enabletoc = 1;
$x->enableinclude = 1;
$x->snippets['**'] = "%s "; # instead of
# run all processing
$x->go();
# get the complete HTML output including
$html = $x->fullhtml;
# alternatively, get some other variables:
$body = $x->bodyhtml; # only the part inside the tags
$fullconfig = $x->config; # what was in the "config" area of the file?
== Notes ==
- This is an early public release, ready to be tested.
- Including disk files is disabled out of the box, because of
potential vulnerabilities. Enabling it on systems where
untrusted people could edit the t2t content is not recommended.
A future version may allow the inclusion of files from the
current directory or only with *.t2t extension.
To enable inclusions, use $x->enableinclude = 1;
*/
class T2T {
# these variables could be read or forced
var $title = ''; # the document title
var $content = ''; # the content of the t2t file
var $headers = ''; # the first 3 lines of the t2t file
var $enableheaders = 0; # enables the first 3 lines headers (default=1)
var $enableproc = 1; # enables the pre and post processor (default=1)
var $enabletagged = 1; # enables the tagged mark (''tagged'') (default=1)
var $enableraw = 1; # enables the raw mark (""raw"") (default=1)
var $enableverbatim = 1; # enables the verbatim mark (``raw``) (default=1)
var $enablehotlinks = 1; # enables hotlinks [http://www.externalserver.com/images.jpg] (default=1) (note: it's not enabled in the python implementation of txt2tags)
var $config = ''; # the full config area, including ext.ref.
var $bodytext = ''; # the full body text after inclusions
var $bodyhtml = ''; # the innerHTML of the body of the output, no ...
var $fullhtml = ''; # the full ... output
var $enabletoc = 0; # automatically enabled if %%toc or %!options: --toc
var $enableinclude = 1; # allow file inclusions
var $maxtoclevels = 5; # h1-h? titles go into toc, same as %!options: --toc-level 1
var $mtime; # last modified timestamp of the input file
var $date; # timestamp of the current date
var $cssfile = ''; # the css file to be included in the HTML header
var $maskemail = 0; # rewrite plaintext e-mail links
var $encoding = "UTF-8"; # assume default encoding if none in file
var $parsetargets = "html|xhtml"; # accept %!command(html) and %!command(xhtml)
var $snippets = array(
'header1' => "%s \n", # text (first line of file)
'header2' => "%s \n", # text
'header3' => "%s \n", # text
'headerwrap' => " \n%s
\n", # headers
'title' => '%s', # level, id, text
'hrule' => '\n%s ", # content
'mono' => '%s', # content
'center' => "%s ", # content
'img' => '%s ', # url, text
'cssfile' => '%s ', # bold content
'//' => '%s ', # italics content
'__' => '%s ', # underlined content
'--' => '%s ', # striked content
'tableopen' => "\n", # align, border
'tableclose' => "
\n",
'tablerow' => " \n%s \n", # tagged cells
'tablehead' => " %s\n", # align, content
'tablecell' => " %s\n", # align, content
# special
'blockquoteopen' => '',
'blockquoteclose' => ' ',
'paraopen' => '',
'paraclose' => '
',
'+listopen' => "\n",
'+listclose' => " \n",
'+itemopen' => "",
'+itemmiddle' => "",
'+itemclose' => " \n",
'-listopen' => "\n",
'-itemopen' => "",
'-itemmiddle' => "",
'-itemclose' => " \n",
':listopen' => "\n",
':listclose' => " \n",
':itemopen' => "",
':itemmiddle' => " \n",
':itemclose' => " \n",
'listindent' => ' ', # nicer HTML output
# title, encoding, version, styles, body
'html' => '
%s
1) $attr .= " colspan=\"$span\"";
