include/plaintext.php

   1 <?php
   2 function get_old_attachment_data($body) {
   3
   4         $post = array();
   5
   6         // Simplify image codes
   7         $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
   8
   9         if (preg_match_all("(\[class=(.*?)\](.*?)\[\/class\])ism",$body, $attached,  PREG_SET_ORDER)) {
  10                 foreach ($attached AS $data) {
  11                         if (!in_array($data[1], array("type-link", "type-video", "type-photo")))
  12                                 continue;
  13
  14                         $post["type"] = substr($data[1], 5);
  15
  16                         $post["text"] = trim(str_replace($data[0], "", $body));
  17
  18                         $attacheddata = $data[2];
  19
  20                         $URLSearchString = "^\[\]";
  21
  22                         if (preg_match("/\[img\]([$URLSearchString]*)\[\/img\]/ism", $attacheddata, $matches))
  23                                 $post["image"] = $matches[1];
  24
  25                         if (preg_match("/\[bookmark\=([$URLSearchString]*)\](.*?)\[\/bookmark\]/ism", $attacheddata, $matches)) {
  26                                 $post["url"] = $matches[1];
  27                                 $post["title"] = $matches[2];
  28                         }
  29
  30                         // Search for description
  31                         if (preg_match("/\[quote\](.*?)\[\/quote\]/ism", $attacheddata, $matches))
  32                                 $post["description"] = $matches[1];
  33
  34                 }
  35         }
  36
  37         return $post;
  38 }
  39
  40 function get_attachment_data($body) {
  41
  42         $data = array();
  43
  44         if (!preg_match("/(.*)\[attachment(.*)\](.*?)\[\/attachment\](.*)/ism", $body, $match))
  45                 return get_old_attachment_data($body);
  46
  47         $attributes = $match[2];
  48
  49         $data["text"] = trim($match[1]);
  50
  51         $type = "";
  52         preg_match("/type='(.*?)'/ism", $attributes, $matches);
  53         if ($matches[1] != "")
  54                 $type = strtolower($matches[1]);
  55
  56         preg_match('/type="(.*?)"/ism', $attributes, $matches);
  57         if ($matches[1] != "")
  58                 $type = strtolower($matches[1]);
  59
  60         if ($type == "")
  61                 return(array());
  62
  63         if (!in_array($type, array("link", "audio", "video")))
  64                 return(array());
  65
  66         if ($type != "")
  67                 $data["type"] = $type;
  68
  69         $url = "";
  70         preg_match("/url='(.*?)'/ism", $attributes, $matches);
  71         if ($matches[1] != "")
  72                 $url = $matches[1];
  73
  74         preg_match('/url="(.*?)"/ism', $attributes, $matches);
  75         if ($matches[1] != "")
  76                 $url = $matches[1];
  77
  78         if ($url != "")
  79                 $data["url"] = $url;
  80
  81         $title = "";
  82         preg_match("/title='(.*?)'/ism", $attributes, $matches);
  83         if ($matches[1] != "")
  84                 $title = $matches[1];
  85
  86         preg_match('/title="(.*?)"/ism', $attributes, $matches);
  87         if ($matches[1] != "")
  88                 $title = $matches[1];
  89
  90         //$title = htmlentities($title, ENT_QUOTES, 'UTF-8', false);
  91         $title = bbcode(html_entity_decode($title, ENT_QUOTES, 'UTF-8'), false, false, true);
  92         $title = str_replace(array("[", "]"), array("&#91;", "&#93;"), $title);
  93
  94         if ($title != "")
  95                 $data["title"] = $title;
  96
  97         $image = "";
  98         if ($type != "video") {
  99                 preg_match("/image='(.*?)'/ism", $attributes, $matches);
 100                 if ($matches[1] != "")
 101                         $image = $matches[1];
 102
 103                 preg_match('/image="(.*?)"/ism', $attributes, $matches);
 104                 if ($matches[1] != "")
 105                         $image = $matches[1];
 106         }
 107
 108         if ($image != "")
 109                 $data["image"] = $image;
 110
 111         $preview = "";
 112         if ($type != "video") {
 113                 preg_match("/preview='(.*?)'/ism", $attributes, $matches);
 114                 if ($matches[1] != "")
 115                         $preview = $matches[1];
 116
 117                 preg_match('/preview="(.*?)"/ism', $attributes, $matches);
 118                 if ($matches[1] != "")
 119                         $preview = $matches[1];
 120         }
 121
 122         if (($image == "") AND ($preview != ""))
 123                 $data["image"] = $preview;
 124         else
 125                 $data["preview"] = $preview;
 126
 127         $data["description"] = trim($match[3]);
 128
 129         $data["after"] = trim($match[4]);
 130
 131         return($data);
 132 }
 133
 134 function get_attached_data($body) {
 135 /*
 136  - text:
 137  - type: link, video, photo
 138  - title:
 139  - url:
 140  - image:
 141  - description:
 142  - (thumbnail)
 143 */
 144
 145         $post = get_attachment_data($body);
 146
 147         // if nothing is found, it maybe having an image.
