Subversion Repositories cheapmusic

Rev

Rev 18 | Rev 58 | Go to most recent revision | 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');
4 
ini_set("memory_limit", "256M");
5 
 
6 
$cnt = 0;
7 
 
8 
class ArtistsXmlStreamer extends XmlStreamer
9 
{
10 
	public function processNode($xmlString, $elementName, $nodeIndex)
11 
	{
12 
		global $cnt;
13 
		global $fh;
14 
		global $fhVariations;
15 
		global $fhAlias;
16 
		global $fhGroups;
17 
		global $fhMembers;
18 
		global $fhUrls;
19 
		global $fhProfiles;
20 
 
21 
		$xml = simplexml_load_string($xmlString, 'simple_xml_extended');
22 
 
23 
		if ($elementName == 'artist') {
24 
			++$cnt;
25 
 
26 
			$id = $xml->id;
27 
			$name = cleanName($xml->name);
28 
			$realname = $xml->realname;
29 
 
30 
			if (isset($xml->profile) && !empty($xml->profile)) {
31 
				$profile = str_replace("\r", '', $xml->profile);
32 
				$profile = rtrim($profile);
44 - 33 
				if (!empty($profile)) {
34 
					fputcsv2($fhProfiles, array($id, $profile), ',', '"');
35 
				}
18 - 36 
			}
37 
 
38 
			if (isset($xml->namevariations)) {
39 
				foreach ($xml->namevariations->name as $n) {
44 - 40 
					$cn = cleanName($n);
41 
					if (!empty($cn)) {
42 
						fputcsv2($fhVariations, array($id, $cn), ',', '"');
43 
					}
18 - 44 
				}
45 
			}
46 
 
47 
			if (isset($xml->aliases)) {
48 
				foreach ($xml->aliases->name as $n) {
44 - 49 
					$cn = cleanName($n);
50 
					if (!empty($cn)) {
51 
						fputcsv2($fhAlias, array($id, $cn), ',', '"');
52 
					}
18 - 53 
				}
54 
			}
55 
 
56 
			if (isset($xml->members)) {
57 
				foreach ($xml->members as $n) {
44 - 58 
					if (!empty($n->id)) {
59 
						fputcsv2($fhMembers, array($id, $n->id), ',', '"');
60 
					}
18 - 61 
				}
62 
			}
63 
 
64 
			if (isset($xml->groups)) {
65 
				foreach ($xml->groups->name as $n) {
44 - 66 
					$cn = cleanName($n);
67 
					if (!empty($cn)) {
68 
						fputcsv2($fhGroups, array($id, $cn), ',', '"');
69 
					}
18 - 70 
				}
71 
			}
72 
 
73 
			if (isset($xml->urls)) {
74 
				foreach ($xml->urls->url as $n) {
44 - 75 
					if (!empty($n)) {
76 
						fputcsv2($fhUrls, array($id, $n), ',', '"');
77 
					}
18 - 78 
				}
79 
			}
80 
 
81 
			$arr = array(
82 
				$id,
83 
				$name,
84 
				$realname
85 
			);
86 
 
44 - 87 
			fputcsv2($fh, $arr, ',', '"');
18 - 88 
		}
89 
 
90 
		return true;
91 
	}
92 
}
93 
 
44 - 94 
$xmlfile = "./in/discogs_20190701_artists.xml.gz";
18 - 95 
 
96 
$fh = fopen("out/artists.load", "w+");
97 
$fhVariations = fopen("out/artistVariations.load", "w+");
98 
$fhAlias = fopen("out/artistAliases.load", "w+");
99 
$fhGroups = fopen("out/artistGroups.load", "w+");
100 
$fhMembers = fopen("out/artistMembers.load", "w+");
101 
$fhUrls = fopen("out/artistUrls.load", "w+");
102 
$fhProfiles = fopen("out/artistProfiles.load", "w+");
103 
 
104 
$xmlstream = "compress.zlib://$xmlfile";
105 
$xmlfileSize = gzfilesize($xmlfile);
106 
 
107 
$streamer = new ArtistsXmlStreamer($xmlstream, $xmlfileSize);
108 
if ($streamer->parse()) {
109 
	echo "Finished $cnt artists." . PHP_EOL;
110 
} else {
111 
	echo "Couldn't find root node" . PHP_EOL;
112 
}
113 
 
114 
fclose($fh);
115 
fclose($fhVariations);
116 
fclose($fhAlias);
117 
fclose($fhGroups);
118 
fclose($fhMembers);
119 
fclose($fhUrls);
120 
fclose($fhProfiles);