JS实现淘宝提取关键词搜索

最近无聊写了个简单的关键词搜索代码，啧啧啧

实现的效果：

用到的工具代码：

/*** 公用方法*/
var utils={//获取滚动条的高度getScrollTop:function(){let scrollTop=0;if(document.documentElement&&document.documentElement.scrollTop){scrollTop=document.documentElement.scrollTop;}else if(document.body) {scrollTop=document.body.scrollTop;}return scrollTop;},//获取标签ele:function(text){if(text!=null&text!=""){if(text.length>1){if (/^#[a-zA-Z0-9_]*$/.test(text)){return document.getElementById(text.substring(1,text.length));}else if(/^\.[a-zA-Z0-9_]*$/.test(text)){return document.getElementsByClassName(text.substring(1,text.length));}else{return document.getElementsByTagName(text);}}}return null;},//设置文本框的最大输入长度inputTextLength:function(ele,textLength){let py=false;ele.addEventListener("input",function(){if(!py){sub();}});ele.addEventListener("compositionstart",function(){py=true;});ele.addEventListener("compositionend",function(){py=false;sub();});function sub(){let text=ele.value;if(text.length>=length){ele.value=text.substring(0,length);}}},//设置cssaddCss:function(el,cssStyles){if(typeof el=="object"){let css={};let oldCss=el.getAttribute("style");if(oldCss!=null){let oldCsss=oldCss.split(";");for(let i=0;i<oldCsss.length;i++){let keyAndValue=oldCsss[i];if(keyAndValue!=""){let k=keyAndValue.substring(0,keyAndValue.indexOf(":"));let v=keyAndValue.substring(keyAndValue.indexOf(":")+1,keyAndValue.length);css[k]=v.trim();}}}for(cssStyle in cssStyles){css[cssStyle]=cssStyles[cssStyle];}if(JSON.stringify(css)!="{}"){oldCss = "";for(let c in css){oldCss+=c+":"+css[c] +";";}}el.setAttribute("style",oldCss);}},//创建标签createElement:function (eleName,attrs,write,parseEle){if(typeof eleName=="string"){let ele=document.createElement(eleName);if(typeof attrs=="object"&attrs!=null){for(let attr in attrs){ele.setAttribute(attr, attrs[attr]);}}if(typeof write=="object"&write!=null){for(let text in write){if(text=="text"){this.InnerText(ele, write[text]);}else if(text=="html"){ele.innerHTML = write[text];}else if(text=="value"){ele.value = write[text];}}}if(typeof parseEle=="object"&parseEle!=null){parseEle.appendChild(ele);}return ele;}},//分割数组splitArray:function (array, arrayLength){let VarTempArray = [];let temp = [];for(let i=0;i<array.length;i++){temp.push(array[i]);if(i%arrayLength==(arrayLength-1)||i==(array.length-1)){VarTempArray.push(temp);temp = [];}}return VarTempArray;},//时间格式化dateTimeFormat:function(timeStamp,timeFormat){let date=new Date(timeStamp);let y=date.getFullYear();let M=(date.getMonth()+1<10?'0'+(date.getMonth()+1):date.getMonth()+1);let d=date.getDate()<10?'0'+date.getDate()+'':date.getDate()+'';let h=date.getHours()<10?'0'+date.getHours():date.getHours();let m=date.getMinutes()<10?'0'+date.getMinutes():date.getMinutes();let s=date.getSeconds()<10?'0'+date.getSeconds():date.getSeconds();if(timeFormat.indexOf("yyyy")!=-1||timeFormat.indexOf("YYYY")!=-1){timeFormat=timeFormat.replace(new RegExp(/yyyy|YYYY/g),y);}if(timeFormat.indexOf("MM")!=-1){timeFormat=timeFormat.replace(new RegExp(/MM/g),M);}if(timeFormat.indexOf("dd")!=-1||timeFormat.indexOf("DD")!=-1){timeFormat=timeFormat.replace(new RegExp(/dd|DD/g),d);}if(timeFormat.indexOf("hh")!=-1||timeFormat.indexOf("HH")!=-1){timeFormat=timeFormat.replace(new RegExp(/hh|HH/g),h);}if(timeFormat.indexOf("mm")!=-1){timeFormat=timeFormat.replace(new RegExp(/mm/g), m);}if(timeFormat.indexOf("ss")!=-1||timeFormat.indexOf("SS")!=-1){timeFormat=timeFormat.replace(new RegExp(/ss|SS/g),s);}return timeFormat;},//获取元素的节点getEle:function(element,text){if(text=="子节点"){return element.children;}else if(text=="父节点"){return element.parentNode;}else if(text=="下一个节点"){return element.nextSibling;}else if(text=="上一个节点"){return element.previousSibling;}else if(text=="第一个子节点"){return element.firstChild;}else if(text=="最后一个子节点 "){return element.lastChild;}else if(text=="兄弟节点"){let r = [];let n = element.parentNode.firstChild;for(;n;n=n.nextSibling){if(n.nodeType===1&&n!