代码提交

liusheng

2024-01-30 7ae67ba965ea680407ae21fedbd035d43ce8c313

 ruoyi-admin/src/main/java/com/ruoyi/web/test/pctest.java

@@ -15,72 +15,96 @@
import java.nio.file.Paths;
import java.util.ArrayList;
import java.util.List;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class pctest {

    public static void main(String[] args) {
        String str = null;
//        for (int i = 51; i <= 189; i++) {
//        System.out.print(i + " ");
        try {
        // 待匹配的字符串
        String sentence = "135465465好46546";

//                long l = System.currentTimeMillis();
////                str = "2263œ189œœ232105105§胸部肿瘤科日间病房患者<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§11¤232334680§胸部肿瘤科患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§19¤232313714§胸部肿瘤科食管癌患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§20¤232294126§胸部肿瘤科肺癌患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§22¤232231895§TKA术后<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§16¤232349695§<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>1月§29¤232303115§高糖<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>工作培训试题§16¤232295017§肿瘤<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>登记培训测试题§15¤232233427§老年健康监测队列<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>培训测试§10¤232215217§24病区IBD患者<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>问卷（8.22-8.28）§20¤232169849§2023.8.25小金人俱乐部MDT门诊<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§9¤232173705§<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>信息反馈表§6¤";
////                str = HttpUtils.sendGet("https://www.wjx.cn/handler/Search.ashx?input=%E9%9A%8F%E8%AE%BF&qc=&cp=" + i + "&nw=1&qt=1&accurate=0&t=" + l);
//                while (true) {
//                    if (str.contains("<font")) {
//                        str = str.substring(0, str.indexOf("<font")) + str.substring(str.indexOf("</font>") + 7);
//                    } else {
////                        System.out.println(str);
        // 定义正则表达式
//        String regex = ".*吃了.*早上好.*";
//        String regex = ".*吃了.*|.*早上好.*";
//        String regex = "^(?!.*很好，吃了吗).*|.*没吃.*早上好.*医院.*";
        String regex = "^(?!.别|不可|不可靠)(.(好|幸福)+.*)$";

            List<String> list = new ArrayList<>();
            String folderPath = "E:/pc";
            try {
                Files.walk(Paths.get(folderPath)).filter(Files::isRegularFile).forEach(path -> list.add(path.getFileName().toString()));
            } catch (IOException e) {
                System.out.println("An error occurred while traversing the folder.");
                e.printStackTrace();
            }
        // 编译正则表达式
        Pattern pattern = Pattern.compile(regex);

            for (String name : list) {
                name = name.substring(0, name.indexOf(".html"));
//                String sd = "https://www.wjx.cn/xz/" + name + ".aspx";
//                String content = HttpUtils.sendGet(sd, "review=1&u_atoken=31d3a0e6-e00e-483c-b956-b6177972e734&u_asession=01691hXKO9p1D59fHgd0NAIGdr-8HrPakCUS6X_v_Z43ohuMPPQJnwrrPl3mly-UFXYrO71Pz35K7wOrV6v8t3g9sq8AL43dpOnCClYrgFm6o&u_asig=052KZ_HPijdQddSitCTcvhNlwy7cGO2t46JgQGTyclFPT7xwZjoK0YBNED-DbuN5gugFYO36lFqDfxMpnRFufeLFk7BSuILL02zWW0QNCM-NWgLElQ9wTw7yMg4BFlXpIhOVzkXl1W45ZWoBPny57uZ3WVmTrcdvNBGM1L1rWev0UwpA9oQuqX4mfRyWspmK8gksmHjM0JOodanL5-M1Qs1Srh04GpT7zItRH0TxDThMR15LDAE7pTBPfvFUyoAgUl5gGKGQH-vEmN2Ae1t4GLLQkS6fjps_twKYxR-u2135jY94r_LXIIil3Y3aVPRGAe&u_aref=gk2k7DQa4vfdgaFwA4iVd7KUI4Q%3D", Constants.UTF8);
//                String sd = "https://www.wjx.cn/xz/214877887.aspx?review=1&u_atoken=7ea94b7e-ff2c-4f9f-a16a-2377e7b932f7&u_asession=01mUmaPv86XmI35R93EFzW4vAn-_bjd_hg8O3aV3Qj_hY4LTkzoq3vcQyaBAt_bXG0L3UvT1OC8yhlQ_Vrlj1nD9sq8AL43dpOnCClYrgFm6o&u_asig=05qqfmDpV5jnzQ3zaOR-kKvhCvox19kVoC-lC8wI68PnJ7lkckawOfhM2Sz6TSlLyxhPCLiM-UkS5JtWfnNrZeTMme2Xl4_nh34yljc8iiZeJybY6mcSFGRriqnFvQfUeuCvV5odDJuWRnGyKMRbx_RypafWqYj328He90miD2j6an7QQBtuwLgtjfNI6ktmk7ksmHjM0JOodanL5-M1Qs1cOGKb4pT6nOz7TsF50O_Y5mwcLzFTlOSTrqpuPzNwho1AC5ZokKrUg3JS1GlbH-sRHfAvD91UjmwUK--AQbESDY94r_LXIIil3Y3aVPRGAe&u_aref=muWANYNZeB7WW%2B%2F7kLKv3uRT8ws%3D";
//                String content = HttpUtils.sendGet(sd);
                HttpRequest get = HttpUtil.createGet("https://www.wjx.cn/xz/215122141.aspx?review=1&u_atoken=ae7fe80c-bf23-4e45-bcda-e1a79738fa28&u_asession=014vQCkRvW7c3i2ofZKQ4cwL_Kbllht28bP58Yv5WaOmib7bTm0yOu3E9ghHeDwRromXsXiqHPUHP-r4N4zkLsNdsq8AL43dpOnCClYrgFm6o&u_asig=05cv8a98zgMB5Y5sCj2tshR0ymwhcsf5o0ZnRdzw_MWEqjWv9qUXA2y-W56hqxwwu62XCEZ1EDfpROWbv7u5K0esMlHJrRxH3x6MobynAbRVK73TZbz-pPzJrSvkDKmiOfptL_k4wm25Oa44SC2LnI-f_gUMOBxNBqpVgCnR54boOBWxWY3t65iSXiQbR5_szmksmHjM0JOodanL5-M1Qs1XAPImYHnI2tMZUJbFSrHeeZf56fnMo9lJFDBHsYIUFUSQYCfCQCLXjf34sWIadxtsAb6Kn2CIzT3kClvRmyaM_Y94r_LXIIil3Y3aVPRGAe&u_aref=OUtfes0VUTyTV7SKqEByRa9VGaU%3D");
                String content = get.execute().body();
                //创建文件夹
                Path folder = Paths.get("E:/pc2/" + name + ".html");
                Files.createFile(folder);
                Files.write(Paths.get("E:/pc2/" + name + ".html"), content.getBytes());
            }
        // 创建 Matcher 对象
        Matcher matcher = pattern.matcher(sentence);


