文章开始之前，我们先来看一个常见的问题：

接到一个任务，需要抓取某个网站上的数据内容，网页上需要输入搜索关键词，然后点击搜索按钮，等待页面加载完毕，获取网页上的搜索结果，而每一个搜索结果项，都需要点击展开才能查看到具体内容。

对于该问题，我们可以从网上找到一些解决问题的途径，但是大都不是很全面。这里小编对所有可能出现的问题做了一次调研，并汇总成如下的解决方案，希望对大家有所帮助。

首先，我们先来汇总一下python生态下爬取网站数据需要用到的工具。

1，selenium chromedriver， 这个标准的自动化工具，可以帮助我们从后台操控google浏览器，并能模拟鼠标移动和点击事件，配合xpath快速定位网页元素，从根本上解放了我们的双手。

2，ajax-hook, 这个是用来拦截xhr的钩子，可以快速获取网站服务器的响应内容，而无需我们被动地从网页上来获取。

明确了我们处理任务的可用资源，下一步就是各种踩坑了。

这里小编先列举一下，我所遇到的坑：

1，selenium的webdriver加载网页，怎么判断页面是否加载完全，使用

wait = WebDriverWait(self.driver, timeout=10)
wait.until(lambda my_driver: my_driver.execute_script('return document.readyState') == 'complete')

发现这种方案不可行，遇到动态网页异步加载就gameover，不通用，可行的方案应该wait for webelement loaded，也就是等待你所关注的网页元素加载完毕才行。

2，xpath的相对路径查找，是带了position的向下查找，也就是说每次调用xpath find, 其position就往下移动一次。所以如果用xpath在同一个流程调用多次只有第一次可以找到元素，后面都是失败的。所以对于需要反复查找的元素，最好要用绝对路径，获取xpath绝对路径很简单

打开浏览器的开发者工具，点击左上角的小箭头，移动鼠标到具体某个按钮或者输入框点击一下就可以定位到html的标签源代码位置，鼠标右键，Copy xpath即可

3，怎么截获并过滤xhr请求，将响应内容保存下来，而不会造成内存暴涨。

from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.support.ui import WebDriverWait
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
chrome_options.add_argument("--disable-gpu")
chrome_options.add_argument("--headless")
chrome_options.add_argument("--no-sandbox")
prefs = {
'download.default_directory': os.getenv('OS_LOG_PATH')
}
chrome_options.add_experimental_option('prefs', prefs)
capabilities = webdriver.DesiredCapabilities().CHROME
capabilities['acceptSslCerts'] = True
driver = webdriver.Chrome(options=chrome_options, desired_capabilities=capabilities)

我们先来初始化一下webdriver，网上有很多资料，但是都不是很全面，上面是我整理的初始化内容，大家按照这个来，就不会有问题。

截获https/http响应body内容，网上大致有三种实现思路：

1，开启browser, performance log，再根据requestId，调用chrome-devtools protocal Network.getResponseBody, 获取响应body。但是问题来了，我们需要大量采集response，而performance log都是存在内存中，还需要我们人为清理，并且performance log采集的信息很杂，没有过滤的入口提供给开发者。而且最烦的是，还需要去解析log，需要为了查询一个repsonse body，有时候需要人为过滤N行日志。。。。。（直接无语凝噎）

2，使用代理，网上传的最多的是browsermob-proxy，但是这个玩意很无语的是只提供response的统计信息，不返回实质性的response body内容，配置各种参数都不会返回content.text字段，气得跳脚，去github上看源码发现，text被下放了，当前稳定版本都会省略这个参数。所以你辛辛苦苦安装proxy，调整ssl参数，在python中配置各种proxy har，设置captureContent，结果却得不到response body，不自觉就口吐芬芳了。所以想要其返回text字段，你需要改它的源码，改源码不难，但是很烦。

3，使用ajax-hook, 这个方案的思路是：在源网页加载之前，实现一个XMLHttpRequest的代理对象，然后覆盖全局的XMLHttpRequest，这样一但上层调用 new XMLHttpRequest这样的代码时，其实创建的是Ajax-hook的代理对象实例

