#! /usr/bin/php
<?php

//Force timezone
date_default_timezone_set('UTC');

//Memcache prefix
define('MEMPFX', 'zipcode_');

//Set first
$first = 0;

//Memcached configuration
$memcfg = array(
	'id' => 'zipcode',
	'address' => 'localhost',
	'port' => 11211
);

//Mysql configuration
$mycfg = array(
	'host' => 'localhost',
	'username' => 'root',
	'passwd' => '',
	'dbname' => 'zipcode'
);

//Ctx configuration
$headers = array(
	'Connection: close',
	'User-agent: Laposte/0.1'
);

//Create mysqli object
$mysqli = new mysqli($mycfg['host'], $mycfg['username'], $mycfg['passwd'], $mycfg['dbname']);

//Set to restart on last zipcode in database
//XXX: use to restart from last zipcode in case of memcache reset since last run
if ($zipcode = $mysqli->query('SELECT MAX(zipcode) AS zipcode FROM zipcode')) {
	if (($row = $zipcode->fetch_row()) && !empty($row[0])) {
		$first = $row[0] + 1;
	}
}

//Search url
$url = 'https://www.laposte.fr/particulier/outils/trouver-un-code-postal';

//Tidy config
$config = array(
	//mostly useless in fact
	'indent' => true,
	//Required to simplify simplexml transition
	'output-xml' => true,
	//Required to avoid xml errors
	'quote-nbsp' => false,
	//Required to fix code
	'clean' => true
);

//Configure memcached
$mem = new Memcached($memcfg['id']);
$mem->setOption(Memcached::OPT_LIBKETAMA_COMPATIBLE, true);
$mem->addServer($memcfg['address'], $memcfg['port']);

//Tidy object
$tidy = new tidy();

//Find out if we have ezxform_token in cache
if (($ezxform_token = $mem->get(MEMPFX.'ezxform_token')) === false || ($cookies = $mem->get(MEMPFX.'cookies')) === false) {
	//Set fake context
	$ctx = stream_context_create(
		array(
			'http' => array(
				'method' => 'GET',
				'max_redirects' => 0,
				'ignore_errors' => 0,
				'header' => $headers+array(
				)
			)
		)
	);

	//Load the page for fetching VilleCP[ezxform_token]
	if (($data = file_get_contents($url, false, $ctx)) !== false) {
		//Parse string
		$tidy->parseString($data, $config, 'utf8');

		//Fix error buffer
		//XXX: don't care about theses errors, tidy is here to fix...
		#if (!empty($tidy->errorBuffer)) {
		#	var_dump($tidy->errorBuffer);
		#	die('Tidy errors');
		#}

		//Load simplexml
		$sx = new SimpleXMLElement($tidy);

		//Set ezxform_token in memcache
		$mem->set(MEMPFX.'ezxform_token', $ezxform_token = (string)$sx->xpath('//input[@id="VilleCP_ezxform_token"]')[0]->attributes()->value, time()+60);

		//Cleanup
		unset($sx);

		//Handle cookies
		$cookies = array();
		foreach($http_response_header as $header) {
			//Add cookie
			if (preg_match('/^Set-Cookie:\s+([^;]+)/', $header, $matches)) {
				parse_str($matches[1], $tmp);
				$cookies += $tmp;
			}
			//Handle cookie reset
			if (preg_match('/^Set-Cookie:\s+([^;]+).*expires=([^;]+)/', $header, $matches) && strtotime($matches[2]) < 100000) {
				parse_str($matches[1], $tmp);
				foreach($tmp as $key => $null) {
					unset($cookies[$key]);
				}
			}
		}

		//Set cookies in memcache
		$mem->set(MEMPFX.'cookies', $cookies, time()+60);
	//Failed get content
	} else {
		die($_SERVER['PHP_SELF'].': file_get_contents(ezxform_token) failed: '.$url);
	}
}

//Find out if we have last code in memcache
if (($last = $mem->get(MEMPFX.'last')) === false) {
	//Set last in memcache
	$mem->set(MEMPFX.'last', ($last = $first), time()+3600*24);
}

//Create insert stmt
if (!($stmt = $mysqli->prepare('INSERT IGNORE INTO zipcode (zipcode, city) VALUES (?, ?)'))) {
	die('Couldn\'t prepare insert');
}

//Loop on all possible postal code
//TODO: see if we need to validate in extra all insee code with letters
for($i = $last; $i <= 10000; $i++) {
	//Build data
	$pdata = http_build_query(
		array(
			'VilleCP[filtre]' => 'ville',
			'VilleCP[communeCode]' => sprintf('%05d', $i),
			'VilleCP[save]' => '',
			'VilleCP[ezxform_token]' => $ezxform_token
		)
	);
	
	echo $pdata.': ';

	//Build cookie
	$cookie = array();
	foreach($cookies as $key => $value) {
		$cookie[] = urlencode($key).'='.urlencode($value);
	}
	$cookie = implode('; ', $cookie);

	//Set post context
	$pctx = stream_context_create(
		array(
			'http' => array(
				'method' => 'POST',
				'max_redirects' => 0,
				'ignore_errors' => 0,
				'header' => $headers+array(
					'Content-type: application/x-www-form-urlencoded',
					'Content-Length: '.strlen($pdata),
					'Cookie: '.$cookie
				),
				'content' => $pdata
			)
		)
	);

	//Load the page for fetching VilleCP[ezxform_token]
	if (($data = file_get_contents($url, false, $pctx)) !== false) {
		//Parse string
		$tidy->parseString($data, $config, 'utf8');

		//Fix error buffer
		//XXX: don't care about theses errors, tidy is here to fix...
		#if (!empty($tidy->errorBuffer)) {
		#	var_dump($tidy->errorBuffer);
		#	die('Tidy errors');
		#}

		//Load simplexml
		$sx = new SimpleXMLElement($tidy);

		//Extract and store ezxform_token in memcache
		$mem->set(MEMPFX.'ezxform_token', $ezxform_token = (string)$sx->xpath('//input[@id="VilleCP_ezxform_token"]')[0]->attributes()->value, time()+30);

		//Handle cookies
		foreach($http_response_header as $header) {
			//Handle cookie reset
			if (preg_match('/^Set-Cookie:\s+([^;]+).*expires=([^;]+)/', $header, $matches) && strtotime($matches[2]) < 100000) {
				parse_str($matches[1], $tmp);
				foreach($tmp as $key => $null) {
					unset($cookies[$key]);
				}
			//Add cookie
			} elseif (preg_match('/^Set-Cookie:\s+([^;]+)/', $header, $matches)) {
				parse_str($matches[1], $tmp);
				$cookies += $tmp;
			}
		}

		//Set cookies in memcache
		$mem->set(MEMPFX.'cookies', $cookies, time()+60);

		//Extract cities
		foreach($sx->xpath('/html/body/div[@id="app"][1]/div[9]/div[3]/div/div/div/div/table/tbody/tr') as $line) {
			$zipcode = trim($line->td[0]);
			$city = trim($line->td[1]);
			if (!$stmt->bind_param('ss', $zipcode, $city)) {
				die('Couldn\'t bind params');
			}
			if (!$stmt->execute()) {
				die('Couldn\'t execute');
			}
		}
	//Failed get content
	} else {
		die($_SERVER['PHP_SELF'].': file_get_contents('.$pdata.') failed: '.$url);
	}

	echo 'OK'."\n";

	//Set last in memcache
	$mem->set(MEMPFX.'last', $i, time()+3600*24);
}

//Close insert request
$stmt->close();

//Close mysqli connection
$mysqli->close();