$cells .= sprintf($fmt, $attr, $c);
}
$table .= sprintf($snippets['tablerow'], $cells);
continue;
}
elseif($table) { # close table
$lines2[] = "\033\033". $this->Keep($table . $snippets['tableclose']); # \033: block that can be nested in lists
$table = '';
}
# horizontal rule (bar1, bar2)
if(preg_match('/^ *(=|-|_){20,}\\s*$/', $line, $m)) {
$class = $m[1] == "=" ? 'heavy':'light';
$lines2[] = "\032\032".$this->Keep(sprintf($snippets['hrule'], $class));
continue;
}
if(preg_match("/^ +\\[([\034\\w_,.+%$#@!?+~\\/-]+\\.(?:png|jpe?g|gif|bmp|svg))\\] +$/i", $line)) {
$lines2[] = "\033\033". $this->Keep(sprintf($snippets['center'], $this->run_inline($line)));
continue;
}
if(trim(strtolower($line))=='%%toc') {
$this->enabletoc = $postoc = 1;
$line = '%%toc';
}
$lines2[] = $line;
} # end foreach line
# close ALL
if($openarea && $openarea != '%%%') # close all areas
$lines2[] = $this->closeRTV($openarea, $areacontent);
if($table)
$lines2[] = "\033\033". $this->Keep($table . $snippets['tableclose']."\n");
if($this->enabletoc && count($toc)) {
if($postoc) { # there is %%toc in the page
array_unshift($toc, "\032\032");
$toc[] = "\032\032";
foreach($lines2 as $k=>$v) {
if($v=='%%toc') array_splice($lines2, $k, 1, $toc);
}
}
else { # before the body: bar, toc-list, bar
$bar = "\032\032".$this->Keep(sprintf($snippets['hrule'], 'light'));
array_unshift($toc, $bar);
$toc[] = $bar;
$lines2 = array_merge($toc, $lines2);
}
}
return $lines2;
}
function secondpass($headers, $lines) { # all other marks
$snippets = $this->snippets;
$lines2 = array();
$html = '';
for($i=0; $i<3; $i++) {
$j = $i+1;
$h = $this->esc($headers[$i]);
if($h) $html .= sprintf($snippets["header$j"], $this->run_macros($h));
}
if($html) $lines2[] = $this->Keep(sprintf($snippets["headerwrap"], $html));
$blockquote = 0;
$blockquotecontent = '';
$para = false;
$openlist = 0;
$listcontent = '';
$ListLevels = array(array(-1, '', '')); # $level, $spaces, $type
foreach($lines as $line) {
# blockquote
if(preg_match('/^(\\t+)([^\\t].*)$/', $line, $m)) {
if($para) {
$para = false;
$lines2[] = $snippets['paraclose'];
}
$level = strlen($m[1]);
$blockquotecontent .= $this->fixblockquote($blockquote, $level)
. $m[1]. $this->run_inline($m[2])."\n";
continue;
}
elseif($blockquote) { # close bq
$lines2[] = "\032\032".$blockquotecontent . $this->fixblockquote($blockquote, 0);
$blockquotecontent = '';
}
# List, Numbered List, Definition List
if(preg_match('/^( *)([+-]) (\\S.*)$/', $line, $m) ||
preg_match( '/^( *)(:) ( *\\S.*)$/', $line, $m)) {
$openlist = 2;
if($para) {
$para = false;
$lines2[] = $snippets['paraclose'];
}
list(, $spaces, $type, $text) = $m;
$text = $this->run_inline($text);
$upped = 0;
while(count($ListLevels)>0) {
list($plevel, $pspaces, $ptype) = array_pop($ListLevels);
## close previous list
if($plevel>=0 && strcmp($spaces, $pspaces)<0) {
$listcontent .= "\n".$this->sp($plevel).$snippets["{$ptype}itemclose"] . $this->sp($plevel).$snippets["{$ptype}listclose"];
$upped ++;
continue;
}
if($upped) $pspaces = $spaces;
$ListLevels[] = array($plevel, $pspaces, $ptype); # restore
## open list
if($plevel<0 || strcmp($spaces, $pspaces)>0) { # new list/sublist
$level = $plevel+1;
$listcontent .= "\n" . $this->sp($level). $snippets["{$type}listopen"];
$ListLevels[] = array($level, $spaces, $type);
}
else { ## close prev item
$listcontent .= "\n".$this->sp($plevel).$snippets["{$ptype}itemclose"];
$level = $plevel;
if($ptype!=$type) {
$listcontent .= "\n".$this->sp($level).$snippets["{$ptype}listclose"]
