Merge pull request 'Fix calculation of the tree data' (#5) from fix_fake into main

Reviewed-on: https://gitea.opensuse.org/importers/git-importer/pulls/5

lib/db.py

@@ -206,6 +206,10 @@ class DB:
             "ALTER TABLE linked_revs ADD COLUMN considered BOOLEAN DEFAULT FALSE",
             "UPDATE scheme SET version=18",
         )
+        schemes[19] = (
+            "CREATE INDEX ON files(revision_id);",
+            "UPDATE scheme SET version=19",
+        )
         schema_version = self.schema_version()
         if (schema_version + 1) not in schemes:
             return

lib/db_revision.py

@@ -200,13 +200,12 @@ class DBRevision:
         return self._files
 
     @staticmethod
-    def requests_to_fetch(db, project, package):
+    def requests_to_fetch(db):
         with db.cursor() as cur:
             cur.execute(
                 """SELECT request_number FROM revisions revs LEFT JOIN requests
                 reqs ON reqs.number=revs.request_number WHERE reqs.id is null AND
-                revs.request_number IS NOT NULL and project=%s AND package=%s;""",
+                revs.request_number IS NOT NULL""",
-                (project, package),
             )
             return [row[0] for row in cur.fetchall()]
 

lib/importer.py

@@ -167,18 +167,6 @@ class Importer:
                     (rev.dbid, linked_rev.dbid),
                 )
 
-    def calculate_file_hashes(self, db):
-        with db.cursor() as cur:
-            cur.execute(
-                "SELECT * from revisions where files_hash IS NULL AND broken is FALSE"
-            )
-            for row in cur.fetchall():
-                rev = DBRevision(row)
-                md5 = rev.calculate_files_hash(db)
-                cur.execute(
-                    "UPDATE revisions SET files_hash=%s WHERE id=%s", (md5, rev.dbid)
-                )
-
     def fetch_all_linked_packages(self, db, project, package):
         with db.cursor() as cur:
             cur.execute(
@@ -193,12 +181,10 @@ class Importer:
     def find_fake_revisions(self, db):
         with db.cursor() as cur:
             cur.execute(
-                """SELECT * from revisions WHERE
+                "SELECT * from revisions WHERE id in (SELECT linked_id from linked_revs WHERE considered=FALSE)"
-                            id in (SELECT revision_id from linked_revs WHERE considered=FALSE) AND
-                            id not in (SELECT revision_id FROM fake_revs) ORDER by project,package,rev"""
             )
             for row in cur.fetchall():
-                DBRevision(row)
+                self._find_fake_revision(db, DBRevision(row))
 
     def _find_fake_revision(self, db, rev):
         prev = rev.previous_commit(db)
@@ -211,7 +197,7 @@ class Importer:
             return
         with db.cursor() as cur:
             cur.execute(
-                """SELECT * from revisions where id in 
+                """SELECT * FROM revisions WHERE id IN
                             (SELECT revision_id from linked_revs WHERE linked_id=%s) 
                             AND commit_time <= %s ORDER BY commit_time""",
                 (prev.dbid, rev.commit_time),
@@ -267,6 +253,13 @@ class Importer:
                 (rev.dbid, linked.dbid),
             )
 
+    def revisions_without_files(self, db):
+        with db.cursor() as cur:
+            cur.execute(
+                "SELECT * FROM revisions WHERE broken=FALSE AND expanded_srcmd5 IS NULL"
+            )
+            return [DBRevision(row) for row in cur.fetchall()]
+
     def import_into_db(self):
         db = DB()
         for project, _, api_url in self.projects:
@@ -275,7 +268,6 @@ class Importer:
             self.fetch_all_linked_packages(db, project, self.package)
             # all remaining, no filtering here
             self.find_linked_revs(db)
-            self.find_fake_revisions(db)
 
         missing_users = User.missing_users(db)
         for userid in missing_users:
@@ -283,10 +275,8 @@ class Importer:
             if missing_user:
                 missing_user.import_into_db(db)
 
-            for rev in DBRevision.all_revisions(db, project, self.package):
+        self.find_fake_revisions(db)
-                # TODO move into SELECT
+        for rev in self.revisions_without_files(db):
-                if rev.broken or rev.expanded_srcmd5:
-                    continue
             with db.cursor() as cur:
                 cur.execute(
                     """SELECT unexpanded_srcmd5 from revisions WHERE
@@ -297,17 +287,22 @@ class Importer:
             if linked_rev:
                 linked_rev = linked_rev[0]
             list = self.obs.list(
-                    project, self.package, rev.unexpanded_srcmd5, linked_rev
+                rev.project, rev.package, rev.unexpanded_srcmd5, linked_rev
             )
             if list:
                 rev.import_dir_list(db, list)
+                md5 = rev.calculate_files_hash(db)
+                with db.cursor() as cur:
+                    cur.execute(
+                        "UPDATE revisions SET files_hash=%s WHERE id=%s",
+                        (md5, rev.dbid),
+                    )
             else:
                 rev.set_broken(db)
 
-            for number in DBRevision.requests_to_fetch(db, project, self.package):
+        for number in DBRevision.requests_to_fetch(db):
             self.obs.request(number).import_into_db(db)
 
-        self.calculate_file_hashes(db)
         db.conn.commit()
         TreeBuilder(db).build(self.package)
 

lib/tree_builder.py

@@ -38,4 +38,4 @@ class TreeBuilder:
                     if rev2.commit_time > rev.commit_time:
                         continue
                     if rev2.files_hash == rev.files_hash:
-                        print("    ", rev2)
+                        print("    ", rev2, rev2.files_hash)