X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2Fbookmark%2FLinkUtils.php;h=8fa2953a78d3e8f18bbcba3824801d859169c16e;hb=9ef8555ad298668bcb8537ccdd2ab6560f44177f;hp=17c379796cd39b239be09459d78e3cad22fdfb0b;hpb=8bbf57a2d0dfe8d8d6416a26f9ff1177c77e806b;p=github%2Fshaarli%2FShaarli.git
diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
index 17c37979..8fa2953a 100644
--- a/application/bookmark/LinkUtils.php
+++ b/application/bookmark/LinkUtils.php
@@ -1,6 +1,7 @@
]+(?:'. $properties .')=(?:'. $orCondition .')[^>]*content=["\'](.*?)["\'].*?>#';
+ $orCondition = '["\']?(?:og:)?' . $tag . '["\']?|["\'][^\'"]*?(?:og:)?' . $tag . '[^\'"]*?[\'"]';
+ // Support quotes in double quoted content, and the other way around
+ $content = 'content=(["\'])((?:(?!\1).)*)\1';
+ // Try to retrieve OpenGraph tag.
+ $ogRegex = '#]+(?:' . $properties . ')=(?:' . $orCondition . ')[^>]*' . $content . '.*?>#';
// If the attributes are not in the order property => content (e.g. Github)
// New regex to keep this readable... more or less.
- $ogRegexReverse = '#]+content=["\'](.*?)["\'][^>]+(?:'. $properties .')=(?:'. $orCondition .').*?>#';
+ $ogRegexReverse = '#]+' . $content . '[^>]+(?:' . $properties . ')=(?:' . $orCondition . ').*?>#';
- if (preg_match($ogRegex, $html, $matches) > 0
+ if (
+ preg_match($ogRegex, $html, $matches) > 0
|| preg_match($ogRegexReverse, $html, $matches) > 0
) {
- return $matches[1];
+ return $matches[2];
}
return false;
@@ -95,7 +99,18 @@ function html_extract_tag($tag, $html)
function text2clickable($text)
{
$regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
- return preg_replace($regex, '$1', $text);
+ $format = function (array $match): string {
+ return '' . $match[1] . ''
+ ;
+ };
+
+ return preg_replace_callback($regex, $format, $text);
}
/**
@@ -108,6 +123,9 @@ function text2clickable($text)
*/
function hashtag_autolink($description, $indexUrl = '')
{
+ $tokens = '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN . ')' .
+ '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE . ')'
+ ;
/*
* To support unicode: http://stackoverflow.com/a/35498078/1484919
* \p{Pc} - to match underscore
@@ -115,9 +133,20 @@ function hashtag_autolink($description, $indexUrl = '')
* \p{L} - letter from any language
* \p{Mn} - any non marking space (accents, umlauts, etc)
*/
- $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
- $replacement = '$1#$2';
- return preg_replace($regex, $replacement, $description);
+ $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}' . $tokens . ']+)/mui';
+ $format = function (array $match) use ($indexUrl): string {
+ $cleanMatch = str_replace(
+ BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN,
+ '',
+ str_replace(BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE, '', $match[2])
+ );
+ return $match[1] . '' .
+ '#' . $match[2] .
+ '';
+ };
+
+ return preg_replace_callback($regex, $format, $description);
}
/**
@@ -176,3 +205,49 @@ function is_note($linkUrl)
{
return isset($linkUrl[0]) && $linkUrl[0] === '?';
}
+
+/**
+ * Extract an array of tags from a given tag string, with provided separator.
+ *
+ * @param string|null $tags String containing a list of tags separated by $separator.
+ * @param string $separator Shaarli's default: ' ' (whitespace)
+ *
+ * @return array List of tags
+ */
+function tags_str2array(?string $tags, string $separator): array
+{
+ // For whitespaces, we use the special \s regex character
+ $separator = $separator === ' ' ? '\s' : $separator;
+
+ return preg_split('/\s*' . $separator . '+\s*/', trim($tags) ?? '', -1, PREG_SPLIT_NO_EMPTY);
+}
+
+/**
+ * Return a tag string with provided separator from a list of tags.
+ * Note that given array is clean up by tags_filter().
+ *
+ * @param array|null $tags List of tags
+ * @param string $separator
+ *
+ * @return string
+ */
+function tags_array2str(?array $tags, string $separator): string
+{
+ return implode($separator, tags_filter($tags, $separator));
+}
+
+/**
+ * Clean an array of tags: trim + remove empty entries
+ *
+ * @param array|null $tags List of tags
+ * @param string $separator
+ *
+ * @return array
+ */
+function tags_filter(?array $tags, string $separator): array
+{
+ $trimDefault = " \t\n\r\0\x0B";
+ return array_values(array_filter(array_map(function (string $entry) use ($separator, $trimDefault): string {
+ return trim($entry, $trimDefault . $separator);
+ }, $tags ?? [])));
+}