."\n".$this->sp($level).$snippets["{$type}listopen"];
$ListLevels[count($ListLevels)-1][2] = $type; # restore
}
}
## open item
$listcontent .= $this->sp($level).$snippets["{$type}itemopen"];
## fill content
$listcontent .= $text . $snippets["{$type}itemmiddle"];
break;
}
continue;
}
elseif($openlist) {
if(trim($line)=='' || strpos($line, "\032\032")===0 ) {
$openlist --;
if(!$openlist || strpos($line, "\032\032")===0) { # second empty line, close ALL
while(count($ListLevels)>1) {
list($plevel, $pspaces, $ptype) = array_pop($ListLevels);
$listcontent .= "\n".$this->sp($plevel).$snippets["{$ptype}itemclose"]
.$this->sp($plevel).$snippets["{$ptype}listclose"];
}
$lines2[] = $this->Keep($listcontent);
$lines2[] = $line;
$listcontent = "";$openlist=0;
continue;
}
else {
list($plevel, ,) = $ListLevels[count($ListLevels)-1];
$listcontent .= $this->sp($plevel+1)."{$snippets["paraopen"]}{$snippets["paraclose"]}\n";
}
}
else {
list($plevel, $pspaces, $ptype) = array_pop($ListLevels);
if(preg_match('/^(( *)([+-:])) *$/', $line, $m) && $m[1]== "$pspaces$ptype") { # close
$listcontent .= "\n".$this->sp($plevel).$snippets["{$ptype}itemclose"]
. $this->sp($plevel).$snippets["{$ptype}listclose"];
if($plevel) $openlist = 2;
else {
$lines2[] = $this->Keep($listcontent);
$listcontent = "";$openlist=0;
continue;
}
}
else {
if($openlist==1)
$listcontent .= $this->sp($plevel+1)."{$snippets["paraopen"]}{$snippets["paraclose"]}\n";
$openlist = 2;
$ListLevels[] = array($plevel, $pspaces, $ptype);
$listcontent .= $this->sp($plevel+1).$this->run_inline($line)."\n";
}
}
continue;
}
if(preg_match("/^[\032\033]{2}/", $line)) {
if($para) {
$para = false;
$lines2[] = $snippets['paraclose'];
}
$lines2[] = $line;
continue;
}
if($para) {
if(trim($line)=='') {
$para = false;
$lines2[] = $snippets['paraclose'];
}
else {
$lines2[] = $this->run_inline($line);
}
continue;
}
# $para = false;
if(trim($line)!='') {
$lines2[] = $snippets['paraopen'];
$lines2[] = $this->run_inline($line);
$para = true;
continue;
}
$lines2[] = $line;
}
# close ALL
if($blockquote)
$lines2[] = $this->Keep($blockquotecontent . $this->fixblockquote($blockquote, 0));
if($openlist) {
while(count($ListLevels)>1) {
list($plevel, $pspaces, $ptype) = array_pop($ListLevels);
$listcontent .= "\n".$this->sp($plevel).$snippets["{$ptype}itemclose"]
.$this->sp($plevel).$snippets["{$ptype}listclose"];
}
$lines2[] = $this->Keep($listcontent);
}
if($para) {
$para = false;
$lines2[] = $snippets['paraclose'];
}
return $lines2;
}
function run_inline($line) { # inline transformations (links, images, bold, mono...)
$snippets = $this->snippets;
# inline Raw, Mono, Tagged
if(preg_match_all('/(\'|"|`){2}([^\\s](.*?[^\\s])?\\1*)\\1\\1/', $line, $m, PREG_SET_ORDER)) {
foreach($m as $a) {
$type = $a[1].$a[1];
$c = $this->PSS($a[2]);
$tmp = $this->closeRTV($type, $c);
$line = preg_replace('/(\'|"|`){2}([^\\s](.*?[^\\s])?\\1*)\\1\\1/', $tmp, $line, 1);
}
}
# macros
$line = $this->run_macros($line);
# <[img]>
$imgrx = "\\[([\034\\w_,.+%$#@!?+~\\/-]+\\.(?:png|jpe?g|gif|bmp))\\]";
$line = preg_replace("/^$imgrx(?=.)/ei",
"\$this->Keep(sprintf(\$snippets['img'], 'left', '$1'))", $line);
$line = preg_replace("/(?<=.)$imgrx$/ei",
"\$this->Keep(sprintf(\$snippets['img'], 'right', '$1'))", $line);
$line = preg_replace("/$imgrx/ei",
"\$this->Keep(sprintf(\$snippets['img'], 'middle', '$1', 'middle'))", $line);
$UEX = '<>"{}|\\\\^`()\\[\\]\''; # UrlExcludeChars
