Dump number of pages per visit

1 files changed, 72 insertions, 20 deletions

diff --git a/admin/stats/dump.py b/admin/stats/dump.py
index 20f7c93..d0bdda3 100755
--- a/admin/stats/dump.py
+++ b/admin/stats/dump.py
@@ -4,12 +4,14 @@ from collections import Counter, defaultdict
 import csv
 from dataclasses import dataclass
 from datetime import datetime, timedelta
+from enum import Enum
+from logging import warning
 from pathlib import Path
 import re
+from statistics import mean, median, stdev
 from sys import argv
 from urllib.parse import urlparse
 from typing import Dict, List, Tuple
-from warnings import warn
 
 import user_agents
 
@@ -86,28 +88,44 @@ def parse(logs_paths):
 Visit = List[Access]
 
 
+class UserAgentKind(Enum):
+    PC = 'pc'
+    MOBILE = 'mobile'
+    TABLET = 'tablet'
+    BOT = 'bot'
+    NA = 'n/a'
+
+    @classmethod
+    def from_useragent(cls, ua_string):
+        ua = user_agents.parse(ua_string)
+        # is_bot is not mutually exclusive with other is_* predicates.
+        if ua.is_bot:
+            return cls.BOT
+        if ua.is_pc:
+            return cls.PC
+        if ua.is_mobile:
+            return cls.MOBILE
+        if ua.is_tablet:
+            return cls.TABLET
+        warning(f'Unknown user agent kind: {ua_string}')
+        return cls.NA
+
+    def is_human(self):
+        return self in {
+            UserAgentKind.PC,
+            UserAgentKind.MOBILE,
+            UserAgentKind.TABLET
+        }
+
+
 @dataclass
 class Visitor:
     address: str
-    useragent: str
+    useragent: UserAgentKind
     referrers: List[str]
     visits: List[Visit]
 
 
-def useragent_kind(ua_string):
-    ua = user_agents.parse(ua_string)
-    if ua.is_pc:
-        return 'pc'
-    if ua.is_mobile:
-        return 'mobile'
-    if ua.is_tablet:
-        return 'tablet'
-    if ua.is_bot:
-        return 'bot'
-    warn(f'Unknown user agent kind: {ua_string}')
-    return 'n/a'
-
-
 def sort_visits(accesses):
     visitors: Dict[Tuple(str, str), Visitor] = {}
 
@@ -118,7 +136,7 @@ def sort_visits(accesses):
         if visitor is None:
             visitor = Visitor(
                 a.address,
-                useragent_kind(a.useragent),
+                UserAgentKind.from_useragent(a.useragent),
                 a.referrer,
                 [[a]]
             )
@@ -155,7 +173,7 @@ def daily_visitors(visitors, output_path):
 
     for v in visitors.values():
         for day in find_days(v.visits):
-            days[day][v.useragent] += 1
+            days[day][v.useragent.value] += 1
 
     with open(output_path, 'w') as f:
         out = csv.writer(f)
@@ -180,7 +198,7 @@ def daily_visits(visitors, output_path):
     for v in visitors.values():
         for visit in v.visits:
             day = datetime_day(visit[0].time)
-            days[day][v.useragent] += 1
+            days[day][v.useragent.value] += 1
 
     with open(output_path, 'w') as f:
         out = csv.writer(f)
@@ -196,11 +214,45 @@ def daily_visits(visitors, output_path):
             print(*values, sep='\t')
 
 
+def daily_pages_per_visit(visitors, output_path):
+    days: Dict[datetime, list] = defaultdict(list)
+    columns = ('min', 'max', 'med', 'avg', 'dev')
+
+    print('Pages/visit:')
+
+    for v in visitors.values():
+        if not v.useragent.is_human():
+            continue
+
+        for visit in v.visits:
+            day = datetime_day(visit[0].time)
+            days[day].append(len(visit))
+
+    with open(output_path, 'w') as f:
+        out = csv.writer(f)
+        out.writerow(('day', *columns))
+        print('day', *columns, sep='\t')
+
+        for day in sorted(days):
+            view_counts = days[day]
+            values = (
+                day.strftime('%F'),
+                min(view_counts),
+                max(view_counts),
+                median(view_counts),
+                mean(view_counts),
+                stdev(view_counts)
+            )
+
+            out.writerow(values)
+            print(*values[:4], *(f'{v:.2f}' for v in values[4:]), sep='\t')
+
+
 def daily_stats(visitors, output_dir):
     output_dir = Path(output_dir)
     daily_visitors(visitors, output_dir.joinpath('dailyvisitors.csv'))
     daily_visits(visitors, output_dir.joinpath('dailyvisits.csv'))
-    # daily_pages_per_visit(visitors, output_dir.joinpath('dailypagespervisit.csv'))
+    daily_pages_per_visit(visitors, output_dir.joinpath('dailypagespervisit.csv'))
     # daily_page_hits(visitors, output_dir.joinpath('dailypagehits.csv'))
     # daily_referrers(visitors, output_dir.joinpath('dailyreferrers.csv'))