix_photos_license_id ON photos USING btree (license_id);
+
+        ALTER TABLE ONLY photos RENAME CONSTRAINT "FlickrUser" TO "User";
+
+        ALTER TABLE ONLY photos ADD CONSTRAINT photos_license_id_fkey FOREIGN KEY (license_id) REFERENCES licenses(id);
+
+        ALTER TABLE ONLY tag_photo_associations
+            ADD CONSTRAINT tag_photo_associations_photo_id_fkey FOREIGN KEY (photo_id) REFERENCES photos(id);
+
+        ALTER TABLE ONLY tag_photo_associations
+            ADD CONSTRAINT tag_photo_associations_tag_tag_fkey FOREIGN KEY (tag_tag) REFERENCES tags(tag);
     """,
 }
 
@@ -94,9 +127,10 @@ def update_to_latest(self):
                 file=sys.stderr,
                 flush=True,  # so that we don’t seem without work
             )
-            with self.engine.connect() as connection:
+            with self.engine.begin() as connection:
                 next_version = self.installed_version + 1
                 connection.execute(sqlalchemy.text(SCHEMA_UPDATES[next_version]))
+                connection.commit()
                 self.set_schema_version(next_version)
             installed_version = self.installed_version
 

src/flickrhistory/database/session.py

@@ -49,6 +49,7 @@ def __new__(cls, *args, **kwargs):
 
             if sqlalchemy.inspect(engine).has_table(Photo.__table__.name):  # data exist
                 DatabaseSchemaUpdater().update_to_latest()
+                Base.metadata.create_all(engine, checkfirst=True)
             else:
                 Base.metadata.create_all(engine)
                 DatabaseSchemaUpdater().set_schema_version(DatabaseSchemaUpdater.LATEST)