优化

Src/Entities/DisCoOrgran.cs

@@ -10,7 +10,7 @@ namespace Njust.Pdf.Analysis.Entities
 {
     [Alias(Configs.TableDisCoOrgan)]
     [Api(Description = "机构合作分布")]
-    public class DisCoOrgran : EntityWithMetadata
+    public class DisCoOrgan : EntityWithMetadata
     {
         [ApiMember(Description = "所属领域")]
         public string DomainName { get; set; }

Src/Entities/DisCountry.cs

@@ -24,6 +24,9 @@ namespace Njust.Pdf.Analysis.Entities
         [ApiMember(Description = "国家")]
         public string Country { get; set; }
 
+        [ApiMember(Description = "英文表示")]
+        public string EnCountry { get; set; }
+
         [ApiMember(Description = "国家发表文献数量")]
         public int Count { get; set; }
     }

Src/Entities/DisTimeTopic.cs

@@ -28,7 +28,7 @@ namespace Njust.Pdf.Analysis.Entities
         public string TopicWord { get; set; }
 
         [ApiMember(Description = "主题词概率TOP10")]
-        public string TopicWordProbability { get; set; }
+        public List<double> TopicWordProbability { get; set; }
 
         [ApiMember(Description = "主题支持文档列表")]
         [StringLength(5000)]

Src/Entities/DisTopicEvolution.cs

@@ -8,8 +8,8 @@ using Kivii.DataAnnotations;
 
 namespace Njust.Pdf.Analysis.Entities
 {
-    [Alias(Configs.TableDisTimeTopic)]
-    [Api(Description = "TableDisTopicEvolution")]
+    [Alias(Configs.TableDisTopicEvolution)]
+    [Api(Description = "主题演化分布")]
     public class DisTopicEvolution : EntityWithMetadata
     {
         [ApiMember(Description = "所属领域")]

Src/Jobs/AnalyseJob.cs

@@ -42,9 +42,6 @@ namespace Njust.Pdf.Analysis.Jobs
         }
         private bool textPdfBox()
         {
-            //var text1 = "\x9f\x0f\x12";
-            Regex regex = new Regex(@"/\\x[\da-zA-Z]+/g");
-            //text1 = regex.Replace(text1, "");
             int succeed = 0, total = 0;
             var conn = KiviiContext.GetOpenedDbConnection<AnalyseAll>();
             try
@@ -73,8 +70,10 @@ namespace Njust.Pdf.Analysis.Jobs
                         if (pdfFile == null) throw new Exception($"未找Pdf文件:{item.Title},HashCode:{item.HashCode}");
                         var physicalStorageFilePath = pdfFile.GetPhysicalPath();
                         if (!File.Exists(physicalStorageFilePath)) throw new Exception($"未找Pdf文件:{item.Title},HashCode:{item.HashCode}");
+                        Console.WriteLine("-------------------------Pre-------------------------");
+                        Console.WriteLine($"-------------------------{physicalStorageFilePath}-------------------------");
                         var text = Kivii.PdfBox.Instance.ReadText(physicalStorageFilePath);
-                        text = regex.Replace(text, "");
+                        Console.WriteLine("-------------------------Ok-------------------------");
                         item.Allbody = text;
                         item.AddOnlyProperties(o => o.Allbody);
                         item.AnlysisTime = DateTime.Now;
@@ -83,15 +82,19 @@ namespace Njust.Pdf.Analysis.Jobs
                         item.AddOnlyProperties(o => o.AnlysisUserName);
                         item.IsAnlysis = true;
                         item.AddOnlyProperties(o => o.IsAnlysis);
+                        Console.WriteLine("-------------------------PreUpdateOnly-------------------------");
                         conn.UpdateOnly(item);
+                        Console.WriteLine("-------------------------UpdateOnly-------------------------");
                         succeed++;
                     }
                     catch (Exception ex)
                     {
+                        Console.WriteLine("-------------------------Ex-------------------------");
                         TaskContext.Message = $"处理消息异常:共处理{succeed}/{total},Kvid:{item.Kvid},Ex:{ex.Message}";
                         item.Status += 1;
                         item.AddOnlyProperties(o => o.Status);
                         conn.UpdateOnly(item);
+                        Console.WriteLine("-------------------------ExUpdate-------------------------");
 
                         return false;
                     }

Src/Njust.Pdf.Analysis.V4.5.csproj

@@ -84,5 +84,8 @@
   <ItemGroup>
     <None Include="packages.config" />
   </ItemGroup>
+  <ItemGroup>
+    <EmbeddedResource Include="Resources\countrytranslation.txt" />
+  </ItemGroup>
   <Import Project="$(MSBuildToolsPath)\Microsoft.CSharp.targets" />
 </Project>
\ No newline at end of file

