0

src/api/miaoguo/literacyController.js

@@ -127,9 +127,82 @@ export async function GetMiaoguoAISearch(ctx) {
         else{
             console.log("New Word");
             isNew=true;
+            //如果是古诗词
+            if (param.ShiciUrl){
+                console.log("古诗词");
+                let url2=param.ShiciUrl;
+                result = await axios.get(url2)
+                    .then(response => {
+                        let htmlString=response.data;
+                        //若是古诗词
+                        if (htmlString.indexOf("poem-detail-body")>0){
+                            let peom=getPeom(htmlString);
+                            param.ShiciTitle = "《" + peom.ShiciTitle + "》";
+                            if (peom.Dynasty)
+                                peom.Dynasty="["+peom.Dynasty+"]";
+
+                            //判断诗词单词是否过长，过长就分段显示
+                            if (peom.PeomContent) {
+                                peom.PeomContentIsLong = false;
+                                if (peom.PeomContent.length > 1 ) {
+                                    for (let i = 0; i < peom.PeomContent.length; i++) {
+                                        if (peom.PeomContent[i]){
+                                            let item = peom.PeomContent[i].join("");
+                                            if (item && item.length>18){
+                                                peom.PeomContentIsLong = true;
+                                                break;
+                                            }
+                                        }
+                                    }
+                                }
+                                else if (peom.PeomContent.length == 1) {
+
+                                    for (let i = 0; i < peom.PeomContent[0].length; i++) {
+                                        let item = peom.PeomContent[0][i];
+                                        if (item && item.length > 18) {
+                                            peom.PeomContentIsLong = true;
+                                            break;
+                                        }
+                                    }
+                                }
+                            }
+
+                            return {
+                                CHN:{
+                                    HanZi: param.Word,
+                                    ShiciTitle:param.ShiciTitle,
+                                    Author:peom.Author,
+                                    Dynasty:peom.Dynasty,
+                                    PeomContent:peom.PeomContent,
+                                    PeomContentIsLong:peom.PeomContentIsLong,
+                                    Translation:peom.Translation,
+                                }
+                            };
+                        }
+                        else{
+                            return null;
+                        }
+                    })
+                    .catch(err => {
+                        return null;
+                    });
+
+                let obj={};
+                obj.Word=param.Word;
+                obj.SearchType='shici';
+                obj.Author=result.Author;
+                obj.ShiciUrl=param.ShiciUrl;
+                obj.JSONString=JSON.stringify(result);
+                let sql3="INSERT INTO `MiaoguoLiteracy` SET ?;";
+                await commonModel.RunSql(obj,sql3);
+                result={"errcode": 10000, result: result};
+            }
             //如果是英文单词
-            if ((stringUtils.IsEnglish(param.Word)) || param.SearchType=="eng"){
-            
+            else if ((stringUtils.IsEnglish(param.Word)) || param.SearchType=="eng"){
+                
+                let result2=await getAiDataEng(param.Word);
+
+                result={"errcode": 10000, result: result2};
             }
             //整句中是否有中文
             else if ((stringUtils.IsChineseSentence(param.Word)) && param.SearchType!="eng") {
@@ -174,15 +247,13 @@ export async function GetMiaoguoAISearch(ctx) {
                     
                     let json=JSON.stringify(result);
 
-                    let sql3="INSERT INTO `MiaoguoLiteracy` SET ?;";
-                    //console.log(sql3);
-                    //console.log(json);
                     obj={};
                     obj.Word=param.Word;
                     obj.SearchType='zici';
                     obj.Author='';
                     obj.ShiciUrl='';
                     obj.JSONString=json;
+                    let sql3="INSERT INTO `MiaoguoLiteracy` SET ?;";
                     await commonModel.RunSql(obj,sql3);
 
                     globalCache.set(cacheKey, result, config.BufferMemoryTimeHigh);
@@ -196,7 +267,7 @@ export async function GetMiaoguoAISearch(ctx) {
 }
 