我们先新建一个hook.js脚本

!function(t,e){for(var n in e)t[n]=e[n]}(window,function(t){function e(r){if(n[r])return n[r].exports;var o=n[r]={i:r,l:!1,exports:{}};return t[r].call(o.exports,o,o.exports,e),o.l=!0,o.exports}var n={};return e.m=t,e.c=n,e.i=function(t){return t},e.d=function(t,n,r){e.o(t,n)||Object.defineProperty(t,n,{configurable:!1,enumerable:!0,get:r})},e.n=function(t){var n=t&&t.__esModule?function(){return t.default}:function(){return t};return e.d(n,"a",n),n},e.o=function(t,e){return Object.prototype.hasOwnProperty.call(t,e)},e.p="",e(e.s=3)}([function(t,e,n){"use strict";function r(t,e){var n={};for(var r in t)n[r]=t[r];return n.target=n.currentTarget=e,n}function o(t){function e(e){return function(){var n=this.hasOwnProperty(e+"_")?this[e+"_"]:this.xhr[e],r=(t[e]||{}).getter;return r&&r(n,this)||n}}function n(e){return function(n){var o=this.xhr,i=this,u=t[e];if("on"===e.substring(0,2))i[e+"_"]=n,o[e]=function(u){u=r(u,i),t[e]&&t[e].call(i,o,u)||n.call(i,u)};else{var s=(u||{}).setter;n=s&&s(n,i)||n,this[e+"_"]=n;try{o[e]=n}catch(t){}}}}function o(e){return function(){var n=[].slice.call(arguments);if(t[e]){var r=t[e].call(this,n,this.xhr);if(r)return r}return this.xhr[e].apply(this.xhr,n)}}return window[s]=window[s]||XMLHttpRequest,XMLHttpRequest=function(){var t=new window[s];for(var r in t){var i="";try{i=u(t[r])}catch(t){}"function"===i?this[r]=o(r):Object.defineProperty(this,r,{get:e(r),set:n(r),enumerable:!0})}var a=this;t.getProxy=function(){return a},this.xhr=t},window[s]}function i(){window[s]&&(XMLHttpRequest=window[s]),window[s]=void 0}Object.defineProperty(e,"__esModule",{value:!0});var u="function"==typeof Symbol&&"symbol"==typeof Symbol.iterator?function(t){return typeof t}:function(t){return t&&"function"==typeof Symbol&&t.constructor===Symbol&&t!==Symbol.prototype?"symbol":typeof t};e.configEvent=r,e.hook=o,e.unHook=i;var s="_rxhr"},function(t,e,n){"use strict";function r(t){if(h)throw"Proxy already exists";return h=new f(t)}function o(){h=null,(0,d.unHook)()}function i(t){return t.replace(/^\s+|\s+$/g,"")}function u(t){return t.watcher||(t.watcher=document.createElement("a"))}function s(t,e){var n=t.getProxy(),r="on"+e+"_",o=(0,d.configEvent)({type:e},n);n[r]&&n[r](o);var i;"function"==typeof Event?i=new Event(e,{bubbles:!1}):(i=document.createEvent("Event"),i.initEvent(e,!1,!0)),u(t).dispatchEvent(i)}function a(t){this.xhr=t,this.xhrProxy=t.getProxy()}function c(t){function e(t){a.call(this,t)}return e[b]=Object.create(a[b]),e[b].next=t,e}function f(t){function e(t,e){var n=new P(t);if(!f)return n.resolve();var r={response:e.response,status:e.status,statusText:e.statusText,config:t.config,headers:t.resHeader||t.getAllResponseHeaders().split("\r\n").reduce(function(t,e){if(""===e)return t;var n=e.split(":");return t[n.shift()]=i(n.join(":")),t},{})};f(r,n)}function n(t,e,n){var r=new H(t),o={config:t.config,error:n};h?h(o,r):r.next(o)}function r(){return!0}function o(t,e){return n(t,this,e),!0}function a(t,n){return 4===t.readyState&&0!==t.status?e(t,n):4!