CFLite: Fix corpus pruning for git store. (#6434)

Fixes #6376.

5 files changed, 18 insertions, 11 deletions

diff --git a/infra/cifuzz/clusterfuzz_deployment.py b/infra/cifuzz/clusterfuzz_deployment.py
index 169c83fb3..fdc3738df 100644
--- a/infra/cifuzz/clusterfuzz_deployment.py
+++ b/infra/cifuzz/clusterfuzz_deployment.py
@@ -63,7 +63,7 @@ class BaseClusterFuzzDeployment:
     """Uploads crashes in |crashes_dir| to filestore."""
     raise NotImplementedError('Child class must implement method.')
 
-  def upload_corpus(self, target_name, corpus_dir):  # pylint: disable=no-self-use,unused-argument
+  def upload_corpus(self, target_name, corpus_dir, replace=False):  # pylint: disable=no-self-use,unused-argument
     """Uploads the corpus for |target_name| to filestore."""
     raise NotImplementedError('Child class must implement method.')
 
@@ -152,12 +152,12 @@ class ClusterFuzzLite(BaseClusterFuzzDeployment):
     """Returns the name of the crashes artifact."""
     return 'current'
 
-  def upload_corpus(self, target_name, corpus_dir):
+  def upload_corpus(self, target_name, corpus_dir, replace=False):
     """Upload the corpus produced by |target_name|."""
     logging.info('Uploading corpus in %s for %s.', corpus_dir, target_name)
     name = self._get_corpus_name(target_name)
     try:
-      self.filestore.upload_corpus(name, corpus_dir)
+      self.filestore.upload_corpus(name, corpus_dir, replace=replace)
       logging.info('Done uploading corpus.')
     except Exception as error:  # pylint: disable=broad-except
       logging.error('Failed to upload corpus for target: %s. Error: %s.',
@@ -272,7 +272,7 @@ class OSSFuzz(BaseClusterFuzzDeployment):
     """Noop Implementation of upload_build."""
     logging.info('Not uploading latest build because on OSS-Fuzz.')
 
-  def upload_corpus(self, target_name, corpus_dir):  # pylint: disable=no-self-use,unused-argument
+  def upload_corpus(self, target_name, corpus_dir, replace=False):  # pylint: disable=no-self-use,unused-argument
     """Noop Implementation of upload_corpus."""
     logging.info('Not uploading corpus because on OSS-Fuzz.')
 
@@ -323,7 +323,7 @@ class NoClusterFuzzDeployment(BaseClusterFuzzDeployment):
     logging.info('Not uploading latest build because no ClusterFuzz '
                  'deployment.')
 
-  def upload_corpus(self, target_name, corpus_dir):  # pylint: disable=no-self-use,unused-argument
+  def upload_corpus(self, target_name, corpus_dir, replace=False):  # pylint: disable=no-self-use,unused-argument
     """Noop Implementation of upload_corpus."""
     logging.info('Not uploading corpus because no ClusterFuzz deployment.')
 
diff --git a/infra/cifuzz/filestore/__init__.py b/infra/cifuzz/filestore/__init__.py
index 49999d081..d112f7b8c 100644
--- a/infra/cifuzz/filestore/__init__.py
+++ b/infra/cifuzz/filestore/__init__.py
@@ -29,7 +29,7 @@ class BaseFilestore:
     """Uploads the crashes at |directory| to |name|."""
     raise NotImplementedError('Child class must implement method.')
 
-  def upload_corpus(self, name, directory):
+  def upload_corpus(self, name, directory, replace=False):
     """Uploads the corpus at |directory| to |name|."""
     raise NotImplementedError('Child class must implement method.')
 
diff --git a/infra/cifuzz/filestore/git/__init__.py b/infra/cifuzz/filestore/git/__init__.py
index 742aebdb7..5414003da 100644
--- a/infra/cifuzz/filestore/git/__init__.py
+++ b/infra/cifuzz/filestore/git/__init__.py
@@ -112,10 +112,13 @@ class GitFilestore(filestore.BaseFilestore):
     """Uploads the crashes at |directory| to |name|."""
     return self._ci_filestore.upload_crashes(name, directory)
 
-  def upload_corpus(self, name, directory):
+  def upload_corpus(self, name, directory, replace=False):
     """Uploads the corpus at |directory| to |name|."""
-    self._upload_to_git('Corpus upload', self.config.git_store_branch,
-                        os.path.join(_CORPUS_DIR, name), directory)
+    self._upload_to_git('Corpus upload',
+                        self.config.git_store_branch,
+                        os.path.join(_CORPUS_DIR, name),
+                        directory,
+                        replace=replace)
 
   def upload_build(self, name, directory):
     """Uploads the build at |directory| to |name|."""
diff --git a/infra/cifuzz/filestore/github_actions/__init__.py b/infra/cifuzz/filestore/github_actions/__init__.py
index 6d04ccfbd..8b9b5986e 100644
--- a/infra/cifuzz/filestore/github_actions/__init__.py
+++ b/infra/cifuzz/filestore/github_actions/__init__.py
@@ -75,8 +75,11 @@ class GithubActionsFilestore(filestore.BaseFilestore):
     """Uploads the crashes at |directory| to |name|."""
     return _raw_upload_directory(self.CRASHES_PREFIX + name, directory)
 
-  def upload_corpus(self, name, directory):
+  def upload_corpus(self, name, directory, replace=False):
     """Uploads the corpus at |directory| to |name|."""
+    # Not applicable as the the entire corpus is uploaded under a single
+    # artifact name.
+    del replace
     return self._upload_directory(self.CORPUS_PREFIX + name, directory)
 
   def upload_build(self, name, directory):
diff --git a/infra/cifuzz/run_fuzzers.py b/infra/cifuzz/run_fuzzers.py
index 347615772..67c4c66fd 100644
--- a/infra/cifuzz/run_fuzzers.py
+++ b/infra/cifuzz/run_fuzzers.py
@@ -178,7 +178,8 @@ class PruneTargetRunner(BaseFuzzTargetRunner):
     result = fuzz_target_obj.prune()
     logging.debug('Corpus path contents: %s.', os.listdir(result.corpus_path))
     self.clusterfuzz_deployment.upload_corpus(fuzz_target_obj.target_name,
-                                              result.corpus_path)
+                                              result.corpus_path,
+                                              replace=True)
     return result
 
   def cleanup_after_fuzz_target_run(self, fuzz_target_obj):  # pylint: disable=no-self-use