 function getWordFrequency(id){
-    var result={};
+    let result={};
     if (id==0){
         result={Min:0,Max:0};
     }
@@ -204,15 +275,15 @@ function getWordFrequency(id){
         result={Min:1,Max:220};
     }
     else if (id<=2000){
-        var num=Math.floor(id/100)*100;
+        let num=Math.floor(id/100)*100;
         result={Min:num,Max:Number(num+100)};
     }
     else if (id<=10000){
-        var num=Math.floor(id/500)*500;
+        let num=Math.floor(id/500)*500;
         result={Min:num,Max:Number(num+500)};
     }
     else{
-        var num=Math.floor(id/1000)*1000;
+        let num=Math.floor(id/1000)*1000;
         result={Min:num,Max:Number(num+1000)};
     }
     return result;
@@ -222,11 +293,11 @@ function tiankongFun(word,pinyin){
     let TianKong = [];
     if (word.length <= 4 && pinyin ) {
         
-        var arrPinyin = pinyin.split(" ");
+        let arrPinyin = pinyin.split(" ");
         //console.log(arrPinyin);
-        for (var i = 0; i < word.length; i++) {
-            var str = stringUtils.ReplaceAllString(word,word[i],arrPinyin[i]);
-            //var str=param.Word.substring(0,i)+arrPinyin[i]+param.Word.substring(i+1);
+        for (let i = 0; i < word.length; i++) {
+            let str = stringUtils.ReplaceAllString(word,word[i],arrPinyin[i]);
+            //let str=param.Word.substring(0,i)+arrPinyin[i]+param.Word.substring(i+1);
             TianKong.push(str);
         }
         TianKong= _.uniq(TianKong);
@@ -260,3 +331,277 @@ async function getAiData(word){
     
