bundle analysis: save custom compare sha to metadata (#327)

* bundle analysis: save custom compare sha to metadata

* fix caching related bug

shared/bundle_analysis/models.py

@@ -112,7 +112,6 @@ def __enter__(self):
 
     def __exit__(self, type, value, traceback):
         self.session.close()
-        return True
 
 
 def _use_modern_sqlalchemy_session_manager():

shared/bundle_analysis/parsers/v1.py

@@ -91,7 +91,7 @@ def reset(self):
         self.chunk_list = []
         self.module_list = []
 
-    def parse(self, path: str) -> int:
+    def parse(self, path: str) -> Tuple[int, str]:
         try:
             self.reset()
 
@@ -152,7 +152,7 @@ def parse(self, path: str) -> int:
                 self._create_associations()
 
                 assert self.session.bundle is not None
-                return self.session.id
+                return self.session.id, self.session.bundle.name
         except Exception as e:
             # Inject the plugin name to the Exception object so we have visibilitity on which plugin
             # is causing the trouble.

shared/bundle_analysis/parsers/v2.py

@@ -92,7 +92,7 @@ def reset(self):
         self.chunk_list = []
         self.module_list = []
 
-    def parse(self, path: str) -> int:
+    def parse(self, path: str) -> Tuple[int, str]:
         try:
             self.reset()
 
@@ -110,20 +110,6 @@ def parse(self, path: str) -> int:
                 for event in ijson.parse(f):
                     self._parse_event(event)
 
-                if self.asset_list:
-                    insert_asset = Asset.__table__.insert().values(self.asset_list)
-                    self.db_session.execute(insert_asset)
-
-                if self.chunk_list:
-                    insert_chunks = Chunk.__table__.insert().values(self.chunk_list)
-                    self.db_session.execute(insert_chunks)
-
-                if self.module_list:
-                    insert_modules = Module.__table__.insert().values(self.module_list)
-                    self.db_session.execute(insert_modules)
-
-                self.db_session.flush()
-
                 # Delete old session/asset/chunk/module with the same bundle name if applicable
                 old_session = (
                     self.db_session.query(Session)
@@ -144,6 +130,20 @@ def parse(self, path: str) -> int:
                     self.db_session.delete(old_session)
                     self.db_session.flush()
 
+                if self.asset_list:
+                    insert_asset = Asset.__table__.insert().values(self.asset_list)
+                    self.db_session.execute(insert_asset)
+
+                if self.chunk_list:
+                    insert_chunks = Chunk.__table__.insert().values(self.chunk_list)
+                    self.db_session.execute(insert_chunks)
+
+                if self.module_list:
+                    insert_modules = Module.__table__.insert().values(self.module_list)
+                    self.db_session.execute(insert_modules)
+
+                self.db_session.flush()
+
                 # save top level bundle stats info
                 self.session.info = json.dumps(self.info)
 
@@ -153,7 +153,7 @@ def parse(self, path: str) -> int:
                 self._create_associations()
 
                 assert self.session.bundle is not None
-                return self.session.id
+                return self.session.id, self.session.bundle.name
         except Exception as e:
             # Inject the plugin name to the Exception object so we have visibilitity on which plugin
             # is causing the trouble.

shared/bundle_analysis/report.py

@@ -228,16 +228,26 @@ def _setup(self, db_session: DbSession):
     def cleanup(self):
         os.unlink(self.db_path)
 
-    def ingest(self, path: str) -> int:
+    def ingest(self, path: str, compare_sha: Optional[str] = None) -> Tuple[int, str]:
         """
         Ingest the bundle stats JSON at the given file path.
         Returns session ID of ingested data.
         """
         with get_db_session(self.db_path) as session:
             parser = Parser(path, session).get_proper_parser()
-            session_id = parser.parse(path)
+            session_id, bundle_name = parser.parse(path)
+
+            # Save custom base commit SHA for doing comparisons if available
+            if compare_sha:
+                session.add(
+                    Metadata(
+                        key="compare_sha",
+                        value=compare_sha,
+                    )
+                )
+
             session.commit()
-            return session_id
+            return session_id, bundle_name
 
     def _associate_bundle_report_assets_by_name(
         self, curr_bundle_report: BundleReport, prev_bundle_report: BundleReport