==element){r.push(n);}}return r;}else{return "输入错误！";}},//动态添加jsaddDynamicJS:function(src, callback){let script=document.createElement("script");script.setAttribute("type", "text/javascript");script.src=src;script.charset='utf-8';let first=document.body.firstChild;document.body.insertBefore(script,first);if(callback!=undefined){script.onload=function(){callback();}}},//ascll码与字符间的转换ascll:function(text,isAscll){if(isAscll==true){return String.fromCodePoint(text);}else{return text.charCodeAt();}},//判断页面是否被嵌入iframe里面isIframeView:function(){return window.self===window.top;},//清除指定的class样式removeClass:function(element,className){let cName=element.className;let arr=cName.split(" ");let newArr=[];for(let i=0;i<arr.length;i++){if(arr[i]!=className){newArr.push(arr[i]);}}let str=newArr.join("");element.className=str;},//深拷贝数组copyArray:function(array){return array.slice(0,array.length);},//纯数组排序sort:function(array){return array.sort(function(a, b){return (a + '').localeCompare(b + '');});},//按照数组中的某个属性排序attrSort:function(array,attr,type){return array.sort(function(a,b){let value1=a[attr];let value2=b[attr];if(type==true){value1=b[attr];value2=a[attr];                }if(typeof value1=="number"&&typeof value2=="number"){return value1-value2;}else{return (value1+'').localeCompare(value2+'');}});},//按照数组元素/元素的属性(文本)长度排序("默认升序")textLengthSort:function(array,attr,type){var array=array.map(obj=>({raw:obj,len:obj[attr].length}));if(type==true){return array.sort((n,p) => n.len - p.len).map(i => i.raw);}return array.sort((p, n) => n.len - p.len).map(i => i.raw);},//随机数 (lower 最小值,upper 最大值)random:function(lower, upper) {return Math.floor(Math.random() * (upper - lower)) + lower;},//判断是否为万恶的IEisIE:function() {if (window.ActiveXObject || "ActiveXObject" in window){return true;}else{return false;}},//判断传入参数是否为dom实例isElement:function(obj){return (typeof HTMLElement === 'object') ?(obj instanceof HTMLElement):!!(obj && typeof obj === 'object' && (obj.nodeType === 1 || obj.nodeType === 9) && typeof obj.nodeName === 'string');},//比较两个对象是否相同equalObject:function(x, y) {var i, l, leftChain, rightChain;function compare2Objects(x, y) {var p;if (isNaN(x) && isNaN(y) && typeof x === 'number' && typeof y === 'number') {return true;}if (x === y) {return true;}if ((typeof x === 'function' && typeof y === 'function') ||(x instanceof Date && y instanceof Date) ||(x instanceof RegExp && y instanceof RegExp) ||(x instanceof String && y instanceof String) ||(x instanceof Number && y instanceof Number)) {return x.toString() === y.toString();}if (!(x instanceof Object && y instanceof Object)) {return false;}if (x.isPrototypeOf(y) || y.isPrototypeOf(x)) {return false;}if (x.constructor !== y.constructor) {return false;}if (x.prototype !== y.prototype) {return false;}if (leftChain.indexOf(x) > -1 || rightChain.indexOf(y) > -1) {return false;}for (p in y) {if (y.hasOwnProperty(p) !== x.hasOwnProperty(p)) {return false;} else if (typeof y[p] !== typeof x[p]) {return false;}}for (p in x) {if (y.hasOwnProperty(p) !== x.hasOwnProperty(p)) {return false;} else if (typeof y[p] !== typeof x[p]) {return false;}switch (typeof(x[p])) {case 'object':case 'function':leftChain.push(x);rightChain.push(y);if (!compare2Objects(x[p], y[p])) {return false;}leftChain.pop();rightChain.pop();break;default:if (x[p] !