ow

Program.cs

@@ -29,11 +29,12 @@ public partial class Program
             // 设置控制台编码为UTF-8，解决中文乱码问题
             Console.OutputEncoding = Encoding.UTF8;
             Console.InputEncoding = Encoding.UTF8;
+            Directory.SetCurrentDirectory(AppDomain.CurrentDomain.BaseDirectory);
 
             string taskName = null;
             Dictionary<string, string> bodyDict = new Dictionary<string, string>();
 
-            //File.AppendAllText("log.txt", "Main:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"));
+            File.AppendAllText("log.txt", "Main:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss") + "\r\n");
 
             if (args != null && args.Length > 0)
             {

Properties/launchSettings.json

@@ -2,7 +2,7 @@
   "profiles": {
     "CloudBuilder.Topshelf": {
       "commandName": "Project",
-      "commandLineArgs": "task:TtsTask book_id:B000008 chapter_id:4 speed:1 voice:Sherpa2 actor:韩立 actor_voice:Sherpa1 save_path:D:\\\\Net8\\\\FileServer\\\\Backup\\\\DmsFile\\\\voice"
+      "commandLineArgs": "task:ChineseNameExtractorTask book_id:B000010 chapter_id:73"
     }
   }
 }

Python/PythonCommand.cs

@@ -92,6 +92,11 @@ namespace CloudBuilder.Topshelf.Python
 import os
 import json
 import logging
+import sys
+# web api通過cmd調用，如果沒有加入下面三行，會出錯
+sys.stdout = open(os.devnull, 'w')
+sys.stderr = open(os.devnull, 'w')
+os.environ['TQDM_DISABLE'] = '1'
 
 # 配置日志
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
@@ -656,8 +661,21 @@ class HanLPProcessor:
         # 直接返回提取的值
         return value
     
+    def get_surnames_json(self):
+        """返回姓氏词典数组的JSON格式字符串，供C#使用"""
+        try:
+            with open(self.surnames_path, 'r', encoding='utf-8') as f:
+                surnames = [line.strip() for line in f if line.strip()]
+            
+            return json.dumps(surnames, ensure_ascii=False)
+        except Exception as e:
+            logger.error(f"获取姓氏词典失败: {e}")
+            return json.dumps([])
+        
 processor=HanLPProcessor()
 
+# 测试split函数
+
 # 测试split函数
 def test_split_function():
     # 测试用例1: 简单的逗号分隔
@@ -838,4 +856,5 @@ if __name__ == "__main__":
     # processor=HanLPProcessor()
     print("")
 
+
 */

Task/AI/ChineseNameExtractorTask.cs

@@ -8,7 +8,8 @@ using static System.Net.Mime.MediaTypeNames;
 
 namespace CloudBuilder.Topshelf.Task
 {
-    //task:ChineseNameExtractorTask book_id:B000008 chapter_id:3
+    //task:ChineseNameExtractorTask book_id:B000010 chapter_id:74 from_chapter_id:74
+    //task:ChineseNameExtractorTask book_id:B000010 from_chapter_id:500
     public class ChineseNameExtractorTask : IScheduleTask
     {
         private readonly IApplicationService service;
@@ -20,8 +21,6 @@ namespace CloudBuilder.Topshelf.Task
 
         public void Run(Dictionary<string, string> bodyDict)
         {
-            //File.AppendAllText("log.txt", "ChineseNameExtractorTask:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"));
-
             if (!bodyDict.ContainsKey("book_id")) return;
 
             string bookId = bodyDict["book_id"];
@@ -29,19 +28,22 @@ namespace CloudBuilder.Topshelf.Task
             if (bodyDict.ContainsKey("chapter_id"))
                 chapterId = bodyDict["chapter_id"];
 
+            string from_chapter_id = string.Empty;
+            if (bodyDict.ContainsKey("from_chapter_id"))
+                from_chapter_id = bodyDict["from_chapter_id"];
+
             PythonCommand python = new PythonCommand(service);
             try
             {
                 Hanlp2 hanlp = new Hanlp2(python);
                 commonSurnames = hanlp.GetSurnames();
-
                 IRepository<AiSentenceViewEntity> repository = service.GetRepository<IRepository<AiSentenceViewEntity>>();
 
