Skip to content

Commit f7b0ebd

Browse files
author
Taois
committed
feat: 适配PHP
1 parent d098ade commit f7b0ebd

File tree

5 files changed

+975
-2
lines changed

5 files changed

+975
-2
lines changed

‎drpy-node-bundle/localDsCoreTest.js‎

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -8,5 +8,7 @@ console.log(getEngine)
88
// console.log(b)
99
// const c = await getEngine('果果[官]', {do: 'cat', ac: 'list', t: '3'})
1010
// console.log(c)
11-
const d = await getEngine('设置中心', {do: 'ds'})
12-
console.log(d)
11+
// const d = await getEngine('设置中心', {do: 'ds'})
12+
// console.log(d)
13+
const e = await getEngine('B站 ᵈᶻ', {do: 'php'})
14+
console.log(e)

‎drpy-node-bundle/spider/php/B站 ᵈᶻ.php‎

Lines changed: 266 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,266 @@
1+
<?php
2+
/**
3+
* B站视频爬虫 - PHP 适配版 (道长重构)
4+
* 按照 BaseSpider 结构重写
5+
*/
6+
7+
require_once __DIR__ . '/lib/spider.php';
8+
9+
class Spider extends BaseSpider {
10+
11+
private $cookie = [];
12+
13+
public function init($extend = '') {
14+
$this->headers['Referer'] = "https://www.bilibili.com";
15+
// 配置初始 Cookie
16+
// 实际使用时，建议通过 ext 传入 cookie
17+
$configCookie = 'buvid3=xxxx; SESSDATA=xxxx;';
18+
19+
// 尝试从 extend 获取 cookie (假设 extend 是 JSON 字符串或直接是 cookie 字符串)
20+
// 这里简化处理：如果 extend 包含 SESSDATA，则认为是 cookie
21+
if (!empty($extend)) {
22+
if (strpos($extend, 'SESSDATA') !== false) {
23+
$configCookie = $extend;
24+
} elseif (is_array($extend) && isset($extend['cookie'])) {
25+
$configCookie = $extend['cookie'];
26+
} else {
27+
// 尝试解析 json
28+
$json = json_decode($extend, true);
29+
if (isset($json['cookie'])) {
30+
$configCookie = $json['cookie'];
31+
}
32+
}
33+
}
34+
35+
$this->cookie = $this->parseCookie($configCookie);
36+
}
37+
38+
private function parseCookie($cookieStr) {
39+
if (empty($cookieStr)) return [];
40+
$cookies = [];
41+
$pairs = explode(';', $cookieStr);
42+
foreach ($pairs as $pair) {
43+
$pair = trim($pair);
44+
if (strpos($pair, '=') !== false) {
45+
list($name, $value) = explode('=', $pair, 2);
46+
$cookies[trim($name)] = trim($value);
47+
}
48+
}
49+
return $cookies;
50+
}
51+
52+
private function buildCookieString() {
53+
$pairs = [];
54+
foreach ($this->cookie as $name => $value) {
55+
$pairs[] = $name . '=' . $value;
56+
}
57+
return implode('; ', $pairs);
58+
}
59+
60+
// 覆盖父类 fetch 以自动添加 cookie
61+
protected function fetch($url, $options = [], $headers = []) {
62+
if (!isset($options['cookie'])) {
63+
$cookieStr = $this->buildCookieString();
64+
if (!empty($cookieStr)) {
65+
$options['cookie'] = $cookieStr;
66+
}
67+
}
68+
return parent::fetch($url, $options, $headers);
69+
}
70+
71+
public function homeContent($filter = []) {
72+
$classes = [
73+
["type_id" => "沙雕仙逆", "type_name" => "傻屌仙逆"],
74+
["type_id" => "沙雕动画", "type_name" => "沙雕动画"],
75+
["type_id" => "纪录片超清", "type_name" => "纪录片"],
76+
["type_id" => "演唱会超清", "type_name" => "演唱会"],
77+
["type_id" => "音乐超清", "type_name" => "流行音乐"],
78+
["type_id" => "美食超清", "type_name" => "美食"],
79+