$PRT = '(?:https?|ftp|news|telnet|gopher|wais|mailto):';
if ($this->enablehotlinks == 0) {
$Links = array(
"{$PRT}[^\\s$UEX]+" =>'',
"www\\d?\\.[^\\s$UEX]+" =>'http://', # lazy links
"ftp\\d?\\.[^\\s$UEX]+" =>'ftp://', # lazy links
"\\w[\\w.-]+@[\\w-.]+[^\\s$UEX]+" =>'mailto:', # lazy links
); #
}
else {
$Links = array(
//"{$PRT}[^\\s$UEX]+" =>'', # allows hotlinks by disabling this part
//"www\\d?\\.[^\\s$UEX]+" =>'http://', # lazy links won't work here
"ftp\\d?\\.[^\\s$UEX]+" =>'ftp://', # lazy links
"\\w[\\w.-]+@[\\w-.]+[^\\s$UEX]+" =>'mailto:', # lazy links
); #
}
# [txt link], [txt #anchor]
foreach($Links as $k=>$v) {
$line = preg_replace("/\\[([^\\]]+?) +($k)\\]/ei",
"\$this->Keep(sprintf(\$snippets['link'], \$this->esc('$v$2'), \$this->esc('$1', 1)))", $line);
}
# local links
$line = preg_replace("/\\[([^\\]]+?) +([^\\s$UEX]+)\\]/ei",
"\$this->Keep(sprintf(\$snippets['link'], \$this->esc('$2'), \$this->esc('$1', 1)))", $line);
# free links www.link, e@mail, http://link
foreach($Links as $k=>$v) {
if($v=='mailto:' && $this->maskemail) {
$line = preg_replace("/\\b({$k}[^\\s.,?!$UEX])/ei",
"\$this->Keep('<' . str_replace(array('@', '.'), array(' (a) ', ' '), '$1') . '>$2')", $line);
}
else {
$line = preg_replace("/\\b({$k}[^\\s.,?!$UEX])/ei",
"\$this->Keep(sprintf(\$snippets['link'], \$this->esc('$v$1'), \$this->esc('$1')))", $line);
}
}
$line = $this->esc($line);
# Bold, Italic, Underline, Strike
$b = array('*', '/', '_', '-');
foreach($b as $c) {
$q = preg_quote($c, '/');
$line = preg_replace("/($q){2}([^\s](?:.*?\\S)?\\1*)\\1\\1/e",
"sprintf(\$snippets['$c$c'], \$this->PSS('$2'))", $line);
}
return $line;
}
function set_macros($input, $isfile = 0) {
$this->date=time();
if($isfile && file_exists($input)) {
$this->mtime = filemtime($input);
$this->infile = $this->fileattr($input);
}
else {
$this->mtime = time();
$this->infile = $this->fileattr('-');
}
$this->outfile = $this->fileattr('-');
}
function run_macros($line) {
$line = preg_replace('/%%(date|mtime)(\\((.+?)\\))?/ie',
'strftime("$2"? $this->PSS("$3"):"%Y%m%d", $this->$1)', $line);
$line = preg_replace('/%%infile(?:\\((.*?)\\))?/ie',
'"$1" ? str_replace(array_keys($this->infile), array_values($this->infile), "$1")
: $this->infile["%f"]', $line);
$line = preg_replace('/%%outfile(?:\\((.*?)\\))?/ie',
'"$1" ? str_replace(array_keys($this->outfile), array_values($this->outfile), "$1")
: $this->outfile["%f"]', $line);
/*$line = preg_replace_callback('/%%rand\([0-9]+,[0-9]+\)/', 'create_function(return(rand($1,$2);))', $line);
$line = preg_replace('/%%rand\([0-9]+,[0-9]+\)/i', '', $line);
$line = preg_replace_callback('/%%rand\\(([0-9]+),([0-9]+)\\)/', 'return(rand($1,$2);)', $line);
*/
return $line;
}
function fixblockquote(&$prev, $curr) { # close open blocks, open sub-blocks
$s = $this->snippets;
$x = '';
while ($prev<$curr) $x .= str_repeat("\t", ++$prev) . $s['blockquoteopen'] . "\n";
while ($prev>$curr) $x .= str_repeat("\t", $prev--) . $s['blockquoteclose'] . "\n";
return $x;
}
function closeRTV(&$type, &$x) { # Raw, Tagged or Verbadim lines/areas
switch($type{0}) {
case '%': $type = $x = ''; return '';
case "'":
if ($this->enabletagged == 1) {
$y = $x;
break;
}
case '"': # raw
if ($this->enableraw == 1) {
$y = $this->esc($x);
break;
}
case '`': # verbatim, mono
if ($this->enableverbatim == 1) {
$s = $this->snippets;
$fmt = (strlen($type)==2) ? $s['mono'] : $s['verbatim'];
$y = sprintf($fmt, $this->esc($x));
break;
}
else {
