From: Mischa POSLAWSKY <perl@shiar.org>
Date: Thu, 27 Apr 2017 00:33:39 +0000 (+0200)
Subject: browser: new mkusage-wikipedia to parse analytics tsv
X-Git-Tag: v1.10~63
X-Git-Url: http://git.shiar.nl/sheet.git/commitdiff_plain/916e4f76f6d478e51b42b25f5c9e0ec7373749e5?ds=sidebyside

browser: new mkusage-wikipedia to parse analytics tsv

Amazing export from analytics.wikimedia.org continuing previous squid stats
but all nicely prepared, only needing translation to caniuse agent identifiers.
Contains all samples, so restrict by year for now.
---

diff --git a/tools/mkusage-wikimedia b/tools/mkusage-wikimedia
new file mode 100755
index 0000000..5e264ed
--- /dev/null
+++ b/tools/mkusage-wikimedia
@@ -0,0 +1,65 @@
+#!/usr/bin/perl
+use 5.014;
+use warnings;
+
+use Data::Dump 'pp';
+
+our $VERSION = '1.00';
+
+my %BROWSERID = qw(
+	IE                      ie
+	IE-Mobile               ie_mob
+	Edge                    edge
+	Edge-Mobile             edge
+	Firefox                 firefox
+	Firefox-Mobile          and_ff
+	Safari                  safari
+	Mobile-Safari           ios_saf
+	Mobile-Safari-UIWebView ios_saf
+	Chrome                  chrome
+	Chromium                chrome
+	Chrome-Mobile           and_chr
+	Chrome-Mobile-iOS       and_chr
+	Android                 android
+	Opera                   opera
+	Opera-Mini              op_mini
+	BlackBerry-WebKit       bb
+	UC-Browser              and_uc
+);
+
+my %count = (
+	-title  => 'Wikimedia',
+	-site   => 'https://analytics.wikimedia.org/',
+);
+
+my $recent = qr/^2017-/;
+
+(readline =~ y/\t//) == 3 or die "unexpected amount of columns in header\n";
+
+while (my $row = readline) {
+	my ($date, $name, $version, $pct) = split /\t/, $row;
+	$date =~ $recent or next;
+	$name =~ y/ /-/;
+	my $browser = $BROWSERID{$name} or next;
+	$count{$browser}{$version} += $pct;
+	$count{-total} += $pct;
+	$count{-date}->{$date}++;
+}
+
+$_ = join ' to ', (sort keys %{$_})[0, -1] for $count{-date};
+
+my $mult = 100 / delete $count{-total};
+for (values %count) {
+	ref $_ eq 'HASH' or next;
+	$_ *= $mult for values %{$_};
+}
+
+say '+', pp(\%count);
+
+__END__
+
+=head1 USAGE
+
+	curl https://analytics.wikimedia.org/datasets/periodic/reports/metrics/browser/all_sites_by_browser_family_and_major_percent.tsv |
+	./mkusage-wikimedia >browser-usage.inc.pl
+