==t.readyState&&s(t,w),!0}var c=t.onRequest,f=t.onResponse,h=t.onError;return(0,d.hook)({onload:r,onloadend:r,onerror:o,ontimeout:o,onabort:o,onreadystatechange:function(t){return a(t,this)},open:function(t,e){var r=this,o=e.config={headers:{}};o.method=t[0],o.url=t[1],o.async=t[2],o.user=t[3],o.password=t[4],o.xhr=e;var i="on"+w;e[i]||(e[i]=function(){return a(e,r)});var u=function(t){n(e,r,(0,d.configEvent)(t,r))};if([x,y,g].forEach(function(t){var n="on"+t;e[n]||(e[n]=u)}),c)return!0},send:function(t,e){var n=e.config;if(n.withCredentials=e.withCredentials,n.body=t[0],c){var r=function(){c(n,new m(e))};return!1===n.async?r():setTimeout(r),!0}},setRequestHeader:function(t,e){return e.config.headers[t[0].toLowerCase()]=t[1],!0},addEventListener:function(t,e){var n=this;if(-1!==l.indexOf(t[0])){var r=t[1];return u(e).addEventListener(t[0],function(e){var o=(0,d.configEvent)(e,n);o.type=t[0],o.isTrusted=!0,r.call(n,o)}),!0}},getAllResponseHeaders:function(t,e){var n=e.resHeader;if(n){var r="";for(var o in n)r+=o+": "+n[o]+"\r\n";return r}},getResponseHeader:function(t,e){var n=e.resHeader;if(n)return n[(t[0]||"").toLowerCase()]}})}Object.defineProperty(e,"__esModule",{value:!0}),e.proxy=r,e.unProxy=o;var h,d=n(0),l=["load","loadend","timeout","error","readystatechange","abort"],v=l[0],p=l[1],y=l[2],x=l[3],w=l[4],g=l[5],b="prototype";a[b]=Object.create({resolve:function(t){var e=this.xhrProxy,n=this.xhr;e.readyState=4,n.resHeader=t.headers,e.response=e.responseText=t.response,e.statusText=t.statusText,e.status=t.status,s(n,w),s(n,v),s(n,p)},reject:function(t){this.xhrProxy.status=0,s(this.xhr,t.type),s(this.xhr,p)}});var m=c(function(t){var e=this.xhr;t=t||e.config,e.withCredentials=t.withCredentials,e.open(t.method,t.url,!1!==t.async,t.user,t.password);for(var n in t.headers)e.setRequestHeader(n,t.headers[n]);e.send(t.body)}),P=c(function(t){this.resolve(t)}),H=c(function(t){this.reject(t)})},,function(t,e,n){"use strict";Object.defineProperty(e,"__esModule",{value:!0}),e.ah=void 0;var r=n(0),o=n(1);e.ah={proxy:o.proxy,unProxy:o.unProxy,hook:r.hook,unHook:r.unHook}}]));
!function(e,t){"object"==typeof exports&&"object"==typeof module?module.exports=t():"function"==typeof define&&define.amd?define([],t):"object"==typeof exports?exports.axios=t():e.axios=t()}(this,function(){return function(e){function t(r){if(n[r])return n[r].exports;var o=n[r]={exports:{},id:r,loaded:!1};return e[r].call(o.exports,o,o.exports,t),o.loaded=!0,o.exports}var n={};return t.m=e,t.c=n,t.p="",t(0)}([function(e,t,n){e.exports=n(1)},function(e,t,n){"use strict";function r(e){var t=new s(e),n=i(s.prototype.request,t);return o.extend(n,s.prototype,t),o.extend(n,t),n}var o=n(2),i=n(3),s=n(4),a=n(22),u=n(10),c=r(u);c.Axios=s,c.create=function(e){return r(a(c.defaults,e))},c.Cancel=n(23),c.CancelToken=n(24),c.isCancel=n(9),c.all=function(e){return Promise.all(e)},c.spread=n(25),e.exports=c,e.exports.default=c},function(e,t,n){"use strict";function r(e){return"[object Array]"===j.call(e)}function o(e){return"undefined"==typeof e}function i(e){return null!==e&&!o(e)&&null!==e.constructor&&!o(e.constructor)&&"function"==typeof e.constructor.isBuffer&&e.constructor.isBuffer(e)}function s(e){return"[object ArrayBuffer]"===j.call(e)}function a(e){return"undefined"!