include/bb2diaspora.php

   1 <?php
   2
   3 require_once("include/oembed.php");
   4 require_once("include/event.php");
   5 require_once("library/markdown.php");
   6 require_once("include/html2bbcode.php");
   7 require_once("include/bbcode.php");
   8 require_once("include/markdownify/markdownify.php");
   9
  10 // we don't want to support a bbcode specific markdown interpreter
  11 // and the markdown library we have is pretty good, but provides HTML output.
  12 // So we'll use that to convert to HTML, then convert the HTML back to bbcode,
  13 // and then clean up a few Diaspora specific constructs.
  14
  15 function diaspora2bb($s) {
  16
  17         // for testing purposes: Collect raw markdown articles
  18         // $file = tempnam("/tmp/friendica/", "markdown");
  19         // file_put_contents($file, $s);
  20
  21         $s = html_entity_decode($s,ENT_COMPAT,'UTF-8');
  22
  23         // Too many new lines. So deactivated the following line
  24         // $s = str_replace("\r","\n",$s);
  25         // Simply remove cr.
  26         $s = str_replace("\r","",$s);
  27
  28         // <br/> is invalid. Replace it with the valid expression
  29         $s = str_replace("<br/>","<br />",$s);
  30
  31         $s = preg_replace('/\@\{(.+?)\; (.+?)\@(.+?)\}/','@[url=https://$3/u/$2]$1[/url]',$s);
  32
  33         // Escaping the hash tags - doesn't always seem to work
  34         // $s = preg_replace('/\#([^\s\#])/','\\#$1',$s);
  35         // This seems to work
  36         $s = preg_replace('/\#([^\s\#])/','&#35;$1',$s);
  37
  38         $s = Markdown($s);
  39
  40         $s = str_replace('&#35;','#',$s);
  41
  42         $s = str_replace("\n",'<br />',$s);
  43
  44         $s = html2bbcode($s);
  45 //      $s = str_replace('&#42;','*',$s);
  46
  47         // Convert everything that looks like a link to a link
  48         $s = preg_replace("/([^\]\=]|^)(https?\:\/\/)([a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url=$2$3]$2$3[/url]',$s);
  49
  50         //$s = preg_replace("/([^\]\=]|^)(https?\:\/\/)(vimeo|youtu|www\.youtube|soundcloud)([a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url=$2$3$4]$2$3$4[/url]',$s);
  51         $s = preg_replace("/\[url\=?(.*?)\]https?:\/\/www.youtube.com\/watch\?v\=(.*?)\[\/url\]/ism",'[youtube]$2[/youtube]',$s);
  52         $s = preg_replace("/\[url\=https?:\/\/www.youtube.com\/watch\?v\=(.*?)\].*?\[\/url\]/ism",'[youtube]$1[/youtube]',$s);
  53         $s = preg_replace("/\[url\=?(.*?)\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/url\]/ism",'[vimeo]$2[/vimeo]',$s);
  54         $s = preg_replace("/\[url\=https?:\/\/vimeo.com\/([0-9]+)\](.*?)\[\/url\]/ism",'[vimeo]$1[/vimeo]',$s);
  55         // remove duplicate adjacent code tags
  56         $s = preg_replace("/(\[code\])+(.*?)(\[\/code\])+/ism","[code]$2[/code]", $s);
  57
  58         // Don't show link to full picture (until it is fixed)
  59         $s = scale_external_images($s, false);
  60
  61         return $s;
  62 }
  63
  64
  65 function stripdcode_br_cb($s) {
  66         return '[code]' . str_replace('<br />', "\n\t", $s[1]) . '[/code]';
  67 }
  68
  69
  70
  71 function bb2diaspora($Text,$preserve_nl = false) {
  72
  73         // Convert it to HTML - don't try oembed
  74         $Text = bbcode($Text, $preserve_nl, false);
  75
  76         // Now convert HTML to Markdown
  77         $md = new Markdownify(false, false, false);
  78         $Text = $md->parseString($Text);
  79
  80         // Remove all unconverted tags
  81         $Text = strip_tags($Text);
  82
  83 /*
  84         $ev = bbtoevent($Text);
  85
  86         // Replace any html brackets with HTML Entities to prevent executing HTML or script
  87         // Don't use strip_tags here because it breaks [url] search by replacing & with amp
  88
  89         $Text = str_replace("<", "&lt;", $Text);
  90         $Text = str_replace(">", "&gt;", $Text);
  91
  92         // If we find any event code, turn it into an event.
