Tryag File Manager

//home/cluster1/data/bu01/1121861/html/xfst_dir_copy/find_verbs.php4

<?
include_once "utf8/utf8.class.php";

$utfConverter = new utf8(); //defaults to CP1250.

$xfst_path = "/spd25/htdocs/hyperlex2/nahuatl/xfst_dir/xfst -utf8 -q -e 'loadd SurfaceForms.fsm' -e 'regex GlSR;'";

$HOST = "wave.ldc.upenn.edu";
$USER = "hyperlex";	
$PASSWORD = "jota";
$DATABASE = "hyperlex";
$TABLE = $language;

@$db = mysql_pconnect($HOST, $USER, $PASSWORD);

if(!$db) {
    echo "Error: Could not connect to database. Please try again later.";
    exit;
}
mysql_select_db($DATABASE);

if($story != "") {
    $in = file($story);
} else {
    $in = array($lines);
}

unset($words);
$verbs = array();
$words = array();
$dict_words = array();
$commands = $xfst_path;
$query = "SELECT lxa_d, lxo_d FROM nahuatl WHERE (";
foreach($in as $line) {
    $line = trim($line);
    //$line = utf8_encode($line);
    $line = ereg_replace("[?,;'!]","",$line);
    $line = $utfConverter->strToUtf8($line);
    $line = explode(" ",$line);
    foreach($line as $word) {
	$word = trim($word);
	if(!strpos($word,"...")) {
	    $word = ereg_replace("[.]","",$word);
	    if(($word != "") && !ereg("=|[0-9]|Roman",$word)){
		$word = ereg_replace("<.*>","",$word);
		$word = ereg_replace("[<>|{}]","",$word);
		$word = str_replace("\"","",$word);
		$word = str_replace("*","",$word);
		$word = str_replace("\\","",$word);
		$word = str_replace("/","",$word);
		$word = str_replace("\"","",$word);
		$word = str_replace("'","",$word);
		$word = str_replace("@","",$word);
		
		if(strpos($word,"]")) {
		    $with_brackets = $word;
		    $word = str_replace("[","",$word);
		    $word = str_replace("]","",$word);
		    $bracketed_words[$with_brackets] = $word;
		}
		$word = trim($word);
		if($word != "") {
		    //echo "word :$word \n";
		    $commands .= " -e 'apply up ".strtolower($word)."'";
		    $word = $utfConverter->utf8ToStr($word);
		    if(ereg("^[a-z]*",$word)) {
		        $words[] = $word;
		    }
		}
	    }
	}
    }
}

$pre = "SELECT lxa_d, lxo_d FROM nahuatl WHERE (";
$count = 0;
foreach($words as $word) {
    if($count == 100) {
	$query = substr($query,0,-4).")";	
	$query_result = mysql_query($query);
	if($query_result) {
	    while($row = mysql_fetch_assoc($query_result)) {
		$lxa = trim($row["lxa_d"]);
		$lxo = trim($row["lxo_d"]);
		if($lxa != "" && !in_array($lxa,$dict_words)) {
		    $dict_words[] = $lxa;
		}
		if($lxo != "" && !in_array($lxo,$dict_words)) {
		    $dict_words[] = $lxo;
		}
	    }
	}
	else {
	    mysql_close();
	    @$db = mysql_pconnect($HOST, $USER, $PASSWORD);
	    mysql_select_db($DATABASE);
	}
	
	$query = $pre;
	$count = 0;
    }
    else {
	$query .= "(lxa_d LIKE '$word' OR lxo_d LIKE '$word') OR ";
	$count++;
    }
}

$query = substr($query,0,-4).")";	
$query_result = mysql_query($query);
if($query_result) {
    while($row = mysql_fetch_assoc($query_result)) {
	$lxa = trim($row["lxa_d"]);
	$lxo = trim($row["lxo_d"]);
	if($lxa != "" && !in_array($lxa,$dict_words)) {
	    $dict_words[] = $lxa;
	}
	if($lxo != "" && !in_array($lxo,$dict_words)) {
	    $dict_words[] = $lxo;
	}
    }
}

$commands .= " -stop";
unset($parsable);
exec($commands,$parsable);

$commands = $xfst_path;
foreach($parsable as $p) {
    $commands .= " -e 'apply down $p'";
}

$commands .= " -stop";
unset($verbs);
exec($commands,$verbs);
sort($verbs);
$verbs = array_unique($verbs);

/*
foreach($verbs as $v) {
    echo "$v \n";
}
*/

foreach($verbs as $v) {
    //$v = utf8_decode($v);
    $v = $utfConverter->utf8ToStr($v);
    $with_brackets = $bracketed[$v];    
}

reset($in);
if($story != "") {
    $out = fopen("linked_story.html","w");
}

foreach($in as $line) {
    $new_line = "";
    $line = trim($line);
    $line = explode(" ",$line);
    foreach($line as $word) {
	if(!strpos($word,"...")) {
	    $trimmed_word = trim($word);
	    $trimmed_word = strtolower($trimmed_word);
	    $trimmed_word = ereg_replace("[.,!?']","",$trimmed_word);
	    
	    if(in_array($trimmed_word,$verbs)) {
		$word = ereg_replace($trimmed_word,"<A href='http://hyperlex2.ldc.upenn.edu/nahuatl/xfst_dir/verb_forms.php4?condition=story&dialect=Oa&word=$trimmed_word' target='parser'>$trimmed_word</A>",$word);
		$upper = ucwords($trimmed_word);
		$word = ereg_replace($upper,"<A href='http://hyperlex2.ldc.upenn.edu/nahuatl/xfst_dir/verb_forms.php4?condition=story&dialect=Oa&word=$trimmed_word' target='parser'>$upper</A>",$word);
	    }
	    else if(in_array($trimmed_word,$dict_words)) {
		$word = ereg_replace($trimmed_word,"<A href='http://hyperlex2.ldc.upenn.edu/nahuatl/search_results.php4?user_lang=english&condition=word_search&entry_template=englishwhole.html&word=$trimmed_word'>$trimmed_word</A>",$word);
	    }	    
	}
	$new_line .= "$word ";
    }
    echo $new_line;
    $new_line .= "\n";
    if($story != "") {
	fwrite($out,$new_line);
    }
}

if($story != "") {
    fclose($out);
}
?>