From 7a124e16ca606eb8e170dafca88a4128a35df9ec Mon Sep 17 00:00:00 2001
From: Jack-Benny Persson <jack-benny@cyberinfo.se>
Date: Mon, 4 Aug 2014 14:05:03 +0200
Subject: [PATCH] Added a check and a function to handle UTF-8 article strings

---
 HISTORY |  9 +++++++--
 wie.php | 38 +++++++++++++++++++++++++++++---------
 2 files changed, 36 insertions(+), 11 deletions(-)

diff --git a/HISTORY b/HISTORY
index 48ede65..93871c6 100644
--- a/HISTORY
+++ b/HISTORY
@@ -1,6 +1,11 @@
-0.1 - 2014-07-28
-    First release of wie
+0.3 - 2014-08-04
+    Added a check to see whatever the article string is in
+    UTF-8 or not. If it is in UTF-8, use a specific function
+    to handle it. Regular 'ucwords' can't handle UTF-8.
 
 0.2 - 2014-07-31
     Changed 'shell_exec(curl -s $URL)' to PHP5 cURL function
     Added HISTORY file
+
+0.1 - 2014-07-28
+    First release of wie
diff --git a/wie.php b/wie.php
index 84dce4c..d7a6b00 100755
--- a/wie.php
+++ b/wie.php
@@ -20,14 +20,9 @@
 
 $defaultLang = "en"; // default language
 $progName = $argv[0];
-function usage()
-{
-    print "Wikipedia ingress extractor (wie), version 0.2\n";
-    print "Usage: $GLOBALS[progName] [--lang=sv] article\n";
-    print "Default language if none specified is $GLOBALS[defaultLang].\n";
-    print "Remember to quote the article if there's more than one word,\n";
-    print "for example Roger Bacon as 'Roger Bacon'.\n";
-}
+
+
+
 // check if no argument was specified
 if (!isset($argv[1]))
 {
@@ -55,7 +50,16 @@ else
     $article = $argv[1];
 }
 
-$article = ucwords($article); // uppercase article
+// check if article is UTF-8 encoded, in which case regular ucwords won't work
+if (mb_check_encoding($article, 'UTF-8'))
+{
+    $article = utf8_ucwords($article);
+}
+else
+{
+    $article = ucwords($article); // uppercase article
+}
+
 $article = preg_replace("/\s/", "_" ,$article); // make spaces to underscore
 $url = "http://$lang.wikipedia.org/wiki/$article"; 
 
@@ -78,4 +82,20 @@ if (!isset($match[1]))
 $string = strip_tags($match[1]); 
 print (wordwrap($string, 65, "\n") . "\n"); 
 
+// misc functions
+function usage()
+{
+    print "Wikipedia ingress extractor (wie), version 0.2\n";
+    print "Usage: $GLOBALS[progName] [--lang=sv] article\n";
+    print "Default language if none specified is $GLOBALS[defaultLang].\n";
+    print "Remember to quote the article if there's more than one word,\n";
+    print "for example Roger Bacon as 'Roger Bacon'.\n";
+}
+
+function utf8_ucwords($str)
+{
+    $str = mb_convert_case($str, MB_CASE_TITLE, "UTF-8");
+    return $str;    
+}
+
 ?>