Changeset 541 for trunk/lib/Utilities.inc.php

Timestamp:

Aug 12, 2015 12:22:54 AM (9 years ago)

Author:

anonymous

Message:

v2.2.0-3: Fixed auth password hashing verification issues. Updated hyperlinkTxt() with option. Updated tests.

File:

• trunk/lib/Utilities.inc.php (modified) (2 diffs)

trunk/lib/Utilities.inc.php

@@ -215 +215 @@
 * @access   public
 * @param    string  $text   Text to search for URLs.
+* @param    bool    $strict True to only include URLs starting with a scheme (http:// ftp:// im://), or false to include URLs starting with 'www.'.
 * @param    mixed   $length Number of characters to truncate URL, or NULL to disable truncating.
 * @param    string  $delim  Delimiter to append, indicate truncation.
 * @return   string          Same input text, but URLs hyperlinked.
 * @author   Quinn Comendant <quinn@strangecode.com>
-* @version  1.0
+* @version  2.0
 * @since    22 Mar 2015 23:29:04
 */
-function hyperlinkTxt($text, $length=null, $delim='
')
-{
-    return preg_replace_callback(
-        // Inspired by @stephenhay's regex from https://mathiasbynens.be/demo/url-regex
-        // Here we capture the full URL into the first match and only the first X characters into the second match.
-        sprintf('@\b(?<!")(?<!\')(?<!=)(((?:https?|s?ftps?)://[^\s/$.?#].[^\s]{0,%s})[^\s]*)@iS', $length),
-        // Use an anonymous function to decide when to append the delim.
-        // Also encode special chars with oTxt().
-        function ($m) use ($length, $delim) {
-            if (is_null($length) || $m[1] == $m[2]) {
-                // If not truncating, or URL was not truncated.
-                return sprintf('<a href="%s">%s</a>', oTxt($m[1]), oTxt($m[1]));
-            } else {
-                // Truncated URL.
-                return sprintf('<a href="%s">%s%s</a>', oTxt($m[1]), oTxt(trim($m[2])), $delim);
-            }
-        },
-        $text
+function hyperlinkTxt($text, $strict=false, $length=null, $delim='
')
+{
+    // Capture the full URL into the first match and only the first X characters into the second match.
+    // This will match URLs not preceeded by " ' or = (URLs inside an attribute) or ` (Markdown quoted) or double-scheme (http://http://www.asdf.com)
+    // Valid URL characters: ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-._~:/?#[]@!$&'()*+,;=
+    $regex = '@
+        \b                              # Start with a word-boundary.
+        (?<!"|\'|=|>|`|[\w-]{2}://)     # Negative look-behind to exclude URLs already in <a> tag, Markdown quoted, or double SCHEME://
+        (                               # Begin match 1
+            (                           # Begin match 2
+                (?:[\w-]{2,}://%s)      # URL starts with SCHEME:// or www. (if strict = false)
+                [^\s/$.?#]+             # Any domain-valid characters
+                \.                      # At least one point
+                [^\s"`<>]{1,%s}         # Match 2 is limited to a maximum of LENGTH valid URL characters
+            )
+            [^\s"`<>]*                  # Match 1 continues with any further valid URL characters
+            [^\P{Any}%s\s
<>«»"—–]      # Final character not a space or common end-of-sentence punctuation (.,:;?!, etc). Using double negation set, see http://stackoverflow.com/a/4786560/277303
+        )
+        @Suxi
+    ';
+    $regex = sprintf($regex,
+        ($strict ? '' : '|www\.'), // Strict=false allows URLs beginning with www.
+        $length,
+        ($strict ? '' : '?!.,:;)\'-') // Strict=false excludes these characters from set of the last character of URL.
     );
+
+    // Use a callback function to decide when to append the delim.
+    // Also encode special chars with oTxt().
+    return preg_replace_callback($regex, function ($m) use ($length, $delim) {
+        $url = $m[1];
+        $truncated_url = $m[2];
+        $absolute_url = preg_replace('!^www\.!', 'http://www.', $url);
+        if (is_null($length) || $url == $truncated_url) {
+            // If not truncating, or URL was not truncated.
+            $display_url = preg_replace('!^[\w-]{2,}://!', '', $url);
+            return sprintf('<a href="%s">%s</a>', oTxt($absolute_url), $display_url);
+        } else {
+            // Truncated URL.
+            $display_url = preg_replace('!^[\w-]{2,}://!', '', trim($truncated_url));
+            return sprintf('<a href="%s">%s%s</a>', oTxt($absolute_url), $display_url, $delim);
+        }
+    }, $text);
 }
 
@@ -452 +476 @@
 function URLSlug($str)
 {
-    $slug = preg_replace(array('/[^\w]+/', '/^-+|-+$/'), array('-', ''), $str);
+    $slug = preg_replace(array('/\W+/u', '/^-+|-+$/'), array('-', ''), $str);
     $slug = strtolower($slug);
     return $slug;