     return result;
 }
+
+async function getAiDataEng(word){
+    let result={};
+    let content = {
+        "instruction": "Generate information about the English \""+ word+"\".",
+        "output_format": {
+            "ENG":{
+                Word: word,
+                Soundmark:{
+                    "Eng":"[English phonetic alphabet]",
+                    "AmE":"[American phonetic alphabet]",
+                },
+                ParaphraseEng:[
+                    {
+                        "PartOfSpeech": "n.",
+                        "ParaphraseList": [
+                            "English Explanation(Noun)1.","English Explanation(Noun)2..."
+                        ]
+                    }
+                ],
+                Paraphrase:[
+                    {
+                        "PartOfSpeech": "n.",
+                        "ParaphraseList": [
+                            "中文翻译1;","中文翻译2..."
+                        ]
+                    }
+                ],
+                ExamplesSentences:[
+                    ["ExamplesSentences1","中文翻译1"],
+                    ["ExamplesSentences2","中文翻译2"],
+                ]
+            }
+        }
+    };
+    content=JSON.stringify(content);
+    //console.log(content);
+
+    let aiProvider="llama-4-maverick-17b-128e-instruct";
+    aiProvider="ali-qwen-plus-2025-07-14";
+    
+    //生成结果
+    result = await aiController.generateArticle(content, aiProvider);
+    
+    result=JSON.parse(result);
+
+    if (result.ENG.Soundmark){
+        result.ENG.Soundmark.EngSound="https://sp0.baidu.com/-rM1hT4a2gU2pMbgoY3K/gettts?lan=uk&text="+result.ENG.Word+"&spd=3&source=alading";
+        result.ENG.Soundmark.AmESound="https://sp0.baidu.com/-rM1hT4a2gU2pMbgoY3K/gettts?lan=en&text="+result.ENG.Word+"&spd=3&source=alading";
+    }
+    //console.log("result1:"+result);
+    
+    return result;
+}
+
+//对古诗词网页的解析
+function getPeom(data){
+    //console.log("古诗词")
+    let resultObject={};
+    let title = data.substr(data.indexOf("poem-detail-item"));
+    title = title.substr(title.indexOf("<h1>")+4);
+    title = title.substr(0,title.indexOf("</h1>"));
+    resultObject.ShiciTitle=title;
+    //console.log(title);
+    let result = data.substr(data.indexOf("poem-detail-header-author"));
+    //console.log(data);
+    let num1, num2;
+    let doWhileTime=Date.now();
+    do {
+        result = result.substr(result.indexOf("poem-info-gray"));
+        result = result.substr(result.indexOf(">"));
+        //console.log(result);
+        let title=result.substr(1,result.indexOf("</span>")-1);
+        //console.log(title);
+
+        let word = result.substring(result.indexOf("span>") + 5, result.indexOf("\n"));
+        if (word) {
+            if (title=="【作者】")
+                resultObject.Author=remove(word);
+            else if (title=="【朝代】")
+                resultObject.Dynasty=remove(word);
+        }
+
+        num1 = result.indexOf("poem-info-gray");
+
+        try {
+            stringUtils.loopBreaker(doWhileTime);
+        }
+        catch(err){
+            break;
+        }
+    }
+    while (num1>0);
+
+    let content=[],translation=[],temp="",tempContent="";
+    result = data.substr(data.indexOf("poem-detail-item-content"));
+    //唯一内容
+    if (result.indexOf("<div class=\"poem-detail-main-text\">")>=0){
+        let contentChild=[];
+        let doWhileTime=Date.now();
+        do {
+            result = result.substring(result.indexOf("<div class=\"poem-detail-main-text\">") + 35);
+            //console.log(result);
+            tempContent = result.substring(0, result.indexOf("</div>"));
+            //console.log(tempContent);
+            contentChild.push(tempContent);
+            result = result.substring(result.indexOf("</div>") + 6);
+            num1 = result.indexOf("<div class=\"poem-detail-main-text\">");
+
+            try {
+                stringUtils.loopBreaker(doWhileTime);
+            }
+            catch(err){
+                break;
+            }
+        }
+        while (num1>0);
+        contentChild=remove(contentChild);
+        content.push(contentChild);
+    }
+    else{
+        let resultContent=result;
+        let doWhileTime2=Date.now();
+        //诗正文
+        do {
+            resultContent = resultContent.substring(resultContent.indexOf("<p class=\"poem-detail-main-text"));
+            resultContent = resultContent.substring(resultContent.indexOf(">")+1);
+            tempContent=resultContent.substring(0,resultContent.indexOf("</p>")+4);
+            tempContent=stringUtils.ReplaceAllString(tempContent,"\n                    ","");
+            //console.log("tempContent:"+tempContent);
+            let contentChild=[],arrTemp=[];
+            arrTemp=tempContent.split("<span id=\"body_");
+
+            //console.log("arrTemp:"+JSON.stringify(arrTemp));
+
+            for(let i=0;i<arrTemp.length;i++){
+                if (arrTemp[i]){
+                    //console.log("tempContentChild:"+arrTemp[i]);
+
