Merge branch 'main' into 18842-arv-mount-disk-config

[arvados.git] / sdk / cwl / arvados_cwl / fsaccess.py

diff --git a/sdk/cwl/arvados_cwl/fsaccess.py b/sdk/cwl/arvados_cwl/fsaccess.py
index d5866e72853181a0fda8412cfd24f6687c27417a..5c09e671fa21eac1952c417e10580d332e3612be 100644 (file)
--- a/sdk/cwl/arvados_cwl/fsaccess.py
+++ b/sdk/cwl/arvados_cwl/fsaccess.py
@@ -70,9 +70,12 @@ class CollectionCache(object):
                 if m:
                     self.cap_cache(int(m.group(2)) * 128)
                 logger.debug("Creating collection reader for %s", locator)
-                cr = arvados.collection.CollectionReader(locator, api_client=self.api_client,
-                                                         keep_client=self.keep_client,
-                                                         num_retries=self.num_retries)
+                try:
+                    cr = arvados.collection.CollectionReader(locator, api_client=self.api_client,
+                                                             keep_client=self.keep_client,
+                                                             num_retries=self.num_retries)
+                except arvados.errors.ApiError as ap:
+                    raise IOError(errno.ENOENT, "Could not access collection '%s': %s" % (locator, str(ap._get_reason())))
                 sz = len(cr.manifest_text()) * 128
                 self.collections[locator] = (cr, sz)
                 self.total += sz
@@ -97,7 +100,8 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
         if p.startswith("keep:") and (arvados.util.keep_locator_pattern.match(p[5:]) or
                                       arvados.util.collection_uuid_pattern.match(p[5:])):
             locator = p[5:]
-            return (self.collection_cache.get(locator), urllib.parse.unquote(sp[1]) if len(sp) == 2 else None)
+            rest = os.path.normpath(urllib.parse.unquote(sp[1])) if len(sp) == 2 else None
+            return (self.collection_cache.get(locator), rest)
         else:
             return (None, path)
 
@@ -125,15 +129,15 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
 
     def glob(self, pattern):
         collection, rest = self.get_collection(pattern)
-        if collection is not None and not rest:
+        if collection is not None and rest in (None, "", "."):
             return [pattern]
         patternsegments = rest.split("/")
         return sorted(self._match(collection, patternsegments, "keep:" + collection.manifest_locator()))
 
-    def open(self, fn, mode):
+    def open(self, fn, mode, encoding=None):
         collection, rest = self.get_collection(fn)
         if collection is not None:
-            return collection.open(rest, mode)
+            return collection.open(rest, mode, encoding=encoding)
         else:
             return super(CollectionFsAccess, self).open(self._abs(fn), mode)
 
@@ -145,6 +149,11 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
                 return False
             else:
                 raise
+        except IOError as err:
+            if err.errno == errno.ENOENT:
+                return False
+            else:
+                raise
         if collection is not None:
             if rest:
                 return collection.exists(rest)
@@ -220,24 +229,26 @@ class CollectionFetcher(DefaultFetcher):
         self.fsaccess = fs_access
         self.num_retries = num_retries
 
-    def fetch_text(self, url):
+    def fetch_text(self, url, content_types=None):
         if url.startswith("keep:"):
-            with self.fsaccess.open(url, "r") as f:
+            with self.fsaccess.open(url, "r", encoding="utf-8") 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
+            definition = yaml.round_trip_load(record["definition"])
+            definition["label"] = record["name"]
+            return yaml.round_trip_dump(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):
+            urld, _ = urllib.parse.urldefrag(url)
+            if urld.startswith("keep:"):
+                return self.fsaccess.exists(urld)
+            if urld.startswith("arvwf:"):
+                if self.fetch_text(urld):
                     return True
         except arvados.errors.NotFoundError:
             return False
@@ -265,8 +276,8 @@ class CollectionFetcher(DefaultFetcher):
             locator = baseparts.pop(0)
 
             if (basesp.scheme == "keep" and
-                (not arvados.util.keep_locator_pattern.match(pdh)) and
-                (not arvados.util.collection_uuid_pattern.match(pdh))):
+                (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("/"):