=typeof FormData&&e instanceof FormData}function u(e){var t;return t="undefined"!=typeof ArrayBuffer&&ArrayBuffer.isView?ArrayBuffer.isView(e):e&&e.buffer&&e.buffer instanceof ArrayBuffer}function c(e){return"string"==typeof e}function f(e){return"number"==typeof e}function p(e){return null!==e&&"object"==typeof e}function d(e){return"[object Date]"===j.call(e)}function l(e){return"[object File]"===j.call(e)}function h(e){return"[object Blob]"===j.call(e)}function m(e){return"[object Function]"===j.call(e)}function y(e){return p(e)&&m(e.pipe)}function g(e){return"undefined"!=typeof URLSearchParams&&e instanceof URLSearchParams}function v(e){return e.replace(/^\s*/,"").replace(/\s*$/,"")}function x(){return("undefined"==typeof navigator||"ReactNative"!==navigator.product&&"NativeScript"!==navigator.product&&"NS"!==navigator.product)&&("undefined"!=typeof window&&"undefined"!=typeof document)}function w(e,t){if(null!==e&&"undefined"!=typeof e)if("object"!=typeof e&&(e=[e]),r(e))for(var n=0,o=e.length;n=200&&e<300}};u.headers={common:{Accept:"application/json, text/plain, */*"}},i.forEach(["delete","get","head"],function(e){u.headers[e]={}}),i.forEach(["post","put","patch"],function(e){u.headers[e]=i.merge(a)}),e.exports=u},function(e,t,n){"use strict";var r=n(2);e.exports=function(e,t){r.forEach(e,function(n,r){r!==t&&r.toUpperCase()===t.toUpperCase()&&(e[t]=n,delete e[r])})}},function(e,t,n){"use strict";var r=n(2),o=n(13),i=n(5),s=n(16),a=n(19),u=n(20),c=n(14);e.exports=function(e){return new Promise(function(t,f){var p=e.data,d=e.headers;r.isFormData(p)&&delete d["Content-Type"];var l=new XMLHttpRequest;if(e.auth){var h=e.auth.username||"",m=e.auth.password||"";d.Authorization="Basic "+btoa(h+":"+m)}var y=s(e.baseURL,e.url);if(l.open(e.method.toUpperCase(),i(y,e.params,e.paramsSerializer),!0),l.timeout=e.timeout,l.onreadystatechange=function(){if(l&&4===l.readyState&&(0!==l.status||l.responseURL&&0===l.responseURL.indexOf("file:"))){var n="getAllResponseHeaders"in l?a(l.getAllResponseHeaders()):null,r=e.responseType&&"text"!==e.responseType?l.response:l.responseText,i={data:r,status:l.status,statusText:l.statusText,headers:n,config:e,request:l};o(t,f,i),l=null}},l.οnabοrt=function(){l&&(f(c("Request aborted",e,"ECONNABORTED",l)),l=null)},l.οnerrοr=function(){f(c("Network Error",e,null,l)),l=null},l.ontimeout=function(){var t="timeout of "+e.timeout+"ms exceeded";e.timeoutErrorMessage&&(t=e.timeoutErrorMessage),f(c(t,e,"ECONNABORTED",l)),l=null},r.isStandardBrowserEnv()){var g=n(21),v=(e.withCredentials||u(y))&&e.xsrfCookieName?g.read(e.xsrfCookieName):void 0;v&&(d[e.xsrfHeaderName]=v)}if("setRequestHeader"in l&&r.forEach(d,function(e,t){"undefined"==typeof p&&"content-type"===t.toLowerCase()?delete d[t]:l.setRequestHeader(t,e)}),r.isUndefined(e.withCredentials)||(l.withCredentials=!!e.withCredentials),e.responseType)try{l.responseType=e.responseType}catch(t){if("json"!