                 IRepository<AiSentenceEntity> repositoryAiSentenceEntity = service.GetRepository<IRepository<AiSentenceEntity>>();
 
                 IRepository<AiChapterPersonEntity> repositoryAiChapterPersonEntity = service.GetRepository<IRepository<AiChapterPersonEntity>>();
 
-                AiChapterPersonEntity[] dels = repositoryAiChapterPersonEntity.DetachedEntities.Where(x => x.BookId == bookId).ToArray();
+                AiChapterPersonEntity[] dels = repositoryAiChapterPersonEntity.DetachedEntities.Where(x => x.BookId == bookId && (string.IsNullOrEmpty(chapterId) || x.ChapterId == Convert.ToInt32(chapterId)) && (string.IsNullOrEmpty(from_chapter_id) || x.ChapterId >= Convert.ToInt32(from_chapter_id))).ToArray();
                 if (dels != null && dels.Length > 0) repositoryAiChapterPersonEntity.DeleteNow(dels);
 
                 string depRoot;
@@ -54,7 +56,7 @@ namespace CloudBuilder.Topshelf.Task
 
                 SpeakerAnalysisHelper speakerAnalysisHelper = new SpeakerAnalysisHelper();
 
-                AiSentenceViewEntity[] ents = repository.DetachedEntities.Where(x => x.BookId == bookId && (string.IsNullOrEmpty(chapterId) || x.ChapterId == Convert.ToInt32(chapterId))).OrderBy(x => x.ChapterId).ThenBy(x => x.ParagraphId).ThenBy(x => x.SentenceIndex).ToArray();
+                AiSentenceViewEntity[] ents = repository.DetachedEntities.Where(x => x.BookId == bookId && (string.IsNullOrEmpty(chapterId) || x.ChapterId == Convert.ToInt32(chapterId)) && (string.IsNullOrEmpty(from_chapter_id) || x.ChapterId >= Convert.ToInt32(from_chapter_id))).OrderBy(x => x.ChapterId).ThenBy(x => x.ParagraphId).ThenBy(x => x.SentenceIndex).ToArray();
                 if (ents == null || ents.Length == 0) return;
 
                 AiSentenceEntity aiSentenceEntity;
@@ -65,36 +67,45 @@ namespace CloudBuilder.Topshelf.Task
                 int processedLines = 0; // 已处理行数计数器
                 AiChapterPersonEntity aiChapterPerson;
                 List<AiChapterPersonEntity> cps = new List<AiChapterPersonEntity>();
+                
                 foreach (var ent in ents)
                 {
                     processedLines++;
-
+                    try
-                    person = speakerAnalysisHelper.GetPerson(hanlp, ent.Content);
-                    paragraphIndex = ent.ParagraphId;
-
-                    if (!string.IsNullOrEmpty(person))
                     {
-                        string[] ps = person.Split(',');
+                        person = speakerAnalysisHelper.GetPerson(hanlp, ent.Content);
-                        int orderIndex = 1;
+                        paragraphIndex = ent.ParagraphId;
-                        foreach (var p in ps)
+
+                        if (!string.IsNullOrEmpty(person))
                         {
-                            if (p.Length == 1) continue;
+                            string[] ps = person.Split(',');
-                            aiChapterPerson = new AiChapterPersonEntity();
+                            int orderIndex = 1;
-                            aiChapterPerson.PersonName = KeepChineseCharactersOnly(p);
+                            foreach (var p in ps)
-                            aiChapterPerson.BookId = ent.BookId;
+                            {
-                            aiChapterPerson.ChapterId = ent.ChapterId;
+                                if (p.Length == 1) continue;
-                            aiChapterPerson.ParagraphIndex = ent.ParagraphId;
+                                aiChapterPerson = new AiChapterPersonEntity();
-                            aiChapterPerson.SentenceIndex = ent.SentenceIndex;
+                                aiChapterPerson.PersonName = KeepChineseCharactersOnly(p);
-                            aiChapterPerson.DialogueIndc = ent.DialogueIndc;
+                                aiChapterPerson.BookId = ent.BookId;
-                            aiChapterPerson.OrderIndex = orderIndex++;
+                                aiChapterPerson.ChapterId = ent.ChapterId;
-                            cps.Add(aiChapterPerson);
+                                aiChapterPerson.ParagraphIndex = ent.ParagraphId;
-                            repositoryAiChapterPersonEntity.InsertNow(aiChapterPerson);
+                                aiChapterPerson.SentenceIndex = ent.SentenceIndex;
+                                aiChapterPerson.DialogueIndc = ent.DialogueIndc;
+                                aiChapterPerson.OrderIndex = orderIndex++;
+                                
+                                repositoryAiChapterPersonEntity.InsertNow(aiChapterPerson);
+                                cps.Add(aiChapterPerson);
+                            }
                         }
                     }
+                    catch (Exception ex)
+                    {
+                        Console.WriteLine(ex.Message);
+                    }
 
