From 8d52295596ac3dab71cbd4bc5b876c7810a0ed59 Mon Sep 17 00:00:00 2001
From: Taha Yassine Kraiem <tahayk2@gmail.com>
Date: Tue, 21 Sep 2021 17:19:12 +0200
Subject: [PATCH] feat(api): insights retention: fill missing values feat(api):
 insights retention: max date restrictions

---
 api/chalicelib/core/insights.py | 61 ++++++++++++++++++++++++++++++---
 api/chalicelib/utils/TimeUTC.py |  9 +++++
 2 files changed, 66 insertions(+), 4 deletions(-)

diff --git a/api/chalicelib/core/insights.py b/api/chalicelib/core/insights.py
index 37042e64d..8e8611dfb 100644
--- a/api/chalicelib/core/insights.py
+++ b/api/chalicelib/core/insights.py
@@ -116,9 +116,61 @@ def __compute_retention_percentage(rows):
     return rows
 
 
+def __complete_retention(rows, start_date, end_date=None):
+    if rows is None or len(rows) == 0:
+        return rows
+    max_week = 10
+    week = 0
+    delta_date = 0
+    while max_week > 0:
+        start_date += TimeUTC.MS_WEEK
+        if end_date is not None and start_date >= end_date:
+            break
+        delta = 0
+        if delta_date + week >= len(rows) \
+                or delta_date + week < len(rows) and rows[delta_date + week]["firstConnexionWeek"] > start_date:
+            for i in range(max_week):
+                if end_date is not None and start_date + i * TimeUTC.MS_WEEK >= end_date:
+                    break
+
+                neutral = {
+                    "firstConnexionWeek": start_date,
+                    "week": i,
+                    "usersCount": 0,
+                    "connectedUsers": [],
+                    "percentage": 0
+                }
+                rows.insert(delta_date + week + i, neutral)
+                delta = i
+        else:
+            for i in range(max_week):
+                if end_date is not None and start_date + i * TimeUTC.MS_WEEK >= end_date:
+                    break
+
+                neutral = {
+                    "firstConnexionWeek": start_date,
+                    "week": i,
+                    "usersCount": 0,
+                    "connectedUsers": [],
+                    "percentage": 0
+                }
+                if delta_date + week + i < len(rows) \
+                        and i != rows[delta_date + week + i]["week"]:
+                    rows.insert(delta_date + week + i, neutral)
+                elif delta_date + week + i >= len(rows):
+                    rows.append(neutral)
+                delta = i
+        week += delta
+        max_week -= 1
+        delta_date += 1
+    return rows
+
+
 @dev.timed
-def get_retention(project_id, startTimestamp=TimeUTC.now(delta_days=-1), endTimestamp=TimeUTC.now(), filters=[],
+def get_retention(project_id, startTimestamp=TimeUTC.now(delta_days=-70), endTimestamp=TimeUTC.now(), filters=[],
                   **args):
+    startTimestamp = TimeUTC.trunc_week(startTimestamp)
+    endTimestamp = startTimestamp + 10 * 7 * 24 * 60 * 60 * 1000
     pg_sub_query = __get_constraints(project_id=project_id, data=args, duration=True, main_table="sessions",
                                      time_constraint=True)
 
@@ -130,6 +182,7 @@ def get_retention(project_id, startTimestamp=TimeUTC.now(delta_days=-1), endTime
                         FROM (SELECT DISTINCT user_id, MIN(DATE_TRUNC('week', to_timestamp(start_ts / 1000))) AS first_connexion_week
                               FROM sessions
                               WHERE {" AND ".join(pg_sub_query)}
+                                AND user_id IS NOT NULL 
                                 AND NOT EXISTS((SELECT 1
                                                 FROM sessions AS bsess
                                                 WHERE bsess.start_ts<EXTRACT('EPOCH' FROM DATE_TRUNC('week', to_timestamp(%(startTimestamp)s / 1000))) * 1000
@@ -144,7 +197,7 @@ def get_retention(project_id, startTimestamp=TimeUTC.now(delta_days=-1), endTime
                                                       AND first_connexion_week <=
                                                           DATE_TRUNC('week', to_timestamp(sessions.start_ts / 1000)::timestamp)
                                                       AND sessions.project_id = 1
-                                                      AND sessions.start_ts < (%(startTimestamp)s + 10 * 7 * 24 * 60 * 60 * 1000::bigint - 1)
+                                                      AND sessions.start_ts < (%(endTimestamp)s - 1)
                                                     GROUP BY connexion_week, user_id) AS connexions_list ON (TRUE)
                         GROUP BY first_connexion_week, week
                         ORDER BY first_connexion_week, week;"""
@@ -154,5 +207,5 @@ def get_retention(project_id, startTimestamp=TimeUTC.now(delta_days=-1), endTime
         # print(cur.mogrify(pg_query, params))
         cur.execute(cur.mogrify(pg_query, params))
         rows = cur.fetchall()
-
-    return __compute_retention_percentage(helper.list_to_camel_case(rows))
+        rows = __compute_retention_percentage(helper.list_to_camel_case(rows))
+    return __complete_retention(rows=rows, start_date=startTimestamp, end_date=TimeUTC.now())
diff --git a/api/chalicelib/utils/TimeUTC.py b/api/chalicelib/utils/TimeUTC.py
index 4b19c45ba..9cd353eb4 100644
--- a/api/chalicelib/utils/TimeUTC.py
+++ b/api/chalicelib/utils/TimeUTC.py
@@ -7,6 +7,7 @@ class TimeUTC:
     MS_MINUTE = 60 * 1000
     MS_HOUR = MS_MINUTE * 60
     MS_DAY = MS_HOUR * 24
+    MS_WEEK = MS_DAY * 7
     MS_MONTH = MS_DAY * 30
     MS_MONTH_TRUE = monthrange(datetime.now(pytz.utc).astimezone(pytz.utc).year,
                                datetime.now(pytz.utc).astimezone(pytz.utc).month)[1] * MS_DAY
@@ -113,3 +114,11 @@ class TimeUTC:
     @staticmethod
     def get_utc_offset():
         return int((datetime.now(pytz.utc).now() - datetime.now(pytz.utc).replace(tzinfo=None)).total_seconds() * 1000)
+
+    @staticmethod
+    def trunc_week(timestamp):
+        dt = TimeUTC.from_ms_timestamp(timestamp)
+        start = dt - timedelta(days=dt.weekday())
+        return TimeUTC.datetime_to_timestamp(start
+                                             .replace(hour=0, minute=0, second=0, microsecond=0)
+                                             .astimezone(pytz.utc))