src/Model/ItemContent.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Model;
  23
  24 use Friendica\Content\Text;
  25 use Friendica\Content\Text\BBCode;
  26 use Friendica\Core\Protocol;
  27 use Friendica\Database\DBA;
  28 use Friendica\DI;
  29
  30 class ItemContent
  31 {
  32         /**
  33          * Search posts for given content
  34          *
  35          * @param string $search
  36          * @param integer $uid
  37          * @param integer $start
  38          * @param integer $limit
  39          * @param integer $last_uriid
  40          * @return array
  41          */
  42         public static function getURIIdListBySearch(string $search, int $uid = 0, int $start = 0, int $limit = 100, int $last_uriid = 0)
  43         {
  44                 $condition = ["`uri-id` IN (SELECT `uri-id` FROM `item-content` WHERE MATCH (`title`, `content-warning`, `body`) AGAINST (? IN BOOLEAN MODE))
  45                         AND (NOT `private` OR (`private` AND `uid` = ?))
  46                         AND `uri-id` IN (SELECT `uri-id` FROM `post-view` WHERE `network` IN (?, ?, ?, ?))",
  47                         $search, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS];
  48
  49                 if (!empty($last_uriid)) {
  50                         $condition = DBA::mergeConditions($condition, ["`uri-id` < ?", $last_uriid]);
  51                 }
  52
  53                 $params = [
  54                         'order' => ['uri-id' => true],
  55                         'group_by' => ['uri-id'],
  56                         'limit' => [$start, $limit]
  57                 ];
  58
  59                 $tags = Post::select(['uri-id'], $condition, $params);
  60
  61                 $uriids = [];
  62                 while ($tag = DBA::fetch($tags)) {
  63                         $uriids[] = $tag['uri-id'];
  64                 }
  65                 DBA::close($tags);
  66
  67                 return $uriids;
  68         }
  69
  70         public static function countBySearch(string $search, int $uid = 0)
  71         {
  72                 $condition = ["`uri-id` IN (SELECT `uri-id` FROM `item-content` WHERE MATCH (`title`, `content-warning`, `body`) AGAINST (? IN BOOLEAN MODE))
  73                         AND (NOT `private` OR (`private` AND `uid` = ?))
  74                         AND `uri-id` IN (SELECT `uri-id` FROM `post-view` WHERE `network` IN (?, ?, ?, ?))",
  75                         $search, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS];
  76                 return Post::count($condition);
  77         }
  78
  79         /**
  80          * Convert a message into plaintext for connectors to other networks
  81          *
  82          * @param array  $item           The message array that is about to be posted
  83          * @param int    $limit          The maximum number of characters when posting to that network
  84          * @param bool   $includedlinks  Has an attached link to be included into the message?
  85          * @param int    $htmlmode       This controls the behavior of the BBCode conversion
  86          * @param string $target_network Name of the network where the post should go to.
  87          *
  88          * @return array Same array structure than \Friendica\Content\Text\BBCode::getAttachedData
  89          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  90          * @see   \Friendica\Content\Text\BBCode::getAttachedData
  91          *
  92          */
  93         public static function getPlaintextPost($item, $limit = 0, $includedlinks = false, $htmlmode = BBCode::API, $target_network = '')
  94         {
  95                 // Remove hashtags
  96                 $URLSearchString = '^\[\]';
  97                 $body = preg_replace("/([#@])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $item['body']);
  98
  99                 // Add an URL element if the text contains a raw link
 100                 $body = preg_replace('/([^\]\=\'"]|^)(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism',
 101                         '$1[url]$2[/url]', $body);
 102
 103                 // Remove the abstract
 104                 $body = Text\BBCode::stripAbstract($body);
 105
 106                 // At first look at data that is attached via "type-..." stuff
 107                 // This will hopefully replaced with a dedicated bbcode later
 108                 //$post = self::getAttachedData($b['body']);
 109                 $post = Text\BBCode::getAttachedData($body, $item);
 110
 111                 if (($item['title'] != '') && ($post['text'] != '')) {
 112                         $post['text'] = trim($item['title'] . "\n\n" . $post['text']);
 113                 } elseif ($item['title'] != '') {
 114                         $post['text'] = trim($item['title']);
 115                 }
 116
 117                 $abstract = '';
 118
 119                 // Fetch the abstract from the given target network
 120                 if ($target_network != '') {
 121                         $default_abstract = Text\BBCode::getAbstract($item['body']);
 122                         $abstract = Text\BBCode::getAbstract($item['body'], $target_network);
 123
 124                         // If we post to a network with no limit we only fetch
 125                         // an abstract exactly for this network
 126                         if (($limit == 0) && ($abstract == $default_abstract)) {
 127                                 $abstract = '';
 128                         }
 129                 } else {// Try to guess the correct target network
 130                         switch ($htmlmode) {
 131                                 case BBCode::TWITTER:
 132                                         $abstract = Text\BBCode::getAbstract($item['body'], Protocol::TWITTER);
 133                                         break;
 134
 135                                 case BBCode::OSTATUS:
 136                                         $abstract = Text\BBCode::getAbstract($item['body'], Protocol::STATUSNET);
 137                                         break;
 138
 139                                 default: // We don't know the exact target.
