Hack in fragile GitLab support

nbviewer/app.py

@@ -86,6 +86,7 @@ class NBViewer(Application):
     github_tree_handler = Unicode(default_value="nbviewer.providers.github.handlers.GitHubTreeHandler", help="The Tornado handler to use for viewing directory trees on GitHub").tag(config=True)
     gist_handler        = Unicode(default_value="nbviewer.providers.gist.handlers.GistHandler",         help="The Tornado handler to use for viewing notebooks stored as GitHub Gists").tag(config=True)
     user_gists_handler  = Unicode(default_value="nbviewer.providers.gist.handlers.UserGistsHandler",    help="The Tornado handler to use for viewing directory containing all of a user's Gists").tag(config=True)
+    gitlab_handler      = Unicode(default_value="nbviewer.providers.gitlab.handlers.GitlabHandler",     help="The Tornado handler to use for viewing notebooks in a GitLab instance").tag(config=True)
 
     client = Any().tag(config=True)
     @default('client')
@@ -245,6 +246,7 @@ def init_tornado_application(self):
                   local_handler=self.local_handler,
                   gist_handler=self.gist_handler,
                   user_gists_handler=self.user_gists_handler,
+                  gitlab_handler=self.gitlab_handler,
         )
         handler_kwargs = {'handler_names' : handler_names, 'handler_settings' : self.handler_settings}
         handlers = init_handlers(self.formats, options.providers, self.base_url, options.localfiles, **handler_kwargs)

nbviewer/providers/__init__.py

@@ -6,10 +6,10 @@
 #-----------------------------------------------------------------------------
 
 default_providers = ['nbviewer.providers.{}'.format(prov)
-                     for prov in ['url', 'github', 'gist']]
+                     for prov in ['url', 'github', 'gist', 'gitlab']]
 
 default_rewrites = ['nbviewer.providers.{}'.format(prov)
-                    for prov in ['gist', 'github', 'dropbox', 'url']]
+                    for prov in ['gitlab', 'gist', 'github', 'dropbox', 'url']]
 
 
 def provider_handlers(providers, **handler_kwargs):

nbviewer/providers/gitlab/__init__.py

@@ -0,0 +1 @@
+from .handlers import default_handlers, uri_rewrites

nbviewer/providers/gitlab/handlers.py

@@ -0,0 +1,107 @@
+import json
+import os
+from tornado import web
+from tornado.log import app_log
+from ..base import RenderingHandler, cached
+from ...utils import response_text
+from .. import _load_handler_from_location
+
+
+class GitlabHandler(RenderingHandler):
+
+    async def get_notebook_data(self, host, group, repo, blob, branch, path):
+
+        token = os.environ.get("GITLAB_TOKEN")
+
+        base_url = "https://{host}/api/v4".format(host=host)
+
+        projects_url = ("{base_url}/projects?private_token={token}"
+                        .format(base_url=base_url, token=token))
+
+        app_log.info("Fetching " + projects_url)
+
+        try:
+            projects_response = await self.fetch(projects_url)
+            projects_text = response_text(projects_response)
+            projects = json.loads(projects_text)
+
+            path_with_namespace = "{group}/{repo}".format(group=group, repo=repo)
+
+            project = None
+            for p in projects:
+                if p["path_with_namespace"] == path_with_namespace:
+                    project = p
+                    break
+            else:
+                raise Exception("Project path not found: " + path_with_namespace)
+
+            prj = project["id"]
+            tree_url = ("{base_url}/projects/{prj}/repository/tree?recursive=true&ref={branch}&per_page=1000&private_token={token}"
+                        .format(base_url=base_url,
+                                prj=prj,
+                                branch=branch,
+                                token=token))
+
+            app_log.info("Fetching " + tree_url)
+
+            tree_response = await self.fetch(tree_url)
+            tree_text = response_text(tree_response)
+            tree = json.loads(tree_text)
+
+            blob = None
+            for item in tree:
+                if item["path"] == path:
+                    blob = item
+                    break
+            else:
+                raise Exception("Blob not found: " + path)
+
+            sha = blob["id"]
+
+            raw_url = "{base_url}/projects/{prj}/repository/blobs/{sha}/raw?private_token={token}"
+            return raw_url.format(base_url=base_url,
+                                  prj=prj,
+                                  sha=sha,
+                                  token=token)
+
+        except Exception as e:
+            app_log.error(e)
+
+
+    async def deliver_notebook(self, remote_url):
+        app_log.info("Fetching notebook: " + remote_url)
+
+        response = await self.fetch(remote_url)
+
+        try:
+            nbjson = response_text(response, encoding='utf-8')
+        except UnicodeDecodeError:
+            app_log.error("Notebook is not utf8: %s", remote_url, exc_info=True)
+            raise web.HTTPError(400)
+
+        await self.finish_notebook(nbjson,
+                                   download_url=remote_url,
+                                   msg="file from url: " + remote_url,
+                                   public=False,
+                                   request=self.request)
+
+
+
+    @cached
+    async def get(self, host, group, repo, blob, branch, path):
+        raw_url = await self.get_notebook_data(host, group, repo, blob, branch, path)
+        await self.deliver_notebook(raw_url)
+
+def uri_rewrites(rewrites=[]):
+    gitlab_rewrites = [
+        (r'^https?://(gitlab\..*)$', r'/gitlab/{0}'),
+        (r'^/url[s]?/(gitlab\..*)$', r'/gitlab/{0}'),
+        (r'^/url[s]?/https?://(gitlab\..*)$', r'/gitlab/{0}'),
+    ]
+    return rewrites + gitlab_rewrites
+
+def default_handlers(handlers=[], **handler_names):
+    gitlab_handler = _load_handler_from_location(handler_names['gitlab_handler'])
+    return handlers + [
+        (r'/gitlab/(?P<host>[\w_\-.]+)/(?P<group>[\w_\-.]+)/(?P<repo>[\w_\-]+)/(?P<blob>blob)/(?P<branch>[\w_\-()]+)/(?P<path>.*)', gitlab_handler, {}),
+    ]