1 |
830652de
|
Vojtěch Bartička
|
using Core.Entities;
|
2 |
|
|
using Models.Enums;
|
3 |
|
|
using Newtonsoft.Json;
|
4 |
|
|
using System;
|
5 |
|
|
using System.Collections.Generic;
|
6 |
|
|
using System.IO.Compression;
|
7 |
|
|
using System.Linq;
|
8 |
|
|
using System.Text;
|
9 |
|
|
using System.Threading.Tasks;
|
10 |
|
|
|
11 |
|
|
namespace Core.ZipUtils
|
12 |
|
|
{
|
13 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
14 |
|
|
/// Static methods for database exporting
|
15 |
|
|
/// </summary>
|
16 |
830652de
|
Vojtěch Bartička
|
public class Export
|
17 |
|
|
{
|
18 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
19 |
|
|
/// Exports the specified annotations
|
20 |
|
|
/// </summary>
|
21 |
|
|
/// <param name="documents">list of documents to export</param>
|
22 |
|
|
/// <param name="documentAnnotations">document-annotations to export mapping</param>
|
23 |
|
|
/// <param name="documentFinalAnnotations">document-final annotations to export mapping</param>
|
24 |
|
|
/// <param name="annotationTags">annotation-annotation tags mapping</param>
|
25 |
|
|
/// <param name="finalAnnotationTags">final annotation-final annotation tags mapping</param>
|
26 |
|
|
/// <returns></returns>
|
27 |
830652de
|
Vojtěch Bartička
|
public static MemoryStream FullExport(List<Document> documents, Dictionary<Document, List<Annotation>> documentAnnotations, Dictionary<Document, FinalAnnotation> documentFinalAnnotations,
|
28 |
|
|
Dictionary<Annotation, List<AnnotationTag>> annotationTags, Dictionary<FinalAnnotation, List<FinalAnnotationTag>> finalAnnotationTags)
|
29 |
|
|
{
|
30 |
|
|
MemoryStream ms = new MemoryStream();
|
31 |
a35cb648
|
Vojtěch Bartička
|
// Set keepOpen to true so that Dispose() does not close the MemoryStream
|
32 |
b875be78
|
Vojtěch Bartička
|
var archive = new ZipArchive(ms, ZipArchiveMode.Create, true);
|
33 |
830652de
|
Vojtěch Bartička
|
|
34 |
|
|
for (int docIndex = 0; docIndex < documents.Count; docIndex++)
|
35 |
|
|
{
|
36 |
|
|
var document = documents[docIndex];
|
37 |
|
|
string documentDir = $"{docIndex:00000}";
|
38 |
|
|
|
39 |
|
|
// Write document content and name into a JSON file
|
40 |
|
|
string documentInfoJson = DumpDocument(document);
|
41 |
|
|
CreateFile(documentDir + "/document.json", documentInfoJson, archive);
|
42 |
|
|
|
43 |
|
|
// Deal with annotations
|
44 |
|
|
var annotations = documentAnnotations[document];
|
45 |
|
|
for (int annotationIndex = 0; annotationIndex < annotations.Count; annotationIndex++)
|
46 |
|
|
{
|
47 |
|
|
var annotation = annotations[annotationIndex];
|
48 |
|
|
var tags = annotationTags[annotation].Select(at => at as AnnotationTagGeneric).ToList();
|
49 |
|
|
string annotationInfoJson = DumpAnnotation(annotation, tags, document.Id);
|
50 |
|
|
CreateFile(documentDir + $"/annotation_{annotationIndex:00000}.json", annotationInfoJson, archive);
|
51 |
|
|
}
|
52 |
|
|
|
53 |
a35cb648
|
Vojtěch Bartička
|
// Deal with final annotation if it exists
|
54 |
830652de
|
Vojtěch Bartička
|
var finalAnnotation = documentFinalAnnotations[document];
|
55 |
baf6fc22
|
Vojtěch Bartička
|
if (finalAnnotation == null)
|
56 |
|
|
{
|
57 |
|
|
continue;
|
58 |
|
|
}
|
59 |
|
|
|
60 |
830652de
|
Vojtěch Bartička
|
