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\DI;
  28
  29 class ItemContent
  30 {
  31         /**
  32          * Convert a message into plaintext for connectors to other networks
  33          *
  34          * @param array  $item           The message array that is about to be posted
  35          * @param int    $limit          The maximum number of characters when posting to that network
  36          * @param bool   $includedlinks  Has an attached link to be included into the message?
  37          * @param int    $htmlmode       This controls the behavior of the BBCode conversion
  38          * @param string $target_network Name of the network where the post should go to.
  39          *
  40          * @return array Same array structure than \Friendica\Content\Text\BBCode::getAttachedData
  41          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  42          * @see   \Friendica\Content\Text\BBCode::getAttachedData
  43          *
  44          */
  45         public static function getPlaintextPost($item, $limit = 0, $includedlinks = false, $htmlmode = BBCode::API, $target_network = '')
  46         {
  47                 // Remove hashtags
  48                 $URLSearchString = '^\[\]';
  49                 $body = preg_replace("/([#@])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $item['body']);
  50
  51                 // Add an URL element if the text contains a raw link
  52                 $body = preg_replace('/([^\]\=\'"]|^)(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism',
  53                         '$1[url]$2[/url]', $body);
  54
  55                 // Remove the abstract
  56                 $body = Text\BBCode::stripAbstract($body);
  57
  58                 // At first look at data that is attached via "type-..." stuff
  59                 // This will hopefully replaced with a dedicated bbcode later
  60                 //$post = self::getAttachedData($b['body']);
  61                 $post = Text\BBCode::getAttachedData($body, $item);
  62
  63                 if (($item['title'] != '') && ($post['text'] != '')) {
  64                         $post['text'] = trim($item['title'] . "\n\n" . $post['text']);
  65                 } elseif ($item['title'] != '') {
  66                         $post['text'] = trim($item['title']);
  67                 }
  68
  69                 $abstract = '';
  70
  71                 // Fetch the abstract from the given target network
  72                 if ($target_network != '') {
  73                         $default_abstract = Text\BBCode::getAbstract($item['body']);
  74                         $abstract = Text\BBCode::getAbstract($item['body'], $target_network);
  75
  76                         // If we post to a network with no limit we only fetch
  77                         // an abstract exactly for this network
  78                         if (($limit == 0) && ($abstract == $default_abstract)) {
  79                                 $abstract = '';
  80                         }
  81                 } else {// Try to guess the correct target network
  82                         switch ($htmlmode) {
  83                                 case BBCode::TWITTER:
  84                                         $abstract = Text\BBCode::getAbstract($item['body'], Protocol::TWITTER);
  85                                         break;
  86
  87                                 case BBCode::OSTATUS:
  88                                         $abstract = Text\BBCode::getAbstract($item['body'], Protocol::STATUSNET);
  89                                         break;
  90
  91                                 default: // We don't know the exact target.
  92                                         // We fetch an abstract since there is a posting limit.
  93                                         if ($limit > 0) {
  94                                                 $abstract = Text\BBCode::getAbstract($item['body']);
  95                                         }
  96                         }
  97                 }
  98
  99                 if ($abstract != '') {
 100                         $post['text'] = $abstract;
 101
 102                         if ($post['type'] == 'text') {
 103                                 $post['type'] = 'link';
 104                                 $post['url'] = $item['plink'];
 105                         }
 106                 }
 107
 108                 $html = Text\BBCode::convert($post['text'] . ($post['after'] ?? ''), false, $htmlmode);
 109                 $msg = Text\HTML::toPlaintext($html, 0, true);
 110                 $msg = trim(html_entity_decode($msg, ENT_QUOTES, 'UTF-8'));
 111
 112                 $link = '';
 113                 if ($includedlinks) {
 114                         if ($post['type'] == 'link') {
 115                                 $link = $post['url'];
 116                         } elseif ($post['type'] == 'text') {
 117                                 $link = $post['url'] ?? '';
 118                         } elseif ($post['type'] == 'video') {
 119                                 $link = $post['url'];
 120                         } elseif ($post['type'] == 'photo') {
 121                                 $link = $post['image'];
 122                         }
 123
 124                         if (($msg == '') && isset($post['title'])) {
 125                                 $msg = trim($post['title']);
 126                         }
 127
 128                         if (($msg == '') && isset($post['description'])) {
 129                                 $msg = trim($post['description']);
 130                         }
 131
 132                         // If the link is already contained in the post, then it neeedn't to be added again
 133                         // But: if the link is beyond the limit, then it has to be added.
 134                         if (($link != '') && strstr($msg, $link)) {
 135                                 $pos = strpos($msg, $link);
 136
 137                                 // Will the text be shortened in the link?
 138                                 // Or is the link the last item in the post?
 139                                 if (($limit > 0) && ($pos < $limit) && (($pos + 23 > $limit) || ($pos + strlen($link) == strlen($msg)))) {
 140                                         $msg = trim(str_replace($link, '', $msg));
 141                                 } elseif (($limit == 0) || ($pos < $limit)) {
 142                                         // The limit has to be increased since it will be shortened - but not now
 143                                         // Only do it with Twitter
 144                                         if (($limit > 0) && (strlen($link) > 23) && ($htmlmode == BBCode::TWITTER)) {
 145                                                 $limit = $limit - 23 + strlen($link);
 146                                         }
 147
 148                                         $link = '';
 149
 150                                         if ($post['type'] == 'text') {
 151                                                 unset($post['url']);
 152                                         }
 153                                 }
 154                         }
 155                 }
 156
 157                 if ($limit > 0) {
 158                         // Reduce multiple spaces
 159                         // When posted to a network with limited space, we try to gain space where possible
 160                         while (strpos($msg, '  ') !== false) {
 161                                 $msg = str_replace('  ', ' ', $msg);
 162                         }
 163
 164                         // Twitter is using its own limiter, so we always assume that shortened links will have this length
 165                         if (iconv_strlen($link, 'UTF-8') > 0) {
 166                                 $limit = $limit - 23;
 167                         }
 168
 169                         if (iconv_strlen($msg, 'UTF-8') > $limit) {
 170                                 if (($post['type'] == 'text') && isset($post['url'])) {
 171                                         $post['url'] = $item['plink'];
 172                                 } elseif (!isset($post['url'])) {
 173                                         $limit = $limit - 23;
 174                                         $post['url'] = $item['plink'];
 175                                 } elseif (strpos($item['body'], '[share') !== false) {
 176                                         $post['url'] = $item['plink'];
 177                                 } elseif (DI::pConfig()->get($item['uid'], 'system', 'no_intelligent_shortening')) {
 178                                         $post['url'] = $item['plink'];
 179                                 }
 180                                 $msg = Text\Plaintext::shorten($msg, $limit);
 181                         }
 182                 }
 183
 184                 $post['text'] = trim($msg);
 185
 186                 return $post;
 187         }
 188 }