include/plaintext.php

   1 <?php
   2
   3 require_once("include/Photo.php");
   4
   5 /**
   6  * @brief Fetches attachment data that were generated the old way
   7  *
   8  * @param string $body Message body
   9  * @return array
  10  * 'type' -> Message type ("link", "video", "photo")
  11  * 'text' -> Text before the shared message
  12  * 'after' -> Text after the shared message
  13  * 'image' -> Preview image of the message
  14  * 'url' -> Url to the attached message
  15  * 'title' -> Title of the attachment
  16  * 'description' -> Description of the attachment
  17  */
  18 function get_old_attachment_data($body) {
  19
  20         $post = array();
  21
  22         // Simplify image codes
  23         $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
  24
  25         if (preg_match_all("(\[class=(.*?)\](.*?)\[\/class\])ism",$body, $attached,  PREG_SET_ORDER)) {
  26                 foreach ($attached AS $data) {
  27                         if (!in_array($data[1], array("type-link", "type-video", "type-photo")))
  28                                 continue;
  29
  30                         $post["type"] = substr($data[1], 5);
  31
  32                         $pos = strpos($body, $data[0]);
  33                         if ($pos > 0) {
  34                                 $post["text"] = trim(substr($body, 0, $pos));
  35                                 $post["after"] = trim(substr($body, $pos + strlen($data[0])));
  36                         } else
  37                                 $post["text"] = trim(str_replace($data[0], "", $body));
  38
  39                         $attacheddata = $data[2];
  40
  41                         $URLSearchString = "^\[\]";
  42
  43                         if (preg_match("/\[img\]([$URLSearchString]*)\[\/img\]/ism", $attacheddata, $matches)) {
  44
  45                                 $picturedata = get_photo_info($matches[1]);
  46
  47                                 if (($picturedata[0] >= 500) AND ($picturedata[0] >= $picturedata[1]))
  48                                         $post["image"] = $matches[1];
  49                                 else
  50                                         $post["preview"] = $matches[1];
  51                         }
  52
  53                         if (preg_match("/\[bookmark\=([$URLSearchString]*)\](.*?)\[\/bookmark\]/ism", $attacheddata, $matches)) {
  54                                 $post["url"] = $matches[1];
  55                                 $post["title"] = $matches[2];
  56                         }
  57
  58                         // Search for description
  59                         if (preg_match("/\[quote\](.*?)\[\/quote\]/ism", $attacheddata, $matches))
  60                                 $post["description"] = $matches[1];
  61
  62                 }
  63         }
  64
  65         return $post;
  66 }
  67
  68 /**
  69  * @brief Fetches attachment data that were generated with the "attachment" element
  70  *
  71  * @param string $body Message body
  72  * @return array
  73  * 'type' -> Message type ("link", "video", "photo")
  74  * 'text' -> Text before the shared message
  75  * 'after' -> Text after the shared message
  76  * 'image' -> Preview image of the message
  77  * 'url' -> Url to the attached message
  78  * 'title' -> Title of the attachment
  79  * 'description' -> Description of the attachment
  80  */
  81 function get_attachment_data($body) {
  82
  83         $data = array();
  84
  85         if (!preg_match("/(.*)\[attachment(.*)\](.*?)\[\/attachment\](.*)/ism", $body, $match))
  86                 return get_old_attachment_data($body);
  87
  88         $attributes = $match[2];
  89
  90         $data["text"] = trim($match[1]);
  91
  92         $type = "";
  93         preg_match("/type='(.*?)'/ism", $attributes, $matches);
  94         if ($matches[1] != "")
  95                 $type = strtolower($matches[1]);
  96
  97         preg_match('/type="(.*?)"/ism', $attributes, $matches);
  98         if ($matches[1] != "")
  99                 $type = strtolower($matches[1]);
 100
 101         if ($type == "")
 102                 return(array());
 103
 104         if (!in_array($type, array("link", "audio", "photo", "video")))
 105                 return(array());
 106
 107         if ($type != "")
 108                 $data["type"] = $type;
 109
 110         $url = "";
 111         preg_match("/url='(.*?)'/ism", $attributes, $matches);
 112         if ($matches[1] != "")
 113                 $url = $matches[1];
 114
 115         preg_match('/url="(.*?)"/ism', $attributes, $matches);
 116         if ($matches[1] != "")
 117                 $url = $matches[1];
 118
 119         if ($url != "")
 120                 $data["url"] = $url;
 121
 122         $title = "";
 123         preg_match("/title='(.*?)'/ism", $attributes, $matches);
 124         if ($matches[1] != "")
 125                 $title = $matches[1];
 126
