#!/usr/bin/env php5
<?php

$opts = getopt('hxGtiUu:o:O:d:w:');
$statusUrl = isset($opts['u']) ? $opts['u'] : 'http://osm.gt.owl.de/Strassenliste/Siegen/Status.html';
$datafile = isset($opts['d']) ? $opts['d'] : 'streets.data';

if (isset($opts['h']))
{
    die("\n  Usage: prg [-i | -d | -G | -t -o outfile | -U [-u url]]
        -i : show info about local stored street data
        -d datafile : datefile to use instead of streets.data
        -x : dump all street data
        -G : geocode incomplete streets
        -t -o outfile : create OpenLayers.Text file outfile
        -U [-u url] : Update streetlist from url like http://osm.gt.owl.de/Strassenliste/Siegen/Status.html
        -O lat,lon : Set Origin for placement of streets without coordinates.
        \n\n");
}
$osmWhitelist = array();
if (isset($opts['w'])) // read whitelist. is in osm only but is ok
{
    
    $osmWhitelist = file($opts['w']);
    if ($osmWhitelist === false)
       throw new Exception('cannot read whitelist from "' . $opts['w'] . '".');
var_dump($osmWhitelist);
}


// Update streetlists and street status from remote url
if (isset($opts['U']))
{
    $countNew = 0;
    $countAll = 0;
    $countUpdated = 0;
    $data = readData($datafile) ;

    // read "Komplettliste" and prozess the data:
    $fh = fopen($statusUrl, 'r');
    if (false === $fh):
        throw new Exception('could not open the file "' . $statusUrl . '".');
    endif; 
    $line = '';
    $cnt = 0;
    while ((! feof($fh)) && (strpos($line, '<h2>Komplettliste</h2>') === false)):
        $line = fgets($fh);
        $cnt++;
    endwhile;

    $line = '';
    while ((! feof($fh)) && (strpos($line, '<table class="list">') === false)):
        $line = fgets($fh);
        $cnt++;
    endwhile;

    $line = '';
    while ((! feof($fh)) && (strpos($line, '<table class="list">') === false)):
        $line = fgets($fh);
        $cnt++;
    endwhile;

    echo "Current line offset in HTML: $cnt\n";
    $stateInTr = 0;
    $stateOutTr = 1;
    $state = $stateOutTr;
    $pois = array();
    while ((! feof($fh)) && (false === strpos($line, '</table>')))
    {
        $line = fgets($fh, 50000);
        $lineNumber++;
        if ($state == $stateOutTr)
        {
            if (preg_match('/<tr class="([a-z]*)">/i', $line, $matches))
            {
                $state = $stateInTr;
                $poi = array();
                $poi['status'] = colorToStatus($matches[1]);
            }
        }
        else
        {
            if (preg_match('/<td class="nw">([^<]+)<\/td>/i', $line, $matches))
            {
                $poi['name'] = $matches[1];
            }
            elseif ((false !== strpos($line, 'Google Maps')) && preg_match('/\&q=([^\&]+)\&/', $line, $matches))
            {
                $poi['q'] = $matches[1];
            }
            elseif (preg_match_all('/mlat=([0-9\.]+)\&mlon=([0-9\.]+)\&/', $line, $matches, PREG_SET_ORDER))
            {
                $poi['coords'] = array();
                for ($i = 0; $i < count($matches); $i++)
                {
                    $poi['coords'][] = array($matches[$i][1], $matches[$i][2]); // lat + lon
                }
            }
            elseif (false !== strpos($line, '</tr>'))
            {
                $countAll++;
                // is poi already known? no? get it:
                if (! isset($data['streets'][$poi['name']]))
                {
                    $data['streets'][$poi['name']] = $poi;
                    $countNew++;
                }
                else // merge into existing data:
                {
                    $countUpdated++;
                    $data['streets'][$poi['name']]['status'] = $poi['status']; // statuswechsel übernehmen
                    if (isset($poi['coords']))
                    {
                        $data['streets'][$poi['name']]['coords'] = $poi['coords'];
                    }
                    if (isset($poi['q']))
                    {
                        $data['streets'][$poi['name']]['q'] = $poi['q'];
                    }
                }
                