["type_id" => "食谱", "type_name" => "食谱"],
80+
["type_id" => "体育超清", "type_name" => "体育"],
81+
["type_id" => "球星", "type_name" => "球星"],
82+
["type_id" => "中小学教育", "type_name" => "教育"],
83+
["type_id" => "幼儿教育", "type_name" => "幼儿教育"],
84+
["type_id" => "旅游", "type_name" => "旅游"],
85+
["type_id" => "风景4K", "type_name" => "风景"],
86+
["type_id" => "说案", "type_name" => "说案"],
87+
["type_id" => "知名UP主", "type_name" => "知名UP主"],
88+
["type_id" => "探索发现超清", "type_name" => "探索发现"],
89+
["type_id" => "鬼畜", "type_name" => "鬼畜"],
90+
["type_id" => "搞笑超清", "type_name" => "搞笑"],
91+
["type_id" => "儿童超清", "type_name" => "儿童"],
92+
["type_id" => "动物世界超清", "type_name" => "动物世界"],
93+
["type_id" => "相声小品超清", "type_name" => "相声小品"],
94+
["type_id" => "戏曲", "type_name" => "戏曲"],
95+
["type_id" => "解说", "type_name" => "解说"],
96+
["type_id" => "演讲", "type_name" => "演讲"],
97+
["type_id" => "小姐姐超清", "type_name" => "小姐姐"],
98+
["type_id" => "荒野求生超清", "type_name" => "荒野求生"],
99+
["type_id" => "健身", "type_name" => "健身"],
100+
["type_id" => "帕梅拉", "type_name" => "帕梅拉"],
101+
["type_id" => "太极拳", "type_name" => "太极拳"],
102+
["type_id" => "广场舞", "type_name" => "广场舞"],
103+
["type_id" => "舞蹈", "type_name" => "舞蹈"],
104+
["type_id" => "音乐", "type_name" => "音乐"],
105+
["type_id" => "歌曲", "type_name" => "歌曲"],
106+
["type_id" => "MV4K", "type_name" => "MV"],
107+
["type_id" => "舞曲超清", "type_name" => "舞曲"],
108+
["type_id" => "4K", "type_name" => "4K"],
109+
["type_id" => "电影", "type_name" => "电影"],
110+
["type_id" => "电视剧", "type_name" => "电视剧"],
111+
["type_id" => "白噪音超清", "type_name" => "白噪音"],
112+
["type_id" => "考公考证", "type_name" => "考公考证"],
113+
["type_id" => "平面设计教学", "type_name" => "平面设计教学"],
114+
["type_id" => "软件教程", "type_name" => "软件教程"],
115+
["type_id" => "Windows", "type_name" => "Windows"]
116+
];
117+
return ['class' => $classes];
118+
}
119+
120+
public function homeVideoContent() {
121+
$url = 'https://api.bilibili.com/x/web-interface/popular?ps=20&pn=1';
122+
$data = json_decode($this->fetch($url), true);
123+
124+
$videos = [];
125+
if (isset($data['data']['list'])) {
126+
foreach ($data['data']['list'] as $item) {
127+
$videos[] = [
128+
'vod_id' => $item['aid'],
129+
'vod_name' => strip_tags($item['title']),
130+
'vod_pic' => $item['pic'],
131+
'vod_remarks' => $this->formatDuration($item['duration'])
132+
];
133+
}
134+
}
135+
return ['list' => $videos];
136+
}
137+
138+
public function categoryContent($tid, $pg = 1, $filter = [], $extend = []) {
139+
$page = max(1, intval($pg));
140+
141+
$url = 'https://api.bilibili.com/x/web-interface/search/type';
142+
$params = [
143+
'search_type' => 'video',
144+
'keyword' => $tid,
145+
'page' => $page
146+
];
147+
$url .= '?' . http_build_query($params);
148+
149+
$data = json_decode($this->fetch($url), true);
150+
151+
$videos = [];
152+
if (isset($data['data']['result'])) {
153+
foreach ($data['data']['result'] as $item) {
154+
if ($item['type'] !== 'video') continue;
155+
156+
$videos[] = [
157+
'vod_id' => $item['aid'],
158+