//            int index = 0;
//            while (true) {
//                int dd = str.indexOf("¤", index) + 1;
//                String newStr = str.substring(dd, str.indexOf("§", dd));
//                index = dd;
//                String sd = "https://www.wjx.cn/xz/" + newStr + ".aspx";
//                String content = HttpUtils.sendGet(sd, "review=1");
//
//                //创建文件夹
//                Path folder = Paths.get("E:/pc/" + newStr + ".html");
//                Files.createFile(folder);
//                Files.write(Paths.get("E:/pc/" + newStr + ".html"), content.getBytes());
//
//                if (index == -1) {
//                    break;
//                }
//            }
//            break;
//                    }

//                }

        } catch (Exception e) {
        // 进行匹配
        if (matcher.matches()) {
            System.out.println("句子中包含\"吃了，早上好\"。");
        } else {
            System.out.println("句子中不包含\"吃了，早上好\"。");
        }
//        String str = null;
////        for (int i = 51; i <= 189; i++) {
////        System.out.print(i + " ");
//        try {
//
////                long l = System.currentTimeMillis();
//////                str = "2263œ189œœ232105105§胸部肿瘤科日间病房患者<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§11¤232334680§胸部肿瘤科患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§19¤232313714§胸部肿瘤科食管癌患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§20¤232294126§胸部肿瘤科肺癌患者出院<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§22¤232231895§TKA术后<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§16¤232349695§<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>1月§29¤232303115§高糖<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>工作培训试题§16¤232295017§肿瘤<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>登记培训测试题§15¤232233427§老年健康监测队列<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>培训测试§10¤232215217§24病区IBD患者<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>问卷（8.22-8.28）§20¤232169849§2023.8.25小金人俱乐部MDT门诊<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>§9¤232173705§<font color='#30a6f5'>随</font><font color='#30a6f5'>访</font>信息反馈表§6¤";
//////                str = HttpUtils.sendGet("https://www.wjx.cn/handler/Search.ashx?input=%E9%9A%8F%E8%AE%BF&qc=&cp=" + i + "&nw=1&qt=1&accurate=0&t=" + l);
////                while (true) {
////                    if (str.contains("<font")) {
////                        str = str.substring(0, str.indexOf("<font")) + str.substring(str.indexOf("</font>") + 7);
////                    } else {
//////                        System.out.println(str);
//
//            List<String> list = new ArrayList<>();
//            String folderPath = "E:/pc";
//            try {
//                Files.walk(Paths.get(folderPath)).filter(Files::isRegularFile).forEach(path -> list.add(path.getFileName().toString()));
//            } catch (IOException e) {
//                System.out.println("An error occurred while traversing the folder.");
//                e.printStackTrace();
//            }
//
//            for (String name : list) {
//                name = name.substring(0, name.indexOf(".html"));
