Subversion Repositories cheapmusic

Rev

Rev 58 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
18 - 1 
<?php
2 
require_once('php/XmlStreamer.php');
3 
require_once('php/tools.php');
78 - 4 
require_once('php/constants.php');
18 - 5 
ini_set("memory_limit", "256M");
6 
 
7 
$cnt = 0;
8 
 
9 
class ArtistsXmlStreamer extends XmlStreamer
10 
{
11 
	public function processNode($xmlString, $elementName, $nodeIndex)
12 
	{
13 
		global $cnt;
14 
		global $fh;
15 
		global $fhVariations;
16 
		global $fhAlias;
17 
		global $fhGroups;
18 
		global $fhMembers;
19 
		global $fhUrls;
20 
		global $fhProfiles;
21 
 
22 
		$xml = simplexml_load_string($xmlString, 'simple_xml_extended');
23 
 
24 
		if ($elementName == 'artist') {
25 
			++$cnt;
26 
 
27 
			$id = $xml->id;
28 
			$name = cleanName($xml->name);
29 
			$realname = $xml->realname;
30 
 
31 
			if (isset($xml->profile) && !empty($xml->profile)) {
32 
				$profile = str_replace("\r", '', $xml->profile);
33 
				$profile = rtrim($profile);
44 - 34 
				if (!empty($profile)) {
35 
					fputcsv2($fhProfiles, array($id, $profile), ',', '"');
36 
				}
18 - 37 
			}
38 
 
39 
			if (isset($xml->namevariations)) {
40 
				foreach ($xml->namevariations->name as $n) {
44 - 41 
					$cn = cleanName($n);
42 
					if (!empty($cn)) {
43 
						fputcsv2($fhVariations, array($id, $cn), ',', '"');
44 
					}
18 - 45 
				}
46 
			}
47 
 
48 
			if (isset($xml->aliases)) {
49 
				foreach ($xml->aliases->name as $n) {
44 - 50 
					$cn = cleanName($n);
51 
					if (!empty($cn)) {
52 
						fputcsv2($fhAlias, array($id, $cn), ',', '"');
53 
					}
18 - 54 
				}
55 
			}
56 
 
57 
			if (isset($xml->members)) {
58 
				foreach ($xml->members as $n) {
44 - 59 
					if (!empty($n->id)) {
60 
						fputcsv2($fhMembers, array($id, $n->id), ',', '"');
61 
					}
18 - 62 
				}
63 
			}
64 
 
65 
			if (isset($xml->groups)) {
66 
				foreach ($xml->groups->name as $n) {
44 - 67 
					$cn = cleanName($n);
68 
					if (!empty($cn)) {
69 
						fputcsv2($fhGroups, array($id, $cn), ',', '"');
70 
					}
18 - 71 
				}
72 
			}
73 
 
74 
			if (isset($xml->urls)) {
75 
				foreach ($xml->urls->url as $n) {
44 - 76 
					if (!empty($n)) {
77 
						fputcsv2($fhUrls, array($id, $n), ',', '"');
78 
					}
18 - 79 
				}
80 
			}
81 
 
82 
			$arr = array(
83 
				$id,
84 
				$name,
85 
				$realname
86 
			);
87 
 
44 - 88 
			fputcsv2($fh, $arr, ',', '"');
18 - 89 
		}
90 
 
91 
		return true;
92 
	}
93 
}
94 
 
78 - 95 
$xmlfile = "./in/discogs_" . XMLFILEDATE . "_artists.xml.gz";
18 - 96 
 
97 
$fh = fopen("out/artists.load", "w+");
98 
$fhVariations = fopen("out/artistVariations.load", "w+");
99 
$fhAlias = fopen("out/artistAliases.load", "w+");
100 
$fhGroups = fopen("out/artistGroups.load", "w+");
101 
$fhMembers = fopen("out/artistMembers.load", "w+");
102 
$fhUrls = fopen("out/artistUrls.load", "w+");
103 
$fhProfiles = fopen("out/artistProfiles.load", "w+");
104 
 
105 
$xmlstream = "compress.zlib://$xmlfile";
106 
$xmlfileSize = gzfilesize($xmlfile);
107 
 
108 
$streamer = new ArtistsXmlStreamer($xmlstream, $xmlfileSize);
109 
if ($streamer->parse()) {
110 
	echo "Finished $cnt artists." . PHP_EOL;
111 
} else {
112 
	echo "Couldn't find root node" . PHP_EOL;
113 
}
114 
 
115 
fclose($fh);
116 
fclose($fhVariations);
117 
fclose($fhAlias);
118 
fclose($fhGroups);
119 
fclose($fhMembers);
120 
fclose($fhUrls);
121 
fclose($fhProfiles);