|
| 1 | +# 浏览器渲染-Playwright |
| 2 | + |
| 3 | +采集动态页面时（Ajax渲染的页面），常用的有两种方案。一种是找接口拼参数，这种方式比较复杂但效率高，需要一定的爬虫功底；另外一种是采用浏览器渲染的方式，直接获取源码，简单方便 |
| 4 | + |
| 5 | +框架支持playwright渲染下载，每个线程持有一个playwright实例 |
| 6 | + |
| 7 | + |
| 8 | +## 使用方式： |
| 9 | + |
| 10 | +1. 修改配置文件的渲染下载器： |
| 11 | + |
| 12 | + ``` |
| 13 | + RENDER_DOWNLOADER="feapder.network.downloader.PlaywrightDownloader" |
| 14 | + ``` |
| 15 | +2. 使用 |
| 16 | +
|
| 17 | + ```python |
| 18 | + def start_requests(self): |
| 19 | + yield feapder.Request("https://news.qq.com/", render=True) |
| 20 | + ``` |
| 21 | +
|
| 22 | +在返回的Request中传递`render=True`即可 |
| 23 | +
|
| 24 | +框架支持`chromium`、`firefox`、`webkit` 三种浏览器渲染，可通过[配置文件](source_code/配置文件)进行配置。相关配置如下： |
| 25 | +
|
| 26 | +```python |
| 27 | +PLAYWRIGHT = dict( |
| 28 | + user_agent=None, # 字符串 或 无参函数，返回值为user_agent |
| 29 | + proxy=None, # xxx.xxx.xxx.xxx:xxxx 或 无参函数，返回值为代理地址 |
| 30 | + headless=False, # 是否为无头浏览器 |
| 31 | + driver_type="chromium", # chromium、firefox、webkit |
| 32 | + timeout=30, # 请求超时时间 |
| 33 | + window_size=(1024, 800), # 窗口大小 |
| 34 | + executable_path=None, # 浏览器路径，默认为默认路径 |
| 35 | + download_path=None, # 下载文件的路径 |
| 36 | + render_time=0, # 渲染时长，即打开网页等待指定时间后再获取源码 |
| 37 | + wait_until="networkidle", # 等待页面加载完成的事件,可选值："commit", "domcontentloaded", "load", "networkidle" |
| 38 | + use_stealth_js=False, # 使用stealth.min.js隐藏浏览器特征 |
| 39 | + page_on_event_callback=None, # page.on() 事件的回调 如 page_on_event_callback={"dialog": lambda dialog: dialog.accept()} |
| 40 | + storage_state_path=None, # 保存浏览器状态的路径 |
| 41 | + url_regexes=None, # 拦截接口，支持正则，数组类型 |
| 42 | + save_all=False, # 是否保存所有拦截的接口, 配合url_regexes使用，为False时只保存最后一次拦截的接口 |
| 43 | +) |
| 44 | +``` |
| 45 | + |
| 46 | + - `feapder.Request` 也支持`render_time`参数， 优先级大于配置文件中的`render_time` |
| 47 | + |
| 48 | + - 代理使用优先级：`feapder.Request`指定的代理 > 配置文件中的`PROXY_EXTRACT_API` > webdriver配置文件中的`proxy` |
| 49 | + |
| 50 | + - user_agent使用优先级：`feapder.Request`指定的header里的`User-Agent` > 框架随机的`User-Agent` > webdriver配置文件中的`user_agent` |
| 51 | + |
| 52 | +## 设置User-Agent |
| 53 | + |
| 54 | +> 每次生成一个新的浏览器实例时生效 |
| 55 | +
|
| 56 | +### 方式1： |
| 57 | + |
| 58 | +通过配置文件的 `user_agent` 参数设置 |
| 59 | + |
| 60 | +### 方式2： |
| 61 | + |
| 62 | +通过 `feapder.Request`携带，优先级大于配置文件, 如： |
| 63 | + |
| 64 | +```python |
| 65 | +def download_midware(self, request): |
| 66 | + request.headers = { |
| 67 | + "User-Agent": "xxxxxxxx" |
| 68 | + } |
| 69 | + return request |
| 70 | +``` |
| 71 | + |
| 72 | +## 设置代理 |
| 73 | + |
| 74 | +> 每次生成一个新的浏览器实例时生效 |
| 75 | +
|
| 76 | +### 方式1： |
| 77 | + |
| 78 | +通过配置文件的 `proxy` 参数设置 |
| 79 | + |
| 80 | +### 方式2： |
| 81 | + |
| 82 | +通过 `feapder.Request`携带，优先级大于配置文件, 如： |
| 83 | + |
| 84 | +```python |
| 85 | +def download_midware(self, request): |
| 86 | + request.proxies = { |
| 87 | + "https": "https://xxx.xxx.xxx.xxx:xxxx" |
| 88 | + } |
| 89 | + return request |
| 90 | +``` |