+                    arrTemp[i]=arrTemp[i].substring(arrTemp[i].indexOf(">")+1);
+                    arrTemp[i]=stringUtils.ReplaceAllString(arrTemp[i],"</span>                </p>","");
+                    arrTemp[i]=stringUtils.ReplaceAllString(arrTemp[i],"</span>","");
+
+                    //console.log("tempContentChild1:"+arrTemp[i]);
+
+                    let doWhileTime=Date.now();
+                    while(arrTemp[i].indexOf("<span")>=0){
+                        temp=arrTemp[i].substring(arrTemp[i].indexOf("<span"));
+                        temp = temp.substring(0,temp.indexOf("'>")+2);
+
+                        arrTemp[i]=arrTemp[i].replace(temp,"");
+                        try {
+                            stringUtils.loopBreaker(doWhileTime);
+                        }
+                        catch(err){
+                            break;
+                        }
+                    }
+
+                    //console.log("tempContentChild2:"+arrTemp[i]);
+                    arrTemp[i]=remove(arrTemp[i]);
+                    contentChild.push(arrTemp[i]);
+                }
+            }
+
+            content.push(contentChild);
+
+            num1 = resultContent.indexOf("<p class=\"poem-detail-main-text");
+            //break;
+
+            try {
+                stringUtils.loopBreaker(doWhileTime2);
+            }
+            catch(err){
+                break;
+            }
+        }
+        while (num1>0);
+
+        resultContent=result;
+        
+        //诗的每句译文
+        doWhileTime2=Date.now();
+        do {
+            if (resultContent.indexOf("<p id=\"means_p\" class=\"poem-detail-main-text body-means-p\">")>=0) {
+                resultContent = resultContent.substring(resultContent.indexOf("<p id=\"means_p\" class=\"poem-detail-main-text body-means-p\">"));
+                resultContent = resultContent.substring(resultContent.indexOf(">") + 1);
+                tempContent = resultContent.substring(0, resultContent.indexOf("</p>") + 4);
+                tempContent = stringUtils.ReplaceAllString(tempContent, "\n                    ", "");
+                //console.log("tempContent:" + tempContent);
+                let translationChild = [], arrTemp = [];
+                arrTemp = tempContent.split("<span id=\"means_");
+                //console.log("arrTemp:" + arrTemp.length);
+                for (let i = 0; i < arrTemp.length; i++) {
+                    if (arrTemp[i]) {
+                        //console.log("tempContentChild:"+arrTemp[i]);
+
+                        arrTemp[i] = arrTemp[i].substring(arrTemp[i].indexOf(">") + 1);
+                        arrTemp[i] = stringUtils.ReplaceAllString(arrTemp[i], "</span>                </p>", "");
+                        arrTemp[i] = stringUtils.ReplaceAllString(arrTemp[i], "</span>", "");
+
+                        let doWhileTime=Date.now();
+                        while (arrTemp[i].indexOf("<span") >= 0) {
+                            temp = arrTemp[i].substring(arrTemp[i].indexOf("<span"));
+                            temp = temp.substring(0, temp.indexOf(">") + 1);
+
+                            arrTemp[i] = arrTemp[i].replace(temp, "");
+
+                            try {
+                                stringUtils.loopBreaker(doWhileTime);
+                            }
+                            catch(err){
+                                break;
+                            }
+
+                        }
+
+                        //console.log("tempContentChild:" + arrTemp[i]);
+                        //console.log(tempContent);
+                        arrTemp[i] = remove(arrTemp[i]);
+                        translationChild.push(arrTemp[i]);
+
+
+                    }
+                }
+
+                translation.push(translationChild);
+            }
+
+            num1 = resultContent.indexOf("<p class=\"poem-detail-main-text\"");
+            //break;
+
+            try {
+                stringUtils.loopBreaker(doWhileTime2);
+            }
+            catch(err){
+                break;
+            }
+
+        }
+        while (num1>0);
+    }
+
+
+    resultObject.PeomContent=content;
+
+    resultObject.Translation=[];
+    if (translation && translation.length>0){
+        resultObject.Translation=translation;
+    }
+    else {
+        result = data.substr(data.indexOf("poem-detail-item-content means-fold"));
+        result = result.substr(result.indexOf(">"));
+        translation=remove(result.substr(1, result.indexOf("</div>") - 1));
+        if (translation)
+            resultObject.Translation.push(translation);
+    }
+
+    //console.log(resultObject);
+
+    return resultObject;
+
+    function remove(str){
+        if (stringUtils.IsArray(str)){
+            str=str.join("\n");
+        }
+        str = stringUtils.ReplaceAllString(str,"</span>","");
+        str = stringUtils.ReplaceAllString(str,"<br>","\n");
+        str = stringUtils.ReplaceAllString(str,"</br>","\n");
+        str = stringUtils.Trim(str);
+        return str;
+    }
+}
+

