Tweak stats script

- move referer normalization earlier
- remove temporary dot files
- draw graphs with arrows

1 files changed, 40 insertions, 30 deletions

diff --git a/admin/stats.py b/admin/stats.py
index f69390b..e3d46cc 100644..100755
--- a/admin/stats.py
+++ b/admin/stats.py
@@ -1,7 +1,10 @@
+#!/usr/bin/env python3
+
 from collections import Counter, defaultdict
 from dataclasses import dataclass
 from datetime import datetime, timedelta
 import re
+from os import remove
 from subprocess import run
 from sys import argv
 
@@ -38,9 +41,17 @@ class Access:
         if resource == '/':
             resource = '/index.html'
 
+        referer = re.sub(
+            r'https://(?:www\.)?quatuorbellefeuille\.(?:fr|com)(/[^?]*)(?:\?.+)?',
+            r'\1',
+            info['referer']
+        )
+        if referer == '/':
+            referer = '/index.html'
+
         return cls(
             info['address'], user_agents.parse(info['useragent']),
-            info['referer'], datetime.strptime(info['date'], DATE_FMT),
+            referer, datetime.strptime(info['date'], DATE_FMT),
             resource
         )
 
@@ -96,16 +107,33 @@ def order(grouped_visits):
 
     return visits
 
-def normalizeref(referer):
-    if referer == '-':
-        return 'inconnu'
-    referer = re.sub(
-        r'https://(?:www\.)?quatuorbellefeuille\.(?:fr|com)(/[^?]*)(?:\?.+)?',
-        r'\1', referer
-    )
-    if referer == '/':
-        return '/index.html'
-    return referer
+def visit_graph(accesses):
+    edges = (f'    "{a.referer}" -> "{a.resource}";'
+             for a in accesses)
+    return '\n'.join((f'digraph visit {{', *edges, '}'))
+
+def graph(visits):
+    date = visits[0][0].time.strftime('%F')
+
+    tempfiles = {
+        f'{date}-{i}.pdf': visit for i, visit in enumerate(visits)
+    }
+
+    for tempfile, visit in tempfiles.items():
+        vgraph = visit_graph(visit)
+
+        with open(tempfile, 'wb') as vfile:
+            vfile.write(
+                run(('dot', '-Tpdf'), text=False, check=True,
+                    capture_output=True, input=vgraph.encode())
+                .stdout
+            )
+
+    run(('qpdf', '--empty', '--pages', *tempfiles, '--', f'{date}.pdf'),
+        check=True)
+
+    for f in tempfiles:
+        remove(f)
 
 def analyze(logs_path):
     accesses = parse(logs_path)
@@ -119,25 +147,7 @@ def analyze(logs_path):
     for page, hits in pagehits.most_common():
         print(hits, page)
 
-    date = accesses[0].time.strftime('%F')
-
-    for i, visit in enumerate(visits_by_time.values()):
-        edges = (f'    "{normalizeref(a.referer)}" -- "{a.resource}";'
-                 for a in visit)
-        graph = '\n'.join(('graph trip {',
-                           *edges,
-                           '}'))
-        dot = run(
-            ('dot', '-Tpdf'), text=False, capture_output=True, check=True,
-            input=graph.encode()
-        )
-        with open(f'{date}-{i}.pdf', 'wb') as visitgraph:
-            visitgraph.write(dot.stdout)
-
-    pages = (f'{date}-{i}.pdf' for i in range(len(visits_by_time.values())))
-    run(('qpdf', '--empty', '--pages', *pages, '--', f'{date}.pdf'),
-        text=False, check=True)
-
+    graph(tuple(visits_by_time.values()))
 
 if __name__ == '__main__':
     analyze(argv[1])