Extension:Fasta.php

From OrganicDesign Wiki

<?php
# [[Category:Private]][[Category:Private]]
# Extension:Fasta
# - Licenced under LGPL (http://www.gnu.org/copyleft/lesser.html)
# - Author: [http://www.organicdesign.co.nz/Sven User:Sven]
 
/*
 * Local variables:
 * tab-width: 2
 * c-basic-offset: 2
 * indent-tabs-mode: nil
 * End:
*/
 
if (!defined('MEDIAWIKI')) die('Not a valid entry point!');
 
define('FASTA_VERSION','1.0.0, 2007-05-07');
 
$wgFastaMagic	        = "tfaFormat";
$wgFastaTag	        = "tfa";
$wgExtensionFunctions[]	= 'wfSetupFasta';
$wgHooks['LanguageGetMagic'][] = 'wfFastaLanguageGetMagic';
 
$wgExtensionCredits['parserhook'][] = array(
	'name'        => 'Fasta',
	'author'      => '[http://www.organicdesign.co.nz/Sven User:Sven]',
	'description' => 'FASTA file rendering extension with regular expression search capability',
	'url'         => 'http://www.organicdesign.co.nz/Extension:Fasta',
	'version'     => FASTA_VERSION
	);
 
class Fasta {
 
	# Properties
	private $dnaColour = array(
		'G' => 'black',
		'A' => 'green',
		'C' => 'blue',
		'T' => 'red'
		);
 
private $aaColour = array(
		 'A' => "C8C8C8", # grey
		 'R' => "145AFF", # blue
		 'N' => "00DCDC", # Turquoise
		 'D' => "E60A0A", # red
		 'C' => "E6E600", # yellow
		 'Q' => "00DCDC", # turquoise
		 'E' => "E60A0A", # red
		 'G' => "EBEBEB", # light grey
		 'H' => "8282D2", # purple
		 'I' => "0F820F", # green
		 'L' => "0F820F", # green
		 'K' => "145AFF", # blue
		 'M' => "E6E600", # yellow
		 'F' => "3232AA", # purple
		 'P' => "DC9682", # pink
		 'S' => "DC9682", # orange
		 'T' => "DC9682", # orange
		 'Y' => "3232AA", # purple
		 'V' => "0F820F", # green
		 );
 
private $url = array('gi' => 'http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?val=');
 
	# Constructor
         function Fasta() {
		global $wgHooks,$wgParser,$wgFastaMagic,$wgFastaTag,$wgJavaScriptFunctions;
		# Add the parser-function
		$wgParser->setFunctionHook($wgFastaMagic,array($this,'magicTfaFormat'));
 
		# Add the tagHook
		$wgParser->setHook($wgFastaTag,array($this,'tagTfa'));
 
		}
function type($bases)
{
  return preg_match("/^[\\n\\d\\sacgtn*-]+$/i", $bases) ? "dna"
       : (preg_match("/^[\\n\\d\\sacgun*-]+$/i", $bases) ? "rna" : "protein" );
} 
 
# DEBUGGING
# Logging to an article from within the MW environment
	function logAdd($article,$msg) {
	  $ts = $GLOBALS['wgLang']->timeanddate(wfTimestampNow(),true);
	  $la = new Article(Title::newFromText($article));
	  $la->quickEdit($la->getContent()."\n*$ts: $msg");
	}
 
        # Logging to a file
	function logFile($file, $msg) {
		$fh = fopen($file,'a');
		$ts = $GLOBALS['wgLang']->timeanddate(wfTimestampNow(),true);
		if(is_array($msg)) $msg = print_r($msg, true);
 
		fwrite($fh, "\n$ts: $msg");
		}
 
	# Expand the tfaFormat-magic
	function magicTfaFormat(&$parser) {
 
		# Populate $argv with both named and numeric parameters
		$argv = array();
		foreach (func_get_args() as $arg) if (!is_object($arg)) {
			if (preg_match('/^(.+?)\\s*=\\s*(.+)$/',$arg,$match)) $argv[$match[1]] = $match[2]; else $argv[] = $arg;
			}
 