 140                                         // We fetch an abstract since there is a posting limit.
 141                                         if ($limit > 0) {
 142                                                 $abstract = Text\BBCode::getAbstract($item['body']);
 143                                         }
 144                         }
 145                 }
 146
 147                 if ($abstract != '') {
 148                         $post['text'] = $abstract;
 149
 150                         if ($post['type'] == 'text') {
 151                                 $post['type'] = 'link';
 152                                 $post['url'] = $item['plink'];
 153                         }
 154                 }
 155
 156                 $html = Text\BBCode::convert($post['text'] . ($post['after'] ?? ''), false, $htmlmode);
 157                 $msg = Text\HTML::toPlaintext($html, 0, true);
 158                 $msg = trim(html_entity_decode($msg, ENT_QUOTES, 'UTF-8'));
 159
 160                 $link = '';
 161                 if ($includedlinks) {
 162                         if ($post['type'] == 'link') {
 163                                 $link = $post['url'];
 164                         } elseif ($post['type'] == 'text') {
 165                                 $link = $post['url'] ?? '';
 166                         } elseif ($post['type'] == 'video') {
 167                                 $link = $post['url'];
 168                         } elseif ($post['type'] == 'photo') {
 169                                 $link = $post['image'];
 170                         }
 171
 172                         if (($msg == '') && isset($post['title'])) {
 173                                 $msg = trim($post['title']);
 174                         }
 175
 176                         if (($msg == '') && isset($post['description'])) {
 177                                 $msg = trim($post['description']);
 178                         }
 179
 180                         // If the link is already contained in the post, then it neeedn't to be added again
 181                         // But: if the link is beyond the limit, then it has to be added.
 182                         if (($link != '') && strstr($msg, $link)) {
 183                                 $pos = strpos($msg, $link);
 184
 185                                 // Will the text be shortened in the link?
 186                                 // Or is the link the last item in the post?
 187                                 if (($limit > 0) && ($pos < $limit) && (($pos + 23 > $limit) || ($pos + strlen($link) == strlen($msg)))) {
 188                                         $msg = trim(str_replace($link, '', $msg));
 189                                 } elseif (($limit == 0) || ($pos < $limit)) {
 190                                         // The limit has to be increased since it will be shortened - but not now
 191                                         // Only do it with Twitter
 192                                         if (($limit > 0) && (strlen($link) > 23) && ($htmlmode == BBCode::TWITTER)) {
 193                                                 $limit = $limit - 23 + strlen($link);
 194                                         }
 195
 196                                         $link = '';
 197
 198                                         if ($post['type'] == 'text') {
 199                                                 unset($post['url']);
 200                                         }
 201                                 }
 202                         }
 203                 }
 204
 205                 if ($limit > 0) {
 206                         // Reduce multiple spaces
 207                         // When posted to a network with limited space, we try to gain space where possible
 208                         while (strpos($msg, '  ') !== false) {
 209                                 $msg = str_replace('  ', ' ', $msg);
 210                         }
 211
 212                         // Twitter is using its own limiter, so we always assume that shortened links will have this length
 213                         if (iconv_strlen($link, 'UTF-8') > 0) {
 214                                 $limit = $limit - 23;
 215                         }
 216
 217                         if (iconv_strlen($msg, 'UTF-8') > $limit) {
 218                                 if (($post['type'] == 'text') && isset($post['url'])) {
 219                                         $post['url'] = $item['plink'];
 220                                 } elseif (!isset($post['url'])) {
 221                                         $limit = $limit - 23;
 222                                         $post['url'] = $item['plink'];
 223                                 } elseif (strpos($item['body'], '[share') !== false) {
 224                                         $post['url'] = $item['plink'];
 225                                 } elseif (DI::pConfig()->get($item['uid'], 'system', 'no_intelligent_shortening')) {
 226                                         $post['url'] = $item['plink'];
 227                                 }
 228                                 $msg = Text\Plaintext::shorten($msg, $limit);
 229                         }
 230                 }
 231
 232                 $post['text'] = trim($msg);
 233
 234                 return $post;
 235         }
 236 }