<?php
// Utility function to convert wiki-like to Markdown format
// Howard Miller, 2005

// state defines
define( "STATE_NONE",1 ); // blank line has been detected, so looking for first line on next para
define( "STATE_PARAGRAPH",2 ); // currently processing vanilla paragraph
define( "STATE_BLOCKQUOTE",3 ); // currently processing blockquote section
define( "STATE_PREFORM",4 ); // currently processing preformatted text
define( "STATE_NOTIKI",5 ); // currently processing preformatted / no formatting

// list defines
define( "LIST_NONE", 1 ); // no lists active
define( "LIST_UNORDERED", 2 ); // unordered list active
define( "LIST_ORDERED", 3 ); // ordered list active
define( "LIST_DEFINITION", 4 ); // definition list active

class WikiToMarkdown {

  var $block_state;
  var $list_state;
  var $list_depth;
  var $list_backtrack;
  var $output; // output buffer
  var $courseid;

  function close_block( $state ) {
    // provide appropriate closure for block according to state
    
    // if in list close this first
    $lclose = "";
    if ($this->list_state != LIST_NONE) {
      $lclose = $this->do_list( " ",true );
    }

    $sclose = "";
    switch ($state) {
      case STATE_PARAGRAPH:
        $sclose =  "\n";
        break;
      case STATE_BLOCKQUOTE:
        $sclose =  "\n";
        break;
      case STATE_PREFORM:
        $sclose =  "</pre>\n";
        break;
      case STATE_NOTIKI:
        $sclose =  "\n";
        break;  
    }

    return $lclose . $sclose;
  }

  function do_replace( $line, $mark, $tag ) {
    // do the regex thingy for things like bold, italic etc
    // $mark is the magic character, and $tag the HTML tag to insert

    // BODGE: replace inline $mark characters in places where we want them ignored
    // they will be put back after main substitutue, stops problems with eg, and/or
    $bodge = chr(1);
    $line = eregi_replace( '([[:alnum:]])'.$mark.'([[:alnum:]])', '\\1'.$bodge.'\\2',$line );

    $regex = '(^| |[(.,])'.$mark.'([^'.$mark.']*)'.$mark.'([^[:alnum:]]|$)';
    $replace = '\\1<'.$tag.'>\\2</'.$tag.'>\\3';
    $line = eregi_replace( $regex, $replace, $line );

    // BODGE: back we go
    $line = eregi_replace( $bodge, $mark, $line );

    return $line;
  }


  function do_replace_markdown( $line, $mark, $tag ) {
    // do the regex thingy for things like bold, italic etc
    // $mark is the magic character, and $tag the HTML tag to insert
    // MARKDOWN version does not generate HTML tags, just straigt replace

    // BODGE: replace inline $mark characters in places where we want them ignored
    // they will be put back after main substitutue, stops problems with eg, and/or
    $bodge = chr(1);
    $line = eregi_replace( '([[:alnum:]])'.$mark.'([[:alnum:]])', '\\1'.$bodge.'\\2',$line );

    $regex = '(^| |[(.,])'.$mark.'([^'.$mark.']*)'.$mark.'([^[:alnum:]]|$)';
    $replace = '\\1'.$tag.'\\2'.$tag.'\\3';
    $line = eregi_replace( $regex, $replace, $line );

    // BODGE: back we go
    $line = eregi_replace( $bodge, $mark, $line );

    return $line;
  }


  function do_replace_sub( $line, $mark, $tag ) {
    // do regex for subscript and superscript (slightly different)
    // $mark is the magic character and $tag the HTML tag to insert

    $regex = $mark.'([^'.$mark.']*)'.$mark;
    $replace = '<'.$tag.'>\\1</'.$tag.'>';

    return eregi_replace( $regex, $replace, $line );
  }

  function do_list( $line, $blank=false ) {
    // handle line with list character on it
    // if blank line implies drop to level 0
    
    // get magic character and then delete it from the line if not blank
    if ($blank) {
      $listchar="";
      $count = 0;
    }
    else {
      $listchar = $line{0};
      $count = strspn( $line, $listchar );
      $line = eregi_replace( "^[".$listchar."]+ ", "", $line );
    }
    
    // find what sort of list this character represents
    $list_tag = "";
    $list_close_tag = "";
    $item_tag = "";
    $item_close_tag = "";
    $list_style = LIST_NONE;
    switch ($listchar) {
      case '*':
        $list_tag = "";
        $list_close_tag = "";
        $item_tag = "*";
        $item_close_tag = "";
        $list_style = LIST_UNORDERED;
        break;
      case '#':
        $list_tag = "";
        $list_close_tag = "";
        $item_tag = "1.";
        $item_close_tag = "";
        $list_style = LIST_ORDERED;
        break;
      case ';':
        $list_tag = "<dl>";
        $list_close_tag = "</dl>";
        $item_tag = "<dd>";
        $item_close_tag = "</dd>";
        $list_style = LIST_DEFINITION;
        break;
      case ':':
        $list_tag = "<dl>";
        $list_close_tag = "</dl>";
        $item_tag = "<dt>";
        $item_close_tag = "</dt>";
        $list_style = LIST_DEFINITION;
        break;  
      }  

