实装调用参数，重构增加代码重用程度

输出文件的命名存在问题

.gitignore

@@ -1 +1,2 @@
 /out/
+*txt

meta/DumpMoegirl/META-INF/MANIFEST.MF

@@ -0,0 +1,3 @@
+Manifest-Version: 1.0
+Main-Class: com.tumuyan.dictspider.DumpMoeGirl
+

src/META-INF/MANIFEST.MF

@@ -1,3 +1,3 @@
 Manifest-Version: 1.0
-Main-Class: com.tumuyan.dictspider.WikiCClean
+Main-Class: com.tumuyan.dictspider.Clean
 

src/com/tumuyan/dictspider/Clean.java

@@ -0,0 +1,134 @@
+package com.tumuyan.dictspider;
+
+import java.io.*;
+import java.util.*;
+
+import static com.tumuyan.dictspider.Utils.OpenCC_T2S;
+import static com.tumuyan.dictspider.Utils.WriteList;
+
+// 把dump的wiki词条，拆解为中文、英文、混合词条
+
+
+/*
+env.engine.context:commit_history().latest_text()
+
+pinyin_simp_wiki.dict.yaml
+
+# Rime dictionary
+# encoding: utf-8
+#
+# 维基百科词库
+# by tumuyan
+---
+name: pinyin_simp_wiki
+version: "20210628"
+sort: by_weight
+use_preset_vocabulary: false
+...
+
+
+
+手动清理
+
+^[^a-zA-Z][a-zA-Z]+\s
+
+^.{1,2}[村县]$
+*/
+
+
+public class Clean {
+
+    public static void main(String[] args) {
+//        System.out.println("args.length="+args.length + ", class="+ Clean.class.getSimpleName());
+//        for(String s :args){
+//            System.out.println("  args:"+s);
+//        }
+
+        Config config = new Config();
+        config.setDefault_path("A:\\ProjectPython\\zhwiki-20220220-all-titles-in-ns0");
+        config.Parse(args);
+
+        if (!config.verifyInputPath()) {
+            return;
+        }
+
+        Dict dict = new Dict();
+        for (String p : config.getInput_files()) {
+            System.out.println("Load file: " + p);
+            dict.add(ReadFile(p));
+        }
+
+        OutputWords(dict, config.getPath_w(), config.auto_delete, true);
+    }
+
+
+    public static void OutputWords(Dict dict, String path_w, boolean auto_delete, boolean t2s) {
+        try {
+            Set<String> chs = dict.getChs();
+
+            if (t2s) {
+                WriteList(chs, path_w + ".cn.dict.txt", auto_delete, false);
+
+                OpenCC_T2S(path_w + ".cn.dict.txt", path_w + ".chs.dict.txt", "A:\\EBookTools\\OpenCC\\bin");
+                chs = Utils.ReadWords(path_w + ".chs.dict.txt");
+                chs.removeAll(ReadWords());
+                WriteList(chs, path_w + ".chs2.dict.txt", auto_delete, false);
+
+            } else {
+                chs.removeAll(ReadWords());
+                WriteList(chs, path_w + ".cn.dict.txt", auto_delete, false);
+            }
+
+            WriteList(dict.getEng(), path_w + ".eng.dict.txt", auto_delete, false);
+            WriteList(dict.getMix(), path_w + ".mix.dict.txt", auto_delete, false);
+            WriteList(dict.getSuffix(), path_w + ".chs.suffix.txt", auto_delete, false);
+        } catch (Exception e) {
+            e.printStackTrace();
+        }
+
+        System.out.println("Finish");
+    }
+
+
+    public static Set<String> ReadWords() {
+        Set<String> words = new HashSet<>();
+
+        words.addAll(Utils.ReadWords("A:\\ProjectOthers\\rime-pinyin-simp\\others\\废词.txt"));
+        words.addAll(Utils.ReadWords("A:\\ProjectOthers\\rime-pinyin-simp\\others\\废词-村县镇乡路村縣鎮鄉路.txt"));
+        return words;
+    }
+
+
+    // path_w为空时，读取path每一行文本,如果包含tab，把第一个字到keys中；并返回key
+//    path_w不为空时，把带拼音的写入path_w并返回key
+    public static Dict ReadFile(String path) {
+        Dict dict = new Dict();
+
+        try {
+            FileInputStream fileInputStream = new FileInputStream(path);
+
+            BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(fileInputStream));
+
+            String line = null;
+
+            StringBuffer buffer = new StringBuffer();
+
+            while ((line = bufferedReader.readLine()) != null) {
+
+//              如果匹配到空行
+                if (line.length() < 2)
+                    continue;
+                dict.add(line);
+            }
+
+            fileInputStream.close();
+
+        } catch (Exception e) {
+            e.printStackTrace();
+        }
+        return dict;
+
+    }
+
+
+}

