| 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);
|
|
|
33 |
fputcsv2($fhProfiles, array($id, $profile), ',', '"', true);
|
|
|
34 |
}
|
|
|
35 |
|
|
|
36 |
if (isset($xml->namevariations)) {
|
|
|
37 |
foreach ($xml->namevariations->name as $n) {
|
|
|
38 |
fputcsv2($fhVariations, array($id, cleanName($n)), ',', '"', true);
|
|
|
39 |
}
|
|
|
40 |
}
|
|
|
41 |
|
|
|
42 |
if (isset($xml->aliases)) {
|
|
|
43 |
foreach ($xml->aliases->name as $n) {
|
|
|
44 |
fputcsv2($fhAlias, array($id, cleanName($n)), ',', '"', true);
|
|
|
45 |
}
|
|
|
46 |
}
|
|
|
47 |
|
|
|
48 |
if (isset($xml->members)) {
|
|
|
49 |
foreach ($xml->members as $n) {
|
|
|
50 |
fputcsv2($fhMembers, array($id, $n->id), ',', '"', true);
|
|
|
51 |
}
|
|
|
52 |
}
|
|
|
53 |
|
|
|
54 |
if (isset($xml->groups)) {
|
|
|
55 |
foreach ($xml->groups->name as $n) {
|
|
|
56 |
fputcsv2($fhGroups, array($id, cleanName($n)), ',', '"', true);
|
|
|
57 |
}
|
|
|
58 |
}
|
|
|
59 |
|
|
|
60 |
if (isset($xml->urls)) {
|
|
|
61 |
foreach ($xml->urls->url as $n) {
|
|
|
62 |
fputcsv2($fhUrls, array($id, $n), ',', '"', true);
|
|
|
63 |
}
|
|
|
64 |
}
|
|
|
65 |
|
|
|
66 |
$arr = array(
|
|
|
67 |
$id,
|
|
|
68 |
$name,
|
|
|
69 |
$realname
|
|
|
70 |
);
|
|
|
71 |
|
|
|
72 |
fputcsv2($fh, $arr, ',', '"', true);
|
|
|
73 |
}
|
|
|
74 |
|
|
|
75 |
return true;
|
|
|
76 |
}
|
|
|
77 |
}
|
|
|
78 |
|
|
|
79 |
$xmlfile = "./in/discogs_20190601_artists.xml.gz";
|
|
|
80 |
|
|
|
81 |
$fh = fopen("out/artists.load", "w+");
|
|
|
82 |
$fhVariations = fopen("out/artistVariations.load", "w+");
|
|
|
83 |
$fhAlias = fopen("out/artistAliases.load", "w+");
|
|
|
84 |
$fhGroups = fopen("out/artistGroups.load", "w+");
|
|
|
85 |
$fhMembers = fopen("out/artistMembers.load", "w+");
|
|
|
86 |
$fhUrls = fopen("out/artistUrls.load", "w+");
|
|
|
87 |
$fhProfiles = fopen("out/artistProfiles.load", "w+");
|
|
|
88 |
|
|
|
89 |
$xmlstream = "compress.zlib://$xmlfile";
|
|
|
90 |
$xmlfileSize = gzfilesize($xmlfile);
|
|
|
91 |
|
|
|
92 |
$streamer = new ArtistsXmlStreamer($xmlstream, $xmlfileSize);
|
|
|
93 |
if ($streamer->parse()) {
|
|
|
94 |
echo "Finished $cnt artists." . PHP_EOL;
|
|
|
95 |
} else {
|
|
|
96 |
echo "Couldn't find root node" . PHP_EOL;
|
|
|
97 |
}
|
|
|
98 |
|
|
|
99 |
fclose($fh);
|
|
|
100 |
fclose($fhVariations);
|
|
|
101 |
fclose($fhAlias);
|
|
|
102 |
fclose($fhGroups);
|
|
|
103 |
fclose($fhMembers);
|
|
|
104 |
fclose($fhUrls);
|
|
|
105 |
fclose($fhProfiles);
|