[MRG] add hydroshare as a repo provider

binderhub/app.py

@@ -33,7 +33,7 @@
 from .main import MainHandler, ParameterizedMainHandler, LegacyRedirectHandler
 from .repoproviders import (GitHubRepoProvider, GitRepoProvider,
                             GitLabRepoProvider, GistRepoProvider,
-                            ZenodoProvider, FigshareProvider)
+                            ZenodoProvider, FigshareProvider, HydroshareProvider)
 from .metrics import MetricsHandler
 
 from .utils import ByteSpecification, url_path_join
@@ -393,6 +393,7 @@ def _add_slash(self, proposal):
             'gl': GitLabRepoProvider,
             'zenodo': ZenodoProvider,
             'figshare': FigshareProvider,
+            'hydroshare': HydroshareProvider,
         },
         config=True,
         help="""

binderhub/event-schemas/launch.json

@@ -12,7 +12,8 @@
                 "GitLab",
                 "Git",
                 "Zenodo",
-                "Figshare"
+                "Figshare",
+                "Hydroshare"
             ],
             "description": "Provider for the repository being launched"
         },

binderhub/main.py

@@ -16,7 +16,8 @@
     "gl": "GitLab",
     "git": "Git repo",
     "zenodo": "Zenodo",
-    "figshare": "Figshare"
+    "figshare": "Figshare",
+    "hydroshare": "Hydroshare"
 }
 
 

binderhub/repoproviders.py

@@ -7,7 +7,7 @@
 Note: When adding a new repo provider, add it to the allowed values for
       repo providers in event-schemas/launch.json.
 """
-from datetime import timedelta
+from datetime import timedelta, datetime, timezone
 import json
 import os
 import time
@@ -250,6 +250,51 @@ def get_build_slug(self):
         return "figshare-{}".format(self.record_id)
 
 
+class HydroshareProvider(RepoProvider):
+    """Provide contents of a Hydroshare resource
+
+    Users must provide a spec consisting of the Hydroshare resource id.
+    """
+    name = Unicode("Hydroshare")
+    url_regex = re.compile(r".*([0-9a-f]{32}).*")
+
+    def _parse_resource_id(self, spec):
+        match = self.url_regex.match(spec)
+        if not match:
+            raise ValueError("The specified Hydroshare resource id was not recognized.")
+        resource_id = match.groups()[0]
+        return resource_id
+
+    @gen.coroutine
+    def get_resolved_ref(self):
+        client = AsyncHTTPClient()
+        self.resource_id = self._parse_resource_id(self.spec)
+        req = HTTPRequest("https://www.hydroshare.org/hsapi/resource/{}/scimeta/elements".format(self.resource_id),
+                          user_agent="BinderHub")
+        r = yield client.fetch(req)
+        def parse_date(json_body):
+            json_response = json.loads(json_body)
+            date = next(
+                item for item in json_response["dates"] if item["type"] == "modified"
+            )["start_date"]
+            # Hydroshare timestamp always returns the same timezone, so strip it
+            date = date.split(".")[0]
+            parsed_date = datetime.strptime(date, "%Y-%m-%dT%H:%M:%S")
+            epoch = parsed_date.replace(tzinfo=timezone(timedelta(0))).timestamp()
+            # truncate the timestamp
+            return str(int(epoch))
+        # date last updated is only good for the day... probably need something finer eventually
+        self.record_id = "{}.v{}".format(self.resource_id, parse_date(r.body))
+        return self.record_id
+
+    def get_repo_url(self):
+        self.resource_id = self._parse_resource_id(self.spec)
+        return "https://www.hydroshare.org/resource/{}".format(self.resource_id)
+
+    def get_build_slug(self):
+        return "hydroshare-{}".format(self.record_id)
+
+
 class GitRepoProvider(RepoProvider):
     """Bare bones git repo provider.
 

binderhub/static/js/index.js

@@ -89,6 +89,11 @@ function updateRepoText() {
     $("#ref").prop("disabled", true);
     $("label[for=ref]").prop("disabled", true);
   }
+  else if (provider === "hydroshare") {
+    text = "Hydroshare resource id or URL";
+    $("#ref").prop("disabled", true);
+    $("label[for=ref]").prop("disabled", true);
+  }
   $("#repository").attr('placeholder', text);
   $("label[for=repository]").text(text);
   $("#ref").attr('placeholder', tag_text);
@@ -112,7 +117,7 @@ function getBuildFormValues() {
   }
 
   var ref = $('#ref').val().trim() || 'master';
-  if (providerPrefix === 'zenodo' || providerPrefix === 'figshare') {
+  if (providerPrefix === 'zenodo' || providerPrefix === 'figshare' || providerPrefix === 'hydroshare') {
     ref = "";
   }
   var path = $('#filepath').val().trim();

binderhub/templates/index.html

@@ -56,6 +56,7 @@ <h4 id="form-header" class='row'>Build and launch a repository</h4>
                 <li class="dropdown-item" value="git"><a href="#">Git repository</a></li>
                 <li class="dropdown-item" value="zenodo"><a href="#">Zenodo DOI</a></li>
                 <li class="dropdown-item" value="figshare"><a href="#">Figshare DOI</a></li>
+                <li class="dropdown-item" value="hydroshare"><a href="#">Hydroshare resource</a></li>
               </ul>
             </div>
           </div>

binderhub/tests/test_repoproviders.py

@@ -6,7 +6,8 @@
 
 from binderhub.repoproviders import (
     tokenize_spec, strip_suffix, GitHubRepoProvider, GitRepoProvider,
-    GitLabRepoProvider, GistRepoProvider, ZenodoProvider, FigshareProvider
+    GitLabRepoProvider, GistRepoProvider, ZenodoProvider, FigshareProvider,
+    HydroshareProvider
 )
 
 
@@ -67,6 +68,34 @@ async def test_figshare():
     assert repo_url == spec
 
 
+async def test_hydroshare():
+    spec = 'https://www.hydroshare.org/resource/142c59757ed54de1816777828c9716e7'
+
+    provider = HydroshareProvider(spec=spec)
+
+    ref = await provider.get_resolved_ref()
+    assert ref == '142c59757ed54de1816777828c9716e7.v1545934606'
+
+    slug = provider.get_build_slug()
+    assert slug == 'hydroshare-142c59757ed54de1816777828c9716e7.v1545934606'
+    repo_url = provider.get_repo_url()
+    assert repo_url == spec
+
+
+async def test_hydroshare_doi():
+    spec = '10.4211/hs.b8f6eae9d89241cf8b5904033460af61'
+
+    provider = HydroshareProvider(spec=spec)
+
+    ref = await provider.get_resolved_ref()
+    assert ref == 'b8f6eae9d89241cf8b5904033460af61.v1565445792'
+
+    slug = provider.get_build_slug()
+    assert slug == 'hydroshare-b8f6eae9d89241cf8b5904033460af61.v1565445792'
+    repo_url = provider.get_repo_url()
+    assert repo_url ==  'https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61'
+
+
 @pytest.mark.github_api
 def test_github_ref():
     provider = GitHubRepoProvider(spec='jupyterhub/zero-to-jupyterhub-k8s/v0.4')