Merge pull request #446 from NBISweden/feature/bioschema

Feature/bioschema

backend/application.py

@@ -12,6 +12,7 @@
 import string
 import uuid
 import math
+import re
 
 import db
 import handlers
@@ -40,6 +41,82 @@ def build_dataset_structure(dataset_version, user=None, dataset=None):
     return r
 
 
+class GetSchema(handlers.UnsafeHandler):
+    """
+    Returns the schema.org, and bioschemas.org, annotation for a given
+    url.
+
+    This function behaves quite differently from the rest of the application as
+    the structured data testing tool had trouble catching the schema inject
+    when it went through AngularJS. The solution for now has been to make this
+    very general function that "re-parses" the 'url' request parameter to
+    figure out what information to return.
+    """
+    def get(self):
+
+        dataset = None
+        version = None
+        try:
+            url = self.get_argument('url')
+            match = re.match(".*/dataset/([^/]+)(/version/([^/]+))?", url)
+            if match:
+                dataset = match.group(1)
+                version = match.group(3)
+        except tornado.web.MissingArgumentError:
+            pass
+
+        base = {"@context": "http://schema.org/",
+                "@type": "DataCatalog",
+                "name": "SweFreq",
+                "alternateName": [ "The Swedish Frequency resource for genomics" ],
+                "description": "The Swedish Frequency resource for genomics (SweFreq) is a website developed to make genomic datasets more findable and accessible in order to promote collaboration, new research and increase public benefit.",
+                "url": "https://swefreq.nbis.se/",
+                "provider": {
+                    "@type": "Organization",
+                    "name": "National Bioinformatics Infrastructure Sweden",
+                    "alternateName": [ "NBIS",
+                                       "ELIXIR Sweden" ],
+                    "logo": "http://nbis.se/assets/img/logos/nbislogo-green.svg",
+                    "url": "https://nbis.se/"
+                },
+                "datePublished": "2016-12-23",
+                "dateModified": "2017-02-01",
+                "license": {
+                    "@type": "CreativeWork",
+                    "name": "GNU General Public License v3.0",
+                    "url": "https://www.gnu.org/licenses/gpl-3.0.en.html"
+                }
+            }
+
+        if dataset:
+            dataset_schema = {'@type':"Dataset"}
+
+            try:
+                dataset_version = db.get_dataset_version(dataset, version)
+
+                if dataset_version.available_from > datetime.now():
+                    # If it's not available yet, only return if user is admin.
+                    if not (self.current_user and self.current_user.is_admin(version.dataset)):
+                        self.send_error(status_code=403)
+
+                base_url = "%s://%s" % (self.request.protocol, self.request.host)
+                dataset_schema['url']         = base_url + "/dataset/" + dataset_version.dataset.short_name
+                dataset_schema['@id']         = dataset_schema['url']
+                dataset_schema['name']        = dataset_version.dataset.short_name
+                dataset_schema['description'] = dataset_version.description
+                dataset_schema['identifier']  = dataset_schema['name']
+                dataset_schema['citation']    = dataset_version.ref_doi
+
+                base["dataset"] = dataset_schema
+
+            except db.DatasetVersion.DoesNotExist as e:
+                logging.error("Dataset version does not exist: {}".format(e))
+            except db.DatasetVersionCurrent.DoesNotExist as e:
+                logging.error("Dataset does not exist: {}".format(e))
+
+        self.finish(base)
+
+
 class ListDatasets(handlers.UnsafeHandler):
     def get(self):
         # List all datasets available to the current user, earliear than now OR

backend/handlers.py

@@ -80,14 +80,8 @@ def write(self, chunk):
 
 def _convert_keys_to_hump_back(chunk):
     """
-    Converts keys given in snake_case to humbBack-case, while preserving the
+    Converts keys given in snake_case to humpBack-case, while preserving the
     capitalization of the first letter.
-
-    This conversion will rewrite a name already in camel, or humpback,
-    i.e. thisIsAKey -> thisisakey.
-    If this is unwanted, the conversion can instead be written as:
-    new_key = k[0] + "".join([a[0].upper() + a[1:] for a in k.split("_")])[1:]
-    to preserve upper-case letters within words.
     """
     if isinstance(chunk, list):
         return [_convert_keys_to_hump_back(e) for e in chunk]
@@ -98,7 +92,7 @@ def _convert_keys_to_hump_back(chunk):
     new_chunk = {}
     for k, v in chunk.items():
         # First character should be the same as in the original string
-        new_key = k[0] + k.title().replace("_", "")[1:]
+        new_key = k[0] + "".join([a[0].upper() + a[1:] for a in k.split("_")])[1:]
         new_chunk[new_key] = _convert_keys_to_hump_back(v)
     return new_chunk
 