== y[p]) {return false;}break;}}return true;}if (arguments.length < 1) {return true;}for (i = 1, l = arguments.length; i < l; i++) {leftChain = [];rightChain = [];if (!compare2Objects(arguments[0],arguments[i])){return false;}}return true;},//是否为数组isArray:function(o){return Object.prototype.toString.call(o)== '[object Array]';},//字符串搜索(exact=false时为模糊搜索)search:function(data, text, exact) {let result = new Array;(function() {dataType();_search();}());function dataType() {let t = typeof data;if(t == "string") {let temp = data;data = new Array;data.push(temp);} else if(t == "object" && utils.isArray(data)) {console.log("data参数类型为数组.");} else {console.log("data参数类型为字符串或者数组！");}}function search_0() {let i = 0,len = data.length,arr=text.split(""),l=arr.length;for(i; i < len; i++) {let d=data[i],dl=d.length,j=0,isOkNumber=0,index=0;for(j;j<l;j++){let a=arr[j],_index=d.indexOf(a,index);if((dl-j)<(l-isOkNumber)){break;}if(_index>=0){isOkNumber++;index=_index;}if(isOkNumber==l){result.push(d);break;}}}}function search_1() {let i = 0,len = data.length;for(i; i < len; i++) {let d = data[i];if(d == text) {result.push(d);}}}function _search() {exact == false ? search_0() : search_1();}return result;},//获得字符串的字节长度getBytesLength:function(text){var totalLength = 0;var charCode;for (var i = 0; i < text.length; i++){charCode = text.charCodeAt(i);if (charCode < 0x007f){totalLength++;} else if ((0x0080 <= charCode) && (charCode <= 0x07ff)){totalLength += 2;} else if ((0x0800 <= charCode) && (charCode <= 0xffff)){totalLength += 3;} else{totalLength += 4;}}return totalLength;},//迭代数组iterator:function(arr){let index=-1,len=-1;(function(){index=0;len=arr.length;}());this.hasNext=function(){return len>index&&len>0?true:false;};this.next=function(){return index<len&&len>0?{"index":index,"data":arr[index++]}:null;};},//获得占用内存getMemory:function(){if(window.performance){return Math.round(window.performance.memory.usedJSHeapSize*100/1024)/100;}return -1;},//读取本地文件readerFile:function(input,fun) {if (window.FileReader) {let file = input.files[0];filename = file.name.split(".")[0];let reader = new FileReader();reader.onload = function() {fun(this.result);};reader.readAsText(file);}else if (typeof window.ActiveXObject != 'undefined'){let xmlDoc;xmlDoc = new ActiveXObject("Microsoft.XMLDOM");xmlDoc.async = false;xmlDoc.load(input.value);fun(xmlDoc.xml);}else if (document.implementation && document.implementation.createDocument) {let xmlDoc;xmlDoc = document.implementation.createDocument("", "", null);xmlDoc.async = false;xmlDoc.load(input.value);fun(xmlDoc.xml);} else {fun(null);}},//定位光标到最后positioningCursor:function(element){var range = document.createRange();range.selectNodeContents(element);range.collapse(false);var sel = window.getSelection(); sel.removeAllRanges();sel.addRange(range);return sel.anchorOffset;},//格式化文件大小formatFileSize:function(fileSize){let fSizeType=["B","K","M","G","T"];function _format(fileSize,index){index=index==undefined?0:index;if(fileSize>=1024&&index<fSizeType.length-1){return _format((fileSize/1024).toFixed(2),++index)}return fileSize+fSizeType[index];}return _format(fileSize,0);},//分词split:function(text,charset,maxLen){if(maxLen==undefined){maxLen=5;}//中文分词function spiltString_UTF8(text){var return_array=new Array,text_array=text.split(""),len=text_array.length,_text="",index=0;for(index;index<len;index++){_text=text_array[index];if(verifyUtils.isChinese(_text)){return_array.push(_text);}else{var _len=return_array.length;if(_len>=1&&return_array[_len-1]!=null){return_array.push(null);}}}return splitString(return_array);}function splitString(array){var i=0,len=array.length,return_chars=new Array;for(i;i<len;i++){var _char=array[i];if(_char!=null){var _i=i,_text=_char;for(_i;_i<len;_i++){var _c=array[_i];if(_c!=null){if(i!=_i){_text=_text+_c;}if(!return_chars.includes(_text)){return_chars.push(_text);if(_text.length==maxLen){break;}}}else{break;}}}else{continue;}}return return_chars;}if(charset=="utf-8"||charset=="UTF-8"){return spiltString_UTF8(text);}}
}

