Fix HTTP link quoting in stored links

trains/backend_interface/metrics/events.py

@@ -11,6 +11,7 @@ from six.moves.urllib.parse import urlparse, urlunparse
 
 from ...backend_api.services import events
 from ...config import config
+from ...storage.util import quote_url
 
 
 @six.add_metaclass(abc.ABCMeta)
@@ -294,7 +295,7 @@ class UploadEvent(MetricsEventAdapter):
             delete_local_file=local_file if self._delete_after_upload else None,
         )
 
-    def get_target_full_upload_uri(self, storage_uri, storage_key_prefix=None):
+    def get_target_full_upload_uri(self, storage_uri, storage_key_prefix=None, quote_uri=True):
         e_storage_uri = self._upload_uri or storage_uri
         # if we have an entry (with or without a stream), we'll generate the URL and store it in the event
         filename = self._upload_filename
@@ -306,6 +307,10 @@ class UploadEvent(MetricsEventAdapter):
         # make sure we preserve local path root
         if e_storage_uri.startswith('/'):
             url = '/'+url
+
+        if quote_uri:
+            url = quote_url(url)
+
         return key, url
 
 
@@ -319,16 +324,7 @@ class ImageEvent(UploadEvent):
 
     def get_api_event(self):
         return events.MetricsImageEvent(
-            # Hack: replace single '%' with quoted value '%25',
+            url=self._url,
-            # allowing the link to be properly unquoted during http serving
-            url=self._url
-            if (
-                not self._url
-                or self._url.startswith("file://")
-                or self._url.startswith("/")
-                or self._url.startswith("\\")
-            )
-            else self._url.replace("%", "%25"),
             key=self._key,
             **self._get_base_dict()
         )

trains/backend_interface/metrics/interface.py

@@ -119,7 +119,7 @@ class Metrics(InterfaceBase):
             entry = ev.get_file_entry()
             kwargs = {}
             if entry:
-                key, url = ev.get_target_full_upload_uri(storage_uri, self.storage_key_prefix)
+                key, url = ev.get_target_full_upload_uri(storage_uri, self.storage_key_prefix, quote_uri=False)
                 kwargs[entry.key_prop] = key
                 kwargs[entry.url_prop] = url
                 if not entry.stream:

trains/backend_interface/metrics/reporter.py

@@ -552,11 +552,6 @@ class Reporter(InterfaceBase, AbstractContextManager, SetupUploadMixin, AsyncMan
             return self.report_image_and_upload(title=title, series=series, iter=iter, path=path, image=matrix,
                                                 upload_uri=upload_uri, max_image_history=max_image_history)
 
-        # Hack: replace single '%' with quoted value '%25',
-        #  allowing the link to be properly unquoted during http serving
-        if url:
-            url = url.replace('%', '%25')
-
         self._report(ev)
         plotly_dict = create_image_plot(
             image_src=url,

trains/storage/helper.py

@@ -30,6 +30,7 @@ from six.moves.queue import Queue, Empty
 from six.moves.urllib.parse import urlparse
 from six.moves.urllib.request import url2pathname
 
+from .util import quote_url
 from ..backend_api.utils import get_http_session_with_retry
 from ..backend_config.bucket_config import S3BucketConfigurations, GSBucketConfigurations, AzureContainerConfigurations
 from ..config import config
@@ -547,6 +548,10 @@ class StorageHelper(object):
         if last_ex:
             raise last_ex
 
+        if self.scheme in _HttpDriver.schemes:
+            # qoute link
+            dest_path = quote_url(dest_path)
+
         return dest_path
 
     def upload(self, src_path, dest_path=None, extra=None, async_enable=False, cb=None, retries=1):

trains/storage/util.py

@@ -1,3 +1,4 @@
+from six.moves.urllib.parse import quote, urlparse, urlunparse
 import six
 import fnmatch
 
@@ -16,3 +17,11 @@ def get_config_object_matcher(**patterns):
             if pat and fnmatch.fnmatch(value, pat):
                 return True
     return _matcher
+
+
+def quote_url(url):
+    parsed = urlparse(url)
+    if parsed.scheme not in ('http', 'https'):
+        return url
+    parsed = parsed._replace(path=quote(parsed.path))
+    return urlunparse(parsed)