 127         preg_match('/title="(.*?)"/ism', $attributes, $matches);
 128         if ($matches[1] != "")
 129                 $title = $matches[1];
 130
 131         $title = bbcode(html_entity_decode($title, ENT_QUOTES, 'UTF-8'), false, false, true);
 132         $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 133         $title = str_replace(array("[", "]"), array("&#91;", "&#93;"), $title);
 134         if ($title != "")
 135                 $data["title"] = $title;
 136
 137         $image = "";
 138         preg_match("/image='(.*?)'/ism", $attributes, $matches);
 139         if ($matches[1] != "")
 140                 $image = $matches[1];
 141
 142         preg_match('/image="(.*?)"/ism', $attributes, $matches);
 143         if ($matches[1] != "")
 144                 $image = $matches[1];
 145
 146         if ($image != "")
 147                 $data["image"] = $image;
 148
 149         $preview = "";
 150         preg_match("/preview='(.*?)'/ism", $attributes, $matches);
 151         if ($matches[1] != "")
 152                 $preview = $matches[1];
 153
 154         preg_match('/preview="(.*?)"/ism', $attributes, $matches);
 155         if ($matches[1] != "")
 156                 $preview = $matches[1];
 157
 158         if ($preview != "")
 159                 $data["preview"] = $preview;
 160
 161         $data["description"] = trim($match[3]);
 162
 163         $data["after"] = trim($match[4]);
 164
 165         return($data);
 166 }
 167
 168 function get_attached_data($body) {
 169 /*
 170  - text:
 171  - type: link, video, photo
 172  - title:
 173  - url:
 174  - image:
 175  - description:
 176  - (thumbnail)
 177 */
 178
 179         $post = get_attachment_data($body);
 180
 181         // if nothing is found, it maybe having an image.
 182         if (!isset($post["type"])) {
 183                 require_once("mod/parse_url.php");
 184                 require_once("include/Photo.php");
 185
 186                 $URLSearchString = "^\[\]";
 187                 if (preg_match_all("(\[url=([$URLSearchString]*)\]\s*\[img\]([$URLSearchString]*)\[\/img\]\s*\[\/url\])ism", $body, $pictures,  PREG_SET_ORDER)) {
 188                         if (count($pictures) == 1) {
 189                                 // Checking, if the link goes to a picture
 190                                 $data = parseurl_getsiteinfo_cached($pictures[0][1], true);
 191                                 if ($data["type"] == "photo") {
 192                                         $post["type"] = "photo";
 193                                         if (isset($data["images"][0])) {
 194                                                 $post["image"] = $data["images"][0]["src"];
 195                                                 $post["url"] = $data["url"];
 196                                         } else
 197                                                 $post["image"] = $data["url"];
 198
 199                                         $post["preview"] = $pictures[0][2];
 200                                         $post["text"] = str_replace($pictures[0][0], "", $body);
 201                                 } else {
 202                                         $imgdata = get_photo_info($pictures[0][1]);
 203                                         if (substr($imgdata["mime"], 0, 6) == "image/") {
 204                                                 $post["type"] = "photo";
 205                                                 $post["image"] = $pictures[0][1];
 206                                                 $post["preview"] = $pictures[0][2];
 207                                                 $post["text"] = str_replace($pictures[0][0], "", $body);
 208                                         }
 209                                 }
 210                         } elseif (count($pictures) > 1) {
 211                                 $post["type"] = "link";
 212                                 $post["url"] = $b["plink"];
 213                                 $post["image"] = $pictures[0][2];
 214                                 $post["text"] = $body;
 215                         }
 216                 } elseif (preg_match_all("(\[img\]([$URLSearchString]*)\[\/img\])ism", $body, $pictures,  PREG_SET_ORDER)) {
 217                         if (count($pictures) == 1) {
 218                                 $post["type"] = "photo";
 219                                 $post["image"] = $pictures[0][1];
 220                                 $post["text"] = str_replace($pictures[0][0], "", $body);
 221                         } elseif (count($pictures) > 1) {
 222                                 $post["type"] = "link";
 223                                 $post["url"] = $b["plink"];
 224                                 $post["image"] = $pictures[0][1];
 225                                 $post["text"] = $body;
 226                         }
 227                 }
 228
 229                 if (preg_match_all("(\[url\]([$URLSearchString]*)\[\/url\])ism", $body, $links,  PREG_SET_ORDER)) {