==e.responseType)throw t}"function"==typeof e.onDownloadProgress&&l.addEventListener("progress",e.onDownloadProgress),"function"==typeof e.onUploadProgress&&l.upload&&l.upload.addEventListener("progress",e.onUploadProgress),e.cancelToken&&e.cancelToken.promise.then(function(e){l&&(l.abort(),f(e),l=null)}),void 0===p&&(p=null),l.send(p)})}},function(e,t,n){"use strict";var r=n(14);e.exports=function(e,t,n){var o=n.config.validateStatus;!o||o(n.status)?e(n):t(r("Request failed with status code "+n.status,n.config,null,n.request,n))}},function(e,t,n){"use strict";var r=n(15);e.exports=function(e,t,n,o,i){var s=new Error(e);return r(s,t,n,o,i)}},function(e,t){"use strict";e.exports=function(e,t,n,r,o){return e.config=t,n&&(e.code=n),e.request=r,e.response=o,e.isAxiosError=!0,e.toJSON=function(){return{message:this.message,name:this.name,description:this.description,number:this.number,fileName:this.fileName,lineNumber:this.lineNumber,columnNumber:this.columnNumber,stack:this.stack,config:this.config,code:this.code}},e}},function(e,t,n){"use strict";var r=n(17),o=n(18);e.exports=function(e,t){return e&&!r(t)?o(e,t):t}},function(e,t){"use strict";e.exports=function(e){return/^([a-z][a-z\d\+\-\.]*:)?\/\//i.test(e)}},function(e,t){"use strict";e.exports=function(e,t){return t?e.replace(/\/+$/,"")+"/"+t.replace(/^\/+/,""):e}},function(e,t,n){"use strict";var r=n(2),o=["age","authorization","content-length","content-type","etag","expires","from","host","if-modified-since","if-unmodified-since","last-modified","location","max-forwards","proxy-authorization","referer","retry-after","user-agent"];e.exports=function(e){var t,n,i,s={};return e?(r.forEach(e.split("\n"),function(e){if(i=e.indexOf(":"),t=r.trim(e.substr(0,i)).toLowerCase(),n=r.trim(e.substr(i+1)),t){if(s[t]&&o.indexOf(t)>=0)return;"set-cookie"===t?s[t]=(s[t]?s[t]:[]).concat([n]):s[t]=s[t]?s[t]+", "+n:n}}),s):s}},function(e,t,n){"use strict";var r=n(2);e.exports=r.isStandardBrowserEnv()?function(){function e(e){var t=e;return n&&(o.setAttribute("href",t),t=o.href),o.setAttribute("href",t),{href:o.href,protocol:o.protocol?o.protocol.replace(/:$/,""):"",host:o.host,search:o.search?o.search.replace(/^\?/,""):"",hash:o.hash?o.hash.replace(/^#/,""):"",hostname:o.hostname,port:o.port,pathname:"/"===o.pathname.charAt(0)?o.pathname:"/"+o.pathname}}var t,n=/(msie|trident)/i.test(navigator.userAgent),o=document.createElement("a");return t=e(window.location.href),function(n){var o=r.isString(n)?e(n):n;return o.protocol===t.protocol&&o.host===t.host}}():function(){return function(){return!0}}()},function(e,t,n){"use strict";var r=n(2);e.exports=r.isStandardBrowserEnv()?function(){return{write:function(e,t,n,o,i,s){var a=[];a.push(e+"="+encodeURIComponent(t)),r.isNumber(n)&&a.push("expires="+new Date(n).toGMTString()),r.isString(o)&&a.push("path="+o),r.isString(i)&&a.push("domain="+i),s===!0&&a.push("secure"),document.cookie=a.join("; ")},read:function(e){var t=document.cookie.match(new RegExp("(^|;\\s*)("+e+")=([^;]*)"));return t?decodeURIComponent(t[3]):null},remove:function(e){this.write(e,"",Date.now()-864e5)}}}():function(){return{write:function(){},read:function(){return null},remove:function(){}}}()},function(e,t,n){"use strict";var r=n(2);e.exports=function(e,t){t=t||{};var n={},o=["url","method","params","data"],i=["headers","auth","proxy"],s=["baseURL","url","transformRequest","transformResponse","paramsSerializer","timeout","withCredentials","adapter","responseType","xsrfCookieName","xsrfHeaderName","onUploadProgress","onDownloadProgress","maxContentLength","validateStatus","maxRedirects","httpAgent","httpsAgent","cancelToken","socketPath"];r.forEach(o,function(e){"undefined"!