		# Build text of expanded result
		$args = '';
		foreach ($argv as $k => $v) $args .= "*'''$k:''' ''$v''\n";
		$text = "=== magicTfaFormat(): ===\n$args";
		# Return result with available parser flags
		return $text;
		}
 
        # Extract first row of a FASTA file
	function removeHeader(&$text) {
		$lines = explode("\n", $text, 2);
                list($header, $text) = $lines;
                return $header;
		}
 
 
	# Convert the <tfa> tags to HTML
	function tagTfa($text,$argv,&$parser) {
	  $file = "/tmp/Fasta.log";
 
	  # If a \n is found before fasta header remove it
	  $text = preg_replace('/^\n/', '', $text);
 
	  # Remove header line '>gb|...| description\n'
	  $header = $this->removeHeader($text); 
 
	  # Determine base type
	  $baseType = $this->type($text);
	  $this->logFile($file, "$baseType");
 
	  if( array_key_exists('regex', $argv) ){
	    # Replace regex match with ^ and $ as marker for beginning and end
	    $text = preg_replace("/($argv[regex])/s", "^$1$", $text);
	  }	  
 
	  $this->logFile($file, $text);
 
	  # Replace regex section with url
	  $header = preg_replace_callback('/(>gi\|)(\d+)(\|.+)/',array($this,'urlRegexpCallback'),$header);
          # Reparse the header through $wgParser
	  $header = $parser->parse($header,$parser->mTitle,$parser->mOptions,false,false)->getText();	
 
	  # Replace bases with font tags
	  if($baseType == "protein") {
	    $text = preg_replace_callback('/[ARNDCQEGHILKMFPSTYV]/i',array($this,'aaRegexpCallback'),$text);
	  } else {
	    $text = preg_replace_callback('/[GACT]/i',array($this,'dnaRegexpCallback'),$text);
	  }
	  if( array_key_exists('regex', $argv) ){
	    $file = "/tmp/Fasta.log";
	    # Replace ^ with span opening tag
	    $text = preg_replace('/\^/', '<span class="inline">', $text);
	    # Replace $ with span closure tag
	    $text = preg_replace('/\$/', '</span>', $text);
	  }  
 
	  $args = '';
	  foreach ($argv as $k => $v) $args .= "<li><b>$k:</b> <i>$v</i></li>\n";
 
	  return "<h3>tagTfa():</h3>\n<ul>$args<li>Content:</b> 
$header<br />$text
</ul>\n";
	}
 
	# Called by the regexp in tagTfa()
	function dnaRegexpCallback($match) {
		$chr = strtoupper($match[0]);
 
		return '<font color="' . $this->dnaColour[$chr] . '">' . $chr . '</font>';
		}
 
	# Called by the regexp in tagTfa()
	function aaRegexpCallback($match) {
		$chr = strtoupper($match[0]);
 
		return '<font color="' . $this->aaColour[$chr] . '">' . $chr . '</font>';
		}
 
	# Called by the regexp in tagTfa()
	function urlRegexpCallback($match) {
	  return $match[1] . "[" . $this->url['gi'] . $match[2] . " " . $match[2] . "]" . $match[3];
		}
	# Needed in some versions to prevent Special:Version from breaking
	function __toString() { return 'Fasta'; }
	}
 
# Called from $wgExtensionFunctions array when initialising extensions
function wfSetupFasta() {
	global $wgFasta;
	$wgFasta = new Fasta();
	}
 
# Needed in MediaWiki >1.8.0 for magic word hooks to work properly
function wfFastaLanguageGetMagic(&$magicWords,$langCode = 0) {
	global $wgFastaMagic;
	$magicWords[$wgFastaMagic] = array(0,$wgFastaMagic);
	return true;
	}

?>