| 91 | + |
| 92 | +## 设置Cookie |
| 93 | + |
| 94 | +通过 `feapder.Request`携带，如： |
| 95 | + |
| 96 | +```python |
| 97 | +def download_midware(self, request): |
| 98 | + request.headers = { |
| 99 | + "Cookie": "key=value; key2=value2" |
| 100 | + } |
| 101 | + return request |
| 102 | +``` |
| 103 | + |
| 104 | +或者 |
| 105 | + |
| 106 | +```python |
| 107 | +def download_midware(self, request): |
| 108 | + request.cookies = { |
| 109 | + "key": "value", |
| 110 | + "key2": "value2", |
| 111 | + } |
| 112 | + return request |
| 113 | +``` |
| 114 | + |
| 115 | +或者 |
| 116 | + |
| 117 | +```python |
| 118 | +def download_midware(self, request): |
| 119 | + request.cookies = [ |
| 120 | + { |
| 121 | + "domain": "xxx", |
| 122 | + "name": "xxx", |
| 123 | + "value": "xxx", |
| 124 | + "expirationDate": "xxx" |
| 125 | + }, |
| 126 | + ] |
| 127 | + return request |
| 128 | +``` |
| 129 | + |
| 130 | +## 拦截数据示例 |
| 131 | + |
| 132 | +> 注意：主函数使用run方法运行，不能使用start |
| 133 | +
|
| 134 | +```python |
| 135 | +from playwright.sync_api import Response |
| 136 | +from feapder.utils.webdriver import ( |
| 137 | + PlaywrightDriver, |
| 138 | + InterceptResponse, |
| 139 | + InterceptRequest, |
| 140 | +) |
| 141 | + |
| 142 | +import feapder |
| 143 | + |
| 144 | + |
| 145 | +def on_response(response: Response): |
| 146 | + print(response.url) |
| 147 | + |
| 148 | + |
| 149 | +class TestPlaywright(feapder.AirSpider): |
| 150 | + __custom_setting__ = dict( |
| 151 | + RENDER_DOWNLOADER="feapder.network.downloader.PlaywrightDownloader", |
| 152 | + PLAYWRIGHT=dict( |
| 153 | + user_agent=None, # 字符串 或 无参函数，返回值为user_agent |
| 154 | + proxy=None, # xxx.xxx.xxx.xxx:xxxx 或 无参函数，返回值为代理地址 |
| 155 | + headless=False, # 是否为无头浏览器 |
| 156 | + driver_type="chromium", # chromium、firefox、webkit |
| 157 | + timeout=30, # 请求超时时间 |
| 158 | + window_size=(1024, 800), # 窗口大小 |
| 159 | + executable_path=None, # 浏览器路径，默认为默认路径 |
| 160 | + download_path=None, # 下载文件的路径 |
| 161 | + render_time=0, # 渲染时长，即打开网页等待指定时间后再获取源码 |
| 162 | + wait_until="networkidle", # 等待页面加载完成的事件,可选值："commit", "domcontentloaded", "load", "networkidle" |
| 163 | + use_stealth_js=False, # 使用stealth.min.js隐藏浏览器特征 |
| 164 | + # page_on_event_callback=dict(response=on_response), # 监听response事件 |
| 165 | + # page.on() 事件的回调 如 page_on_event_callback={"dialog": lambda dialog: dialog.accept()} |
| 166 | + storage_state_path=None, # 保存浏览器状态的路径 |
| 167 | + url_regexes=["wallpaper/list"], # 拦截接口，支持正则，数组类型 |
| 168 | + save_all=True, # 是否保存所有拦截的接口 |
| 169 | + ), |
| 170 | + ) |
| 171 | + |
| 172 | + def start_requests(self): |
| 173 | + yield feapder.Request( |
| 174 | + "http://www.soutushenqi.com/image/search/?searchWord=%E6%A0%91%E5%8F%B6", |
| 175 | + render=True, |
| 176 | + ) |
| 177 | + |
| 178 | + def parse(self, reqeust, response): |
| 179 | + driver: PlaywrightDriver = response.driver |
| 180 | + |