var finalTags = finalAnnotationTags[finalAnnotation].Select(ft => ft as AnnotationTagGeneric).ToList();
|
61 |
|
|
string finalAnnotationInfoJson = DumpAnnotation(finalAnnotation, finalTags, document.Id);
|
62 |
|
|
CreateFile(documentDir + "/final.json", finalAnnotationInfoJson, archive);
|
63 |
|
|
}
|
64 |
|
|
|
65 |
a35cb648
|
Vojtěch Bartička
|
// Dispose of archive to close the header
|
66 |
b875be78
|
Vojtěch Bartička
|
archive.Dispose();
|
67 |
830652de
|
Vojtěch Bartička
|
ms.Position = 0;
|
68 |
|
|
return ms;
|
69 |
|
|
}
|
70 |
|
|
|
71 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
72 |
|
|
/// Creates a file and writes to it
|
73 |
|
|
/// </summary>
|
74 |
|
|
/// <param name="path">path to file</param>
|
75 |
|
|
/// <param name="fileContent">string to write</param>
|
76 |
|
|
/// <param name="archive">archive to write it to</param>
|
77 |
830652de
|
Vojtěch Bartička
|
private static void CreateFile(string path, string fileContent, ZipArchive archive)
|
78 |
|
|
{
|
79 |
|
|
var entry = archive.CreateEntry(path);
|
80 |
|
|
var entryStream = entry.Open();
|
81 |
|
|
entryStream.Write(Encoding.UTF8.GetBytes(fileContent));
|
82 |
|
|
entryStream.Close();
|
83 |
|
|
}
|
84 |
|
|
|
85 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
86 |
|
|
/// Dumps a Document into JSON-serialized ExportDocumentInfo
|
87 |
|
|
/// </summary>
|
88 |
|
|
/// <param name="document">Document to serialize</param>
|
89 |
|
|
/// <returns>JSON-serialized ExportDocumentInfo</returns>
|
90 |
830652de
|
Vojtěch Bartička
|
private static string DumpDocument(Document document)
|
91 |
|
|
{
|
92 |
|
|
ExportDocumentInfo documentInfo = new();
|
93 |
|
|
documentInfo.Name = document.Name;
|
94 |
|
|
documentInfo.Content = document.Content.Content;
|
95 |
|
|
documentInfo.DocumentId = document.Id;
|
96 |
|
|
|
97 |
|
|
return JsonConvert.SerializeObject(documentInfo);
|
98 |
|
|
}
|
99 |
|
|
|
100 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
101 |
|
|
/// Dumps annotation into JSON-serialized ExportAnnotationInfo
|
102 |
|
|
/// </summary>
|
103 |
|
|
/// <param name="annotation">Annotation to dump</param>
|
104 |
|
|
/// <param name="tags">Tags related to the annotation</param>
|
105 |
|
|
/// <param name="documentId">Id of the document the annotation is related to</param>
|
106 |
|
|
/// <returns>JSON-serialized ExportAnnotationInfo</returns>
|
107 |
830652de
|
Vojtěch Bartička
|
private static string DumpAnnotation(Annotation annotation, List<AnnotationTagGeneric> tags, Guid documentId)
|
108 |
|
|
{
|
109 |
|
|
ExportAnnotationInfo annotationInfo = new();
|
110 |
|
|
annotationInfo.AnnotatorId = annotation.User.Id;
|
111 |
|
|
annotationInfo.DocumentId = documentId;
|
112 |
|
|
|
113 |
|
|
foreach (var tag in tags)
|
114 |
|
|
{
|
115 |
|
|
ExportTagInstanceInfo tagInfo = new()
|
116 |
|
|
{
|
117 |
|
|
|
118 |
|
|
Category = tag.Tag.Category.Name,
|
119 |
|
|
TagName = tag.Tag.Name,
|
120 |
|
|
InstanceId = tag.Instance,
|
121 |
|
|
FirstCharPosition = tag.Position,
|
122 |
|
|
Length = tag.Length,
|
123 |
|
|
};
|
124 |
|
|
|
125 |
|
|
if (tag.Tag.SentimentEnabled && tagInfo.Sentiment != null)
|
126 |
|
|
{
|
127 |
|
|
tagInfo.Sentiment = tag.Sentiment.ToString().ToLower();
|
128 |
|
|
