2016-01-04 10:45:54 +01:00
|
|
|
<?php
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Extract title from an HTML document.
|
|
|
|
*
|
|
|
|
* @param string $html HTML content where to look for a title.
|
|
|
|
*
|
|
|
|
* @return bool|string Extracted title if found, false otherwise.
|
|
|
|
*/
|
|
|
|
function html_extract_title($html)
|
|
|
|
{
|
2016-04-06 22:00:52 +02:00
|
|
|
if (preg_match('!<title.*?>(.*?)</title>!is', $html, $matches)) {
|
|
|
|
return trim(str_replace("\n", '', $matches[1]));
|
2016-01-04 10:45:54 +01:00
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine charset from downloaded page.
|
|
|
|
* Priority:
|
|
|
|
* 1. HTTP headers (Content type).
|
|
|
|
* 2. HTML content page (tag <meta charset>).
|
|
|
|
* 3. Use a default charset (default: UTF-8).
|
|
|
|
*
|
|
|
|
* @param array $headers HTTP headers array.
|
|
|
|
* @param string $htmlContent HTML content where to look for charset.
|
|
|
|
* @param string $defaultCharset Default charset to apply if other methods failed.
|
|
|
|
*
|
|
|
|
* @return string Determined charset.
|
|
|
|
*/
|
|
|
|
function get_charset($headers, $htmlContent, $defaultCharset = 'utf-8')
|
|
|
|
{
|
|
|
|
if ($charset = headers_extract_charset($headers)) {
|
|
|
|
return $charset;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ($charset = html_extract_charset($htmlContent)) {
|
|
|
|
return $charset;
|
|
|
|
}
|
|
|
|
|
|
|
|
return $defaultCharset;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Extract charset from HTTP headers if it's defined.
|
|
|
|
*
|
|
|
|
* @param array $headers HTTP headers array.
|
|
|
|
*
|
|
|
|
* @return bool|string Charset string if found (lowercase), false otherwise.
|
|
|
|
*/
|
|
|
|
function headers_extract_charset($headers)
|
|
|
|
{
|
|
|
|
if (! empty($headers['Content-Type']) && strpos($headers['Content-Type'], 'charset=') !== false) {
|
|
|
|
preg_match('/charset="?([^; ]+)/i', $headers['Content-Type'], $match);
|
|
|
|
if (! empty($match[1])) {
|
|
|
|
return strtolower(trim($match[1]));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Extract charset HTML content (tag <meta charset>).
|
|
|
|
*
|
|
|
|
* @param string $html HTML content where to look for charset.
|
|
|
|
*
|
|
|
|
* @return bool|string Charset string if found, false otherwise.
|
|
|
|
*/
|
|
|
|
function html_extract_charset($html)
|
|
|
|
{
|
|
|
|
// Get encoding specified in HTML header.
|
2016-04-06 22:00:52 +02:00
|
|
|
preg_match('#<meta .*charset=["\']?([^";\'>/]+)["\']? */?>#Usi', $html, $enc);
|
2016-01-04 10:45:54 +01:00
|
|
|
if (!empty($enc[1])) {
|
|
|
|
return strtolower($enc[1]);
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
2016-05-11 00:05:22 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Count private links in given linklist.
|
|
|
|
*
|
2016-10-20 11:31:52 +02:00
|
|
|
* @param array|Countable $links Linklist.
|
2016-05-11 00:05:22 +02:00
|
|
|
*
|
|
|
|
* @return int Number of private links.
|
|
|
|
*/
|
|
|
|
function count_private($links)
|
|
|
|
{
|
|
|
|
$cpt = 0;
|
|
|
|
foreach ($links as $link) {
|
2017-01-06 18:54:29 +01:00
|
|
|
if ($link['private']) {
|
|
|
|
$cpt += 1;
|
|
|
|
}
|
2016-05-11 00:05:22 +02:00
|
|
|
}
|
2016-05-10 23:18:04 +02:00
|
|
|
|
2016-05-11 00:05:22 +02:00
|
|
|
return $cpt;
|
|
|
|
}
|
2016-05-10 23:18:04 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* In a string, converts URLs to clickable links.
|
|
|
|
*
|
|
|
|
* @param string $text input string.
|
|
|
|
* @param string $redirector if a redirector is set, use it to gerenate links.
|
|
|
|
*
|
|
|
|
* @return string returns $text with all links converted to HTML links.
|
|
|
|
*
|
|
|
|
* @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
|
|
|
|
*/
|
|
|
|
function text2clickable($text, $redirector = '')
|
|
|
|
{
|
2017-09-29 18:52:38 +02:00
|
|
|
$regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
|
2016-05-10 23:18:04 +02:00
|
|
|
|
|
|
|
if (empty($redirector)) {
|
|
|
|
return preg_replace($regex, '<a href="$1">$1</a>', $text);
|
|
|
|
}
|
|
|
|
// Redirector is set, urlencode the final URL.
|
|
|
|
return preg_replace_callback(
|
|
|
|
$regex,
|
|
|
|
function ($matches) use ($redirector) {
|
|
|
|
return '<a href="' . $redirector . urlencode($matches[1]) .'">'. $matches[1] .'</a>';
|
|
|
|
},
|
|
|
|
$text
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Auto-link hashtags.
|
|
|
|
*
|
|
|
|
* @param string $description Given description.
|
|
|
|
* @param string $indexUrl Root URL.
|
|
|
|
*
|
|
|
|
* @return string Description with auto-linked hashtags.
|
|
|
|
*/
|
|
|
|
function hashtag_autolink($description, $indexUrl = '')
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* To support unicode: http://stackoverflow.com/a/35498078/1484919
|
|
|
|
* \p{Pc} - to match underscore
|
|
|
|
* \p{N} - numeric character in any script
|
|
|
|
* \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<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
|
|
|
|
return preg_replace($regex, $replacement, $description);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* This function inserts where relevant so that multiple spaces are properly displayed in HTML
|
|
|
|
* even in the absence of <pre> (This is used in description to keep text formatting).
|
|
|
|
*
|
|
|
|
* @param string $text input text.
|
|
|
|
*
|
|
|
|
* @return string formatted text.
|
|
|
|
*/
|
|
|
|
function space2nbsp($text)
|
|
|
|
{
|
|
|
|
return preg_replace('/(^| ) /m', '$1 ', $text);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Format Shaarli's description
|
|
|
|
*
|
|
|
|
* @param string $description shaare's description.
|
|
|
|
* @param string $redirector if a redirector is set, use it to gerenate links.
|
2016-10-20 11:31:52 +02:00
|
|
|
* @param string $indexUrl URL to Shaarli's index.
|
2016-05-10 23:18:04 +02:00
|
|
|
*
|
|
|
|
* @return string formatted description.
|
|
|
|
*/
|
|
|
|
function format_description($description, $redirector = '', $indexUrl = '') {
|
|
|
|
return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector), $indexUrl)));
|
|
|
|
}
|
2016-11-28 18:24:15 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Generate a small hash for a link.
|
|
|
|
*
|
|
|
|
* @param DateTime $date Link creation date.
|
|
|
|
* @param int $id Link ID.
|
|
|
|
*
|
|
|
|
* @return string the small hash generated from link data.
|
|
|
|
*/
|
|
|
|
function link_small_hash($date, $id)
|
|
|
|
{
|
|
|
|
return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id);
|
|
|
|
}
|