aboutsummaryrefslogtreecommitdiffhomepage
path: root/application/http/MetadataRetriever.php
diff options
context:
space:
mode:
authorArthurHoaro <arthur@hoa.ro>2020-11-12 13:11:07 +0100
committerArthurHoaro <arthur@hoa.ro>2020-11-12 13:11:07 +0100
commitaf50eba28a7bd286de4c8c9ee6dc5216b915d149 (patch)
treeffa30a9358e82d27be75d8fc5e57f3c8820dc6d3 /application/http/MetadataRetriever.php
parentb6f678a5a1d15acf284ebcec16c905e976671ce1 (diff)
parent1409f1c89a7ca01456ae2dcd6357d296e2b99f5a (diff)
downloadShaarli-af50eba28a7bd286de4c8c9ee6dc5216b915d149.tar.gz
Shaarli-af50eba28a7bd286de4c8c9ee6dc5216b915d149.tar.zst
Shaarli-af50eba28a7bd286de4c8c9ee6dc5216b915d149.zip
Merge tag 'v0.12.1' into latestlatest
v0.12.1
Diffstat (limited to 'application/http/MetadataRetriever.php')
-rw-r--r--application/http/MetadataRetriever.php69
1 files changed, 69 insertions, 0 deletions
diff --git a/application/http/MetadataRetriever.php b/application/http/MetadataRetriever.php
new file mode 100644
index 00000000..2e1401ec
--- /dev/null
+++ b/application/http/MetadataRetriever.php
@@ -0,0 +1,69 @@
1<?php
2
3declare(strict_types=1);
4
5namespace Shaarli\Http;
6
7use Shaarli\Config\ConfigManager;
8
9/**
10 * HTTP Tool used to extract metadata from external URL (title, description, etc.).
11 */
12class MetadataRetriever
13{
14 /** @var ConfigManager */
15 protected $conf;
16
17 /** @var HttpAccess */
18 protected $httpAccess;
19
20 public function __construct(ConfigManager $conf, HttpAccess $httpAccess)
21 {
22 $this->conf = $conf;
23 $this->httpAccess = $httpAccess;
24 }
25
26 /**
27 * Retrieve metadata for given URL.
28 *
29 * @return array [
30 * 'title' => <remote title>,
31 * 'description' => <remote description>,
32 * 'tags' => <remote keywords>,
33 * ]
34 */
35 public function retrieve(string $url): array
36 {
37 $charset = null;
38 $title = null;
39 $description = null;
40 $tags = null;
41
42 // Short timeout to keep the application responsive
43 // The callback will fill $charset and $title with data from the downloaded page.
44 $this->httpAccess->getHttpResponse(
45 $url,
46 $this->conf->get('general.download_timeout', 30),
47 $this->conf->get('general.download_max_size', 4194304),
48 $this->httpAccess->getCurlHeaderCallback($charset),
49 $this->httpAccess->getCurlDownloadCallback(
50 $charset,
51 $title,
52 $description,
53 $tags,
54 $this->conf->get('general.retrieve_description'),
55 $this->conf->get('general.tags_separator', ' ')
56 )
57 );
58
59 if (!empty($title) && strtolower($charset) !== 'utf-8') {
60 $title = mb_convert_encoding($title, 'utf-8', $charset);
61 }
62
63 return [
64 'title' => $title,
65 'description' => $description,
66 'tags' => $tags,
67 ];
68 }
69}