                $state = $stateOutTr;
            }
        }
    }
    fclose($fh);
    echo 'Updated streetlist from url ' . $statusUrl . "\n";
    $data['last_update_status'] = strftime('%F %T %s');
    storeData($datafile, $data);
    echo "Streets in url: $countAll, new: $countNew, updated: $countUpdated\n";
    echo "Done.\n\n";
}

// Gecode incomplete streets
if (isset($opts['G']))
{
    $countAll = 0;
    $countHadCoords = 0;
    $countUpdated = 0;
    $countFailed = 0;
    $countGeocoding = 0;
    $countWhitelisted = 0;
    $data = readData($datafile);
    foreach ($data['streets'] as $streetname => $street)
    {
        $countAll++;
        if (isset($street['coords']) && count($street['coords']))
        {
            $countHadCoords++;
        }
        else
        {
            $street['coords'] = array();
            // try geocoder:
            if (isset($street['q']))
            {
                $countGeocoding++;
                $coords = getCoords($street['q'], 500000);
                if ($coords !== null)
                {
                    $countUpdated++;
                    $street['coords'][] = array($coords->lat, $coords->lon);
                    $data['streets'][ $streetname ] = $street; // update
                    echo $countGeocoding . ': Successful geocoded: ' . $streetname . ' ; searched for "' . $street['q'] . '".' . "\n";
                }
                else
                {
                    echo $countGeocoding . ': Failed geocoding: ' . $streetname . ' ; searched for "' . $street['q'] . '".' . "\n";
                    $countFailed++;
                }
            }
        }
    }
    $data['last_update_geocoder'] = strftime('%F %T %s');
    storeData($datafile, $data);
    echo "Updated: $countUpdated, failed: $countFailed.\n";
    echo "Count all: $countAll, completeness: " . ((100*($countUpdated + $countHadCoords))/$countAll) . "%\n";
    echo "Count whitelisted: $countWhitelisted\n";
    echo "\n";
    echo "Done.\n\n";
}

// Export layertext:
if (isset($opts['t']) && isset($opts['o']))
{
    $offsetLat = 51; // lon
    $offsetLon = 8; // lat
    if (isset($opts['O']))
    {
        $tmp = explode(",", $opts['O']);
        if (count($tmp) == 2)
        {
            $offsetLat = $tmp[0];
            $offsetLon = $tmp[1];
        }
    }

    $countOrphans = 0;
    $countWhitelisted = 0;
    $outfile = $opts['o'];
    $data = readData($datafile);
    $fh = fopen($outfile, 'w');
    if ($fh === false)
    {
        throw new Exception('Could not create output file "' . $outfile . '".');
    }
    fwrite($fh, "lat\tlon\ttitle\tdescription\ticonSize\ticonOffset\ticon\n");
    $dt = strftime("%d.%m.%Y %H:%M Uhr");
    foreach ($data['streets'] as $streetname => $street)
    {
        // Idea: group without-coords around city
        if (isset($street['coords']) && count($street['coords']))
        {
            if (in_array($street['name'], $osmWhitelist))
            {
                $countWhitelisted++;
                continue;
            }
            $title = str_replace(array("\n", "\t", "\r"), '', $street['name']);
            $description = $street['status'] . '<br>q=' . $street['q']. '<br>updated: ' . $dt;
            $icon = false;
            switch($street['status'])
            {
                case 'not_in_osm':
                    $icon = 'http://www.openlayers.org/dev/img/marker.png';
                    break;
                case 'in_osm_only':
                    $icon = 'http://www.openlayers.org/dev/img/marker-gold.png';
                    break;
                default: continue; // skip others
            }
            if ($icon !== false)
            {
                fwrite($fh, sprintf("%f\t%f\t%s\t%s\t%s\t%s\t%s\n", $street['coords'][0][0], $street['coords'][0][1], $title, $description, "21,25", "-10,-25", $icon));
            }
        }
        else
        {
            $countOrphans++;
            $title = str_replace(array("\n", "\t", "\r"), '', $street['name']) . ' (#' . $countOrphans . ')';
            $description = 'Gibt es diese Straße?';
            $icon = 'http://www.openlayers.org/dev/img/drag-rectangle-on.png';
            $spiral = getSpiralCoords($countOrphans, $offsetLat, $offsetLon);
            fwrite($fh, sprintf("%f\t%f\t%s\t%s\t%s\t%s\t%s\n", $spiral->lat, $spiral->lon, 'Ohne Koordinaten: ' . $title, $description, "21,25", "-10,-25", $icon));
        }
    }
    fclose($fh);
    echo "Written to tonew file '$outfile'.\n";
    echo "countOrphans: $countOrphans, countWhitelisted: $countWhitelisted\n";
    echo "Done.\n\n";
}

