From: Mischa POSLAWSKY <perl@shiar.org>
Date: Mon, 27 May 2019 20:33:38 +0000 (+0200)
Subject: browser: dynamic include of last 3 wikimedia dates
X-Git-Tag: v1.12~33
X-Git-Url: http://git.shiar.nl/sheet.git/commitdiff_plain/efd1521b9205af6af3427c8191eaa7bc2a6cc137

browser: dynamic include of last 3 wikimedia dates

Replace hardcoded year by a comparison of the 3rd most recent day.

Loads the entire input in memory for simplicity, which is currently only 1MB
so do not care.
---

diff --git a/tools/mkusage-wikimedia b/tools/mkusage-wikimedia
index 7a59ba5..4470c50 100755
--- a/tools/mkusage-wikimedia
+++ b/tools/mkusage-wikimedia
@@ -34,14 +34,21 @@ my %count = (
 	-site   => 'https://analytics.wikimedia.org/',
 );
 
-my $recent = qr/^2019-/;
-
 (readline =~ y/\t//) == 3 or die "unexpected amount of columns in header\n";
+my @lines = readline;
+
+my $recent;  # minimum date to include
+for (reverse @lines) {
+	my ($date) = /(\S+)/;
+	$recent ne $date or next if $recent;  # same day
+	$recent = $date;  # override older date
+	last if state $i++ >= 2;  # repeat twice
+}
 
-while (my $row = readline) {
+for my $row (@lines) {
 	$row =~ s/\r?\n\z//;
 	my ($date, $name, $version, $pct) = split /\t/, $row;
-	$date =~ $recent or next;
+	$date ge $recent or next;
 	$name =~ y/ /-/;
 	my $browser = $BROWSERID{$name};
 	if (not $browser) {