 148         if (!isset($post["type"])) {
 149                 require_once("mod/parse_url.php");
 150                 require_once("include/Photo.php");
 151
 152                 $URLSearchString = "^\[\]";
 153                 if (preg_match_all("(\[url=([$URLSearchString]*)\]\s*\[img\]([$URLSearchString]*)\[\/img\]\s*\[\/url\])ism", $body, $pictures,  PREG_SET_ORDER)) {
 154                         if (count($pictures) == 1) {
 155                                 // Checking, if the link goes to a picture
 156                                 $data = parseurl_getsiteinfo_cached($pictures[0][1], true);
 157                                 if ($data["type"] == "photo") {
 158                                         $post["type"] = "photo";
 159                                         if (isset($data["images"][0])) {
 160                                                 $post["image"] = $data["images"][0]["src"];
 161                                                 $post["url"] = $data["url"];
 162                                         } else
 163                                                 $post["image"] = $data["url"];
 164
 165                                         $post["preview"] = $pictures[0][2];
 166                                         $post["text"] = str_replace($pictures[0][0], "", $body);
 167                                 } else {
 168                                         $imgdata = get_photo_info($pictures[0][1]);
 169                                         if (substr($imgdata["mime"], 0, 6) == "image/") {
 170                                                 $post["type"] = "photo";
 171                                                 $post["image"] = $pictures[0][1];
 172                                                 $post["preview"] = $pictures[0][2];
 173                                                 $post["text"] = str_replace($pictures[0][0], "", $body);
 174                                         }
 175                                 }
 176                         } elseif (count($pictures) > 1) {
 177                                 $post["type"] = "link";
 178                                 $post["url"] = $b["plink"];
 179                                 $post["image"] = $pictures[0][2];
 180                                 $post["text"] = $body;
 181                         }
 182                 } elseif (preg_match_all("(\[img\]([$URLSearchString]*)\[\/img\])ism", $body, $pictures,  PREG_SET_ORDER)) {
 183                         if (count($pictures) == 1) {
 184                                 $post["type"] = "photo";
 185                                 $post["image"] = $pictures[0][1];
 186                                 $post["text"] = str_replace($pictures[0][0], "", $body);
 187                         } elseif (count($pictures) > 1) {
 188                                 $post["type"] = "link";
 189                                 $post["url"] = $b["plink"];
 190                                 $post["image"] = $pictures[0][1];
 191                                 $post["text"] = $body;
 192                         }
 193                 }
 194
 195                 if (preg_match_all("(\[url\]([$URLSearchString]*)\[\/url\])ism", $body, $links,  PREG_SET_ORDER)) {
 196                         if (count($links) == 1) {
 197                                 $post["type"] = "text";
 198                                 $post["url"] = $links[0][1];
 199                                 $post["text"] = $body;
 200                         }
 201                 }
 202                 if (!isset($post["type"])) {
 203                         $post["type"] = "text";
 204                         $post["text"] = trim($body);
 205                 }
 206         } elseif (isset($post["url"]) AND ($post["type"] == "video")) {
 207                 require_once("mod/parse_url.php");
 208                 $data = parseurl_getsiteinfo_cached($post["url"], true);
 209
 210                 if (isset($data["images"][0]))
 211                         $post["image"] = $data["images"][0]["src"];
 212         }
 213
 214         return($post);
 215 }
 216
 217 function shortenmsg($msg, $limit, $twitter = false) {
 218         /// @TODO
 219         /// For Twitter URLs aren't shortened, but they have to be calculated as if.
