#!/usr/bin/php
<?
/*****************************************************************************
 * utdcourtsuck - Copyright (c)2004 badpenguins.com                          *
 * Distributed under the terms of the GNU General Public License v2          *
 * Created 2004-09-09 Mike Green <mikey at badpenguins dot com>              *
 *****************************************************************************/
$court_base_href = 'http://www.utd.uscourts.gov';
$court_dump_url  = $court_base_href . '/reports/media/';
$court_dump_file = '/tmp/utdcourt.txt';
$pdftotext       = '/usr/bin/pdftotext';
$wget            = '/usr/bin/wget';
$download_dir    = '/tmp';

$pdf_keywords = array('shaughessey','chris kao','sco v ibm','cravath',
'international business machines','the sco group','caldera',
'david r. marriott','david marriott','brent o. hatch','363-6363',
'boies, schiller','boies schiller','hatch james','david boies',
'mark j. heise','darl mcbride','kevin mcbride','kevin p mcbride',
'2:03cv00294','2:03-cv-00294');

if (is_executable($pdftotext) and is_executable($wget)) {
	if (is_dir($download_dir) and is_writable($download_dir)) {
		$parse_pdfs = true;
		} else {
		echo "Warning: will not attempt to parse pdfs - $download_dir is ";
		echo " not writable or does not exist\n";
		$parse_pdfs = false;
		}
	} else {
	$parse_pdfs = false;
	}

if (! $fd = fopen($court_dump_file,'a+')) {
	echo "Cannot open file for writing: $court_dump_file\n";
	exit(1);
	} else {
	$old_files_ar = unserialize(fgets($fd));
	fclose($fd);
	}

if (!$html_ar = file($court_dump_url)) {
	echo "Cannot retrieve list from $court_dump_url\n";
	exit(1);
	}

$last_line = count($html_ar) - 1;
$text      = $html_ar[$last_line];

$marker = '<br><br>';
$ptr    = strlen($marker);
$pos    = strpos(strtolower($text),$marker);
if ($pos === false) {
	echo "Cannot locate marker: $marker\n";
	exit(1);
	} else {
	$text = substr($text,($pos + $ptr));
	}

$len    = strlen($text);
$marker = '<br></pre>';
$pos    = strpos(strtolower($text),$marker);
if ($pos === false) {
	echo "Cannot locate marker: $marker\n";
	exit(1);
	} else {
	$ctr  = $len - ($len - $pos);
	$text = substr($text,0,$ctr);
	}
$text_ar    = explode("<br>",$text);
$file_dates = array();
$file_times = array();
$file_sizes = array();
$file_urls  = array();
$file_names = array();
$file_exts  = array();
foreach($text_ar as $line) {
	$line      = trim($line);
	$marker    = strpos($line,' ') + 1;
	$date_str  = trim(substr($line,0,$marker));
	$line      = trim(substr($line,$marker));
	$marker    = strpos($line,'  ');
	$time_str  = trim(substr($line,0,$marker));
	$line      = trim(substr($line,$marker));
	$marker    = strpos($line,' ') + 1;
	$size_str  = trim(substr($line,0,$marker));
	$line      = trim(substr($line,$marker));
	$junk      = explode('"',$line);
	$url_str   = $court_base_href . trim($junk[1]);
	$file_name = trim(basename($url_str));
	$junk      = explode('.',$file_name);
	$file_ext  = trim($junk[1]);

	$file_dates[] = $date_str;
	$file_times[] = $time_str;
	$file_sizes[] = $size_str;
	$file_urls[]  = $url_str;
	$file_names[] = $file_name;
	$file_exts[]  = $file_ext;
	}

if (count($file_dates) !== count($file_times) or
    count($file_dates) !== count($file_sizes) or
    count($file_dates) !== count($file_urls) or
    count($file_dates) !== count($file_names) or
    count($file_dates) !== count($file_exts)) {
	echo "Error: mismatched count for new file stats\n";
	exit(1);
	}