tests/unit/bundle_analysis/test_bundle_analysis.py

@@ -12,6 +12,7 @@
     AssetType,
     Bundle,
     Chunk,
+    Metadata,
     MetadataKey,
     Module,
     Session,
@@ -58,8 +59,9 @@ def _table_rows_count(db_session: DbSession) -> Tuple[int]:
 def test_create_bundle_report():
     try:
         report = BundleAnalysisReport()
-        session_id = report.ingest(sample_bundle_stats_path)
+        session_id, bundle_name = report.ingest(sample_bundle_stats_path)
         assert session_id == 1
+        assert bundle_name == "sample"
 
         assert report.metadata() == {
             MetadataKey.SCHEMA_VERSION: SCHEMA_VERSION,
@@ -143,8 +145,9 @@ def test_create_bundle_report():
 def test_bundle_report_asset_filtering():
     try:
         report = BundleAnalysisReport()
-        session_id = report.ingest(sample_bundle_stats_path)
+        session_id, bundle_name = report.ingest(sample_bundle_stats_path)
         assert session_id == 1
+        assert bundle_name == "sample"
 
         assert report.metadata() == {
             MetadataKey.SCHEMA_VERSION: SCHEMA_VERSION,
@@ -418,8 +421,9 @@ def test_bundle_file_save_unknown_error():
 def test_create_bundle_report_v1():
     try:
         report = BundleAnalysisReport()
-        session_id = report.ingest(sample_bundle_stats_path_4)
+        session_id, bundle_name = report.ingest(sample_bundle_stats_path_4)
         assert session_id == 1
+        assert bundle_name == "sample"
 
         assert report.metadata() == {
             MetadataKey.SCHEMA_VERSION: SCHEMA_VERSION,
@@ -506,11 +510,17 @@ def test_create_bundle_report_v1():
 def test_bundle_is_cached():
     try:
         bundle_analysis_report = BundleAnalysisReport()
-        session_id = bundle_analysis_report.ingest(sample_bundle_stats_path)
+        session_id, bundle_name = bundle_analysis_report.ingest(
+            sample_bundle_stats_path
+        )
         assert session_id == 1
+        assert bundle_name == "sample"
 
-        session_id = bundle_analysis_report.ingest(sample_bundle_stats_path_5)
+        session_id, bundle_name = bundle_analysis_report.ingest(
+            sample_bundle_stats_path_5
+        )
         assert session_id == 2
+        assert bundle_name == "sample2"
 
         assert bundle_analysis_report.metadata() == {
             MetadataKey.SCHEMA_VERSION: SCHEMA_VERSION,
@@ -543,11 +553,17 @@ def test_bundle_deletion():
     try:
         bundle_analysis_report = BundleAnalysisReport()
         with get_db_session(bundle_analysis_report.db_path) as db_session:
-            session_id = bundle_analysis_report.ingest(sample_bundle_stats_path)
+            session_id, bundle_name = bundle_analysis_report.ingest(
+                sample_bundle_stats_path
+            )
             assert session_id == 1
+            assert bundle_name == "sample"
 
-            session_id = bundle_analysis_report.ingest(sample_bundle_stats_path_5)
+            session_id, bundle_name = bundle_analysis_report.ingest(
+                sample_bundle_stats_path_5
+            )
             assert session_id == 2
+            assert bundle_name == "sample2"
 
             assert _table_rows_count(db_session) == (2, 2, 10, 6, 62)
 
@@ -576,3 +592,27 @@ def test_bundle_deletion():
             assert len(res) == 0
     finally:
         bundle_analysis_report.cleanup()
+
+
+def test_create_bundle_report_without_and_with_compare_sha():
+    try:
+        report = BundleAnalysisReport()
+        with get_db_session(report.db_path) as db_session:
+            session_id, bundle_name = report.ingest(sample_bundle_stats_path)
+            assert session_id == 1
+            assert bundle_name == "sample"
+
+            res = db_session.query(Metadata).filter_by(key="compare_sha").all()
+            assert len(res) == 0
+
+            session_id, bundle_name = report.ingest(
+                sample_bundle_stats_path, "compare_sha_123"
+            )
+            assert session_id == 2
+            assert bundle_name == "sample"
+
+            res = db_session.query(Metadata).filter_by(key="compare_sha").all()
+            assert len(res) == 1
+            assert res[0].value == "compare_sha_123"
+    finally:
+        report.cleanup()