|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace BilginPro\Agency\Iha; |
|
4
|
|
|
|
|
5
|
|
|
use Carbon\Carbon; |
|
6
|
|
|
use GuzzleHttp; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class Crawler |
|
10
|
|
|
* @package BilginPro\Ajans\Iha |
|
11
|
|
|
*/ |
|
12
|
|
|
class Crawler |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* @var string |
|
16
|
|
|
*/ |
|
17
|
|
|
protected $user_code = ''; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var string |
|
21
|
|
|
*/ |
|
22
|
|
|
protected $user_name = ''; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var string |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $password = ''; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @var int |
|
31
|
|
|
*/ |
|
32
|
|
|
protected $summary_lenth = 150; |
|
33
|
|
|
|
|
34
|
|
|
protected $attributes = [ |
|
35
|
|
|
'limit' => '5', |
|
36
|
|
|
]; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Create a new Crawler Instance |
|
40
|
|
|
*/ |
|
41
|
|
|
public function __construct($config) |
|
42
|
|
|
{ |
|
43
|
|
|
$this->setParameters($config); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Does the magic. |
|
48
|
|
|
* @return array |
|
49
|
|
|
*/ |
|
50
|
|
|
public function crawl($attributes = []) |
|
51
|
|
|
{ |
|
52
|
|
|
$this->setAttributes($attributes); |
|
53
|
|
|
|
|
54
|
|
|
$response = $this->fetchUrl($this->getUrl()); |
|
55
|
|
|
$xml = new \SimpleXMLElement($response); |
|
56
|
|
|
$result = []; |
|
57
|
|
|
$i = 0; |
|
58
|
|
|
foreach ($xml->channel->item as $item) { |
|
59
|
|
|
if ($this->attributes['limit'] > $i) { |
|
60
|
|
|
$news = new \stdClass; |
|
61
|
|
|
$news->code = (string)$item->HaberKodu; |
|
62
|
|
|
$news->title = (string)$item->title; |
|
63
|
|
|
$news->summary = (string)$this->shortenString($item->description, $this->summary_lenth); |
|
64
|
|
|
$news->content = (string)$item->description; |
|
65
|
|
|
$news->created_at = (new Carbon($item->pubDate))->format('d.m.Y H:i:s'); |
|
66
|
|
|
$news->category = $this->titleCase($item->Kategori); |
|
67
|
|
|
$news->city = (!empty($item->Sehir) ? $this->titleCase($item->Sehir) : ''); |
|
68
|
|
|
$news->images = []; |
|
69
|
|
|
if (isset($item->images) && count($item->images->image) > 0) { |
|
70
|
|
|
foreach ($item->images->image as $image) { |
|
71
|
|
|
$news->images[] = (string)$image; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
$result[] = $news; |
|
76
|
|
|
$i++; |
|
77
|
|
|
} |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
return $result; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Sets config parameters. |
|
85
|
|
|
*/ |
|
86
|
|
|
protected function setParameters($config) |
|
87
|
|
|
{ |
|
88
|
|
|
if (!is_array($config)) { |
|
89
|
|
|
throw new \InvalidArgumentException('$config variable must be an array.'); |
|
90
|
|
|
} |
|
91
|
|
|
if (array_key_exists('user_code', $config)) { |
|
92
|
|
|
$this->user_code = $config['user_code']; |
|
93
|
|
|
} |
|
94
|
|
|
if (array_key_exists('user_name', $config)) { |
|
95
|
|
|
$this->user_name = $config['user_name']; |
|
96
|
|
|
} |
|
97
|
|
|
if (array_key_exists('password', $config)) { |
|
98
|
|
|
$this->password = $config['password']; |
|
99
|
|
|
} |
|
100
|
|
|
if (array_key_exists('summary_lenth', $config)) { |
|
101
|
|
|
$this->summary_lenth = $config['summary_lenth']; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* Sets filter attributes. |
|
107
|
|
|
* @param $attributes array |
|
108
|
|
|
*/ |
|
109
|
|
|
protected function setAttributes($attributes) |
|
110
|
|
|
{ |
|
111
|
|
|
foreach ($attributes as $key => $value) { |
|
112
|
|
|
$this->attributes[$key] = $value; |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Returns full url for crawling. |
|
118
|
|
|
* @return string |
|
119
|
|
|
*/ |
|
120
|
|
|
protected function getUrl() |
|
121
|
|
|
{ |
|
122
|
|
|
$url = 'http://abone.iha.com.tr/yeniabone/rss2.aspx?Sehir=0&UserCode=' |
|
123
|
|
|
. $this->user_code |
|
124
|
|
|
. '&UserName=' |
|
125
|
|
|
. $this->user_name |
|
126
|
|
|
. '&UserPassword=' |
|
127
|
|
|
. $this->password; |
|
128
|
|
|
|
|
129
|
|
|
return $url; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Fethches given url and returns response as string. |
|
135
|
|
|
* @param $url |
|
136
|
|
|
* @param string $method |
|
137
|
|
|
* @param array $options |
|
138
|
|
|
* |
|
139
|
|
|
* @return string |
|
140
|
|
|
*/ |
|
141
|
|
|
protected function fetchUrl($url, $method = 'GET', $options = []) |
|
142
|
|
|
{ |
|
143
|
|
|
$client = new GuzzleHttp\Client(); |
|
144
|
|
|
$res = $client->request($method, $url, $options); |
|
145
|
|
|
if ($res->getStatusCode() == 200) { |
|
146
|
|
|
return (string)$res->getBody(); |
|
147
|
|
|
} |
|
148
|
|
|
return ''; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* Cuts the given string from the end of the appropriate word. |
|
153
|
|
|
* @param $str |
|
154
|
|
|
* @param $len |
|
155
|
|
|
* @return string |
|
156
|
|
|
*/ |
|
157
|
|
|
protected function shortenString($str, $len) |
|
158
|
|
|
{ |
|
159
|
|
|
if (strlen($str) > $len) { |
|
160
|
|
|
$str = rtrim(mb_substr($str, 0, $len, 'UTF-8')); |
|
161
|
|
|
$str = substr($str, 0, strrpos($str, ' ')); |
|
162
|
|
|
$str .= '...'; |
|
163
|
|
|
$str = str_replace(',...', '...', $str); |
|
164
|
|
|
} |
|
165
|
|
|
return $str; |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
|
|
/** |
|
169
|
|
|
* Converts a string to "Title Case" |
|
170
|
|
|
* @param $str |
|
171
|
|
|
* @return string |
|
172
|
|
|
*/ |
|
173
|
|
|
protected function titleCase($str) |
|
174
|
|
|
{ |
|
175
|
|
|
$str = mb_convert_case($str, MB_CASE_TITLE, 'UTF-8'); |
|
176
|
|
|
return $str; |
|
177
|
|
|
} |
|
178
|
|
|
} |
|
179
|
|
|
|
bilginpro /
agency-iha
