Fetch test data for the benchmark

diff --git a/includes/normal/Makefile b/includes/normal/Makefile
index 0443560..0cb5da8 100644 (file)
--- a/includes/normal/Makefile
+++ b/includes/normal/Makefile
@@ -15,7 +15,7 @@ test : testutf8 UtfNormalTest.php UtfNormalData.inc NormalizationTest.txt
 testutf8 : Utf8Test.php UTF-8-test.txt
        $(PHP) Utf8Test.php
 
-bench : UtfNormalData.inc
+bench : UtfNormalData.inc testdata/washington.txt testdata/berlin.txt testdata/tokyo.txt testdata/sociology.txt
        $(PHP) UtfNormalBench.php
 
 clean :
@@ -42,3 +42,15 @@ UnicodeData.txt :
 
 UTF-8-test.txt :
        $(FETCH) http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
+
+testdata/berlin.txt :
+       mkdir -p testdata && wget -U MediaWiki/test -O testdata/berlin.txt "http://de.wikipedia.org/w/wiki.phtml?title=Berlin&oldid=2775712&action=raw"
+
+testdata/washington.txt :
+       mkdir -p testdata && wget -U MediaWiki/test -O testdata/washington.txt "http://en.wikipedia.org/w/wiki.phtml?title=Washington%2C_DC&oldid=6370218&action=raw"
+
+testdata/tokyo.txt :
+       mkdir -p testdata && wget -U MediaWiki/test -O testdata/tokyo.txt "http://ja.wikipedia.org/w/wiki.phtml?title=%E6%9D%B1%E4%BA%AC&oldid=823926&action=raw"
+
+testdata/sociology.txt :
+       mkdir -p testdata && wget -U MediaWiki/test -O testdata/sociology.txt "http://ko.wikipedia.org/w/wiki.phtml?title=%EC%82%AC%ED%9A%8C%ED%95%99&oldid=16409&action=raw"

diff --git a/includes/normal/UtfNormalBench.php b/includes/normal/UtfNormalBench.php
index c37fb46..2e1740c 100644 (file)
--- a/includes/normal/UtfNormalBench.php
+++ b/includes/normal/UtfNormalBench.php
@@ -36,7 +36,7 @@ $testfiles = array(
        'testdata/washington.txt' => 'English text',
        'testdata/berlin.txt' => 'German text',
        'testdata/tokyo.txt' => 'Japanese text',
-       'testdata/byzantium.txt' => 'Korean text'
+       'testdata/sociology.txt' => 'Korean text'
 );
 $normalizer = new UtfNormal;
 foreach( $testfiles as $file => $desc ) {