| 181 | + intercept_response: InterceptResponse = driver.get_response("wallpaper/list") |
| 182 | + intercept_request: InterceptRequest = intercept_response.request |
| 183 | + |
| 184 | + req_url = intercept_request.url |
| 185 | + req_header = intercept_request.headers |
| 186 | + req_data = intercept_request.data |
| 187 | + print("请求url", req_url) |
| 188 | + print("请求header", req_header) |
| 189 | + print("请求data", req_data) |
| 190 | + |
| 191 | + data = driver.get_json("wallpaper/list") |
| 192 | + print("接口返回的数据", data) |
| 193 | + |
| 194 | + print("------ 测试save_all=True ------- ") |
| 195 | + |
| 196 | + # 测试save_all=True |
| 197 | + all_intercept_response: list = driver.get_all_response("wallpaper/list") |
| 198 | + for intercept_response in all_intercept_response: |
| 199 | + intercept_request: InterceptRequest = intercept_response.request |
| 200 | + req_url = intercept_request.url |
| 201 | + req_header = intercept_request.headers |
| 202 | + req_data = intercept_request.data |
| 203 | + print("请求url", req_url) |
| 204 | + print("请求header", req_header) |
| 205 | + print("请求data", req_data) |
| 206 | + |
| 207 | + all_intercept_json = driver.get_all_json("wallpaper/list") |
| 208 | + for intercept_json in all_intercept_json: |
| 209 | + print("接口返回的数据", intercept_json) |
| 210 | + |
| 211 | + # 千万别忘了 |
| 212 | + driver.clear_cache() |
| 213 | + |
| 214 | + |
| 215 | +if __name__ == "__main__": |
| 216 | + TestPlaywright(thread_count=1).run() |
| 217 | +``` |
| 218 | +可通过配置的`page_on_event_callback`参数自定义事件的回调，如设置`on_response`的事件回调，亦可直接使用`url_regexes`设置拦截的接口 |
| 219 | + |
| 220 | +## 操作浏览器对象示例 |
| 221 | + |
| 222 | +> 注意：主函数使用run方法运行，不能使用start |
| 223 | +
|
| 224 | +```python |
| 225 | +import time |
| 226 | + |
| 227 | +from playwright.sync_api import Page |
| 228 | + |
| 229 | +import feapder |
| 230 | +from feapder.utils.webdriver import PlaywrightDriver |
| 231 | + |
| 232 | + |
| 233 | +class TestPlaywright(feapder.AirSpider): |
| 234 | + __custom_setting__ = dict( |
| 235 | + RENDER_DOWNLOADER="feapder.network.downloader.PlaywrightDownloader", |
| 236 | + ) |
| 237 | + |
| 238 | + def start_requests(self): |
| 239 | + yield feapder.Request("https://www.baidu.com", render=True) |
| 240 | + |
| 241 | + def parse(self, reqeust, response): |
| 242 | + driver: PlaywrightDriver = response.driver |
| 243 | + page: Page = driver.page |
| 244 | + |
| 245 | + page.type("#kw", "feapder") |
| 246 | + page.click("#su") |
| 247 | + page.wait_for_load_state("networkidle") |
| 248 | + time.sleep(1) |
| 249 | + |
| 250 | + html = page.content() |
| 251 | + response.text = html # 使response加载最新的页面 |
| 252 | + for data_container in response.xpath("//div[@class='c-container']"): |
| 253 | + print(data_container.xpath("string(.//h3)").extract_first()) |
| 254 | + |
| 255 | + |
| 256 | +if __name__ == "__main__": |
| 257 | + TestPlaywright(thread_count=1).run() |
| 258 | +``` |
0 commit comments