+
+class CollectionFetcher(DefaultFetcher):
+ def __init__(self, cache, session, api_client=None, fs_access=None, num_retries=4):
+ super(CollectionFetcher, self).__init__(cache, session)
+ self.api_client = api_client
+ self.fsaccess = fs_access
+ self.num_retries = num_retries
+
+ def fetch_text(self, url):
+ if url.startswith("keep:"):
+ with self.fsaccess.open(url, "r") as f:
+ return f.read()
+ if url.startswith("arvwf:"):
+ record = self.api_client.workflows().get(uuid=url[6:]).execute(num_retries=self.num_retries)
+ definition = record["definition"] + ('\nlabel: "%s"\n' % record["name"].replace('"', '\\"'))
+ return definition
+ return super(CollectionFetcher, self).fetch_text(url)
+
+ def check_exists(self, url):
+ try:
+ if url.startswith("http://arvados.org/cwl"):
+ return True
+ if url.startswith("keep:"):
+ return self.fsaccess.exists(url)
+ if url.startswith("arvwf:"):
+ if self.fetch_text(url):
+ return True
+ except arvados.errors.NotFoundError:
+ return False
+ except Exception:
+ logger.exception("Got unexpected exception checking if file exists")
+ return False
+ return super(CollectionFetcher, self).check_exists(url)
+
+ def urljoin(self, base_url, url):
+ if not url:
+ return base_url
+
+ urlsp = urllib.parse.urlsplit(url)
+ if urlsp.scheme or not base_url:
+ return url
+
+ basesp = urllib.parse.urlsplit(base_url)
+ if basesp.scheme in ("keep", "arvwf"):
+ if not basesp.path:
+ raise IOError(errno.EINVAL, "Invalid Keep locator", base_url)
+
+ baseparts = basesp.path.split("/")
+ urlparts = urlsp.path.split("/") if urlsp.path else []
+
+ locator = baseparts.pop(0)
+
+ if (basesp.scheme == "keep" and
+ (not arvados.util.keep_locator_pattern.match(locator)) and
+ (not arvados.util.collection_uuid_pattern.match(locator))):
+ raise IOError(errno.EINVAL, "Invalid Keep locator", base_url)
+
+ if urlsp.path.startswith("/"):
+ baseparts = []
+ urlparts.pop(0)
+
+ if baseparts and urlsp.path:
+ baseparts.pop()
+
+ path = "/".join([locator] + baseparts + urlparts)
+ return urllib.parse.urlunsplit((basesp.scheme, "", path, "", urlsp.fragment))
+
+ return super(CollectionFetcher, self).urljoin(base_url, url)
+
+ schemes = [u"file", u"http", u"https", u"mailto", u"keep", u"arvwf"]
+
+ def supported_schemes(self): # type: () -> List[Text]
+ return self.schemes
+
+
+workflow_uuid_pattern = re.compile(r'[a-z0-9]{5}-7fd4e-[a-z0-9]{15}')
+pipeline_template_uuid_pattern = re.compile(r'[a-z0-9]{5}-p5p6p-[a-z0-9]{15}')
+
+def collectionResolver(api_client, document_loader, uri, num_retries=4):
+ if uri.startswith("keep:") or uri.startswith("arvwf:"):
+ return str(uri)
+
+ if workflow_uuid_pattern.match(uri):
+ return u"arvwf:%s#main" % (uri)
+
+ if pipeline_template_uuid_pattern.match(uri):
+ pt = api_client.pipeline_templates().get(uuid=uri).execute(num_retries=num_retries)
+ return u"keep:" + viewvalues(pt["components"])[0]["script_parameters"]["cwl:tool"]
+
+ p = uri.split("/")
+ if arvados.util.keep_locator_pattern.match(p[0]):
+ return u"keep:%s" % (uri)
+
+ if arvados.util.collection_uuid_pattern.match(p[0]):
+ return u"keep:%s%s" % (api_client.collections().
+ get(uuid=p[0]).execute()["portable_data_hash"],
+ uri[len(p[0]):])
+
+ return cwltool.resolver.tool_resolver(document_loader, uri)