update to 3.2 version of full-text-rss, issue #694

[github/wallabag/wallabag.git] / inc / 3rdparty / makefulltextfeedHelpers.php

diff --git a/inc/3rdparty/makefulltextfeedHelpers.php b/inc/3rdparty/makefulltextfeedHelpers.php
index 1c11b8f6b2345ea15adb93496ed97b2686a9cda6..4e985372c27b7f0780d54535acf3401b6c24087b 100755 (executable)
--- a/inc/3rdparty/makefulltextfeedHelpers.php
+++ b/inc/3rdparty/makefulltextfeedHelpers.php
@@ -66,6 +66,38 @@ class DummySingleItem {
 // HELPER FUNCTIONS
 ///////////////////////////////
 
+// Adapted from WordPress\r
+// http://core.trac.wordpress.org/browser/tags/3.5.1/wp-includes/formatting.php#L2173\r
+function get_excerpt($text, $num_words=55, $more=null) {\r
+       if (null === $more) $more = '…';\r
+       $text = strip_tags($text);\r
+       //TODO: Check if word count is based on single characters (East Asian characters)\r
+       /*\r
+       if (1==2) {\r
+       $text = trim(preg_replace("/[\n\r\t ]+/", ' ', $text), ' ');\r
+       preg_match_all('/./u', $text, $words_array);\r
+       $words_array = array_slice($words_array[0], 0, $num_words + 1);\r
+       $sep = '';\r
+       } else {\r
+       $words_array = preg_split("/[\n\r\t ]+/", $text, $num_words + 1, PREG_SPLIT_NO_EMPTY);\r
+       $sep = ' ';\r
+       }\r
+       */\r
+       $words_array = preg_split("/[\n\r\t ]+/", $text, $num_words + 1, PREG_SPLIT_NO_EMPTY);\r
+       $sep = ' ';\r
+       if (count($words_array) > $num_words) {\r
+               array_pop($words_array);\r
+               $text = implode($sep, $words_array);\r
+               $text = $text.$more;\r
+       } else {\r
+               $text = implode($sep, $words_array);\r
+       }\r
+       // trim whitespace at beginning or end of string\r
+       // See: http://stackoverflow.com/questions/4166896/trim-unicode-whitespace-in-php-5-2\r
+       $text = preg_replace('/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $text);\r
+       return $text;\r
+}\r
+\r
 function url_allowed($url) {
        global $options;
        if (!empty($options->allowed_urls)) {
@@ -165,14 +197,6 @@ function convert_to_utf8($html, $header=null)
                        if (strtolower($encoding) != 'utf-8') {
                                debug('Converting to UTF-8');
                                $html = SimplePie_Misc::change_encoding($html, $encoding, 'utf-8');
-                               /*
-                               if (function_exists('iconv')) {
-                                       // iconv appears to handle certain character encodings better than mb_convert_encoding
-                                       $html = iconv($encoding, 'utf-8', $html);
-                               } else {
-                                       $html = mb_convert_encoding($html, 'utf-8', $encoding);
-                               }
-                               */
                        }
                }
        }
@@ -196,7 +220,7 @@ function makeAbsolute($base, $elem) {
 }
 function makeAbsoluteAttr($base, $e, $attr) {
        if ($e->hasAttribute($attr)) {
-               // Trim leading and trailing white space. I don't really like this but 
+               // Trim leading and trailing white space. I don't really like this but
                // unfortunately it does appear on some sites. e.g.  <img src=" /path/to/image.jpg" />
                $url = trim(str_replace('%20', ' ', $e->getAttribute($attr)));
                $url = str_replace(' ', '%20', $url);