backend/route.py

@@ -58,6 +58,7 @@ def __init__(self, settings):
             (r"/api/users/me",                                                        application.GetUser),
             (r"/api/users/datasets",                                                  application.UserDatasetAccess),
             (r"/api/users/sftp_access",                                               application.SFTPAccess),
+            (r"/api/schema",                                                          application.GetSchema),
             ### Dataset Api
             (r"/api/datasets",                                                        application.ListDatasets),
             (r"/api/datasets/(?P<dataset>[^\/]+)",                                    application.GetDataset),

frontend/assets/js/schema-injector.js

@@ -0,0 +1,14 @@
+/*
+ *  This function could not be included in angular, due to what we believe is
+ *  a timing issue. When we had angular inject the schema.org annotation, it
+ *  simply would not register in the testing tool - so now we inject it before
+ *  loading the angular app.
+ *
+ *  The function passes the current url to the backend, which uses it to
+ *  the correct schema.org annotation from the database.
+ */
+(function() {
+  $.getJSON( "/api/schema?url=" + $(location).attr("href"), function(data) {
+    $("#ldJsonTarget").html( JSON.stringify(data, null, 2) );
+  });
+})();

frontend/src/js/app.routes.js

@@ -7,8 +7,8 @@
             .when("/",                                           { templateUrl: "static/templates/ng-templates/home.html"             })
             .when("/login",                                      { templateUrl: "static/templates/ng-templates/login.html"            })
             .when("/profile",                                    { templateUrl: "static/templates/ng-templates/profile.html"          })
-            .when("/error",                                      { templateUrl: "static/templates/ng-templates/error.html"          })
-            .when("/security_warning",                           { templateUrl: "static/templates/ng-templates/security_warning.html"          })
+            .when("/error",                                      { templateUrl: "static/templates/ng-templates/error.html"            })
+            .when("/security_warning",                           { templateUrl: "static/templates/ng-templates/security_warning.html" })
             .when("/dataset/:dataset",                           { templateUrl: "static/templates/ng-templates/dataset.html"          })
             .when("/dataset/:dataset/terms",                     { templateUrl: "static/templates/ng-templates/dataset-terms.html"    })
             .when("/dataset/:dataset/download",                  { templateUrl: "static/templates/ng-templates/dataset-download.html" })

frontend/src/js/controller.mainController.js

@@ -1,6 +1,7 @@
 (function() {
     angular.module("App")
-    .controller("mainController", ["$location", "$cookies", "User", function($location, $cookies, User) {
+    .controller("mainController", ["$location", "$cookies", "$scope", "User",
+                          function( $location,   $cookies,   $scope,   User) {
         var localThis = this;
         localThis.url = function() { return $location.path(); };
         localThis.loggedIn = false;
@@ -9,6 +10,16 @@
         activate();
 
         function activate() {
+            // This function is the same as is run from schema-injector.js.
+            // It re-runs every time the route changes to keep the schema.org
+            // annotation updated.
+            // It passes the new url to the backend to update the the schema tag
+            // based on what is currently being browsed.
+            $scope.$on("$routeChangeStart", function() {
+                $.getJSON( "/api/schema?url=" + $location.path(), function(data) {
+                  $("#ldJsonTarget").html( JSON.stringify(data, null, 2) );
+                });
+            });
             User.getUser().then(function(data) {
                 localThis.user = data;
                 localThis.loginType = data.loginType;

frontend/src/js/controller.profileController.js

@@ -1,6 +1,6 @@
 (function() {
     angular.module("App")
-    .controller("profileController", ["User", "Countries", "SFTPAccess", 
+    .controller("profileController", ["User", "Countries", "SFTPAccess",
                               function(User,   Countries,   SFTPAccess) {
     var localThis = this;
     localThis.sftp = {"user":"", "password":"", "expires":null};

frontend/templates/index.html

@@ -3,6 +3,7 @@
 <html ng-app="App">
     <head>
         <title>SweFreq</title>
+        <script type="application/ld+json" id="ldJsonTarget"></script>
         <!-- SCROLLS -->
         <meta charset="utf-8" />
         <base href="/">
@@ -131,7 +132,10 @@
          <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.0/angular.min.js"></script>
          <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.0/angular-route.min.js"></script>
          <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.0/angular-cookies.min.js"></script>
+
 [% endif %]
+        <!-- Schema.org injector -->
+        <script src="/static/js/schema-injector.js"></script>
         <!-- The application -->
         <script src="/static/js/vendor.js"></script>
         <script src="/static/js/app.js"></script>

frontend/templates/ng-templates/login.html

@@ -15,8 +15,8 @@ <h1>Your account has been updated!</h1>
                     <h1>Login to swefreq is changing</h1>
                 <div class="panel-body">
                     <p>
-                        To transfer your login credentials from a google account to an elixir account, 
-                        just use the buttons below to log in to both google and elixir, we will then 
+                        To transfer your login credentials from a google account to an elixir account,
+                        just use the buttons below to log in to both google and elixir, we will then
                         give you the option to update your account with the elixir information.
                     </p>
                     <center>
@@ -36,7 +36,7 @@ <h1>Login to swefreq is changing</h1>
                             </button>
                         </a>
                     </center>
-            
+
                 </div>
             </div>
         </div>