$y = $this->esc($x);
}
}
$block = (strlen($type)==3) ? "\033\033" : '';
$type = $x = '';
return $block. $this->Keep($y);
}
function PSS($x) { return str_replace('\\"','"',$x); } # Strip RegExp slashes
function Keep($x) { # preserves a string from being processed by wiki markups
$x = $this->Restore($x);
$this->KPCount++; $this->KPV[$this->KPCount]=$x;
return "\034\034{$this->KPCount}\034\034";
}
function Restore($x) { # recovers all hidden strings
return preg_replace("/\034\034(\\d.*?)\034\034/e", "\$this->KPV['\$1']", $x);
}
function sp($n){ # add spaces for nicer indented HTML source code
return str_repeat($this->snippets['listindent'], $n);
}
function esc($x, $pss=0) { # htmlspecialchars
if($pss) $x = $this->PSS($x);
return str_replace(
array('&', '<', '>', '$'),
array('&', '<', '>', '$'),
$x);
}
function fileattr($fname) { # variables that can be in %%infile()
if($fname == '-') return array(
'' => '-', '%f'=> '-', '%F'=> '-', '%e'=> '',
'%p'=> '-', '%d'=> '.', '%D'=> '.', '%%'=>'%');
preg_match('/\\.([^.\\/]+)$/',$fname, $m); $ext=@$m[1];
return array(
'' => basename($fname),
'%f'=> basename($fname),
'%F'=> preg_replace('/\\.[^.]+$/', '', basename($fname)),
'%e'=> $m[1],
'%p'=>realpath($fname),
'%d'=>dirname(realpath($fname)),
'%D'=>basename(dirname(realpath($fname))),
'%%'=>'%');
}
function read($filename, $allowed = false) { # get a file content
if(!$allowed) return '';
if(!file_exists($filename)) return '';
return str_replace("\r", '', implode('', @file($filename)));
}
# get the Header, Config area and Body of a t2t file
# the function will include the content of any included files
function head_conf_body($content) {
$lines = explode("\n", $content);
$R = array();
$R['header'] = $R['config'] = $R['body'] = array();
# headers
if ($this->enableheaders == 0) {
$R['header'][0] = $R['header'][1] = $R['header'][2] = '';
}
else if($lines[0]=='') {
$R['header'][0] = $R['header'][1] = $R['header'][2] = '';
array_shift($lines);
}
else {
for ($i=0; $i<3 && isset($lines[0]); $i++) {
$R["header"][$i] = array_shift($lines);
}
}
# config
$mlcomment = false;
while(isset($lines[0])) {
$line = array_shift($lines);
if(rtrim($line) == '%%%') { # comment areas in Config
$mlcomment = ! $mlcomment;
continue;
}
if($mlcomment || trim($line)=='') continue;
if(preg_match('/^%!\\s*includeconf\\s*:\\s*(.+)$/', $line, $m)) {
$f = trim($m[1]);
if($f{0}!='/') $f = $this->infile['%d'] . DIRECTORY_SEPARATOR . $f;
$r = $this->head_conf_body($this->read($f, $this->enableinclude));
for($i=count($r['config'])-1; $i>=0; $i--)
array_unshift($lines, $r['config'][$i]);
continue;
}
if($line{0} != '%' || preg_match('/^%(%(date|mtime|toc|infile|outfile|rand)|! *include)/i', $line)) {
array_unshift($lines, $line);
break;
}
$R["config"][] = $line;
}
# body
$mlcomment = false;
while(isset($lines[0])) {
$line = array_shift($lines);
if(rtrim($line) == '%%%') { # comment areas
$mlcomment = ! $mlcomment;
continue;
}
if($mlcomment) continue;
if(preg_match('/^%!\\s*include(?:\\(x?html\\))?\\s*:\\s*(``|\'\'|""|)(.+)\\1\\s*$/', $line, $m)) {
$f = trim($m[2]);
if($f{0}!='/') $f = $this->infile['%d'] . DIRECTORY_SEPARATOR . $f;
$r = $this->head_conf_body($this->read($f, $this->enableinclude));
if($m[1]) {
$q = implode("\n", $r['body'])."\n";
$type = str_repeat($m[1]{1}, 3);
$line = $this->closeRTV($type, $q);
}
else {
for($i=count($r['body'])-1; $i>=0; $i--)
array_unshift($lines, $r['body'][$i]);
continue;
}
}
$R['body'][] = $line;
}
return $R;
}
}