add Gitlab registry compatibility

Author: Nicho Kappel

binderhub/builder.py

@@ -7,6 +7,7 @@
 from http.client import responses
 import json
 import string
+import re
 import time
 import escapism
 
@@ -55,6 +56,30 @@
 LAUNCHES_INPROGRESS = Gauge('binderhub_inprogress_launches', 'Launches currently in progress')
 
 
+def _get_image_basename_and_tag(full_name):
+    """Get a supposed image name and tag without the registry part
+    :param full_name: full image specification, e.g. "gitlab.com/user/project:tag"
+    :return: tuple of image name and tag, e.g. ("user/project", "tag")
+    """
+    # the tag is either after the last (and only) colon, or not given at all,
+    # in which case "latest" is implied
+    tag_splits = full_name.rsplit(':', 1)
+    if len(tag_splits) == 2:
+        image_name = tag_splits[0]
+        tag = tag_splits[1]
+    else:
+        image_name = full_name
+        tag = 'latest'
+
+    if re.fullmatch('[a-z0-9]{4,40}/[a-z0-9\._-]{2,255}', image_name):
+        # if it looks like a Docker Hub image name, we're done
+        return image_name, tag
+    else:
+        # if the image isn't implied to origin at Docker Hub, the first part has to be a registry
+        image_basename = '/'.join(image_name.split('/')[1:])
+        return image_basename, tag
+
+
 def _generate_build_name(build_slug, ref, prefix='', limit=63, ref_length=6):
     """Generate a unique build name with a limited character length.
 
@@ -310,7 +335,7 @@ async def get(self, provider_prefix, _unescaped_spec):
         if self.settings['use_registry']:
             for _ in range(3):
                 try:
-                    image_manifest = await self.registry.get_image_manifest(*'/'.join(image_name.split('/')[-2:]).split(':', 1))
+                    image_manifest = await self.registry.get_image_manifest(*_get_image_basename_and_tag(image_name))
                     image_found = bool(image_manifest)
                     break
                 except HTTPClientError:

binderhub/registry.py

@@ -191,7 +191,10 @@ async def get_image_manifest(self, image, tag):
         # first, get a token to perform the manifest request
         if self.token_url:
             auth_req = httpclient.HTTPRequest(
-                url_concat(self.token_url, {"scope": "repository:{}:pull".format(image)}),
+                url_concat(self.token_url, {
+                    "scope": "repository:{}:pull".format(image),
+                    "service": "container_registry"
+                }),
                 auth_username=self.username,
                 auth_password=self.password,
             )

binderhub/tests/test_builder.py

@@ -1,8 +1,22 @@
 import pytest
 
-from binderhub.builder import _generate_build_name
+from binderhub.builder import _generate_build_name, _get_image_basename_and_tag
 
 
+@pytest.mark.parametrize("fullname,basename,tag", [
+    ("jupyterhub/k8s-binderhub:0.2.0-a2079a5", "jupyterhub/k8s-binderhub", "0.2.0-a2079a5"),
+    ("jupyterhub/jupyterhub", "jupyterhub/jupyterhub", "latest"),
+    ("gcr.io/project/image:tag", "project/image", "tag"),
+    ("weirdregistry.com/image:tag", "image", "tag"),
+    ("gitlab-registry.example.com/group/project:some-tag", "group/project", "some-tag"),
+    ("gitlab-registry.example.com/group/project/image:latest", "group/project/image", "latest"),
+    ("gitlab-registry.example.com/group/project/my/image:rc1", "group/project/my/image", "rc1")
+])
+def test_image_basename_resolution(fullname, basename, tag):
+    result_basename, result_tag = _get_image_basename_and_tag(fullname)
+    assert result_basename == basename
+    assert result_tag == tag
+
 @pytest.mark.parametrize('ref,build_slug', [
     # a long ref, no special characters at critical positions
     ('3035124.v3.0', 'dataverse-dvn-2ftjclkp'),