Importing data

<slide>
<title>Importing data</title>

<example><![CDATA[<?php
$file = $argv[1];
$m = new MongoClient( 'mongodb://localhost:27017' );
$d = $m->selectCollection( 'demo', 'poi' );
$d->drop();
$d->ensureIndex( array( 'loc' => '2d' ) );
$d->ensureIndex( array( 'tags_indexed.v' => 1, 'tags_indexed.k' => 1 ) );

$locations = array();

$z = new XMLReader();
$z->open( $argv[1]);
while ($z->read() && $z->name !== 'node' );
$c = 0;

while ($z->name === 'node') {
    $dom = new DomDocument;
    $node = simplexml_import_dom($dom->importNode($z->expand(), true));

    $q = array();
    $q['id'] = "n" . (string) $node['id'];
    $q['type'] = 1;
    $q['loc'] = array( (float) $node['lon'], (float) $node['lat'] );
    parseNode($q, $node);
	$d->insert( $q );

    // also keep a cache of all nodes and locations
    $locations[(string) $node['id']] = "{$node['lon']}|{$node['lat']}";

    $z->next('node');
    $c++;
    if ($c % 1000 === 0) {
        echo ".";
    }
    if ($c % 100000 === 0) {
        echo "\n", $c, "\n";
    }
}
echo "\n";

$z = new XMLReader();
$z->open( $argv[1]);
while ($z->read() && $z->name !== 'way' );
$c = 0;

while ($z->name === 'way') {
    $dom = new DomDocument;
    $way = simplexml_import_dom($dom->importNode($z->expand(), true));

    $q = array();
    $q['id'] = "w" . (string) $node['id'];
    $q['type'] = 2;
    parseNode($q, $way);
    fetchLocations($q, $locations, $way);
    if (count($q['tags']) > 0) {
        $d->insert( $q );
    }

    $z->next('way');
    if (++$c % 100 === 0) {
        echo ".";
    }
    if ($c % 10000 === 0) {
        echo "\n", $c, "\n";
    }
}
echo "\n";


function fetchLocations(&$q, $nodes, $node)
{
    $locations = array();
    foreach ($node->nd as $nd) {
        if (isset( $nodes[(string) $nd['ref']])) {
            $nr = explode('|', $nodes[(string) $nd['ref']]);
            $locations[] = array( (float) $nr[0], (float) $nr[1] );
        }
    }
    $q['loc'] = $locations;
}

function parseNode(&$q, $sxml)
{
    $standAloneNode = false;
    $street = $postcode = $nr = $cuisine = false;
    $tags = $tagsIndexed = array();
    $ignoreTags = array( 'created_by', 'abutters' );

    foreach( $sxml->tag as $tag )
    {
        if ($tag['k'] == 'name') {
            $standAloneNode = true;
        } else {
            if (!in_array( $tag['k'], $ignoreTags)) {
                $tags[(string) $tag['k']]  = (string) $tag['v'];
                $tagsIndexed[] = array( 'k' => (string) $tag['k'], 'v' => (string) $tag['v'] );
            }
        }
        if ($tag['k'] == 'addr:street') {
            $street = $tag['v'];
        }
        if ($tag['k'] == 'addr:postcode') {
            $postcode = strtoupper( (string) $tag['v'] );
        }
        if ($tag['k'] == 'addr:housenumber') {
            $nr = $tag['v'];
        }
    }

    if ( $street || $nr || $postcode )
    {
        $q['address'] = trim( "{$street} {$nr} {$postcode}");
    }
    $q['tags'] = $tags;
    $q['tags_indexed'] = $tagsIndexed;
}
]]></example>
</slide>