                     ConsoleOutput.UpdateProgress(processedLines, totalLines);
                 }
 
+                
                 speakerAnalysisHelper.AiChapterPersons = cps.ToArray();
                 speakerAnalysisHelper.CommonSurnames = commonSurnames;
                 speakerAnalysisHelper.MatchPerson(ents, repositoryAiSentenceEntity);

Task/TTS/TtsTask.cs

@@ -110,7 +110,7 @@ namespace CloudBuilder.Topshelf.Task.TTS
                                 voice = personMapping[ent.PersonName];
                             else
                             {
-                                voice = AI.Policy.ChineseTtsVoiceRandom.GetRandom(80).ToString();
+                                voice = CloudBuilder.AI.Policy.ChineseTtsVoiceRandom.GetRandom(80).ToString();
                                 personMapping.Add(ent.PersonName, voice);
                             }
 

Utility/SpeakerAnalysisHelper.cs

@@ -68,12 +68,15 @@ namespace CloudBuilder.Topshelf.Utility
                             {
                                 HanlpConstituencyNode[] nodesSon;
                                 nodef = HanlpResultHelper.GetNodeById(cons, node.ItemId + 1);
-                                if (nps.Contains(nodef.Label))
+                                if (nodef != null)
                                 {
-                                    nodesSon = HanlpResultHelper.GetNodeSonById(cons, nodef.ItemId);
+                                    if (nps.Contains(nodef.Label))
-                                    if (nodesSon != null && nodesSon.Where(x => x.Children == 0).Any())
                                     {
-                                        feature += string.Concat(nodesSon.Select(x => x.Label));
+                                        nodesSon = HanlpResultHelper.GetNodeSonById(cons, nodef.ItemId);
+                                        if (nodesSon != null && nodesSon.Where(x => x.Children == 0).Any())
+                                        {
+                                            feature += string.Concat(nodesSon.Select(x => x.Label));
+                                        }
                                     }
                                 }
                             }
@@ -152,7 +155,7 @@ namespace CloudBuilder.Topshelf.Utility
                             }
                         }
                     }
-
+                    /*暂不处理arg1
                     foreach (var mrp in srl)
                     {
                         if (mrp.Type.ToUpper() != "ARG0") continue;
@@ -170,6 +173,7 @@ namespace CloudBuilder.Topshelf.Utility
                             arg1 = mrp.Text;
                         }
                     }
+                    */
                 }
             }
 
@@ -208,7 +212,8 @@ namespace CloudBuilder.Topshelf.Utility
 
                         if (aiSentenceView.PersonName.Contains("【"))
                         {
-                            aiSentenceView.PersonName = AiChapterPersons.Where(x => !x.PersonName.Contains("【") && x.PersonName.Contains(aiSentenceView.PersonName.Replace("【", "").Replace("】", ""))).FirstOrDefault()!.PersonName;
+                            if (AiChapterPersons.Where(x => !x.PersonName.Contains("【") && x.PersonName.Contains(aiSentenceView.PersonName.Replace("【", "").Replace("】", ""))).Any())
+                                aiSentenceView.PersonName = AiChapterPersons.Where(x => !x.PersonName.Contains("【") && x.PersonName.Contains(aiSentenceView.PersonName.Replace("【", "").Replace("】", ""))).FirstOrDefault()!.PersonName;
                         }
                     }
                     //说话者在前面