第一版：

/*模拟淘宝关键词搜索arrayData:标题数据例如:[{id:0,text:"测试文本"}]//实现功能://按照搜索词长度对结果集进行排序 √//对结果集按照自定义属性进行排序 ×//对结果集按照搜索词频进行排行   ×//按照禁用词剔除无用关键词       ×
*/
function search_system(arrayData){var max_search_text_length=8;//最大搜索文本,超出此长度不再使用关键词搜索，进行精准搜索var search_Text=new Array;//记录搜索记录var search_result_Texts=[];//记录上一次搜索的内容/*搜索text:输入的搜索文本*/this.search=function(text){var search_text_length=search_Text.length;var search_go=search_text_length>0?(search_Text[search_text_length-1]==text):false;if(!search_go){var result_Array=new Array;search_Text.push(text);if(text.length>max_search_text_length){for(var i=0;i<arrayData.length;i++){var obj=arrayData[i];if(text==obj["text"]){result_Array.push([obj["id"]]);result_Array.push([obj["text"]]);}}search_result_Texts=result_Array;return result_Array;}result_Array=_search(text)search_result_Texts=result_Array;return result_Array;}return search_result_Texts;}//关键词搜索function _search(text){var result_array_id=new Array;var result_array_text=new Array;var array=utils.split(text,"utf-8",max_search_text_length);array=array.map(i => ({raw: i, len: i.length})).sort((p, n) => n.len - p.len).map(i => i.raw)var _array=splitArray(arrayData);var i=0,len=array.length,len1=_array.length;for(i;i<len;i++){var _text=array[i],j=0;for(j;j<len1;j++){var o=_array[j];var search_text=o["text"];if(search_text==_text){var indexs=o["ids"],k=0;for(k;k<indexs.length;k++){var indexAddress=indexs[k];if(result_array_id.indexOf(indexAddress)==-1){if(search_text==text){result_array_id.unshift(indexAddress);result_array_text.unshift(_text);                }else{result_array_id.push(indexAddress);result_array_text.push(_text);}}}}}}return [result_array_id,result_array_text];}//分割关键词function splitArray(textArray){var search_Array=new Array;for(var i=0;i<textArray.length;i++){var texts=textArray[i];var array=utils.split(texts["text"],"utf-8",max_search_text_length-2);for(var j=0;j<array.length;j++){var len=search_Array.length,x=0,text=array[j],isExist=false;for(x;x<len;x++){var _o=search_Array[x];if(text==_o["text"]){_o["ids"].push(texts["id"]);isExist=true;}}if(!isExist){search_Array.push({"text":text,"ids":[texts["id"]]});}}}return search_Array;}
}

第二版：

/*模拟淘宝关键词搜索(返回结果加上词频统计)arrayData:所有的数据(必须含有id这一项)  例如:[{"id":0,"name":"tome"},{"id":0,"name":"tome"}...]attr:搜索词的属性名   例如:"name"uselessKeywords:无用关键词,提高检索准确度以及效率    例如:["的","是","我"...]
*/
//实现功能:
//按照搜索词长度对结果集进行排序 √
//对结果集按照自定义属性进行排序 ×
//对结果集按照搜索词频进行排行   ×
//按照禁用词剔除无用关键词       √
function search_System(arrayData,searchAttr,uselessKeywords){var max_search_text_length=8;//最大搜索词长度,超出长度使用精准搜索var text_array=new Array;//词频集var search_Text_Array=new Array;//历史搜索词集var result_search_array=null;//返回结果对象集//初始化(function(){split_Array();}())//从数据集中提取搜索词function split_Array(){var i=0,len=arrayData.length;for(i;i<len;i++){var text=arrayData[i][searchAttr];split_Data(i,text);}}this.search=function(text){var len=search_Text_Array.length;if(len==0||search_Text_Array[len-1]!=text){search_Start(text);}return result_search_array;}//开始搜索function search_Start(text){//实例化返回结果集result_search_array=new Array;var data_array=new Array;//临时数据集if(isNull(text)){//对检索词进行分词获得检索词集var search_Array=utils.split(text,"utf-8",max_search_text_length-2);var i=0,len=search_Array.length;//对检索词长度进行排序search_Array=search_Array.map(i => ({raw: i, len: i.length})).sort((p, n) => n.len - p.len).map(i => i.raw);for(i;i<len;i++){//获取每一个检索词对象var obj=search_Array[i];//对每一个检索词对象进行检索var ids=search_ObjectToArray(obj);var j=0,_len=ids.length;for(j;j<_len;j++){var _obj=ids[j];if(data_array.indexOf(_obj)==-1){data_array.push(_obj);var obj_data=select_id_Array(_obj);obj_data["search_text"]=obj;if(text==obj_data["text"]){result_search_array.unshift(obj_data);}else{result_search_array.push(obj_data);}}}}}}//根据id从原数组查询数据function select_id_Array(id){var i=0,len=arrayData.length;for(i;i<len;i++){var obj=arrayData[i];if(obj["id"]==id){return obj;}}return null;}//搜索检索词对象function search_ObjectToArray(obj){var i=0,len=text_array.length;for(i;i<len;i++){var _obj=text_array[i];var text=_obj["text"];if(text==obj){return _obj["ids"];}}return [];}//分词临时数据集function split_Data(index,text){//获得数据集的临时分词集var temp_data_array=utils.split(text,"utf-8",max_search_text_length-2);var i=0,len=temp_data_array.length;for(i;i<len;i++){//获得分词集对象var obj=temp_data_array[i];//进行词库统计/词频统计data_statistics(index,obj);}}//进行词库统计/词频统计function data_statistics(index,obj){//剔除禁用词if(uselessKeywords==undefined||uselessKeywords.indexOf(obj)==-1){var len=text_array.length;if(len>0){var i=0,isCreate=false;//词频插入\修改for(i;i<len;i++){var _obj=text_array[i];var ids=_obj["ids"];var text=_obj["text"];if(text==obj){if(ids.indexOf(index)==-1){_obj["ids"].push(index);_obj["size"]+=1;isCreate=true;break;}}}if(!isCreate){text_array.push({"text":obj,"ids":[index],"size":1});}}else{text_array.push({"text":obj,"ids":[index],"size":1});}}}//空值判断function isNull(text){return (text==undefined)||(text==null)||(text.trim()=="")?false:true;}
}