 230                         if (count($links) == 1) {
 231                                 $post["type"] = "text";
 232                                 $post["url"] = $links[0][1];
 233                                 $post["text"] = $body;
 234                         }
 235                 }
 236                 if (!isset($post["type"])) {
 237                         $post["type"] = "text";
 238                         $post["text"] = trim($body);
 239                 }
 240         } elseif (isset($post["url"]) AND ($post["type"] == "video")) {
 241                 require_once("mod/parse_url.php");
 242                 $data = parseurl_getsiteinfo_cached($post["url"], true);
 243
 244                 if (isset($data["images"][0]))
 245                         $post["image"] = $data["images"][0]["src"];
 246         }
 247
 248         return($post);
 249 }
 250
 251 function shortenmsg($msg, $limit, $twitter = false) {
 252         /// @TODO
 253         /// For Twitter URLs aren't shortened, but they have to be calculated as if.
 254
 255         $lines = explode("\n", $msg);
 256         $msg = "";
 257         $recycle = html_entity_decode("&#x2672; ", ENT_QUOTES, 'UTF-8');
 258         foreach ($lines AS $row=>$line) {
 259                 if (iconv_strlen(trim($msg."\n".$line), "UTF-8") <= $limit)
 260                         $msg = trim($msg."\n".$line);
 261                 // Is the new message empty by now or is it a reshared message?
 262                 elseif (($msg == "") OR (($row == 1) AND (substr($msg, 0, 4) == $recycle)))
 263                         $msg = iconv_substr(iconv_substr(trim($msg."\n".$line), 0, $limit, "UTF-8"), 0, -3, "UTF-8")."...";
 264                 else
 265                         break;
 266         }
 267         return($msg);
 268 }
 269
 270 /**
 271  * @brief Convert a message into plaintext for connectors to other networks
 272  *
 273  * @param App $a The application class
 274  * @param array $b The message array that is about to be posted
 275  * @param int $limit The maximum number of characters when posting to that network
 276  * @param bool $includedlinks Has an attached link to be included into the message?
 277  * @param int $htmlmode This triggers the behaviour of the bbcode conversion
 278  * @param string $target_network Name of the network where the post should go to.
 279  *
 280  * @return string The converted message
 281  */
 282 function plaintext($a, $b, $limit = 0, $includedlinks = false, $htmlmode = 2, $target_network = "") {
 283         require_once("include/bbcode.php");
 284         require_once("include/html2plain.php");
 285         require_once("include/network.php");
 286
 287         // Remove the hash tags
 288         $URLSearchString = "^\[\]";
 289         $body = preg_replace("/([#@])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $b["body"]);
 290
 291         // Add an URL element if the text contains a raw link
 292         $body = preg_replace("/([^\]\='".'"'."]|^)(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url]$2[/url]', $body);
 293
 294         // Remove the abstract
 295         $body = remove_abstract($body);
 296
 297         // At first look at data that is attached via "type-..." stuff
 298         // This will hopefully replaced with a dedicated bbcode later
 299         //$post = get_attached_data($b["body"]);
 300         $post = get_attached_data($body);
 301
 302         if (($b["title"] != "") AND ($post["text"] != ""))
 303                 $post["text"] = trim($b["title"]."\n\n".$post["text"]);
 304         elseif ($b["title"] != "")
 305                 $post["text"] = trim($b["title"]);
 306
 307         $abstract = "";
 308
 309         // Fetch the abstract from the given target network
 310         if ($target_network != "") {
 311                 $default_abstract = fetch_abstract($b["body"]);
 312                 $abstract = fetch_abstract($b["body"], $target_network);
 313
 314                 // If we post to a network with no limit we only fetch
 315                 // an abstract exactly for this network
 316                 if (($limit == 0) AND ($abstract == $default_abstract))
 317                         $abstract = "";
 318
 319         } else // Try to guess the correct target network
 320                 switch ($htmlmode) {
 321                         case 8:
 322                                 $abstract = fetch_abstract($b["body"], NETWORK_TWITTER);
 323                                 break;
 324                         case 7:
 325                                 $abstract = fetch_abstract($b["body"], NETWORK_STATUSNET);
 326                                 break;
 327                         case 6:
 328                                 $abstract = fetch_abstract($b["body"], NETWORK_APPNET);
 329                                 break;
 330                         default: // We don't know the exact target.