    // tag opening/closing regime now - fun bit :-)
    $tags = "";

    // if depth has reduced do number of closes to restore level
    for ($i=$this->list_depth; $i>$count; $i-- ) {
      $close_tag = array_pop( $this->list_backtrack );
      $tags = $tags . $close_tag;
      }

    // if depth has increased do number of opens to balance
    for ($i=$this->list_depth; $i<$count; $i++ ) {
      array_push( $this->list_backtrack, "$list_close_tag" );
      $tags = $tags . "$list_tag";
    }

    // ok, so list state is now same as style and depth same as count
    $this->list_state = $list_style;
    $this->list_depth = $count;

    // get indent
    $indent = substr( "                      ",1,$count-1 );
 
    if ($blank) {
      $newline = $tags;
    }
    else {  
      $newline = $tags . $indent . "$item_tag " . $line . "$item_close_tag";
    }

    return $newline;
  } 


  function line_replace( $line ) {
    // return line after various formatting replacements
    // have been made - order is vital to stop them interfering with each other
  
    global $CFG;
 
    // ---- (at least) means a <hr />
    // MARKDOWN: no change so leave

    // is this a list line (starts with * # ; :)    
    if (eregi( "^([*]+|[#]+|[;]+|[:]+) ", $line )) {
      $line = $this->do_list( $line );        
    } 

   // typographic conventions
   // MARKDOWN: no equiv. so convert to entity as before
    // $line = str_replace( "--", "&#8212;", $line );
    // $line = str_replace( " - ", " &#8211; ", $line );
    $line = str_replace( "...", " &#8230; ", $line );
    $line = str_replace( "(R)", "&#174;", $line );
    $line = str_replace( "(r)", "&#174;", $line );
    $line = str_replace( "(TM)", "&#8482;", $line );
    $line = str_replace( "(tm)", "&#8482;", $line );
    $line = str_replace( "(C)", "&#169;", $line );
    $line = str_replace( "1/4", "&#188;", $line );
    $line = str_replace( "1/2", "&#189;", $line );
    $line = str_replace( "3/4", "&#190;", $line );
    $line = eregi_replace( "([[:digit:]]+[[:space:]]*)x([[:space:]]*[[:digit:]]+)", "\\1&#215;\\2", $line ); // (digits) x (digits) - multiply    
    // do formatting tags
    // NOTE: The / replacement  *has* to be first, or it will screw the 
    //    HTML tags that are added by the other ones
    // MARKDOWN: only bold and italic change, rest are just HTML
    $line = $this->do_replace_markdown( $line, "\*", "**" );
    $line = $this->do_replace_markdown( $line, "/", "*" );
    $line = $this->do_replace( $line, "\+", "ins" );
    // $line = $this->do_replace( $line, "-", "del" );
    $line = $this->do_replace_sub( $line, "~", "sub" );
    $line = $this->do_replace_sub( $line, "\^", "sup" );
    $line = $this->do_replace( $line, "%", "code" );
    $line = $this->do_replace( $line, "@", "cite" );
    
    // convert urls into proper link with optional link text URL(text)
    // MARDOWN: HTML conversion should work fine
    $line = eregi_replace("([[:space:]]|^)([[:alnum:]]+)://([^[:space:]]*)([[:alnum:]#?/&=])\(([^)]+)\)",
      "\\1[\\5](\\2://\\3\\4)", $line);
    $line = eregi_replace("([[:space:]])www\.([^[:space:]]*)([[:alnum:]#?/&=])\(([^)]+)\)", 
      "\\1[\\5](http://www.\\2\\3)", $line);

    // make urls (with and without httpd) into proper links
    $line = eregi_replace("([[:space:]]|^)([[:alnum:]]+)://([^[:space:]]*)([[:alnum:]#?/&=])",
      "\\1<\\2://\\3\\4>", $line);
    $line = eregi_replace("([[:space:]])www\.([^[:space:]]*)([[:alnum:]#?/&=])", 
      "\\1<http://www.\\2\\3\>", $line);

    // make email addresses into mailtos....
    // MARKDOWN doesn't quite support this, so do as html
    $line = eregi_replace("([[:space:]]|^)([[:alnum:]._-]+@[[:alnum:]._-]+)\(([^)]+)\)",
       "\\1<a href=\"mailto:\\2\">\\3</a>", $line);

    // !# at the beginning of any lines means a heading
    // MARKDOWN: value (1-6) becomes number of hashes
    if (eregi( "^!([1-6]) (.*)$", $line, $regs )) {
      $depth = substr( $line, 1, 1 );
      $out = substr( '##########', 0, $depth);
      $line = eregi_replace( "^!([1-6]) (.*)$", "$out \\2", $line );
    }