 220
 221         $lines = explode("\n", $msg);
 222         $msg = "";
 223         $recycle = html_entity_decode("&#x2672; ", ENT_QUOTES, 'UTF-8');
 224         foreach ($lines AS $row=>$line) {
 225                 if (iconv_strlen(trim($msg."\n".$line), "UTF-8") <= $limit)
 226                         $msg = trim($msg."\n".$line);
 227                 // Is the new message empty by now or is it a reshared message?
 228                 elseif (($msg == "") OR (($row == 1) AND (substr($msg, 0, 4) == $recycle)))
 229                         $msg = iconv_substr(iconv_substr(trim($msg."\n".$line), 0, $limit, "UTF-8"), 0, -3, "UTF-8")."...";
 230                 else
 231                         break;
 232         }
 233         return($msg);
 234 }
 235
 236 /**
 237  * @brief Convert a message into plaintext for connectors to other networks
 238  *
 239  * @param App $a The application class
 240  * @param array $b The message array that is about to be posted
 241  * @param int $limit The maximum number of characters when posting to that network
 242  * @param bool $includedlinks Has an attached link to be included into the message?
 243  * @param int $htmlmode This triggers the behaviour of the bbcode conversion
 244  * @param string $target_network Name of the network where the post should go to.
 245  *
 246  * @return string The converted message
 247  */
 248 function plaintext($a, $b, $limit = 0, $includedlinks = false, $htmlmode = 2, $target_network = "") {
 249         require_once("include/bbcode.php");
 250         require_once("include/html2plain.php");
 251         require_once("include/network.php");
 252
 253         // Remove the hash tags
 254         $URLSearchString = "^\[\]";
 255         $body = preg_replace("/([#@])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $b["body"]);
 256
 257         // Add an URL element if the text contains a raw link
 258         $body = preg_replace("/([^\]\='".'"'."]|^)(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url]$2[/url]', $body);
 259
 260         // Remove the abstract
 261         $body = remove_abstract($body);
 262
 263         // At first look at data that is attached via "type-..." stuff
 264         // This will hopefully replaced with a dedicated bbcode later
 265         //$post = get_attached_data($b["body"]);
 266         $post = get_attached_data($body);
 267
 268         if (($b["title"] != "") AND ($post["text"] != ""))
 269                 $post["text"] = trim($b["title"]."\n\n".$post["text"]);
 270         elseif ($b["title"] != "")
 271                 $post["text"] = trim($b["title"]);
 272
 273         $abstract = "";
 274
 275         // Fetch the abstract from the given target network
 276         if ($target_network != "") {
 277                 $default_abstract = fetch_abstract($b["body"]);
 278                 $abstract = fetch_abstract($b["body"], $target_network);
 279
 280                 // If we post to a network with no limit we only fetch
 281                 // an abstract exactly for this network
 282                 if (($limit == 0) AND ($abstract == $default_abstract))
 283                         $abstract = "";
 284
 285         } else // Try to guess the correct target network
 286                 switch ($htmlmode) {
 287                         case 8:
 288                                 $abstract = fetch_abstract($b["body"], NETWORK_TWITTER);
 289                                 break;
 290                         case 7:
 291                                 $abstract = fetch_abstract($b["body"], NETWORK_STATUSNET);
 292                                 break;
 293                         case 6:
 294                                 $abstract = fetch_abstract($b["body"], NETWORK_APPNET);
 295                                 break;
 296                         default: // We don't know the exact target.
