def update_license_and_kind(self):
metadata = json.loads(self.library.metadata)
kind = 'element'
bower_json = None
default_branch = metadata.get('default_branch', 'master')
if not self.scope.startswith('@'):
response = urlfetch.fetch(util.content_url(self.owner, self.repo, default_branch, 'bower.json'), validate_certificate=True)
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("Could not parse master/bower.json", ErrorCodes.Library_parse_bower)
elif response.status_code == 404:
bower_json = None
else:
return self.retry('error fetching master/bower.json (%d)' % response.status_code)
if bower_json is not None and 'element-collection' in bower_json.get('keywords', []):
kind = 'collection'
if self.library.kind != kind:
self.library.kind = kind
self.library_dirty = True
spdx_identifier = None
github_license = metadata.get('license')
# GitHub may now return as a license object instead.
if isinstance(github_license, dict):
github_license = github_license.get('spdx_id', 'MISSING')
if github_license is not None:
spdx_identifier = licenses.validate_spdx(github_license)
if spdx_identifier is None and bower_json is not None:
license_name = bower_json.get('license')
if license_name is not None:
spdx_identifier = licenses.validate_spdx(license_name)
if spdx_identifier is None and self.scope.startswith('@'):
registry_metadata = json.loads(self.library.registry_metadata)
spdx_identifier = licenses.validate_spdx(registry_metadata.get('license', ''))
if self.library.spdx_identifier != spdx_identifier:
self.library.spdx_identifier = spdx_identifier
self.library_dirty = True
if self.library.spdx_identifier is None:
if self.scope.startswith('@'):
return self.error('Could not detect an OSI approved license on GitHub or in package info', ErrorCodes.Library_license)
return self.error('Could not detect an OSI approved license on GitHub or in %s/bower.json' % default_branch, ErrorCodes.Library_license)
def update_license_and_kind(self):
metadata = json.loads(self.library.metadata)
kind = 'element'
bower_json = None
default_branch = metadata.get('default_branch', 'master')
if not self.scope.startswith('@'):
response = urlfetch.fetch(util.content_url(self.owner, self.repo, default_branch, 'bower.json'), validate_certificate=True)
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("Could not parse master/bower.json", ErrorCodes.Library_parse_bower)
elif response.status_code == 404:
bower_json = None
else:
return self.retry('error fetching master/bower.json (%d)' % response.status_code)
if bower_json is not None and 'element-collection' in bower_json.get('keywords', []):
kind = 'collection'
if self.library.kind != kind:
self.library.kind = kind
self.library_dirty = True
spdx_identifier = None
github_license = metadata.get('license')
# GitHub may now return as a license object instead.
if isinstance(github_license, dict):
github_license = github_license.get('spdx_id', 'MISSING')
if github_license is not None:
spdx_identifier = licenses.validate_spdx(github_license)
if spdx_identifier is None and bower_json is not None:
license_name = bower_json.get('license')
if license_name is not None:
spdx_identifier = licenses.validate_spdx(license_name)
if spdx_identifier is None and self.scope.startswith('@'):
registry_metadata = json.loads(self.library.registry_metadata)
spdx_identifier = licenses.validate_spdx(registry_metadata.get('license', ''))
if self.library.spdx_identifier != spdx_identifier:
self.library.spdx_identifier = spdx_identifier
self.library_dirty = True
if self.library.spdx_identifier is None:
if self.scope.startswith('@'):
return self.error('Could not detect an OSI approved license on GitHub or in package info', ErrorCodes.Library_license)
return self.error('Could not detect an OSI approved license on GitHub or in %s/bower.json' % default_branch, ErrorCodes.Library_license)
def update_bower(self):
response = urlfetch.fetch(util.content_url(self.owner, self.repo, self.sha, 'bower.json'), validate_certificate=True)
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("could not parse bower.json")
Content(parent=self.version_key, id='bower', content=response.content,
status=Status.ready, etag=response.headers.get('ETag', None)).put()
return bower_json
elif response.status_code == 404:
return self.error("missing bower.json")
else:
return self.retry('could not access bower.json (%d)' % response.status_code)
def update_bower(self):
response = urlfetch.fetch(util.content_url(self.owner, self.repo, self.sha, 'bower.json'), validate_certificate=True)
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("could not parse bower.json", ErrorCodes.Version_parse_bower)
Content(parent=self.version_key, id='bower', json=bower_json,
status=Status.ready, etag=response.headers.get('ETag', None)).put()
return bower_json
elif response.status_code == 404:
return self.error("missing bower.json", ErrorCodes.Version_missing_bower)
else:
return self.retry('could not access bower.json (%d)' % response.status_code)
def update_license_and_kind(self):
metadata = json.loads(self.library.metadata)
default_branch = metadata.get('default_branch', 'master')