}
|
129 |
|
|
|
130 |
|
|
if (tag.SubTag != null)
|
131 |
|
|
{
|
132 |
|
|
tagInfo.SubTagName = tag.SubTag.Name;
|
133 |
|
|
}
|
134 |
|
|
|
135 |
|
|
annotationInfo.Tags.Add(tagInfo);
|
136 |
|
|
}
|
137 |
|
|
|
138 |
|
|
return JsonConvert.SerializeObject(annotationInfo);
|
139 |
|
|
}
|
140 |
|
|
}
|
141 |
|
|
|
142 |
|
|
public class ExportDocumentInfo
|
143 |
|
|
{
|
144 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
145 |
|
|
/// Document name
|
146 |
|
|
/// </summary>
|
147 |
830652de
|
Vojtěch Bartička
|
public string Name { get; set; }
|
148 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
149 |
|
|
/// Original document content
|
150 |
|
|
/// </summary>
|
151 |
830652de
|
Vojtěch Bartička
|
public string Content { get; set; }
|
152 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
153 |
|
|
/// Id of the document
|
154 |
|
|
/// </summary>
|
155 |
830652de
|
Vojtěch Bartička
|
public Guid DocumentId { get; set; }
|
156 |
|
|
}
|
157 |
|
|
|
158 |
|
|
public class ExportAnnotationInfo
|
159 |
|
|
{
|
160 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
161 |
|
|
/// Annotator Id
|
162 |
|
|
/// </summary>
|
163 |
830652de
|
Vojtěch Bartička
|
public Guid AnnotatorId { get; set; }
|
164 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
165 |
|
|
/// Document Id
|
166 |
|
|
/// </summary>
|
167 |
830652de
|
Vojtěch Bartička
|
public Guid DocumentId { get; set; }
|
168 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
169 |
|
|
/// Tags
|
170 |
|
|
/// </summary>
|
171 |
830652de
|
Vojtěch Bartička
|
public List<ExportTagInstanceInfo> Tags { get; set; } = new();
|
172 |
|
|
}
|
173 |
|
|
|
174 |
|
|
public class ExportTagInstanceInfo
|
175 |
|
|
{
|
176 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
177 |
|
|
/// Tag category
|
178 |
|
|
/// </summary>
|
179 |
830652de
|
Vojtěch Bartička
|
public string Category { get; set; }
|
180 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
181 |
|
|
/// Name of the tag (the primary identifier)
|
182 |
|
|
/// </summary>
|
183 |
830652de
|
Vojtěch Bartička
|
public string TagName { get; set; }
|
184 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
185 |
|
|
/// Name of the subtag (the primary identifier of a subtag)
|
186 |
|
|
/// </summary>
|
187 |
830652de
|
Vojtěch Bartička
|
public string? SubTagName { get; set; }
|
188 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
189 |
|
|
/// Id of the instance (in case of multipart tags multiple tags share the same Id)
|
190 |
|
|
/// </summary>
|
191 |
830652de
|
Vojtěch Bartička
|
public Guid InstanceId { get; set; }
|
192 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
193 |
|
|
/// Position in the original document
|
194 |
|
|
/// </summary>
|
195 |
830652de
|
Vojtěch Bartička
|
public int FirstCharPosition { get; set; }
|
196 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
197 |
|
|
/// Length of the section in chars
|
198 |
|
|
/// </summary>
|
199 |
830652de
|
Vojtěch Bartička
|
public int Length { get; set; }
|
200 |
a35cb648
|
Vojtěch Bartička
|
/// <summary>
|
201 |
|
|
/// Optional sentiment
|
202 |
|
|
/// </summary>
|
203 |
830652de
|
Vojtěch Bartička
|
public string? Sentiment { get; set; }
|
204 |
|
|
}
|
205 |
|
|
}
|