    // acronym handing, example HTML(Hypertext Markyp Language)
    // MARKDOWN: no equiv. so just leave as HTML
    $line = ereg_replace( "([A-Z]+)\(([^)]+)\)", "<acronym title=\"\\2\">\\1</acronym>", $line );

    // Replace resource link >>##(Description Text)
    // MARKDOWN: change to MD web link style
    $line = eregi_replace( " ([a-zA-Z]+):([0-9]+)\(([^)]+)\)",
       " [\\3](".$CFG->wwwroot."/mod/\\1/view.php?id=\\2) ", $line );

    require_once($CFG->libdir.'/filelib.php');
    $coursefileurl = get_file_url($this->courseid);

    // Replace picture resource link 
    $line = eregi_replace( "/([a-zA-Z0-9./_-]+)(png|gif|jpg)\(([^)]+)\)",
            "![\\3](".$coursefileurl."/\\1\\2)", $line );

    // Replace file resource link
    $line = eregi_replace( "file:/([[:alnum:]/._-]+)\(([^)]+)\)",
            "[\\2](".$coursefileurl."/\\1)", $line );

    return $line;
  }

  function convert( $content,$courseid ) {

    // main entry point for processing Wiki-like text
    // $content is string containing text with Wiki-Like formatting
    // return: string containing Markdown formatting

    // initialisation stuff
    $this->output = "";
    $this->block_state = STATE_NONE;
    $this->list_state = LIST_NONE;
    $this->list_depth = 0;
    $this->list_backtrack = array();
    $this->spelling_on = false;
    $this->courseid = $courseid;

    // split content into array of single lines
    $lines = explode( "\n",$content );
    $buffer = "";

    // run through lines
    foreach( $lines as $line ) {
      // is this a blank line?
      $blank_line = eregi( "^[[:blank:]\r]*$", $line );
      if ($blank_line) {
        // first end current block according to state
        $buffer = $buffer . $this->close_block( $this->block_state );
        $this->block_state = STATE_NONE;
        continue;
      }

      // act now depending on current block state
      if ($this->block_state == STATE_NONE) {
        // first character of line defines block type
        if (eregi( "^> ",$line )) {
          // blockquote
          $buffer = $buffer . $this->line_replace( $line ). "\n";
          $this->block_state = STATE_BLOCKQUOTE;
        }
        else
        if (eregi( "^  ",$line) ) {
          // preformatted text
          // MARKDOWN: no real equiv. so just use <pre>
          $buffer = $buffer . "<pre>\n";
          $buffer = $buffer . $this->line_replace($line) . "\n";
          $this->block_state = STATE_PREFORM;
        }
        else 
        if (eregi("^\% ",$line) ) {
                // preformatted text - no processing
                // MARKDOWN: this is MD code form of a paragraph
                $buffer = $buffer . "    " . eregi_replace( "^\%","",$line) . "\n";
                $this->block_state = STATE_NOTIKI;
        }
        else {
          // ordinary paragraph
          $buffer = $buffer . $this->line_replace($line) . "\n";
          $this->block_state = STATE_PARAGRAPH; 
        }
        continue;
      }

      if (($this->block_state == STATE_PARAGRAPH) |
          ($this->block_state == STATE_BLOCKQUOTE) |
          ($this->block_state == STATE_PREFORM) ) {
        $buffer = $buffer . $this->line_replace($line) . "\n";
        continue;
      }
      elseif ($this->block_state == STATE_NOTIKI) {
        $buffer = $buffer . "    " .$line . "\n";
      }
    }

    // close off any block level tags
    $buffer = $buffer . $this->close_block( $this->block_state );

    //return $buffer;    
    return $buffer;
  }

  function update( $thing, $textfield, $formatfield, $coursesql='' ) {
    // converts the text in a particular activity (or sub-activity)
    // $thing = the database name for that 'thing' (eg, resource, choice)
    // $textfield = the name of the field that might hold the wiki-text
    // $formatfield = the name of the field that contains the format type
    // $coursesql = if supplied, the query to get the courseid, if not get from the 'course' field 
    //   ($id of record is tacked on right at the end, so phrase accordingly)
    // returns a count of records converted 
    $count = 0;
    if ($records = get_records( $thing,$formatfield,FORMAT_WIKI )) {
        foreach( $records as $record ) {
          $text = $record->$textfield;
          $id = $record->id;
          if (!$coursesql) {
            $courseid = $record->course;
          } else {
            $r = get_record_sql( $coursesql . "$id" );
            $courseid = $r->course;
          }
          $newtext = $this->convert( $text,$courseid );
      
          $record->$textfield = $newtext;
          $record->$formatfield = FORMAT_MARKDOWN;
          update_record( $thing, addslashes_object( $record ) );
          $count++;
        }
    }
    return $count; 
  }
}
?>