 331                                  // We fetch an abstract since there is a posting limit.
 332                                 if ($limit > 0)
 333                                         $abstract = fetch_abstract($b["body"]);
 334                 }
 335
 336         if ($abstract != "") {
 337                 $post["text"] = $abstract;
 338
 339                 if ($post["type"] == "text") {
 340                         $post["type"] = "link";
 341                         $post["url"] = $b["plink"];
 342                 }
 343         }
 344
 345         $html = bbcode($post["text"].$post["after"], false, false, $htmlmode);
 346         $msg = html2plain($html, 0, true);
 347         $msg = trim(html_entity_decode($msg,ENT_QUOTES,'UTF-8'));
 348
 349         $link = "";
 350         if ($includedlinks) {
 351                 if ($post["type"] == "link")
 352                         $link = $post["url"];
 353                 elseif ($post["type"] == "text")
 354                         $link = $post["url"];
 355                 elseif ($post["type"] == "video")
 356                         $link = $post["url"];
 357                 elseif ($post["type"] == "photo")
 358                         $link = $post["image"];
 359
 360                 if (($msg == "") AND isset($post["title"]))
 361                         $msg = trim($post["title"]);
 362
 363                 if (($msg == "") AND isset($post["description"]))
 364                         $msg = trim($post["description"]);
 365
 366                 // If the link is already contained in the post, then it neeedn't to be added again
 367                 // But: if the link is beyond the limit, then it has to be added.
 368                 if (($link != "") AND strstr($msg, $link)) {
 369                         $pos = strpos($msg, $link);
 370
 371                         // Will the text be shortened in the link?
 372                         // Or is the link the last item in the post?
 373                         if (($limit > 0) AND ($pos < $limit) AND (($pos + 23 > $limit) OR ($pos + strlen($link) == strlen($msg))))
 374                                 $msg = trim(str_replace($link, "", $msg));
 375                         elseif (($limit == 0) OR ($pos < $limit)) {
 376                                 // The limit has to be increased since it will be shortened - but not now
 377                                 // Only do it with Twitter (htmlmode = 8)
 378                                 if (($limit > 0) AND (strlen($link) > 23) AND ($htmlmode == 8))
 379                                         $limit = $limit - 23 + strlen($link);
 380
 381                                 $link = "";
 382
 383                                 if ($post["type"] == "text")
 384                                         unset($post["url"]);
 385                         }
 386                 }
 387         }
 388
 389         if ($limit > 0) {
 390                 // Reduce multiple spaces
 391                 // When posted to a network with limited space, we try to gain space where possible
 392                 while (strpos($msg, "  ") !== false)
 393                         $msg = str_replace("  ", " ", $msg);
 394
 395                 // Twitter is using its own limiter, so we always assume that shortened links will have this length
 396                 if (iconv_strlen($link, "UTF-8") > 0)
 397                         $limit = $limit - 23;
 398
 399                 if (iconv_strlen($msg, "UTF-8") > $limit) {
 400
 401                         if (($post["type"] == "text") AND isset($post["url"]))
 402                                 $post["url"] = $b["plink"];
 403                         elseif (!isset($post["url"])) {
 404                                 $limit = $limit - 23;
 405                                 $post["url"] = $b["plink"];
 406                         } elseif (strpos($b["body"], "[share") !== false)
 407                                 $post["url"] = $b["plink"];
 408                         elseif (get_pconfig($b["uid"], "system", "no_intelligent_shortening"))
 409                                 $post["url"] = $b["plink"];
 410
 411                         $msg = shortenmsg($msg, $limit);
 412                 }
 413         }
 414
 415         $post["text"] = trim($msg);
 416
 417         return($post);
 418 }
 419 ?>