if (!is_array($old_files_ar)) {
	$old_file_names = array();
	} else {
	$old_file_dates = $old_files_ar['file_dates'];
	$old_file_times = $old_files_ar['file_times'];
	$old_file_sizes = $old_files_ar['file_sizes'];
	$old_file_urls  = $old_files_ar['file_urls'];
	$old_file_names = $old_files_ar['file_names'];
	$old_file_exts  = $old_files_ar['file_exts'];
	if (!is_array($old_file_dates) or
	    !is_array($old_file_times) or
	    !is_array($old_file_sizes) or
	    !is_array($old_file_urls) or
	    !is_array($old_file_names) or
	    !is_array($old_file_exts) or
	    count($old_file_dates) !== count($old_file_times) or
	    count($old_file_dates) !== count($old_file_sizes) or
	    count($old_file_dates) !== count($old_file_urls) or
	    count($old_file_dates) !== count($old_file_names) or
	    count($old_file_dates) !== count($old_file_exts)) {
		echo "Error reading old file stats\n";
		exit(1);
		}
	}

$new_file_dates = array();
$new_file_times = array();
$new_file_sizes = array();
$new_file_urls  = array();
$new_file_names = array();
$new_file_exts  = array();
for($j=0; $j < count($file_names); $j++) {
	$file_name = $file_names[$j];
	if (in_array($file_name,$old_file_names)) {
		continue;
		} else {
		$new_file_dates[] = $file_dates[$j];
		$new_file_times[] = $file_times[$j];
		$new_file_sizes[] = $file_sizes[$j];
		$new_file_urls[]  = $file_urls[$j];
		$new_file_names[] = $file_names[$j];
		$new_file_exts[]  = $file_exts[$j];
		}
	}

$new = count($new_file_names);
if ($new > 0) {
	$msg = "\n$new new files at $court_dump_url:\n";
	for ($j=0; $j < count($new_file_names); $j++) {
		$f_date = $new_file_dates[$j];
		$f_time = $new_file_times[$j];
		$f_size = $new_file_sizes[$j];
		$f_url  = $new_file_urls[$j];
		$f_name = $new_file_names[$j];
		$f_ext  = $new_file_exts[$j];
		$old_file_dates[] = $f_date;
		$old_file_times[] = $f_time;
		$old_file_sizes[] = $f_size;
		$old_file_urls[]  = $f_url;
		$old_file_names[] = $f_name;
		$old_file_exts[]  = $f_ext;

		$related = "Unknown";
		if (strtolower($f_ext) == "pdf" and $parse_pdfs) {
			echo "Fetching: $f_url\n";
			$out = "$download_dir/$f_name";
			$cmd = "$wget -t 5 -q '$f_url' -O $out";
			exec($cmd,$junk,$rc);
			if ($rc !== 0 or !is_file($out)) {
				echo "Error: failed to fetch $f_url\n";
				} else {
				echo "Scanning: $out\n";
				//$cmd = "$pdftotext -nopgbrk -q $out $out.txt";
				//exec($cmd,$junk,$rc);
				$cmd = "$pdftotext -nopgbrk -q $out -";
				exec($cmd,$output,$rc);
				if ($rc !== 0 or !is_array($output)) {
					echo "Error: failed on $cmd\n";
					} else {
					$hit = false;
					foreach($output as $line) {
						if ($hit === true) { break; }
						$line = strtolower($line);
						foreach($pdf_keywords as $keyword) {
							if ($hit === true) { break; }
							$pos = strpos($line,$keyword);
							if (! $pos === false) {
								$hit = true;
								$related = "**YES** keyword: $keyword";
								}
							}
						}
					if ($hit === false) { @unlink($out); }
					}
				}
			}

		$msg .= "\nFile URL: $f_url\n";
		$msg .= "   File Name:   $f_name\n";
		$msg .= "   File Date:   $f_date\n";
		$msg .= "   File Size:   $f_size\n";
		$msg .= "   SCO Related: $related\n";
		}
	$new_ar = array('file_dates'=>$old_file_dates,
	                'file_times'=>$old_file_times,
	                'file_sizes'=>$old_file_sizes,
	                'file_urls'=>$old_file_urls,
	                'file_names'=>$old_file_names,
	                'file_exts'=>$old_file_exts);
	$new_ar = serialize($new_ar);
	echo $msg;

	if (! $fd = fopen($court_dump_file,'w')) {
		echo "\nError: Cannot re-open file for writing: $court_dump_file\n";
		exit(1);
		} else {
		if (!fputs($fd,$new_ar)) {
			echo "\nWARNING: Could not re-create $court_dump_file\n";
			} else {
			fclose($fd);
			}
		}
	}

?>