[Fix] Add engine field to coverage information to remove ambiguity

src/clusterfuzz/_internal/bot/tasks/utasks/corpus_pruning_task.py

@@ -750,9 +750,10 @@ def do_corpus_pruning(context, revision) -> CorpusPruningResult:
 
   # Store corpus stats into CoverageInformation entity.
   project_qualified_name = context.fuzz_target.project_qualified_name()
+  target_engine = context.fuzz_target.engine
   today = datetime.datetime.utcnow()
   coverage_info = data_types.CoverageInformation(
-      fuzzer=project_qualified_name, date=today)
+      fuzzer=project_qualified_name, date=today, engine=target_engine)
 
   quarantine_corpus_size = shell.get_directory_file_count(
       context.quarantine_corpus_path)
@@ -996,13 +997,19 @@ def _save_coverage_information(output):
     return
 
   cov_info = output.corpus_pruning_task_output.coverage_info
+  # Retrieve fuzz target engine, as it is not in the cov_info proto.
+  fuzz_target = uworker_io.entity_from_protobuf(
+      output.uworker_input.corpus_pruning_task_input.fuzz_target,
+      data_types.FuzzTarget)
+  target_engine = fuzz_target.engine
 
   # Use ndb.transaction with retries below to mitigate risk of a race condition.
   def _try_save_coverage_information():
     """Implements save_coverage_information function."""
     coverage_info = data_handler.get_coverage_information(
         cov_info.project_name,
         cov_info.timestamp.ToDatetime().date(),
+        engine=target_engine,
         create_if_needed=True)
 
     # Intentionally skip edge and function coverage values as those would come

src/clusterfuzz/_internal/datastore/data_handler.py

@@ -1822,15 +1822,18 @@ def get_all_job_type_names(project=None):
   return sorted([job.name for job in query])
 
 
-def get_coverage_information(fuzzer_name, date, create_if_needed=False):
+def get_coverage_information(fuzzer_name,
+                             date,
+                             engine=None,
+                             create_if_needed=False):
   """Get coverage information, or create if it doesn't exist."""
   coverage_info = ndb.Key(
       data_types.CoverageInformation,
       data_types.coverage_information_key(fuzzer_name, date)).get()
 
   if not coverage_info and create_if_needed:
     coverage_info = data_types.CoverageInformation(
-        fuzzer=fuzzer_name, date=date)
+        fuzzer=fuzzer_name, date=date, engine=engine)
 
   return coverage_info
 

src/clusterfuzz/_internal/datastore/data_types.py

@@ -1389,6 +1389,7 @@ class CoverageInformation(Model):
   """Coverage info."""
   date = ndb.DateProperty(auto_now_add=True)
   fuzzer = ndb.StringProperty()
+  engine = ndb.StringProperty()
 
   # Function coverage information.
   functions_covered = ndb.IntegerProperty()

src/clusterfuzz/_internal/metrics/fuzzer_stats.py

@@ -368,10 +368,12 @@ def get_coverage_info(self, fuzzer, date=None):
       return get_coverage_info(project, date)
 
     fuzz_target = data_handler.get_fuzz_target(fuzzer)
+    engine = None
     if fuzz_target:
       fuzzer = fuzz_target.project_qualified_name()
+      engine = fuzz_target.engine
 
-    return get_coverage_info(fuzzer, date)
+    return get_coverage_info(fuzzer, date, engine)
 
 
 class BaseCoverageField:
@@ -965,11 +967,17 @@ def build(self):
     return ' '.join(result)
 
 
-def get_coverage_info(fuzzer, date=None):
-  """Returns a CoverageInformation entity for a given fuzzer and date. If date
-  is not specified, returns the latest entity available."""
+def get_coverage_info(fuzzer, date=None, engine=None):
+  """Returns a CoverageInformation entity for a given fuzzer, engine and date.
+
+  If date is not specified, returns the latest entity available.
+  """
   query = data_types.CoverageInformation.query(
       data_types.CoverageInformation.fuzzer == fuzzer)
+
+  if engine:
+    query = query.filter(data_types.CoverageInformation.engine == engine)
+
   if date:
     # Return info for specific date.
     query = query.filter(data_types.CoverageInformation.date == date)

src/clusterfuzz/_internal/tests/appengine/handlers/fuzzer_stats_test.py

@@ -108,7 +108,9 @@ def setUp(self):
         fuzz_target_name='testFuzzer2_1_fuzzer', job='job', last_run=now).put()
 
     cov_info = data_types.CoverageInformation(
-        fuzzer='2_fuzzer', date=datetime.date(2016, 10, 19))
+        fuzzer='2_fuzzer',
+        engine='testFuzzer',
+        date=datetime.date(2016, 10, 19))
     cov_info.edges_covered = 11
     cov_info.edges_total = 30
     cov_info.functions_covered = 10
@@ -124,7 +126,9 @@ def setUp(self):
     cov_info.put()
 
     cov_info = data_types.CoverageInformation(
-        fuzzer='2_fuzzer', date=datetime.date(2016, 10, 21))
+        fuzzer='2_fuzzer',
+        engine='testFuzzer',
+        date=datetime.date(2016, 10, 21))
     cov_info.edges_covered = 15
     cov_info.edges_total = 30
     cov_info.functions_covered = 11
@@ -140,7 +144,9 @@ def setUp(self):
     cov_info.put()
 
     cov_info = data_types.CoverageInformation(
-        fuzzer='1_fuzzer', date=datetime.date(2016, 10, 20))
+        fuzzer='1_fuzzer',
+        engine='testFuzzer',
+        date=datetime.date(2016, 10, 20))
     cov_info.edges_covered = 17
     cov_info.edges_total = 38
     cov_info.functions_covered = 12

src/clusterfuzz/_internal/tests/core/bot/tasks/utasks/corpus_pruning_task_test.py

@@ -283,6 +283,8 @@ def test_prune(self):
                 None,
             'fuzzer':
                 'test_fuzzer',
+            'engine':
+                'libFuzzer',
             'html_report_url':
                 None,
             'quarantine_location':