 297                                  // We fetch an abstract since there is a posting limit.
 298                                 if ($limit > 0)
 299                                         $abstract = fetch_abstract($b["body"]);
 300                 }
 301
 302         if ($abstract != "") {
 303                 $post["text"] = $abstract;
 304
 305                 if ($post["type"] == "text") {
 306                         $post["type"] = "link";
 307                         $post["url"] = $b["plink"];
 308                 }
 309         }
 310
 311         $html = bbcode($post["text"], false, false, $htmlmode);
 312         $msg = html2plain($html, 0, true);
 313         $msg = trim(html_entity_decode($msg,ENT_QUOTES,'UTF-8'));
 314
 315         $link = "";
 316         if ($includedlinks) {
 317                 if ($post["type"] == "link")
 318                         $link = $post["url"];
 319                 elseif ($post["type"] == "text")
 320                         $link = $post["url"];
 321                 elseif ($post["type"] == "video")
 322                         $link = $post["url"];
 323                 elseif ($post["type"] == "photo")
 324                         $link = $post["image"];
 325
 326                 if (($msg == "") AND isset($post["title"]))
 327                         $msg = trim($post["title"]);
 328
 329                 if (($msg == "") AND isset($post["description"]))
 330                         $msg = trim($post["description"]);
 331
 332                 // If the link is already contained in the post, then it neeedn't to be added again
 333                 // But: if the link is beyond the limit, then it has to be added.
 334                 if (($link != "") AND strstr($msg, $link)) {
 335                         $pos = strpos($msg, $link);
 336
 337                         // Will the text be shortened in the link?
 338                         // Or is the link the last item in the post?
 339                         if (($limit > 0) AND ($pos < $limit) AND (($pos + 23 > $limit) OR ($pos + strlen($link) == strlen($msg))))
 340                                 $msg = trim(str_replace($link, "", $msg));
 341                         elseif (($limit == 0) OR ($pos < $limit)) {
 342                                 // The limit has to be increased since it will be shortened - but not now
 343                                 // Only do it with Twitter (htmlmode = 8)
 344                                 if (($limit > 0) AND (strlen($link) > 23) AND ($htmlmode == 8))
 345                                         $limit = $limit - 23 + strlen($link);
 346
 347                                 $link = "";
 348
 349                                 if ($post["type"] == "text")
 350                                         unset($post["url"]);
 351                         }
 352                 }
 353         }
 354
 355         if ($limit > 0) {
 356                 // Reduce multiple spaces
 357                 // When posted to a network with limited space, we try to gain space where possible
 358                 while (strpos($msg, "  ") !== false)
 359                         $msg = str_replace("  ", " ", $msg);
 360
 361                 // Twitter is using its own limiter, so we always assume that shortened links will have this length
 362                 if (iconv_strlen($link, "UTF-8") > 0)
 363                         $limit = $limit - 23;
 364
 365                 if (iconv_strlen($msg, "UTF-8") > $limit) {
 366
 367                         if (($post["type"] == "text") AND isset($post["url"]))
 368                                 $post["url"] = $b["plink"];
 369                         elseif (!isset($post["url"])) {
 370                                 $limit = $limit - 23;
 371                                 $post["url"] = $b["plink"];
 372                         } elseif (strpos($b["body"], "[share") !== false)
 373                                 $post["url"] = $b["plink"];
 374                         elseif (get_pconfig($b["uid"], "system", "no_intelligent_shortening"))
 375                                 $post["url"] = $b["plink"];
 376
 377                         $msg = shortenmsg($msg, $limit);
 378                 }
 379         }
 380
 381         $post["text"] = trim($msg);
 382
 383         return($post);
 384 }
 385 ?>