Src/Njust.Pdf.Analysis.V4.5.csproj.user

+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="Current" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <PropertyGroup>
+    <ProjectView>ProjectFiles</ProjectView>
+  </PropertyGroup>
+</Project>
\ No newline at end of file

Src/Resources/countrytranslation.txt

+Singapore Rep.:新加坡
+Dominican Rep.:多米尼加
+Palestine:巴勒斯坦
+Bahamas:巴哈马
+Timor-Leste:东帝汶
+Afghanistan:阿富汗
+Guinea-Bissau:几内亚比绍
+Côte dIvoire:科特迪瓦
+Siachen Glacier:锡亚琴冰川
+Br. Indian Ocean Ter.:英属印度洋领土
+Angola:安哥拉
+Albania:阿尔巴尼亚
+United Arab Emirates:阿联酋
+Argentina:阿根廷
+Armenia:亚美尼亚
+French Southern and Antarctic Lands:法属南半球和南极领地
+Australia:澳大利亚
+Austria:奥地利
+Azerbaijan:阿塞拜疆
+Burundi:布隆迪
+Belgium:比利时
+Benin:贝宁
+Burkina Faso:布基纳法索
+Bangladesh:孟加拉国
+Bulgaria:保加利亚
+The Bahamas:巴哈马
+Bosnia and Herz.:波斯尼亚和黑塞哥维那
+Belarus:白俄罗斯
+Belize:伯利兹
+Bermuda:百慕大
+Bolivia:玻利维亚
+Brazil:巴西
+Brunei:文莱
+Bhutan:不丹
+Botswana:博茨瓦纳
+Central African Rep.:中非
+Canada:加拿大
+Switzerland:瑞士
+Chile:智利
+China:中国
+Ivory Coast:象牙海岸
+Cameroon:喀麦隆
+Dem. Rep. Congo:刚果民主共和国
+Congo:刚果
+Colombia:哥伦比亚
+Costa Rica:哥斯达黎加
+Cuba:古巴
+N. Cyprus:北塞浦路斯
+Cyprus:塞浦路斯
+Czech Rep.:捷克
+Germany:德国
+Djibouti:吉布提
+Denmark:丹麦
+Algeria:阿尔及利亚
+Ecuador:厄瓜多尔
+Egypt:埃及
+Eritrea:厄立特里亚
+Spain:西班牙
+Estonia:爱沙尼亚
+Ethiopia:埃塞俄比亚
+Finland:芬兰
+Fiji:斐
+Falkland Islands:福克兰群岛
+France:法国
+Gabon:加蓬
+United Kingdom:英国
+Georgia:格鲁吉亚
+Ghana:加纳
+Guinea:几内亚
+Gambia:冈比亚
+Guinea Bissau:几内亚比绍
+Eq. Guinea:赤道几内亚
+Greece:希腊
+Greenland:格陵兰
+Guatemala:危地马拉
+French Guiana:法属圭亚那
+Guyana:圭亚那
+Honduras:洪都拉斯
+Croatia:克罗地亚
+Haiti:海地
+Hungary:匈牙利
+Indonesia:印度尼西亚
+India:印度
+Ireland:爱尔兰
+Iran:伊朗
+Iraq:伊拉克
+Iceland:冰岛
+Israel:以色列
+Italy:意大利
+Jamaica:牙买加
+Jordan:约旦
+Japan:日本
+Japan:日本本土
+Kazakhstan:哈萨克斯坦
+Kenya:肯尼亚
+Kyrgyzstan:吉尔吉斯斯坦
+Cambodia:柬埔寨
+Korea:韩国
+Kosovo:科索沃
+Kuwait:科威特
+Lao PDR:老挝
+Lebanon:黎巴嫩
+Liberia:利比里亚
+Libya:利比亚
+Sri Lanka:斯里兰卡
+Lesotho:莱索托
+Lithuania:立陶宛
+Luxembourg:卢森堡
+Latvia:拉脱维亚
+Morocco:摩洛哥
+Moldova:摩尔多瓦
+Madagascar:马达加斯加
+Mexico:墨西哥
+Macedonia:马其顿
+Mali:马里
+Myanmar:缅甸
+Montenegro:黑山
+Mongolia:蒙古
+Mozambique:莫桑比克
+Mauritania:毛里塔尼亚
+Malawi:马拉维
+Malaysia:马来西亚
+Namibia:纳米比亚
+New Caledonia:新喀里多尼亚
+Niger:尼日尔
+Nigeria:尼日利亚
+Nicaragua:尼加拉瓜
+Netherlands:荷兰
+Norway:挪威
+Nepal:尼泊尔
+New Zealand:新西兰
+Oman:阿曼
+Pakistan:巴基斯坦
+Panama:巴拿马
+Peru:秘鲁
+Philippines:菲律宾
+Papua New Guinea:巴布亚新几内亚
+Poland:波兰
+Puerto Rico:波多黎各
+Dem. Rep. Korea:朝鲜
+Portugal:葡萄牙
+Paraguay:巴拉圭
+Qatar:卡塔尔
+Romania:罗马尼亚
+Russia:俄罗斯
+Rwanda:卢旺达
+W. Sahara:西撒哈拉
+Saudi Arabia:沙特阿拉伯
+Sudan:苏丹
+S. Sudan:南苏丹
+Senegal:塞内加尔
+Solomon Is.:所罗门群岛
+Sierra Leone:塞拉利昂
+El Salvador:萨尔瓦多
+Somaliland:索马里兰
+Somalia:索马里
+Serbia:塞尔维亚
+Suriname:苏里南
+Slovakia:斯洛伐克
+Slovenia:斯洛文尼亚
+Sweden:瑞典
+Swaziland:斯威士兰
+Syria:叙利亚
+Chad:乍得
+Togo:多哥
+Thailand:泰国
+Tajikistan:塔吉克斯坦
+Turkmenistan:土库曼斯坦
+East Timor:东帝汶
+Trinidad and Tobago:特里尼达和多巴哥
+Tunisia:突尼斯
+Turkey:土耳其
+Tanzania:坦桑尼亚
+Uganda:乌干达
+Ukraine:乌克兰
+Uruguay:乌拉圭
+United States:美国
+Uzbekistan:乌兹别克斯坦
+Venezuela:委内瑞拉
+Vietnam:越南
+Vanuatu:瓦努阿图
+West Bank:西岸
+Yemen:也门
+South Africa:南非
+Zambia:赞比亚
+Zimbabwe:津巴布韦
\ No newline at end of file