'vod_name' => strip_tags($item['title']),
159+
'vod_pic' => 'https:' . $item['pic'],
160+
'vod_remarks' => $this->formatSearchDuration($item['duration'])
161+
];
162+
}
163+
}
164+
165+
$pageCount = $data['data']['numPages'] ?? 1;
166+
$total = $data['data']['numResults'] ?? count($videos);
167+
168+
return $this->pageResult($videos, $page, $total, 20);
169+
}
170+
171+
public function searchContent($key, $quick = false, $pg = 1) {
172+
return $this->categoryContent($key, $pg);
173+
}
174+
175+
public function detailContent($ids) {
176+
if (empty($ids)) return ['list' => []];
177+
$vid = $ids[0];
178+
179+
$url = 'https://api.bilibili.com/x/web-interface/view?aid=' . $vid;
180+
$data = json_decode($this->fetch($url), true);
181+
182+
if (!isset($data['data'])) {
183+
return ['list' => []];
184+
}
185+
186+
$video = $data['data'];
187+
188+
// 构建播放列表
189+
$playUrl = '';
190+
foreach ($video['pages'] as $index => $page) {
191+
$part = $page['part'] ?: '第' . ($index + 1) . '集';
192+
// 构造 playId: avid_cid
193+
$playUrl .= "{$part}\${$vid}_{$page['cid']}#";
194+
}
195+
196+
$vod = [
197+
"vod_id" => $vid,
198+
"vod_name" => strip_tags($video['title']),
199+
"vod_pic" => $video['pic'],
200+
"vod_content" => $video['desc'],
201+
"vod_play_from" => "B站视频",
202+
"vod_play_url" => rtrim($playUrl, '#')
203+
];
204+
205+
return ['list' => [$vod]];
206+
}
207+
208+
public function playerContent($flag, $id, $vipFlags = []) {
209+
if (strpos($id, '_') !== false) {
210+
list($avid, $cid) = explode('_', $id);
211+
} else {
212+
return ['parse' => 0, 'url' => '', 'error' => '无效的视频ID格式'];
213+
}
214+
215+
$url = 'https://api.bilibili.com/x/player/playurl';
216+
$params = [
217+
'avid' => $avid,
218+
'cid' => $cid,
219+
'qn' => 112, // 原画质量
220+
'fnval' => 0,
221+
];
222+
$url .= '?' . http_build_query($params);
223+
224+
$data = json_decode($this->fetch($url), true);
225+
226+
if (!isset($data['data']) || $data['code'] !== 0) {
227+
return ['parse' => 0, 'url' => '', 'error' => '获取播放地址失败'];
228+
}
229+
230+
// 直接返回第一个播放地址
231+
if (isset($data['data']['durl'][0]['url'])) {
232+
$playUrl = $data['data']['durl'][0]['url'];
233+
234+
$headers = $this->headers;
235+
$headers['Referer'] = 'https://www.bilibili.com/video/av' . $avid;
236+
$headers['Origin'] = 'https://www.bilibili.com';
237+
238+
return [
239+
'parse' => 0,
240+
'url' => $playUrl,
241+
'header' => $headers,
242+
'danmaku' => "https://api.bilibili.com/x/v1/dm/list.so?oid={$cid}"
243+
];
244+
}
245+
246+
return ['parse' => 0, 'url' => '', 'error' => '无法获取播放地址'];
247+
}
248+
249+
// 工具函数
250+
private function formatDuration($seconds) {
251+
if ($seconds <= 0) return '00:00';
252+
$minutes = floor($seconds / 60);
253+
$secs = $seconds % 60;
254+
return sprintf('%02d:%02d', $minutes, $secs);
255+
}
256+
257+
private function formatSearchDuration($duration) {
258+
$parts = explode(':', $duration);
259+
if (count($parts) === 2) {
260+
return $duration;
261+
}
262+
return '00:00';
263+
}
264+
}
265+
266+
(new Spider())->run();

