git-svn-id: https://scm.gforge.inria.fr/authscm/cfourrie/svn/lingwb/MElt/trunk@5739 dc05b511-7f1d-0410-9f1c-d6f32a2df9e4

sxpipe-melt/gl_url.pl

@@ -40,13 +40,13 @@ $a         = qr/[a-zàáâäåãßçèéêëìíîïòóôöøùúûüýÿąćč
 $pre       = qr/(?<=[^\/／\{\@\.．0-9a-zàáâäåãßçèéêëìíîïòóôöøùúûüýÿąćčďęěĺľłńňŕřśšťůźżñA-ZÀÁÂÄÅÆÃÇÈÉÊËÌÍÎÏÒÓÔÖØÙÚÛÜÝŸĄĆČĎĘĚĹĽŁŃŇŔŘŚŠŤŮŹŻÑ_<－-])/o;
 $post      = qr/(?=[^\'\/／\}a-zàáâäåãßçèéêëìíîïòóôöøùúûüýÿąćčďęěĺľłńňŕřśšťůźżñA-ZÀÁÂÄÅÆÃÇÈÉÊËÌÍÎÏÒÓÔÖØÙÚÛÜÝŸĄĆČĎĘĚĹĽŁŃŇŔŘŚŠŤŮŹŻÑ_>－-])/o;
 $dirname   = qr/\s*[\/／]\s*(?:\s(?:_UNDERSCORE|_|\.|．|\~|\?|=)|(?:_UNDERSCORE|_|\.|．|\~|\?|=)\s|[\w－\-\.．\~_\?=%&#])+/o;
-$dom2      = qr/${a}[\w－\-_]+[\.．]/o; # was: $dom2      = qr/${a}[\w\-_]{2,}\./o;
+$dom1      = qr/${a}[\w－\-_]*[\.．]/o; # was: $dom2      = qr/${a}[\w\-_]{2,}\./o;
 $dom2_allow_spaces      = qr/${a}[\w－\-_]+\s*[\.．]\s*/o;
 $domN      = qr/${a}[\w－\-_]+[\.．](?:[\w－\-_]{2,}[\.．])+/o;
 $domN_allow_spaces      = qr/${a}[\w－\-_]+\s*[\.．]\s*(?:[\w－\-_]{2,}\s*[\.．]\s*)+/o;
 $POSTvar   = qr/(?:[a-zA-Z0-9\?=－\-\_]+\s*=(?:\s*[a-zA-Z0-9－\-\?=\_]+)?)/o;
-$refOK     = qr/(?:$domN(?:$safeExt|$unsafeExt)|$dom2$safeExt)(?:$dirname)*(?:\s*[\/／])?(?:\s*\?\s*$POSTvar(?:\s*\&amp;\s*$POSTvar)*)?/o;
-$refOK_allow_spaces     = qr/(?:$domN_allow_spaces(?:$safeExt|$unsafeExt)|$dom2$safeExt)/o;
+$refOK     = qr/(?:$domN(?:$safeExt|$unsafeExt)|$dom1$safeExt)(?:$dirname)*(?:\s*[\/／])?(?:\s*\?\s*$POSTvar(?:\s*\&amp;\s*$POSTvar)*)?/o;
+$refOK_allow_spaces     = qr/(?:$domN_allow_spaces(?:$safeExt|$unsafeExt)|$dom1$safeExt)/o;
 $op        = qr/(?:\<|\&lt;)/o;
 $cl        = qr/(?:\>|\&gt;)/o;
 