src/model/miaoguo.js

@@ -9,7 +9,7 @@ class Miaoguo {
     static async GetLiteracyItem(obj) {
         try {
             if (obj.ShiciUrl) {
-                sql="select * from MiaoguoLiteracy where Word=? and SearchType=? and Author=? and ShiciUrl=? order by ID;";
+                let sql="select * from MiaoguoLiteracy where Word=? and SearchType=? and Author=? and ShiciUrl=? order by ID;";
                 let param=[obj.Word,obj.SearchType,obj.Author,obj.ShiciUrl];
                 return await query(sql, param);
             }

src/test/build.test41.js

@@ -8,7 +8,7 @@ async function runScript(){
     try {
         
         //按照高频单词的使用频率排序，列出所有单词
-        const sql="select * from MiaoguoLiteracy where ID>0 order by ID desc limit 100;";
+        const sql="select * from MiaoguoLiteracy where ID<=114758 order by ID desc;";
         
         let list = await commonModel.RunSql(null,sql);
         let count=list.length;
@@ -18,33 +18,49 @@ async function runScript(){
         
         for(let i=start;i<count;i++){
             let item=list[i];
-            let json=JSON.parse(item.JSONString);
             let b=false;
-            if (json.ENG){
-                if (!stringUtils.IsArray(json.ENG.Paraphrase)){
-                    json.ENG.Paraphrase=[json.ENG.Paraphrase];
-                    console.log(json.ENG.Paraphras);
-                    b=true;
-                }
-                if (!stringUtils.IsArray(json.ENG.Paraphrase[0].ParaphraseList)){
-                    json.ENG.Paraphrase[0].ParaphraseList=[json.ENG.Paraphrase[0].ParaphraseList];
-                    console.log(json.ENG.Paraphrase[0].ParaphraseList);
-                    b=true;
+            let json;
+            try{
+                json=JSON.parse(item.JSONString);
+                
+                if (json.ENG){
+                    if (json.ENG.Paraphrase && !stringUtils.IsArray(json.ENG.Paraphrase)){
+                        json.ENG.Paraphrase=[json.ENG.Paraphrase];
+                        //console.log(json.ENG.Paraphras);
+                        b=true;
+
+                        if (json.ENG.Paraphrase[0].PartOfSpeech=="")
+                            json.ENG.Paraphrase[0].PartOfSpeech="释义";
+                        
+
+                        if (json.ENG.Paraphrase[0].ParaphraseList && !stringUtils.IsArray(json.ENG.Paraphrase[0].ParaphraseList)){
+                            json.ENG.Paraphrase[0].ParaphraseList=[json.ENG.Paraphrase[0].ParaphraseList];
+                            //console.log(json.ENG.Paraphrase[0].ParaphraseList);
+                            b=true;
+                        }
+                    }
+                    
                 }
             }
+            catch(err2){
+                console.error('Error executing script:', err2);
+                console.log(i+ " "+item.Word);
+                process.exit(1);
+            }
             
             if (b){
                 json=JSON.stringify(json);
                 
                 let sql3="update MiaoguoLiteracy set JSONString=? where ID="+item.ID+";";
                 //console.log(sql3);
-                console.log(json);
-                
-                //await commonModel.RunSql(json,sql3);
-                process.exit(1);
+                //console.log(json);
+                console.log( i +"/"+ list.length+" "+ item.ID);
+
+                await commonModel.RunSql(json,sql3);
+                //process.exit(1);
             }
             
-            console.log( i +"/"+ list.length+" "+ item.ID);
+            
         }
         
         console.log("完成");

src/util/stringClass.js

@@ -795,7 +795,7 @@ export const stringUtils = {
     },
     //判断是否是英文
     IsEnglish: (temp) => {
-        const reg = new RegExp("^[A-Za-z]+(?:['-][A-Za-z]+)*$");
+        const reg = new RegExp("^[A-Za-z]+(?:['-][A-Za-z]+)*(?:\\s+[A-Za-z]+(?:['-][A-Za-z]+)*)*$");
         return reg.test(temp);     /*进行验证*/
     },
     //循环熔断函数