  93         // After we're finished processing the bbcode we'll
  94         // replace all of the event code with a reformatted version.
  95
  96         if($preserve_nl)
  97                 $Text = str_replace(array("\n","\r"), array('',''),$Text);
  98
  99
 100         // Set up the parameters for a URL search string
 101         $URLSearchString = "^\[\]";
 102         // Set up the parameters for a MAIL search string
 103         $MAILSearchString = $URLSearchString;
 104
 105         // Perform URL Search
 106
 107         // [img]pathtoimage[/img]
 108
 109         // the following was added on 10-January-2012 due to an inability of Diaspora's
 110         // new javascript markdown processor to handle links with images as the link "text"
 111         // It is not optimal and may be removed if this ability is restored in the future
 112
 113         $Text = preg_replace("/\[url\=([$URLSearchString]*)\]\[img\](.*?)\[\/img\]\[\/url\]/ism",
 114                 '![' . t('image/photo') . '](' . '$2' . ')' . "\n" . '[' . t('link') . '](' . '$1' . ')', $Text);
 115
 116         $Text = preg_replace("/\[bookmark\]([$URLSearchString]*)\[\/bookmark\]/ism", '[$1]($1)', $Text);
 117         $Text = preg_replace("/\[bookmark\=([$URLSearchString]*)\](.*?)\[\/bookmark\]/ism", '[$2]($1)', $Text);
 118
 119         $Text = preg_replace("/\[url\]([$URLSearchString]*)\[\/url\]/ism", '[$1]($1)', $Text);
 120         $Text = preg_replace("/\#\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[#$2]($1)', $Text);
 121         $Text = preg_replace("/\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[$2]($1)', $Text);
 122
 123
 124         $Text = preg_replace("/\[img\](.*?)\[\/img\]/", '![' . t('image/photo') . '](' . '$1' . ')', $Text);
 125         $Text = preg_replace("/\[img\=(.*?)\](.*?)\[\/img\]/", '![' . t('image/photo') . '](' . '$2' . ')', $Text);
 126
 127         // Perform MAIL Search
 128         $Text = preg_replace("(\[mail\]([$MAILSearchString]*)\[/mail\])", '[$1](mailto:$1)', $Text);
 129         $Text = preg_replace("/\[mail\=([$MAILSearchString]*)\](.*?)\[\/mail\]/", '[$2](mailto:$1)', $Text);
 130
 131         $Text = str_replace('*', '\\*', $Text);
 132         $Text = str_replace('_', '\\_', $Text);
 133
 134         $Text = str_replace('`','\\`', $Text);
 135
 136         // Check for bold text
 137         $Text = preg_replace("(\[b\](.*?)\[\/b\])is",'**$1**',$Text);
 138
 139         // Check for Italics text
 140         $Text = preg_replace("(\[i\](.*?)\[\/i\])is",'_$1_',$Text);
 141
 142         // Check for Underline text
 143 //      $Text = preg_replace("(\[u\](.*?)\[\/u\])is",'<u>$1</u>',$Text);
 144
 145         // Check for strike-through text
 146 //      $Text = preg_replace("(\[s\](.*?)\[\/s\])is",'<strike>$1</strike>',$Text);
 147
 148         // Check for over-line text
 149 //      $Text = preg_replace("(\[o\](.*?)\[\/o\])is",'<span class="overline">$1</span>',$Text);
 150
 151         // Check for colored text
 152 //      $Text = preg_replace("(\[color=(.*?)\](.*?)\[\/color\])is","<span style=\"color: $1;\">$2</span>",$Text);
 153
 154         // Check for sized text
 155 //      $Text = preg_replace("(\[size=(.*?)\](.*?)\[\/size\])is","<span style=\"font-size: $1;\">$2</span>",$Text);
 156
 157         // Check for list text
 158 //      $Text = preg_replace("/\[list\](.*?)\[\/list\]/is", '<ul class="listbullet">$1</ul>' ,$Text);
 159 //      $Text = preg_replace("/\[list=1\](.*?)\[\/list\]/is", '<ul class="listdecimal">$1</ul>' ,$Text);