response = urlfetch.fetch(util.content_url(self.owner, self.repo,
default_branch,
'bower.json'),
validate_certificate=True)
bower_json = None
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("Could not parse master/bower.json")
elif response.status_code == 404:
bower_json = None
else:
return self.retry('error fetching master/bower.json' %
response.status_code)
kind = 'element'
if bower_json is not None and 'element-collection' in bower_json.get(
'keywords', []):
kind = 'collection'
if self.library.kind != kind:
self.library.kind = kind
self.library_dirty = True
spdx_identifier = None
github_license = metadata.get('license')
if github_license is not None:
spdx_identifier = licenses.validate_spdx(
github_license.get('key', 'MISSING'))
if spdx_identifier is None and bower_json is not None:
license_name = bower_json.get('license')
if license_name is not None:
spdx_identifier = licenses.validate_spdx(license_name)
if self.library.spdx_identifier != spdx_identifier:
self.library.spdx_identifier = spdx_identifier
self.library_dirty = True
if self.library.spdx_identifier is None:
return self.error(
'Could not detect an OSI approved license on GitHub or in %s/bower.json'
% default_branch)
def update_license_and_kind(self):
metadata = json.loads(self.library.metadata)
default_branch = metadata.get('default_branch', 'master')
response = urlfetch.fetch(util.content_url(self.owner, self.repo, default_branch, 'bower.json'), validate_certificate=True)
bower_json = None
if response.status_code == 200:
try:
bower_json = json.loads(response.content)
except ValueError:
return self.error("Could not parse master/bower.json")
elif response.status_code == 404:
bower_json = None
else:
return self.retry('error fetching master/bower.json' % response.status_code)
kind = 'element'
if bower_json is not None and 'element-collection' in bower_json.get('keywords', []):
kind = 'collection'
if self.library.kind != kind:
self.library.kind = kind
self.library_dirty = True
spdx_identifier = None
github_license = metadata.get('license')
if github_license is not None:
spdx_identifier = licenses.validate_spdx(github_license.get('key', 'MISSING'))
if spdx_identifier is None and bower_json is not None:
license_name = bower_json.get('license')
if license_name is not None:
spdx_identifier = licenses.validate_spdx(license_name)
if self.library.spdx_identifier != spdx_identifier:
self.library.spdx_identifier = spdx_identifier
self.library_dirty = True
if self.library.spdx_identifier is None:
return self.error('Could not detect an OSI approved license on GitHub or in %s/bower.json' % default_branch)
def get(self, owner, repo, version):
logging.info('ingesting version %s/%s/%s', owner, repo, version)
github = quota.GitHub()
if not github.reserve(1):
self.response.set_status(500)
return
key = ndb.Key(Library, '%s/%s' % (owner, repo), Version, version)
response = urlfetch.fetch(util.content_url(owner, repo, version, 'README.md'))
readme = response.content
try:
content = Content(parent=key, id='readme', content=readme)
content.etag = response.headers.get('ETag', None)
content.put()
except db.BadValueError:
ver = key.get()
ver.error = "Could not store README.md as a utf-8 string"
ver.put()
self.response.set_status(200)
return
response = github.markdown(readme)
content = Content(parent=key, id='readme.html', content=response.content)
content.put()
response = urlfetch.fetch(util.content_url(owner, repo, version, 'bower.json'))
try:
json.loads(response.content)
except ValueError:
ver = key.get()
ver.error = "This version has a missing or broken bower.json"
ver.put()
self.response.set_status(200)
return
content = Content(parent=key, id='bower', content=response.content)
content.etag = response.headers.get('ETag', None)
content.put()
versions = Library.versions_for_key(key.parent())
if versions[-1] == version:
library = key.parent().get()
if library.kind == "collection":
task_url = util.ingest_dependencies_task(owner, repo, version)
util.new_task(task_url)
bower = json.loads(response.content)
metadata = json.loads(library.metadata)
logging.info('adding search index for %s', version)
description = bower.get("description", metadata.get("description", ""))
document = search.Document(doc_id='%s/%s' % (owner, repo), fields=[
search.AtomField(name='full_name', value=metadata['full_name']),
search.TextField(name='owner', value=owner),
search.TextField(name='repo', value=repo),
search.TextField(name='version', value=version),
search.TextField(name='repoparts', value=' '.join(repo.split('-'))),
search.TextField(name='description', value=description),
search.TextField(name='keywords', value=' '.join(bower.get('keywords', []))),
search.NumberField(name='stars', value=metadata.get('stargazers_count')),
search.NumberField(name='subscribers', value=metadata.get('subscribers_count')),
search.NumberField(name='forks', value=metadata.get('forks')),
search.NumberField(name='contributors', value=library.contributor_count),
search.DateField(name='updated_at', value=datetime.datetime.strptime(metadata.get('updated_at'), TIME_FORMAT))
])
index = search.Index('repo')
index.put(document)
self.response.set_status(200)
