Revision history for AduC812
Additions:
This is a diff code to add in a 1.3-gettext version a link markup of camelcase WikiNames, plain URLs and interwiki links in unicode, such as ** ВикиИмя **, ** http://кремль.рф ** and ** ВикиПедия:Формат **. For a latter ones to work one should add a desired interwiki header to a [[http://docs.wikkawiki.org/interwiki interwiki list ]]
%%(diff)
--- 1.3_gettext_original/formatters/wakka.php 2010-11-16 19:37:02.000000000 +0200
+++ 1.3_gettext/formatters/wakka.php 2010-11-21 11:02:30.000000000 +0200
@@ -573,7 +573,7 @@
return (++$trigger_center % 2 ? "<div class=\"center\">\n" : "\n</div>\n");
}
// urls (see RFC 1738 <http://www.ietf.org/rfc/rfc1738.txt>)
- elseif (preg_match("/^([a-z]+:\/\/[[:alnum:]\/?;:@&=\.]+[[:alnum:]\/])(.*)$/", $thing, $matches))
+ elseif (preg_match("/^([a-z]+:\/\/[[:alnum:]\p{L}\/?;:@&=\.]+[[:alnum:]\p{L}\/])(.*)$/u", $thing, $matches))
{
$url = $matches[1];
/* Inline images are disabled for security reason, use {{image action}} #142
@@ -788,7 +788,7 @@
// Case 2: Deprecated...(first part is a string
// followed by one or more whitespaces)
else if(preg_match("/^(.*?)\s+([^|]+)$/s", $contents, $matches) &&
- preg_match("/^[A-ZÄÖÜa-zßäöü][A-Za-z0-9ÄÖÜßäöü]*$/", $matches[1]))
+ preg_match("/^[\p{Lu}][\p{L}0-9]*$/u", $matches[1]))
{
$url = $matches[1];
$text = $matches[2];
@@ -984,12 +984,12 @@
}
}
// interwiki links!
- elseif (preg_match("/^[A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+[:]\S*$/s", $thing))
+ else if (preg_match("/^[\p{Lu}][\p{L}]+[:]\S*$/su", $thing))
{
return $wakka->Link($thing);
}
// wiki links!
- elseif (preg_match("/^[A-ZÄÖÜ]+[a-zßäöü]+[A-Z0-9ÄÖÜ][A-Za-z0-9ÄÖÜßäöü]*$/s", $thing))
+ else if (preg_match("/^[\p{Lu}]+[\p{Ll}]+[\p{Lu}0-9][\p{L}0-9]*$/su", $thing))
{
return $wakka->Link($thing);
}
@@ -1088,8 +1088,8 @@
"\[\[[^\[]*?\]\]|".
# forced linebreak and hr
"-{3,}|".
- # URL
- "\b[a-z]+:\/\/\S+|".
+ # URL - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)[\p{Ll}]+:\/\/\S+|".
# Wiki markup
"\*\*|\'\'|\#\#|\#\%|@@|::c::|\>\>|\<\<|££|¥¥|\+\+|__|<|>|\/\/|".
# headings
@@ -1100,15 +1100,15 @@
"\|(?:[^\|])?\|(?:\(.*?\))?(?:\{[^\{\}]*?\})?(?:\n)?|".
# action
"\{\{.*?\}\}|".
- # InterWiki link
- "\b[A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+[:](?![=_])\S*\b|".
- # CamelWords
- "\b([A-ZÄÖÜ]+[a-zßäöü]+[A-Z0-9ÄÖÜ][A-Za-z0-9ÄÖÜßäöü]*)\b|".
+ # InterWiki link - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)[\p{Lu}][\p{L}]+[:](?![=_])\S*(?!\pL)|".
+ # CamelWords - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)([\p{Lu}]+[\p{Ll}]+[\p{Lu}0-9][\p{L}0-9]*)(?!\pL)|".
#ampersands! Track single ampersands or any htmlentity-like (&...;)
'\\&([#a-zA-Z0-9]+;)?|'.
# newline
"\n".
- "/ms", "wakka2callback", $text."\n"); #append \n (#444)
+ "/msu", "wakka2callback", $text."\n"); #append \n (#444) #append /u to enable unicode
// we're cutting the last <br />
$text = preg_replace("/<br \/>$/","", $text);
--- 1.3_gettext_original/libs/Wakka.class.php 2010-11-16 19:37:05.000000000 +0200
+++ 1.3_gettext/libs/Wakka.class.php 2010-11-16 20:20:07.000000000 +0200
@@ -2313,7 +2313,7 @@
// is this an interwiki link?
