def _key(
git_url: str, commit_hash: str, specified_analyzer: SpecifiedAnalyzer
) -> str:
analyzer_name = specified_analyzer.versioned_analyzer.name
version = specified_analyzer.versioned_analyzer.version
repo_id = url_to_repo_id(git_url)
key = f"{analyzer_name}/{version}/{repo_id}/{commit_hash}/container.log"
return key.replace("/", "___")
def container_log_key(git_url: str, commit_hash: str, image_id: str) -> str:
"""
Returns key that docker container log is stored with
"""
analyzer = VersionedAnalyzer.from_image_id(image_id)
repo_id = url_to_repo_id(git_url)
return (
f"{analyzer.name}/{analyzer.version}/{repo_id}/{commit_hash}/container.log"
)
def analysis_key(
self,
git_url: str,
commit_hash: str,
specified_analyzer: SpecifiedAnalyzer,
output_type: Optional[AnalyzerOutputType] = None,
) -> str:
"""
Key analysis report was uploaded with
Args:
git_url: Url of repo analyzed
commit_hash: hash analyzed
specified_analyzer: unique identifier of analysis container w/ parameters
output_type: type of output we want to retrieve from SPECIFIED_ANALYZER if None then
we lookup the output type of the analyzer in the registry
Returns:
key of report in S3 bucket
"""
manifest = self._registry_data.manifest_for(
specified_analyzer.versioned_analyzer
)
analyzer_type = manifest.analyzer_type
repo_id = url_to_repo_id(git_url)
if output_type is None:
output_type = manifest.output_type
extension = self._get_analyzer_output_extension(output_type)
if len(specified_analyzer.parameters) == 0:
analyzer_part = f"{manifest.analyzer_name}/{manifest.version}"
else:
param_part = ""
for param_name in sorted(specified_analyzer.parameters):
param_part += (
f"{param_name}:{specified_analyzer.parameters[param_name]}"
)
analyzer_part = f"{manifest.analyzer_name}/{manifest.version}/{param_part}"
if analyzer_type == AnalyzerType.git:
# for now, we also include the commit_hash to make other parts of the pipelie
# treat git analyzers similar to commit analyzers.
# TODO: figure out a good way of actually deisgning for this,fingerprinting
# the repo and keeping our determinism guarantee
target_part = f"{repo_id}/{commit_hash}/output{extension}"
elif analyzer_type == AnalyzerType.commit:
target_part = f"{repo_id}/{commit_hash}/output{extension}"
else:
raise UnsupportedAnalyzerType(analyzer_type)
key = f"{analyzer_part}/{target_part}"
return key
def get_local_git_origin_and_commit(dir: str) -> Tuple[str, str]:
try:
repo = (subprocess.check_output(
["git", "config", "--get", "remote.origin.url"],
cwd=dir).strip().decode("utf-8"))
commit = (subprocess.check_output(
["git", "show", '--format="%H"', "--no-patch"],
cwd=dir).strip().decode("utf-8"))
return repo, commit.replace('"', "")
except subprocess.CalledProcessError as ex:
logger.error(
f"failed to determine source git repo or commit for {dir}")
# use same util function, but treat local relative dir path as repo
hash_of_dir = url_to_repo_id(dir)
logger.debug(f"Using {hash_of_dir}, {hash_of_dir} as repo, commit")
return hash_of_dir, hash_of_dir
def _key(
git_url: str, commit_hash: str, specified_analyzer: SpecifiedAnalyzer
) -> str:
repo_id = url_to_repo_id(git_url)
analyzer_name = specified_analyzer.versioned_analyzer.name
version = specified_analyzer.versioned_analyzer.version
if len(specified_analyzer.parameters) == 0:
analyzer_part = f"{analyzer_name}/{version}"
else:
param_part = ""
for param_name in sorted(specified_analyzer.parameters):
param_part += (
f"{param_name}:{specified_analyzer.parameters[param_name]}"
)
analyzer_part = f"{analyzer_name}/{version}/{param_part}"
target_part = f"{repo_id}/{commit_hash}/output.tar.gz"
key = f"{analyzer_part}/{target_part}"
return key.replace("/", "___")