add field tracking #110

biothings/web/analytics/channels.py

@@ -45,7 +45,11 @@ def handles(self, event):
         return isinstance(event, Event)
 
     def send(self, payload):
-        yield HTTPRequest(
-            'http://www.google-analytics.com/batch', method='POST',
-            body='\n'.join(payload.to_GA_payload(self.tracking_id, self.uid_version))
-        )
+        events = payload.to_GA_payload(self.tracking_id, self.uid_version)
+        # #batch-limitations section of the URL above
+        # A maximum of 20 hits can be specified per request.
+        for i in range(0, len(events), 20):
+            yield HTTPRequest(
+                'http://www.google-analytics.com/batch', method='POST',
+                body='\n'.join(events[i: i+20])
+            )

biothings/web/analytics/events.py

@@ -113,6 +113,9 @@ def to_GA_payload(self, tracking_id, cid_version=1):
     def __str__(self):  # to facilitate logging
         return f"{type(self).__name__}({pformat(self)})"
 
+def _clean(dict):
+    return {k: v for k, v in dict.items() if v}
+
 class GAEvent(Event):
 
     # GA Event
@@ -127,7 +130,7 @@ def to_GA_payload(self, tracking_id, cid_version=1):
 
         payloads = super().to_GA_payload(tracking_id, cid_version)
         if self.get("category") and self.get("action"):
-            payloads.append(urlencode({
+            payloads.append(urlencode(_clean({
                 "v": 1,  # protocol version
                 "t": "event",
                 "tid": tracking_id,
@@ -136,7 +139,14 @@ def to_GA_payload(self, tracking_id, cid_version=1):
                 "ea": self["action"],
                 "el": self.get("label", ""),
                 "ev": self.get("value", "")
-            }))
+            })))
+        for event in self.get("__secondary__", []):
+            event["__request__"] = self["__request__"]
+            payloads.extend(
+                event.to_GA_payload(
+                    tracking_id, cid_version)[1:])
+            # ignore the first event (pageview)
+            # which is already generated once
         return payloads
 
 

biothings/web/handlers/query.py

@@ -69,12 +69,26 @@ def prepare(self):
         self.args.biothing_type = self.biothing_type
 
         self.event = GAEvent({
+            '__secondary__': [],  # secondary analytical objective: field tracking
             'category': '{}_api'.format(self.biothings.config.APP_VERSION),  # eg.'v1_api'
             'action': '_'.join((self.name, self.request.method.lower())),  # eg.'query_get'
             # 'label': 'fetch_all', etc.
             # 'value': 100, # number of queries
         })
 
+        if self.args._source:
+            for _source in self.args._source:
+                self.event['__secondary__'].append(GAEvent({
+                    'category': 'field_filter',
+                    'action': 'enabled',
+                    'label': _source.split('.', 1)[0]  # root key
+                }))
+        else:
+            self.event['__secondary__'].append(GAEvent({
+                'category': 'field_filter',
+                'action': 'disabled'
+            }))
+
     def write(self, chunk):
         # add an additional header to the JSON formatter
         # with a header image and a title-like section