// before the : should be a WikiName; anything after can be (nearly) anything that's allowed in a URL
- if (preg_match('/^([A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+)[:](\S*)$/', $tag, $matches)) // @@@ FIXME #34 (inconsistent with Formatter)
+ if (preg_match('/^([\p{Lu}_][\p{Lu}\p{Ll}_]+)[:](\S*)$/u', $tag, $matches)) // @@@ FIXME #34 (inconsistent with Formatter)
{
$url = $this->GetInterWikiUrl($matches[1], $matches[2]);
$class = 'interwiki';
%%
%%(diff)
--- 1.3_gettext_original/formatters/wakka.php 2010-11-16 19:37:02.000000000 +0200
+++ 1.3_gettext/formatters/wakka.php 2010-11-21 11:02:30.000000000 +0200
@@ -573,7 +573,7 @@
return (++$trigger_center % 2 ? "<div class=\"center\">\n" : "\n</div>\n");
}
// urls (see RFC 1738 <http://www.ietf.org/rfc/rfc1738.txt>)
- elseif (preg_match("/^([a-z]+:\/\/[[:alnum:]\/?;:@&=\.]+[[:alnum:]\/])(.*)$/", $thing, $matches))
+ elseif (preg_match("/^([a-z]+:\/\/[[:alnum:]\p{L}\/?;:@&=\.]+[[:alnum:]\p{L}\/])(.*)$/u", $thing, $matches))
{
$url = $matches[1];
/* Inline images are disabled for security reason, use {{image action}} #142
@@ -788,7 +788,7 @@
// Case 2: Deprecated...(first part is a string
// followed by one or more whitespaces)
else if(preg_match("/^(.*?)\s+([^|]+)$/s", $contents, $matches) &&
- preg_match("/^[A-ZÄÖÜa-zßäöü][A-Za-z0-9ÄÖÜßäöü]*$/", $matches[1]))
+ preg_match("/^[\p{Lu}][\p{L}0-9]*$/u", $matches[1]))
{
$url = $matches[1];
$text = $matches[2];
@@ -984,12 +984,12 @@
}
}
// interwiki links!
- elseif (preg_match("/^[A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+[:]\S*$/s", $thing))
+ else if (preg_match("/^[\p{Lu}][\p{L}]+[:]\S*$/su", $thing))
{
return $wakka->Link($thing);
}
// wiki links!
- elseif (preg_match("/^[A-ZÄÖÜ]+[a-zßäöü]+[A-Z0-9ÄÖÜ][A-Za-z0-9ÄÖÜßäöü]*$/s", $thing))
+ else if (preg_match("/^[\p{Lu}]+[\p{Ll}]+[\p{Lu}0-9][\p{L}0-9]*$/su", $thing))
{
return $wakka->Link($thing);
}
@@ -1088,8 +1088,8 @@
"\[\[[^\[]*?\]\]|".
# forced linebreak and hr
"-{3,}|".
- # URL
- "\b[a-z]+:\/\/\S+|".
+ # URL - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)[\p{Ll}]+:\/\/\S+|".
# Wiki markup
"\*\*|\'\'|\#\#|\#\%|@@|::c::|\>\>|\<\<|££|¥¥|\+\+|__|<|>|\/\/|".
# headings
@@ -1100,15 +1100,15 @@
"\|(?:[^\|])?\|(?:\(.*?\))?(?:\{[^\{\}]*?\})?(?:\n)?|".
# action
"\{\{.*?\}\}|".
- # InterWiki link
- "\b[A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+[:](?![=_])\S*\b|".
- # CamelWords
- "\b([A-ZÄÖÜ]+[a-zßäöü]+[A-Z0-9ÄÖÜ][A-Za-z0-9ÄÖÜßäöü]*)\b|".
+ # InterWiki link - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)[\p{Lu}][\p{L}]+[:](?![=_])\S*(?!\pL)|".
+ # CamelWords - can also be unicode - ADuC812 16.11.2010
+ "(?<!\pL)([\p{Lu}]+[\p{Ll}]+[\p{Lu}0-9][\p{L}0-9]*)(?!\pL)|".
#ampersands! Track single ampersands or any htmlentity-like (&...;)
'\\&([#a-zA-Z0-9]+;)?|'.
# newline
"\n".
- "/ms", "wakka2callback", $text."\n"); #append \n (#444)
+ "/msu", "wakka2callback", $text."\n"); #append \n (#444) #append /u to enable unicode
// we're cutting the last <br />
$text = preg_replace("/<br \/>$/","", $text);
--- 1.3_gettext_original/libs/Wakka.class.php 2010-11-16 19:37:05.000000000 +0200
+++ 1.3_gettext/libs/Wakka.class.php 2010-11-16 20:20:07.000000000 +0200
@@ -2313,7 +2313,7 @@
// is this an interwiki link?
// before the : should be a WikiName; anything after can be (nearly) anything that's allowed in a URL
- if (preg_match('/^([A-ZÄÖÜ][A-Za-zÄÖÜßäöü]+)[:](\S*)$/', $tag, $matches)) // @@@ FIXME #34 (inconsistent with Formatter)
+ if (preg_match('/^([\p{Lu}_][\p{Lu}\p{Ll}_]+)[:](\S*)$/u', $tag, $matches)) // @@@ FIXME #34 (inconsistent with Formatter)
{
$url = $this->GetInterWikiUrl($matches[1], $matches[2]);
$class = 'interwiki';
%%