// Show info about status:
if (isset($opts['i']))
{
    $data = readData($datafile);
    $countAll = 0;
    $countHadCoords = 0;
    $countColor = array();
    echo 'Last status update: ' . $data['last_update_status'] . "\n";
    echo 'Last status geocoder: ' . $data['last_update_geocoder'] . "\n";
    echo 'Num streets in the database: ' . count($data['streets']) . "\n";
    foreach ($data['streets'] as $streetname => $street)
    {
        $key = strlen($street['status']) > 0 ? $street['status'] : 'unknown';
        if (! isset($countColor[$key]))
        {
            $countColor[$key] = 0;
        }
        $countColor[$key]++;
    }
    echo "Status of streets:\n";
    var_dump($countColor);
    echo "Done.\n\n";
}

// Dump out all data:
if (isset($opts['x']))
{
    $data = readData($datafile);
    echo var_export($data, 1);
}
exit(0);
// ====================================================================
// functions:

/**
 * gecoding
 *
 * see http://code.google.com/support/bin/answer.py?answer=80200&topic=11364
 */
function getCoords($q, $sleepDelayMicroSeconds=1000000)
{
    usleep($sleepDelayMicroSeconds);
    $request_url = 'http://maps.google.de/maps/geo?output=xml&q=' . $q;
    //$xmlCode = file_get_contents($request_url); // simplexml_load_file($request_url) or die("url $request_url not loading");
    $xmlCode = curlDoRequest($request_url);
    if (preg_match('/<coordinates>([0-9\.\-]+),([0-9\.\-]+)/', $xmlCode, $matches))
    {
        $out = new stdClass();
        $out->lat = $matches[2];
        $out->lon = $matches[1];
        return $out;
    }

    // echo 'XML: ' . $xmlCode . "\n";
    return null;
}

function curlDoRequest($url)
{
    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $url);
    curl_setopt($ch, CURLOPT_HEADER, 0);
    curl_setopt($ch, CURLOPT_USERAGENT, "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.0.11) Gecko/2009060308 Ubuntu/9.04 (jaunty) Firefox/3.0.11");
    curl_setopt($ch, CURLOPT_NOBODY, 0);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    $headers = array(
            "Content-type: text/xml;charset=\"utf-8\"",
            "Accept: text/xml",
            "Cache-Control: no-cache",
            "Pragma: no-cache",
    );
    curl_setopt($ch, CURLOPT_HTTPHEADER, $headers); 
    $response = curl_exec($ch);
    curl_close($ch);
    return $response;
}

function storeData($fname, $data)
{
    file_put_contents($fname, serialize($data));
}
function readData($fname)
{
    if (! file_exists($fname))
    {
        storeData($fname, array('streets' => array())); //create empty data
    }
    if (! file_exists($fname)) throw new Exception('datafile "' . $fname . '" not found.');
    $data = unserialize(file_get_contents($fname));
    if ($data === false) throw new Exception('datafile "' . $fname . '" is broken.');
    return $data;
}

function colorToStatus($color)
{
    switch ($color)
    {
        case 'red' : return 'not_in_osm';
        case 'yellow' : return 'in_osm_only';
    }
    return 'ok';
}

/**
 * To place unknown places...
 *
 */
function getSpiralCoords($t, $lat0, $lon0) // num, latOffset, lonOffset
{
    $a = 0.2;
    $b = 0.01;
    $c = 0.001;
    $o = new stdClass();
    $o->lat = $lat0 + sin($t * $a) * ($b + ($c * $t));
    $o->lon = $lon0 + cos($t * $a) * ($b + ($c * $t));    
    return $o; // lat + lon
}