X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/07a2b2c0e743f36be03e746a7e265986db555d3e..d99fc30eb4a6e3f798dfb80ea010e313490683ae:/sdk/cwl/arvados_cwl/fsaccess.py diff --git a/sdk/cwl/arvados_cwl/fsaccess.py b/sdk/cwl/arvados_cwl/fsaccess.py index 5e3d685911..3744b4a93a 100644 --- a/sdk/cwl/arvados_cwl/fsaccess.py +++ b/sdk/cwl/arvados_cwl/fsaccess.py @@ -2,10 +2,16 @@ # # SPDX-License-Identifier: Apache-2.0 +from future import standard_library +standard_library.install_aliases() +from builtins import object +from builtins import str +from future.utils import viewvalues + import fnmatch import os import errno -import urlparse +import urllib.parse import re import logging import threading @@ -22,43 +28,54 @@ import arvados.collection import arvados.arvfile import arvados.errors +from googleapiclient.errors import HttpError + from schema_salad.ref_resolver import DefaultFetcher logger = logging.getLogger('arvados.cwl-runner') +pdh_size = re.compile(r'([0-9a-f]{32})\+(\d+)(\+\S+)*') + class CollectionCache(object): def __init__(self, api_client, keep_client, num_retries, cap=256*1024*1024, min_entries=2): self.api_client = api_client self.keep_client = keep_client + self.num_retries = num_retries self.collections = OrderedDict() self.lock = threading.Lock() self.total = 0 self.cap = cap self.min_entries = min_entries - def cap_cache(self): - if self.total > self.cap: - # ordered list iterates from oldest to newest - for pdh, v in self.collections.items(): - if self.total < self.cap or len(self.collections) < self.min_entries: - break - # cut it loose - logger.debug("Evicting collection reader %s from cache", pdh) - del self.collections[pdh] - self.total -= v[1] + def set_cap(self, cap): + self.cap = cap + + def cap_cache(self, required): + # ordered dict iterates from oldest to newest + for pdh, v in list(self.collections.items()): + available = self.cap - self.total + if available >= required or len(self.collections) < self.min_entries: + return + # cut it loose + logger.debug("Evicting collection reader %s from cache (cap %s total %s required %s)", pdh, self.cap, self.total, required) + del self.collections[pdh] + self.total -= v[1] def get(self, pdh): with self.lock: if pdh not in self.collections: + m = pdh_size.match(pdh) + if m: + self.cap_cache(int(m.group(2)) * 128) logger.debug("Creating collection reader for %s", pdh) cr = arvados.collection.CollectionReader(pdh, api_client=self.api_client, - keep_client=self.keep_client) + keep_client=self.keep_client, + num_retries=self.num_retries) sz = len(cr.manifest_text()) * 128 self.collections[pdh] = (cr, sz) self.total += sz - self.cap_cache() else: cr, sz = self.collections[pdh] # bump it to the back @@ -79,7 +96,7 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): p = sp[0] if p.startswith("keep:") and arvados.util.keep_locator_pattern.match(p[5:]): pdh = p[5:] - return (self.collection_cache.get(pdh), sp[1] if len(sp) == 2 else None) + return (self.collection_cache.get(pdh), urllib.parse.unquote(sp[1]) if len(sp) == 2 else None) else: return (None, path) @@ -110,7 +127,7 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): if collection is not None and not rest: return [pattern] patternsegments = rest.split("/") - return self._match(collection, patternsegments, "keep:" + collection.manifest_locator()) + return sorted(self._match(collection, patternsegments, "keep:" + collection.manifest_locator())) def open(self, fn, mode): collection, rest = self.get_collection(fn) @@ -120,7 +137,13 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): return super(CollectionFsAccess, self).open(self._abs(fn), mode) def exists(self, fn): - collection, rest = self.get_collection(fn) + try: + collection, rest = self.get_collection(fn) + except HttpError as err: + if err.resp.status == 404: + return False + else: + raise if collection is not None: if rest: return collection.exists(rest) @@ -129,6 +152,17 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): else: return super(CollectionFsAccess, self).exists(fn) + def size(self, fn): # type: (unicode) -> bool + collection, rest = self.get_collection(fn) + if collection is not None: + if rest: + arvfile = collection.find(rest) + if isinstance(arvfile, arvados.arvfile.ArvadosFile): + return arvfile.size() + raise IOError(errno.EINVAL, "Not a path to a file %s" % (fn)) + else: + return super(CollectionFsAccess, self).size(fn) + def isfile(self, fn): # type: (unicode) -> bool collection, rest = self.get_collection(fn) if collection is not None: @@ -160,7 +194,7 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): raise IOError(errno.ENOENT, "Directory '%s' in '%s' not found" % (rest, collection.portable_data_hash())) if not isinstance(dir, arvados.collection.RichCollectionBase): raise IOError(errno.ENOENT, "Path '%s' in '%s' is not a Directory" % (rest, collection.portable_data_hash())) - return [abspath(l, fn) for l in dir.keys()] + return [abspath(l, fn) for l in list(dir.keys())] else: return super(CollectionFsAccess, self).listdir(fn) @@ -179,16 +213,13 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess): return os.path.realpath(path) class CollectionFetcher(DefaultFetcher): - def __init__(self, cache, session, api_client=None, fs_access=None, num_retries=4, overrides=None): + 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 - self.overrides = overrides if overrides else {} def fetch_text(self, url): - if url in self.overrides: - return self.overrides[url] if url.startswith("keep:"): with self.fsaccess.open(url, "r") as f: return f.read() @@ -199,8 +230,6 @@ class CollectionFetcher(DefaultFetcher): return super(CollectionFetcher, self).fetch_text(url) def check_exists(self, url): - if url in self.overrides: - return True try: if url.startswith("http://arvados.org/cwl"): return True @@ -211,8 +240,8 @@ class CollectionFetcher(DefaultFetcher): return True except arvados.errors.NotFoundError: return False - except: - logger.exception("Got unexpected exception checking if file exists:") + except Exception: + logger.exception("Got unexpected exception checking if file exists") return False return super(CollectionFetcher, self).check_exists(url) @@ -220,11 +249,11 @@ class CollectionFetcher(DefaultFetcher): if not url: return base_url - urlsp = urlparse.urlsplit(url) + urlsp = urllib.parse.urlsplit(url) if urlsp.scheme or not base_url: return url - basesp = urlparse.urlsplit(base_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) @@ -245,30 +274,36 @@ class CollectionFetcher(DefaultFetcher): baseparts.pop() path = "/".join([pdh] + baseparts + urlparts) - return urlparse.urlunsplit((basesp.scheme, "", path, "", urlsp.fragment)) + 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 uri + return str(uri) if workflow_uuid_pattern.match(uri): - return "arvwf:%s#main" % (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 "keep:" + pt["components"].values()[0]["script_parameters"]["cwl:tool"] + return u"keep:" + viewvalues(pt["components"])[0]["script_parameters"]["cwl:tool"] p = uri.split("/") if arvados.util.keep_locator_pattern.match(p[0]): - return "keep:%s" % (uri) + return u"keep:%s" % (uri) if arvados.util.collection_uuid_pattern.match(p[0]): - return "keep:%s%s" % (api_client.collections(). + return u"keep:%s%s" % (api_client.collections(). get(uuid=p[0]).execute()["portable_data_hash"], uri[len(p[0]):])