////                String sd = "https://www.wjx.cn/xz/" + name + ".aspx";
////                String content = HttpUtils.sendGet(sd, "review=1&u_atoken=31d3a0e6-e00e-483c-b956-b6177972e734&u_asession=01691hXKO9p1D59fHgd0NAIGdr-8HrPakCUS6X_v_Z43ohuMPPQJnwrrPl3mly-UFXYrO71Pz35K7wOrV6v8t3g9sq8AL43dpOnCClYrgFm6o&u_asig=052KZ_HPijdQddSitCTcvhNlwy7cGO2t46JgQGTyclFPT7xwZjoK0YBNED-DbuN5gugFYO36lFqDfxMpnRFufeLFk7BSuILL02zWW0QNCM-NWgLElQ9wTw7yMg4BFlXpIhOVzkXl1W45ZWoBPny57uZ3WVmTrcdvNBGM1L1rWev0UwpA9oQuqX4mfRyWspmK8gksmHjM0JOodanL5-M1Qs1Srh04GpT7zItRH0TxDThMR15LDAE7pTBPfvFUyoAgUl5gGKGQH-vEmN2Ae1t4GLLQkS6fjps_twKYxR-u2135jY94r_LXIIil3Y3aVPRGAe&u_aref=gk2k7DQa4vfdgaFwA4iVd7KUI4Q%3D", Constants.UTF8);
////                String sd = "https://www.wjx.cn/xz/214877887.aspx?review=1&u_atoken=7ea94b7e-ff2c-4f9f-a16a-2377e7b932f7&u_asession=01mUmaPv86XmI35R93EFzW4vAn-_bjd_hg8O3aV3Qj_hY4LTkzoq3vcQyaBAt_bXG0L3UvT1OC8yhlQ_Vrlj1nD9sq8AL43dpOnCClYrgFm6o&u_asig=05qqfmDpV5jnzQ3zaOR-kKvhCvox19kVoC-lC8wI68PnJ7lkckawOfhM2Sz6TSlLyxhPCLiM-UkS5JtWfnNrZeTMme2Xl4_nh34yljc8iiZeJybY6mcSFGRriqnFvQfUeuCvV5odDJuWRnGyKMRbx_RypafWqYj328He90miD2j6an7QQBtuwLgtjfNI6ktmk7ksmHjM0JOodanL5-M1Qs1cOGKb4pT6nOz7TsF50O_Y5mwcLzFTlOSTrqpuPzNwho1AC5ZokKrUg3JS1GlbH-sRHfAvD91UjmwUK--AQbESDY94r_LXIIil3Y3aVPRGAe&u_aref=muWANYNZeB7WW%2B%2F7kLKv3uRT8ws%3D";
////                String content = HttpUtils.sendGet(sd);
//                HttpRequest get = HttpUtil.createGet("https://www.wjx.cn/xz/215122141.aspx?review=1&u_atoken=ae7fe80c-bf23-4e45-bcda-e1a79738fa28&u_asession=014vQCkRvW7c3i2ofZKQ4cwL_Kbllht28bP58Yv5WaOmib7bTm0yOu3E9ghHeDwRromXsXiqHPUHP-r4N4zkLsNdsq8AL43dpOnCClYrgFm6o&u_asig=05cv8a98zgMB5Y5sCj2tshR0ymwhcsf5o0ZnRdzw_MWEqjWv9qUXA2y-W56hqxwwu62XCEZ1EDfpROWbv7u5K0esMlHJrRxH3x6MobynAbRVK73TZbz-pPzJrSvkDKmiOfptL_k4wm25Oa44SC2LnI-f_gUMOBxNBqpVgCnR54boOBWxWY3t65iSXiQbR5_szmksmHjM0JOodanL5-M1Qs1XAPImYHnI2tMZUJbFSrHeeZf56fnMo9lJFDBHsYIUFUSQYCfCQCLXjf34sWIadxtsAb6Kn2CIzT3kClvRmyaM_Y94r_LXIIil3Y3aVPRGAe&u_aref=OUtfes0VUTyTV7SKqEByRa9VGaU%3D");
//                String content = get.execute().body();
//                //创建文件夹
//                Path folder = Paths.get("E:/pc2/" + name + ".html");
//                Files.createFile(folder);
//                Files.write(Paths.get("E:/pc2/" + name + ".html"), content.getBytes());
//            }
//
//
////            int index = 0;
////            while (true) {
////                int dd = str.indexOf("¤", index) + 1;
////                String newStr = str.substring(dd, str.indexOf("§", dd));
////                index = dd;
////                String sd = "https://www.wjx.cn/xz/" + newStr + ".aspx";
////                String content = HttpUtils.sendGet(sd, "review=1");
////
////                //创建文件夹
////                Path folder = Paths.get("E:/pc/" + newStr + ".html");
////                Files.createFile(folder);
////                Files.write(Paths.get("E:/pc/" + newStr + ".html"), content.getBytes());
////
////                if (index == -1) {
////                    break;
////                }
////            }
////            break;
////                    }
//
////                }
//
//        } catch (Exception e) {
//        }
    }
}
//}