|
|
1
|
+#!/usr/bin/perl -w
|
|
|
2
|
+#
|
|
|
3
|
+# This script takes logs archive months as arguments.
|
|
|
4
|
+# Example: count-mar-downloads 2025-08 2025-07
|
|
|
5
|
+# Check which months are available on https://collector.torproject.org/archive/webstats/
|
|
|
6
|
+#
|
|
|
7
|
+# The script will dowload the log archives for the selected months,
|
|
|
8
|
+# extract them, parse the logs and print download numbers per version
|
|
|
9
|
+# numbers, OS, and incremental/full updates.
|
|
|
10
|
+#
|
|
|
11
|
+# The log files archives are stored in the tools/web-logs-count-mar-downloads
|
|
|
12
|
+# directory.
|
|
|
13
|
+#
|
|
|
14
|
+use strict;
|
|
|
15
|
+use FindBin;
|
|
|
16
|
+use lib "$FindBin::Bin/../rbm/lib";
|
|
|
17
|
+use RBM::CaptureExec qw(capture_exec);
|
|
|
18
|
+use File::Temp qw/tempdir/;
|
|
|
19
|
+use File::Copy qw/move/;
|
|
|
20
|
+use File::Find;
|
|
|
21
|
+
|
|
|
22
|
+if (!@ARGV) {
|
|
|
23
|
+ print "Usage: count-mar-downloads <month>\n";
|
|
|
24
|
+ print "Example: count-mar-downloads 2025-08 2025-07\n";
|
|
|
25
|
+ print "Check which months are available on https://collector.torproject.org/archive/webstats/\n";
|
|
|
26
|
+ exit 1;
|
|
|
27
|
+}
|
|
|
28
|
+
|
|
|
29
|
+my @months = @ARGV;
|
|
|
30
|
+
|
|
|
31
|
+my %downloads;
|
|
|
32
|
+
|
|
|
33
|
+my $weblogsdir = "$FindBin::Bin/web-logs-count-mar-downloads";
|
|
|
34
|
+mkdir $weblogsdir;
|
|
|
35
|
+chdir $weblogsdir;
|
|
|
36
|
+
|
|
|
37
|
+sub exit_error {
|
|
|
38
|
+ print STDERR "Error: ", $_[0], "\n";
|
|
|
39
|
+ exit (exists $_[1] ? $_[1] : 1);
|
|
|
40
|
+}
|
|
|
41
|
+
|
|
|
42
|
+sub download_log_files {
|
|
|
43
|
+ my $tmpdir = tempdir(CLEANUP => 1);
|
|
|
44
|
+ foreach my $month (@months) {
|
|
|
45
|
+ my $file = "webstats-$month.tar";
|
|
|
46
|
+ if (-f $file) {
|
|
|
47
|
+ print STDERR "Using existing file $file (remove it if you want to re-download it).\n";
|
|
|
48
|
+ next;
|
|
|
49
|
+ }
|
|
|
50
|
+ my $url = "https://collector.torproject.org/archive/webstats/$file";
|
|
|
51
|
+ exit_error "Error downloading $url" unless
|
|
|
52
|
+ system('wget', '-O', "$tmpdir/$file", $url) == 0;
|
|
|
53
|
+ move("$tmpdir/$file", "$weblogsdir/$file");
|
|
|
54
|
+ }
|
|
|
55
|
+}
|
|
|
56
|
+
|
|
|
57
|
+sub parse_log_file {
|
|
|
58
|
+ return unless -f $File::Find::name;
|
|
|
59
|
+ return unless $File::Find::name =~ m/\.xz$/;
|
|
|
60
|
+
|
|
|
61
|
+ print STDERR "Reading $_\n";
|
|
|
62
|
+
|
|
|
63
|
+ my ($stdout, undef, $success) = capture_exec('xzcat', $File::Find::name);
|
|
|
64
|
+ foreach my $line (split /\n/, $stdout) {
|
|
|
65
|
+ my ($version, $os) =
|
|
|
66
|
