def build_cache_db(conf, versions_to_include):
s = scanner.Scanner(conf)
notes = s.get_notes_by_version()
# Default to including all versions returned by the scanner.
if not versions_to_include:
versions_to_include = list(notes.keys())
# Build a cache data structure including the file contents as well
# as the basic data returned by the scanner.
file_contents = {}
for version in versions_to_include:
for filename, sha in notes[version]:
body = s.get_file_at_commit(filename, sha)
# We want to save the contents of the file, which is YAML,
# inside another YAML file. That looks terribly ugly with
# all of the escapes needed to format it properly as
# embedded YAML, so parse the input and convert it to a
# data structure that can be serialized cleanly.
y = yaml.safe_load(body)
file_contents[filename] = y
cache = {
'notes': [
{'version': k, 'files': v}
for k, v in notes.items()
],
'file-contents': file_contents,
}
return cache
def _load_data(self):
cache_file_exists = os.path.exists(self._cache_filename)
if self._ignore_cache and cache_file_exists:
LOG.debug('ignoring cache file %s', self._cache_filename)
if (not self._ignore_cache) and cache_file_exists:
LOG.debug('loading cache file %s', self._cache_filename)
with open(self._cache_filename, 'r') as f:
self._cache = yaml.safe_load(f.read())
# Save the cached scanner output to the same attribute
# it would be in if we had loaded it "live". This
# simplifies some of the logic in the other methods.
self._scanner_output = collections.OrderedDict(
(n['version'], n['files']) for n in self._cache['notes'])
else:
self._scanner = scanner.Scanner(self._config)
self._scanner_output = self._scanner.get_notes_by_version()