用到的工具代码：

JS实现淘宝提取关键词搜索相关推荐

手机淘宝APP关键词搜索采集方案
自2018年12月中旬之后,淘宝升级了反爬虫策略,淘宝Web版必须要登录之后才能使用搜索功能(直接搜索会被重定向到登录界面,登录之后才能看到搜索结果).然而登录之后如果访问稍微频繁,就会出现滑块验证码 ...
淘宝/天猫关键词搜索采集接口分析商品价格走势（商品列表，商品销量，商品价格，分类ID采集精准商品数据）接口代码对接流程
淘宝/天猫关键词搜索采集接口分析商品价格走势(商品列表,商品销量,商品价格)接口代码对接流程如下: 1.公共参数名称类型必须描述(接口代码教程wx19970108018) key String ...
淘宝/天猫关键词搜索店铺列表 API接口，店铺列表API接口，店铺商品API接口
一.淘宝/天猫关键词搜索店铺列表 API接口,店铺列表API接口,店铺商品API接口参数如下: 1.公共参数: 名称类型必须描述 key String 是调用key(必须以GET方式拼接在UR ...
淘宝按关键词搜索天猫商品接口调用展示
以下为淘宝某商品的调用数据作为展示案例更多API调用示例展示及获取Key和secret请移步以下内容查看 [登录 - 跨境电商平台接口提供商数据采集公司数据接口定制服务企业级数据服务商] { ...
淘宝/天猫关键词搜索最新接口
onebound.taobao.item_search 公共参数请求地址:https://open.onebound.cn/test/?api_type=taobao&api_name=it ...
淘宝API关键词搜索接口调用示例
以下是行业内了解到的一些情况,本帖只展示部分代码,需要更多API调试请移步注册API账号 { "items": { "page": &q ...
淘宝按关键词搜索示例
以下是行业内了解到的一些情况,本帖只展示部分代码,需要更多API调试请移步注册API账号}{{ "items": { "page": & ...
JS仿淘宝网顶部的导航菜单
代码简介: JS仿淘宝网顶部的导航菜单.整合了购物车,收藏夹,搜索,搜索,网站导航等功能的导航. 代码内容: <!DOCTYPE html PUBLIC "-//W3C//DTD XH ...
使用selenium模块自动打开淘宝并进行搜索
Selenium是一个自动化测试工具,可以驱动浏览器器执行特定的动作,如点击,下拉等.可以用来进行模拟人工操作浏览器进行测试,爬虫等. 如果没有安装环境可以先pip install Selenium, ...

JS实现淘宝提取关键词搜索

JS实现淘宝提取关键词搜索相关推荐

最新文章

热门文章