‎drpy-node-bundle/spider/php/_bridge.php‎

Lines changed: 101 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,101 @@
1+
<?php
2+
// _bridge.php
3+
// Bridge script to call PHP spider methods from Node.js
4+
// Usage: php _bridge.php <file_path> <method_name> <env_json> <arg1_json> <arg2_json> ...
5+
6+
// Disable error output to stdout to avoid breaking JSON
7+
ini_set('display_errors', 0);
8+
error_reporting(E_ALL);
9+
date_default_timezone_set('Asia/Shanghai');
10+
11+
define('DRPY_BRIDGE', true);
12+
13+
// Helper to send JSON response
14+
function sendResponse($data) {
15+
// Ensure data is UTF-8 encoded
16+
// echo json_encode($data, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES);
17+
echo json_encode($data, JSON_UNESCAPED_UNICODE);
18+
exit(0);
19+
}
20+
21+
// Helper to send Error response
22+
function sendError($message, $trace = '') {
23+
echo json_encode([
24+
'error' => $message,
25+
'traceback' => $trace
26+
], JSON_UNESCAPED_UNICODE);
27+
exit(1);
28+
}
29+
30+
// Set global error handler to catch warnings/notices and prevent them from corrupting stdout
31+
set_error_handler(function($errno, $errstr, $errfile, $errline) {
32+
// We can log errors to stderr so they don't mess up stdout JSON
33+
fwrite(STDERR, "PHP Error [$errno]: $errstr in $errfile on line $errline\n");
34+
return false; // Let normal error handler continue (but display_errors is 0 so no stdout)
35+
});
36+
37+
// Set exception handler
38+
set_exception_handler(function($e) {
39+
sendError($e->getMessage(), $e->getTraceAsString());
40+
});
41+
42+
try {
43+
// 1. Parse Arguments
44+
if ($argc < 4) {
45+
throw new Exception("Invalid arguments. Usage: php _bridge.php <file> <method> <env> [args...]");
46+
}
47+
48+
$filePath = $argv[1];
49+
$methodName = $argv[2];
50+
$envJson = $argv[3];
51+
$env = json_decode($envJson, true) ?? [];
52+
53+
$args = [];
54+
for ($i = 4; $i < $argc; $i++) {
55+
// Args are passed as individual JSON strings
56+
$args[] = json_decode($argv[$i], true);
57+
}
58+
59+
// 2. Load File
60+
if (!file_exists($filePath)) {
61+
throw new Exception("File not found: $filePath");
62+
}
63+
64+
// Capture any output during require (e.g. trailing newlines or echoes in file)
65+
ob_start();
66+
require_once $filePath;
67+
$output = ob_get_clean();
68+
if (trim($output) !== '') {
69+
fwrite(STDERR, "Output during require: $output\n");
70+
}
71+
72+
if (!class_exists('Spider')) {
73+
throw new Exception("Class 'Spider' not found in $filePath");
74+
}
75+
76+
// 3. Instantiate Spider
77+
$spider = new Spider();
78+
79+
// AUTO-INIT: Call init() before any other method if it's not init itself
80+
if ($methodName !== 'init' && method_exists($spider, 'init')) {
81+
$extend = $env['ext'] ?? '';
82+
$spider->init($extend);
83+
}
84+
85+
// 4. Check Method
86+
if (!method_exists($spider, $methodName)) {
87+
// If the method doesn't exist, we might be calling a mapped method that isn't implemented.
88+
// Or maybe we should check for magic method __call?
89+
// For now, throw error.
90+
throw new Exception("Method '$methodName' not found in Spider class");
91+
}
92+
93+
// 5. Call Method
94+
$result = call_user_func_array([$spider, $methodName], $args);
95+
96+
// 6. Return Result
97+
sendResponse($result);
98+
99+
} catch (Throwable $e) {
100+
sendError($e->getMessage(), $e->getTraceAsString());
101+
}

0 commit comments

Comments
 (0)