X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=parse%2Fparse-wiki-countrylist.pl;fp=parse%2Fparse-wiki-countrylist.pl;h=fc0437b7140d7502eab971f59eb8da877f2426b3;hb=9f7b3c1a46383496e6a30e0aad8d4252e1a94fda;hp=0000000000000000000000000000000000000000;hpb=eb9ee3742ba220deebc83771b03ffdc7f6a4d59c;p=ccbs

diff --git a/parse/parse-wiki-countrylist.pl b/parse/parse-wiki-countrylist.pl
new file mode 100644
index 0000000..fc0437b
--- /dev/null
+++ b/parse/parse-wiki-countrylist.pl
@@ -0,0 +1,34 @@
+#! /usr/bin/perl
+use strict;
+use warnings;
+
+# Parses country list from
+# http://en.wikipedia.org/w/index.php?title=List_of_IOC_country_codes&action=edit
+
+print "begin;\n";
+
+while (<>) {
+	m/
+	  ^ \* \s*
+	    ( [A-Z]{3} )                  # country code
+	    \s* - \s*
+	    \[\[
+	    ( ?: .*? \| ) ?               # optional article lookup
+	    ( .*? )                       # country name
+	    \]\]
+	 /x or next;
+	 
+	my ($countrycode, $countryname) = ($1, $2);
+
+	# fix some wikisyntax ickyness :-)
+	$countryname =~ s/\]\]//g;
+	$countryname =~ s/\[\[//g;
+
+	# minimal SQL escaping
+	$countryname =~ s/'/\\'/g;
+
+	printf "INSERT INTO countries (countryname,countrycode) VALUES ('%s','%s');\n",
+		$countryname, $countrycode;
+}
+
+printf "commit;\n";