 160 //      $Text = preg_replace("/\[list=i\](.*?)\[\/list\]/s",'<ul class="listlowerroman">$1</ul>' ,$Text);
 161 //      $Text = preg_replace("/\[list=I\](.*?)\[\/list\]/s", '<ul class="listupperroman">$1</ul>' ,$Text);
 162 //      $Text = preg_replace("/\[list=a\](.*?)\[\/list\]/s", '<ul class="listloweralpha">$1</ul>' ,$Text);
 163 //      $Text = preg_replace("/\[list=A\](.*?)\[\/list\]/s", '<ul class="listupperalpha">$1</ul>' ,$Text);
 164 //      $Text = preg_replace("/\[li\](.*?)\[\/li\]/s", '<li>$1</li>' ,$Text);
 165
 166 //      $Text = preg_replace("/\[td\](.*?)\[\/td\]/s", '<td>$1</td>' ,$Text);
 167 //      $Text = preg_replace("/\[tr\](.*?)\[\/tr\]/s", '<tr>$1</tr>' ,$Text);
 168 //      $Text = preg_replace("/\[table\](.*?)\[\/table\]/s", '<table>$1</table>' ,$Text);
 169
 170 //      $Text = preg_replace("/\[table border=1\](.*?)\[\/table\]/s", '<table border="1" >$1</table>' ,$Text);
 171 //      $Text = preg_replace("/\[table border=0\](.*?)\[\/table\]/s", '<table border="0" >$1</table>' ,$Text);
 172
 173
 174 //      $Text = str_replace("[*]", "<li>", $Text);
 175
 176         // Check for font change text
 177 //      $Text = preg_replace("(\[font=(.*?)\](.*?)\[\/font\])","<span style=\"font-family: $1;\">$2</span>",$Text);
 178
 179
 180     $Text = preg_replace_callback("/\[code\](.*?)\[\/code\]/is",'stripdcode_br_cb',$Text);
 181
 182         // Check for [code] text
 183         $Text = preg_replace("/(\[code\])+(.*?)(\[\/code\])+/is","\t$2\n", $Text);
 184
 185
 186
 187
 188         // Declare the format for [quote] layout
 189         //      $QuoteLayout = '<blockquote>$1</blockquote>';
 190         // Check for [quote] text
 191         $Text = preg_replace("/\[quote\](.*?)\[\/quote\]/is",">$1\n\n", $Text);
 192
 193         // Images
 194
 195         // html5 video and audio
 196
 197         $Text = preg_replace("/\[video\](.*?)\[\/video\]/", '$1', $Text);
 198
 199         $Text = preg_replace("/\[audio\](.*?)\[\/audio\]/", '$1', $Text);
 200
 201 //      $Text = preg_replace("/\[iframe\](.*?)\[\/iframe\]/", '<iframe src="$1" width="425" height="350"><a href="$1">$1</a></iframe>', $Text);
 202
 203         // [img=widthxheight]image source[/img]
 204 //      $Text = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/", '<img src="$3" style="height:{$2}px; width:{$1}px;" >', $Text);
 205
 206     $Text = preg_replace("/\[youtube\]https?:\/\/www.youtube.com\/watch\?v\=(.*?)\[\/youtube\]/ism",'http://www.youtube.com/watch?v=$1',$Text);
 207     $Text = preg_replace("/\[youtube\]https?:\/\/www.youtube.com\/embed\/(.*?)\[\/youtube\]/ism",'http://www.youtube.com/watch?v=$1',$Text);
 208     $Text = preg_replace("/\[youtube\]https?:\/\/youtu.be\/(.*?)\[\/youtube\]/ism",'http://www.youtube.com/watch?v=$1',$Text);
 209         $Text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism", 'http://www.youtube.com/watch?v=$1', $Text);
 210
 211         $Text = preg_replace("/\[vimeo\]https?:\/\/player.vimeo.com\/video\/([0-9]+)(.*?)\[\/vimeo\]/ism",'http://vimeo.com/$1',$Text);
 212         $Text = preg_replace("/\[vimeo\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/vimeo\]/ism",'http://vimeo.com/$1',$Text);
 213         $Text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism", 'http://vimeo.com/$1',$Text);
 214
 215
 216         $Text = str_replace('[nosmile]','',$Text);
 217
 218         // oembed tag
 219         //      $Text = oembed_bbcode2html($Text);
 220
 221         // If we found an event earlier, strip out all the event code and replace with a reformatted version.
