From 6c4eaf3d14554284bc777b8508fa72f474d177e9 Mon Sep 17 00:00:00 2001
From: Tim Abbott <tabbott@zulipchat.com>
Date: Sun, 5 Feb 2017 16:17:31 -0800
Subject: [PATCH] analytics: Map client names to user-facing versions.

This makes the pie charts on /stats more readable.
---
 analytics/tests/test_views.py | 26 +++++++++++++++++++++++++
 analytics/views.py            | 36 +++++++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+)

diff --git a/analytics/tests/test_views.py b/analytics/tests/test_views.py
index e682a44cd9..a85ee45c39 100644
--- a/analytics/tests/test_views.py
+++ b/analytics/tests/test_views.py
@@ -3,6 +3,7 @@ from zerver.lib.test_classes import ZulipTestCase
 
 from analytics.lib.counts import CountStat
 from analytics.lib.time_utils import time_range
+from analytics.views import rewrite_client_arrays
 
 from datetime import datetime, timedelta
 
@@ -34,3 +35,28 @@ class TestTimeRange(ZulipTestCase):
                          [floor_hour-2*HOUR, floor_hour-HOUR, floor_hour, floor_hour+HOUR])
         self.assertEqual(time_range(floor_day, floor_day+DAY, CountStat.DAY, 4),
                          [floor_day-2*DAY, floor_day-DAY, floor_day, floor_day+DAY])
+
+class TestMapArrays(ZulipTestCase):
+    def test_map_arrays(self):
+        # type: () -> None
+        a = {'desktop app 1.0': [1, 2, 3],
+             'desktop app 2.0': [10, 12, 13],
+             'desktop app 3.0': [21, 22, 23],
+             'website': [1, 2, 3],
+             'ZulipiOS': [1, 2, 3],
+             'ZulipMobile': [1, 5, 7],
+             'ZulipPython': [1, 2, 3],
+             'API: Python': [1, 2, 3],
+             'SomethingRandom': [4, 5, 6],
+             'ZulipGitHubWebhook': [7, 7, 9],
+             'ZulipAndroid': [64, 63, 65]}
+        result = rewrite_client_arrays(a)
+        self.assertEqual(result,
+                         {'Old desktop app': [32, 36, 39],
+                          'Old iOS app': [1, 2, 3],
+                          'New iOS app': [1, 5, 7],
+                          'Website': [1, 2, 3],
+                          'Python API': [2, 4, 6],
+                          'SomethingRandom': [4, 5, 6],
+                          'GitHub webhook': [7, 7, 9],
+                          'Android app': [64, 63, 65]})
diff --git a/analytics/views.py b/analytics/views.py
index 49a3137114..5b977a54f8 100644
--- a/analytics/views.py
+++ b/analytics/views.py
@@ -108,6 +108,36 @@ def table_filtered_to_id(table, key_id):
     else:
         raise ValueError("Unknown table: %s" % (table,))
 
+def client_label_map(name):
+    # type: (str) -> str
+    if name == "website":
+        return "Website"
+    if name.startswith("desktop app"):
+        return "Old desktop app"
+    if name == "ZulipAndroid":
+        return "Android app"
+    if name == "ZulipiOS":
+        return "Old iOS app"
+    if name == "ZulipMobile":
+        return "New iOS app"
+    if name in ["ZulipPython", "API: Python"]:
+        return "Python API"
+    if name.startswith("Zulip") and name.endswith("Webhook"):
+        return name[len("Zulip"):-len("Webhook")] + " webhook"
+    return name
+
+def rewrite_client_arrays(value_arrays):
+    # type: (Dict[str, List[int]]) -> Dict[str, List[int]]
+    mapped_arrays = {} # type: Dict[str, List[int]]
+    for label, array in value_arrays.items():
+        mapped_label = client_label_map(label)
+        if mapped_label in mapped_arrays:
+            for i in range(0, len(array)):
+                mapped_arrays[mapped_label][i] += value_arrays[label][i]
+        else:
+            mapped_arrays[mapped_label] = [value_arrays[label][i] for i in range(0, len(array))]
+    return mapped_arrays
+
 def get_time_series_by_subgroup(stat, table, key_id, end_times, subgroups, labels, include_empty_subgroups):
     # type: (CountStat, Type[BaseCount], Optional[int], List[datetime], List[str], Optional[List[str]], bool) -> Dict[str, List[int]]
     if labels is None:
@@ -124,6 +154,12 @@ def get_time_series_by_subgroup(stat, table, key_id, end_times, subgroups, label
     for subgroup, label in zip(subgroups, labels):
         if (subgroup in value_dicts) or include_empty_subgroups:
             value_arrays[label] = [value_dicts[subgroup][end_time] for end_time in end_times]
+
+    if stat == COUNT_STATS['messages_sent:client:day']:
+        # HACK: We rewrite these arrays to collapse the Client objects
+        # with similar names into a single sum, and generally give
+        # them better names
+        return rewrite_client_arrays(value_arrays)
     return value_arrays