Website Scraper

scraper.php

<?php

include 'phpQuery-onefile.php';

$file = 'http://www.usbr.gov/lc/region/g4000/riverdata/gage-month-table.cfm?GAGE=3'; // see below for source

// loads the file
// basically think of your php script as a regular HTML page running client side with jQuery.  This loads whatever file you want to be the current page
phpQuery::newDocumentFileHTML($file);

//Work our way through the DOM tree.
$lastTr = pq("table tr:last-child")->prev("tr")->prev("tr");

$testTd = $lastTr[0]->find("td");

//Is the last row empty? If so use the row before it.
if($testTd[1]->text() == "") {
	$lastTr = $lastTr->prev('tr');
}

//Get the td's inside the table row.
$infoTd = $lastTr[0]->find("td");

//We have to do it this way for some reason (instead of just accessing the indexes directly?)

foreach($infoTd as $k=>$td) {
	$content = trim(pq($td)->text());
	if($k == 1) {
		$elevation = $content;
	}
	else if($k == 3) {
		$pctfull = $content;
	}
	else if($k == 4) {
		$release = $content;
	}

}

//$elevation = trim(pq($infoTd[1])->text());
//$pctfull = trim(pq($infoTd[3])->text());
//$release = trim(pq($infoTd[4])->text());

$avgTr = pq("table tr:last-child")->prev("tr");

$avgTd = $avgTr[0]->find("th");

foreach($avgTd as $k=>$td) {
	$content = trim(pq($td)->text());
	if($k == 1) {
		$avgelevation = $content;
	}
	
	else if($k == 4) {
		$avgrelease = $content;
	}
}

print("Elevation: " . $elevation);
print("Percent Full" . $pctfull);
print("Release" . $release);
print("Avg. Elevation" . $avgelevation);
print("Avg. Release" . $avgrelease);

?>