 222
 223         if(x($ev,'desc') && x($ev,'start')) {
 224
 225                 $sub = format_event_diaspora($ev);
 226
 227                 $Text = preg_replace("/\[event\-description\](.*?)\[\/event\-description\]/is",$sub,$Text);
 228                 $Text = preg_replace("/\[event\-start\](.*?)\[\/event\-start\]/is",'',$Text);
 229                 $Text = preg_replace("/\[event\-finish\](.*?)\[\/event\-finish\]/is",'',$Text);
 230                 $Text = preg_replace("/\[event\-location\](.*?)\[\/event\-location\]/is",'',$Text);
 231                 $Text = preg_replace("/\[event\-adjust\](.*?)\[\/event\-adjust\]/is",'',$Text);
 232         }
 233
 234         $Text = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism",'<$1$2=$3&$4>',$Text);
 235
 236         $Text = preg_replace_callback('/\[(.*?)\]\((.*?)\)/ism','unescape_underscores_in_links',$Text);
 237 */
 238         call_hooks('bb2diaspora',$Text);
 239
 240         return $Text;
 241 }
 242
 243 function unescape_underscores_in_links($m) {
 244         $y = str_replace('\\_','_', $m[2]);
 245         return('[' . $m[1] . '](' . $y . ')');
 246 }
 247
 248 function format_event_diaspora($ev) {
 249
 250         $a = get_app();
 251
 252         if(! ((is_array($ev)) && count($ev)))
 253                 return '';
 254
 255         $bd_format = t('l F d, Y \@ g:i A') ; // Friday January 18, 2011 @ 8 AM
 256
 257         $o = 'Friendica event notification:' . "\n";
 258
 259         $o .= '**' . bb2diaspora($ev['desc']) .  '**' . "\n";
 260
 261         $o .= t('Starts:') . ' ' . '['
 262                 . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC',
 263                         $ev['start'] , $bd_format ))
 264                         :  day_translate(datetime_convert('UTC', 'UTC',
 265                         $ev['start'] , $bd_format)))
 266                 .  '](' . $a->get_baseurl() . '/localtime/?f=&time=' . urlencode(datetime_convert('UTC','UTC',$ev['start'])) . ")\n";
 267
 268         if(! $ev['nofinish'])
 269                 $o .= t('Finishes:') . ' ' . '['
 270                         . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC',
 271                                 $ev['finish'] , $bd_format ))
 272                                 :  day_translate(datetime_convert('UTC', 'UTC',
 273                                 $ev['finish'] , $bd_format )))
 274                         . '](' . $a->get_baseurl() . '/localtime/?f=&time=' . urlencode(datetime_convert('UTC','UTC',$ev['finish'])) . ")\n";
 275
 276         if(strlen($ev['location']))
 277                 $o .= t('Location:') . bb2diaspora($ev['location'])
 278                         . "\n";
 279
 280         $o .= "\n";
 281         return $o;
 282 }