src/com/tumuyan/dictspider/Config.java

@@ -0,0 +1,146 @@
+package com.tumuyan.dictspider;
+
+import java.io.File;
+import java.util.ArrayList;
+import java.util.List;
+
+public class Config {
+    private String path_w = "";
+    String default_path_w = "";
+    String default_path = "";
+    boolean auto_delete = false;
+    boolean debug = false;
+    Integer pageLimit = Integer.MAX_VALUE;
+    List<String> input_files = new ArrayList<>();
+
+    public void setDefault_path(String default_path) {
+        this.default_path = default_path;
+    }
+
+    public void setDefault_path_w(String default_path_w) {
+        this.default_path_w = default_path_w;
+    }
+
+    public List<String> getInput_files() {
+        return input_files;
+    }
+
+    public Integer getPageLimit() {
+        return pageLimit;
+    }
+
+    public String getPath_w() {
+        return path_w;
+    }
+
+    public boolean isAuto_delete() {
+        return auto_delete;
+    }
+
+    public Config() {
+
+    }
+
+    public void Parse(String[] args) {
+
+        for (int i = 0; i < args.length; i++) {
+            final String arg = args[i];
+            if (arg.equals("-h") || arg.equals("-help")) {
+                System.out.println("help\n");
+                return;
+            } else if (arg.equals("-a")) {
+                auto_delete = true;
+            } else if (arg.equals("-d") || arg.equals("-debug")) {
+                debug = true;
+            } else if (arg.equals("-o") || arg.equals("-output")) {
+                i++;
+                if (args.length > i) {
+
+                    path_w = args[i].trim().replaceFirst("(\\.[^./\\\\]+)?[/\\\\]?$", ".");
+                    File file = new File(path_w).getParentFile();
+//                    File file=new File(path_w.replaceFirst("[^/\\\\]+$",""));
+                    if (file == null) {
+//                        File f = new File(new File(System.getProperty("user.dir")),path_w);
+//                        File f = new File(DumpMoeGirl.class.getClassLoader().getResource("").getFile());
+                        File f = new File(DumpMoeGirl.class.getProtectionDomain().getCodeSource().getLocation().getFile());
+                        f = new File(f.getParentFile(), path_w);
+
+                        file = f.getParentFile();
+                        path_w = f.getPath();
+                    }
+
+                    if (file.exists()) {
+                        System.out.println("Output to: " + path_w);
+                    } else {
+                        path_w = "";
+                        System.out.println("[Err]Output folder not exist: " + file.getPath());
+                    }
+
+                } else {
+                    System.out.println("[Err]Output arg not exist.");
+                }
+            } else if (arg.equals("-p") || arg.equals("-pagelimit")) {
+                i++;
+                if (args.length > i) {
+                    pageLimit = Integer.parseInt(args[i]);
+                    if (pageLimit == null) {
+                        pageLimit = Integer.MAX_VALUE;
+                        System.out.println("[Err]unexpected pageLimit arg and disable pageLimit: " + args[i]);
+                    }
+                } else {
+                    System.out.println("[Err]pageLimit arg not exist.");
+                }
+            } else if (arg.equals("-i") || arg.equals("-input")) {
+                i++;
+                if (!input_files.contains(args[i])) {
+                    File file = new File(args[i]);
+                    if (file.exists()) {
+                        input_files.add(args[i]);
+                        System.out.println("Input: " + args[i]);
+                    } else
+                        System.out.println("[Err]Input file not exist: " + args[i]);
+
+                } else {
+                    System.out.println("[Err]Input arg not exist.");
+                }
+            }
+        }
+
+    }
+
+    //   验证是否设置了输出路径
+    public boolean verifyOutputPath() {
+        if (path_w.length() < 1) {
+            if (debug && default_path_w.length() > 0)
+                path_w = default_path_w;
+            else {
+                System.out.println("[Err]Output path missing.");
+                return false;
+            }
+        }
+        return true;
+    }
+
+
+    //   验证是否设置了输入路径
+    public boolean verifyInputPath() {
+        String path = "";
+
+        if (input_files.size() < 1) {
+            if (debug && default_path.length() > 1) {
+                path = default_path;
+                input_files = new ArrayList<>();
+                input_files.add(path);
+            } else {
+                System.out.println("[Err]Input path missing.");
+                return false;
+            }
+        }
+
+        if (path_w.length() < 0) {
+            path_w = path.replace(".dict.txt", "");
+        }
+        return true;
+
+    }
+}