Src/Tranforms/RestfulAnalyse.cs

@@ -192,6 +192,8 @@ namespace Njust.Pdf.Analysis.Tranforms
             var allAuthors = conn.Select<Author>(o => Sql.In(o.AnalyseKvid, Items.ConvertAll(p => p.Kvid)));
             var allKeywords = conn.Select<Keyword>(o => Sql.In(o.AnalyseKvid, Items.ConvertAll(p => p.Kvid)));
             var allOrgans = conn.Select<Organ>(o => Sql.In(o.AnalyseKvid, Items.ConvertAll(p => p.Kvid)));
+            var allItems = conn.Select<Analyse>(o => Sql.In(o.Kvid, Items.ConvertAll(p => p.Kvid)));
+            allItems.ThrowIfNullOrEmpty("未找到要复核的项!");
 
             var trans = conn.OpenTransaction();
 
@@ -201,6 +203,8 @@ namespace Njust.Pdf.Analysis.Tranforms
                 {
                     if (item == null) continue;
                     if (item.Kvid == Guid.Empty) continue;
+                    var exist = allItems.FirstOrDefault(o => o.Kvid == item.Kvid);
+                    if (exist == null) continue;
                     item.RemoveAllOnlyProperties();
                     #region Author
                     var currentAuthors = allAuthors.Where(o => o.AnalyseKvid == item.Kvid).ToList();
@@ -212,9 +216,9 @@ namespace Njust.Pdf.Analysis.Tranforms
                             if (currentAuthors.Exists(o => o.Name == auth)) continue;
                             var author = new Author();
                             author.Name = auth;
-                            author.DomainName = item.DomainName;
-                            author.Language = item.Language;
-                            author.AnalyseKvid = item.Kvid;
+                            author.DomainName = exist.DomainName;
+                            author.Language = exist.Language;
+                            author.AnalyseKvid = exist.Kvid;
                             conn.Insert(author);
                         }
                         item.Author = string.Join(";", authors);
@@ -233,9 +237,9 @@ namespace Njust.Pdf.Analysis.Tranforms
                             if (currentKeywords.Exists(o => o.Title == key)) continue;
                             var keyword = new Keyword();
                             keyword.Title = key;
-                            keyword.DomainName = item.DomainName;
-                            keyword.Language = item.Language;
-                            keyword.AnalyseKvid = item.Kvid;
+                            keyword.DomainName = exist.DomainName;
+                            keyword.Language = exist.Language;
+                            keyword.AnalyseKvid = exist.Kvid;
                             conn.Insert(keyword);
                         }
                         item.Keywords = string.Join(";", keywords);
@@ -253,9 +257,9 @@ namespace Njust.Pdf.Analysis.Tranforms
                             if (currentOrgans.Exists(o => o.Title == key)) continue;
                             var organ = new Organ();
                             organ.Title = key;
-                            organ.DomainName = item.DomainName;
-                            organ.Language = item.Language;
-                            organ.AnalyseKvid = item.Kvid;
+                            organ.DomainName = exist.DomainName;
+                            organ.Language = exist.Language;
+                            organ.AnalyseKvid = exist.Kvid;
                             conn.Insert(organ);
                         }
                         item.Organization = string.Join(";", organs);