Add option to force update of database records

comma/cli/__init__.py

@@ -92,7 +92,7 @@ class Session:
 
         if options.upstream:
             LOGGER.info("Begin monitoring upstream")
-            Upstream(self.config, self.database, repo).process_commits()
+            Upstream(self.config, self.database, repo).process_commits(options.force_update)
             LOGGER.info("Finishing monitoring upstream")
 
         if options.downstream:

comma/cli/parser.py

@@ -89,6 +89,11 @@ def get_run_parser():
         metavar="APPROXIDATE",
         help="Passed to underlying git commands. By default, the history is not limited",
     )
+    parser.add_argument(
+        "--force-update",
+        action="store_true",
+        help="Force update to existing upstream patch records",
+    )
 
     return parser
 

comma/upstream.py

@@ -22,13 +22,14 @@ class Upstream:
         self.database = database
         self.repo = repo
 
-    def process_commits(self):
+    def process_commits(self, force_update=False):
         """
         Generate patches for commits affecting tracked paths
         """
 
         paths = self.repo.get_tracked_paths(self.config.upstream.sections)
         added = 0
+        updated = 0
         total = 0
 
         # We use `--min-parents=1 --max-parents=1` to avoid both merges and graft commits.
@@ -42,14 +43,42 @@ class Upstream:
         ):
             total += 1
             with self.database.get_session() as session:
-                # See if commit ID is in database
-                if (
-                    session.query(PatchData.commitID)
-                    .filter_by(commitID=commit.hexsha)
-                    .one_or_none()
-                    is None
-                ):
+                # Query database for commit
+                if force_update:
+                    query = session.query(PatchData)
+                else:
+                    query = session.query(PatchData.commitID)
+                patch = query.filter_by(commitID=commit.hexsha).one_or_none()
+
+                # If commit is missing, add it
+                if patch is None:
                     session.add(PatchData.create(commit, paths))
                     added += 1
 
+                # If commit is present, optionally update
+                elif force_update:
+                    # Get a local patch object
+                    patch_data = PatchData.create(commit, paths)
+
+                    # Iterate through the columns
+                    record_updated = False
+                    for column in (
+                        col.name for col in patch_data.__table__.columns if not col.primary_key
+                    ):
+                        # Skip commit ID
+                        if column == "commitID":
+                            continue
+
+                        # If the new value is different, update it
+                        new_value = getattr(patch_data, column)
+                        if getattr(patch, column) != new_value:
+                            LOGGER.info("Updating %s for %s", column, commit.hexsha)
+                            setattr(patch, column, new_value)
+                            record_updated = True
+
+                    if record_updated:
+                        updated += 1
+
         LOGGER.info("%d of %d patches added to database.", added, total)
+        if force_update:
+            LOGGER.info("%d of %d patches updated in database.", updated, total)

comma/util/tracking.py

@@ -26,7 +26,7 @@ def get_filenames(commit: git.Commit) -> List[str]:
         return []
     diffs = commit.tree.diff(commit.parents[0])
     # Sometimes a path is in A and not B but we want all filenames.
-    return list(
+    return sorted(
         {diff.a_path for diff in diffs if diff.a_path is not None}
         | {diff.b_path for diff in diffs if diff.b_path is not None}
     )