@@ -76,7 +76,7 @@ while (<>) {
       s/($protocole)(:\/\/$a+\.$a+\.)\s([a-z]{2,})\b/$1$2$3/go;
       s/([\'\/\.\wàâäãéêèëîïöôùûüÿ-])($protocole):\/\//$1 $2:\/\//go;
       s/\. *(org|net|com|fr)\b/.$1/go;
-      s/((?:$dom2$safeExt|$domN(?:$safeExt|$unsafeExt))[\w-\/\~]*)\s([\w-]*(?:\/[\w-\/]*|[\w-\/]*\.html?))\b/$1$2/go;
+      s/((?:$dom1$safeExt|$domN(?:$safeExt|$unsafeExt))[\w-\/\~]*)\s([\w-]*(?:\/[\w-\/]*|[\w-\/]*\.html?))\b/$1$2/go;
     }
     s/($protocole)(:\/\/[^\s,;]*)[âàäã]([^\s,;]*\.$a)/$1$2a$3/go;
     s/($protocole)(:\/\/[^\s,;]*)[éèêë]([^\s,;]*\.$a)/$1$2e$3/go;
@@ -97,9 +97,9 @@ while (<>) {
     ### urls faciles (deux points dans le nom de domaine ou alors extension autre que .et et .de)
     s/(^|\s)($op?(?:(?:$protocole)\s*:\s*\/\/\s*)?$refOK$cl?)$post/$1\{$2\}_URL/go;
     ### urls de type (http://)?nomsanspoint.(et|de)/deschosesobligatoirement
-    s/(^|\s)($op?(?:(?:$protocole)\s*:\s*\/\/\s*)?$dom2$unsafeExt(?:$dirname)+\/?$cl?)$post/$1\{$2\}_URL/go;
+    s/(^|\s)($op?(?:(?:$protocole)\s*:\s*\/\/\s*)?$dom1$unsafeExt(?:$dirname)+\/?$cl?)$post/$1\{$2\}_URL/go;
     ### urls de type http://nomsanspoint.(et|de)(/deschosesfacultatives)?
-    s/(^|\s)($op?(?:(?:$protocole)\s*:\s*\/\/\s*)$dom2$unsafeExt(?:$dirname)*\/?$cl?)$post/$1\{$2\}_URL/go;
+    s/(^|\s)($op?(?:(?:$protocole)\s*:\s*\/\/\s*)$dom1$unsafeExt(?:$dirname)*\/?$cl?)$post/$1\{$2\}_URL/go;
     s/(\s)\.\}_URL/\}_URL$1\./go || s/\.\}_URL/\}_URL\./go;
     ### twitter
     s/^ (.)([\@\#][a-zA-Z0-9])/ {$1$2}_URL/go; # so-called "fix-replies"
@@ -112,10 +112,10 @@ while (<>) {
     s/$pre($op?(?:(?:$protocole)\s*:\s*[\/／][\/／]\s*)?$refOK$cl?)$post/\{$1\}_URL/go;
     s/$pre($op?(?:$protocole)\s*:\s*[\/／][\/／]\s*$refOK_allow_spaces(?: [\/／])?$cl?)$post/\{$1\}_URL/go;
     ### urls de type (http://)?nomsanspoint.(et|de)/deschosesobligatoirement
-    s/$pre($op?(?:(?:$protocole)\s*:\s*[\/／][\/／]\s*)?$dom2$unsafeExt(?:$dirname)+[\/／]?$cl?)$post/\{$1\}_URL/go;
+    s/$pre($op?(?:(?:$protocole)\s*:\s*[\/／][\/／]\s*)?$dom1$unsafeExt(?:$dirname)+[\/／]?$cl?)$post/\{$1\}_URL/go;
     s/$pre($op?(?:$protocole)\s*:\s*[\/／][\/／]\s*$dom2_allow_spaces$unsafeExt(?: [\/／])?$cl?)$post/\{$1\}_URL/go;
     ### urls de type http://nomsanspoint.(et|de)(/deschosesfacultatives)?
-    s/$pre($op?(?:(?:$protocole)\s*:\s*[\/／][\/／]\s*)$dom2$unsafeExt(?:$dirname)*[\/／]?$cl?)$post/\{$1\}_URL/go;
+    s/$pre($op?(?:(?:$protocole)\s*:\s*[\/／][\/／]\s*)$dom1$unsafeExt(?:$dirname)*[\/／]?$cl?)$post/\{$1\}_URL/go;
     s/(\s)([\.．])\}_URL/\}_URL$1$2/go || s/([\.．])\}_URL/\}_URL$1/go;
     ### twitter
     s/^ (.)([\@\#][a-zA-Z0-9])/ \1{\2}_URL/go; # so-called "fix-replies"