=typeof t[e]&&(n[e]=t[e])}),r.forEach(i,function(o){r.isObject(t[o])?n[o]=r.deepMerge(e[o],t[o]):"undefined"!=typeof t[o]?n[o]=t[o]:r.isObject(e[o])?n[o]=r.deepMerge(e[o]):"undefined"!=typeof e[o]&&(n[o]=e[o])}),r.forEach(s,function(r){"undefined"!=typeof t[r]?n[r]=t[r]:"undefined"!=typeof e[r]&&(n[r]=e[r])});var a=o.concat(i).concat(s),u=Object.keys(t).filter(function(e){return a.indexOf(e)===-1});return r.forEach(u,function(r){"undefined"!=typeof t[r]?n[r]=t[r]:"undefined"!=typeof e[r]&&(n[r]=e[r])}),n}},function(e,t){"use strict";function n(e){this.message=e}n.prototype.toString=function(){return"Cancel"+(this.message?": "+this.message:"")},n.prototype.__CANCEL__=!0,e.exports=n},function(e,t,n){"use strict";function r(e){if("function"!=typeof e)throw new TypeError("executor must be a function.");var t;this.promise=new Promise(function(e){t=e});var n=this;e(function(e){n.reason||(n.reason=new o(e),t(n.reason))})}var o=n(23);r.prototype.throwIfRequested=function(){if(this.reason)throw this.reason},r.source=function(){var e,t=new r(function(t){e=t});return{token:t,cancel:e}},e.exports=r},function(e,t){"use strict";e.exports=function(e){return function(t){return e.apply(null,t)}}}])});
(function(console){
console.save = function(data, filename){
if(!data) {
console.error('Console.save: No data')
return;
}
if(!filename) filename = 'console.json'
if(typeof data === "object"){
data = JSON.stringify(data, undefined, 4)
}
var blob = new Blob([data], {type: 'text/json'}),
e = document.createEvent('MouseEvents'),
a = document.createElement('a')
a.download = filename
a.href = window.URL.createObjectURL(blob)
a.dataset.downloadurl = ['text/json', a.download, a.href].join(':')
e.initMouseEvent('click', true, false, window, 0, 0, 0, 0, 0, false, false, false, false, 0, null)
a.dispatchEvent(e)
}
})(console)
ah.proxy({
//请求发起前进入
onRequest: (config, handler) => {
console.log(config.url)
handler.next(config);
},
//请求发生错误时进入，比如超时；注意，不包括http状态码错误，如404仍然会认为请求成功
onError: (err, handler) => {
console.log(err.type)
handler.next(err)
},
//请求成功后进入
onResponse: (response, handler) => {
if (response.config.url.startsWith('api/apps') && response.config.url.endsWith('/search')) {
timestamp = new Date().getTime().toString()
console.save(response.response, timestamp+'.json')
}
handler.next(response)
}
})

该hook.js是我改良之后的，可以拿来直接用，功能就是根据response.config.url过滤我们需要的请求，并将response body保存到本地的json文件中，注意到上文中webdriver初始化的字段中有个选项，download.default_directory

prefs = {
'download.default_directory': os.getenv('OS_LOG_PATH')
}

os.getenv('OS_LOGPATH')这个是获取环境变量，环境变量的值为我们需要保存response body的文件夹路径，每一个response body都是一个json文件。大功告成！！简洁明了，不需要依赖于其他任何东西，就可以实现了，长舒一口气。

总结经验：

1，不要盲目使用网上的源码，也不要死扣一种解决方案，不要轻易就相信网上的博文资料而不亲自实践，人云亦云。

2，解决问题最有效的途径是：看清问题的本质，从问题的内部挖掘出解决文图的思路，拆解成更容易解决的小问题。当一个问题关联了太多变量因素时，要学会控制单一变量，用排除法一步步解决，省时省力，不容易走弯路。