+ ($line =~ m|^[^\s]+ - - \[.+\] "GET /aus1/torbrowser/([^/]+)/tor-browser-(.*)-\d.*_ALL\.mar HTTP/|);
|
|
|
67
|
+ if ($version) {
|
|
|
68
|
+ $downloads{$version}{total} += 1;
|
|
|
69
|
+ $downloads{$version}{full_update} += 1;
|
|
|
70
|
+ $downloads{$version}{OS}{$os} += 1;
|
|
|
71
|
+ next;
|
|
|
72
|
+ }
|
|
|
73
|
+ my $incremental_from;
|
|
|
74
|
+ ($version, $os, $incremental_from) =
|
|
|
75
|
+ ($line =~ m|^[^\s]+ - - \[.+\] "GET /aus1/torbrowser/([^/]+)/tor-browser-(.*)--(\d[^-]+)-.*_ALL\.incremental\.mar HTTP/|);
|
|
|
76
|
+ if ($incremental_from) {
|
|
|
77
|
+ $downloads{$version}{total} += 1;
|
|
|
78
|
+ $downloads{$version}{OS}{$os} += 1;
|
|
|
79
|
+ $downloads{$version}{incremental_update_total} += 1;
|
|
|
80
|
+ $downloads{$version}{incremental_update}{$incremental_from} += 1;
|
|
|
81
|
+ }
|
|
|
82
|
+ }
|
|
|
83
|
+}
|
|
|
84
|
+
|
|
|
85
|
+sub parse_log_files {
|
|
|
86
|
+ my $tmpdir = tempdir(CLEANUP => 1);
|
|
|
87
|
+ foreach my $month (@months) {
|
|
|
88
|
+ my $file = "webstats-$month.tar";
|
|
|
89
|
+ print STDERR "Extracting $file\n";
|
|
|
90
|
+ my (undef, undef, $success) = capture_exec('tar', '-C', $tmpdir, '-xf',
|
|
|
91
|
+ $file, "webstats-$month/cdn.torproject.org");
|
|
|
92
|
+
|
|
|
93
|
+ find(\&parse_log_file, "$tmpdir/webstats-$month/cdn.torproject.org");
|
|
|
94
|
+ }
|
|
|
95
|
+}
|
|
|
96
|
+
|
|
|
97
|
+sub print_results {
|
|
|
98
|
+ print 'Download numbers for months ', join(' ', @months), ":\n\n";
|
|
|
99
|
+ my @versions = sort { $downloads{$b}{total} <=> $downloads{$a}{total} }
|
|
|
100
|
+ keys %downloads;
|
|
|
101
|
+ foreach my $version (@versions) {
|
|
|
102
|
+ print "Version: $version\n";
|
|
|
103
|
+ print " Total: $downloads{$version}{total}\n";
|
|
|
104
|
+ print " Per OS:\n";
|
|
|
105
|
+ foreach my $os (sort keys %{$downloads{$version}{OS}}) {
|
|
|
106
|
+ print " $os: $downloads{$version}{OS}{$os}\n";
|
|
|
107
|
+ }
|
|
|
108
|
+ print " Full updates: $downloads{$version}{full_update}\n"
|
|
|
109
|
+ if $downloads{$version}{full_update};
|
|
|
110
|
+ if ($downloads{$version}{incremental_update_total}) {
|
|
|
111
|
+ print " Incremental updates (total): $downloads{$version}{incremental_update_total}\n";
|
|
|
112
|
+ foreach my $incremental_from (sort keys %{$downloads{$version}{incremental_update}}) {
|
|
|
113
|
+ print " Incremental updates from $incremental_from: $downloads{$version}{incremental_update}{$incremental_from}\n";
|
|
|
114
|
+ }
|
|
|
115
|
+ }
|
|
|
116
|
+ print "\n";
|
|
|
117
|
+ }
|
|
|
118
|
+}
|
|
|
119
|
+
|
|
|
120
|
+download_log_files;
|
|
|
121
|
+parse_log_files;
|
|
|
122
|
+print_results; |