1
|
using Core.Contexts;
|
2
|
using Core.Entities;
|
3
|
using Models.Annotations;
|
4
|
using Models.Enums;
|
5
|
using Serilog;
|
6
|
using System;
|
7
|
using System.Collections.Generic;
|
8
|
using System.Linq;
|
9
|
using System.Text;
|
10
|
using System.Threading.Tasks;
|
11
|
using Microsoft.EntityFrameworkCore;
|
12
|
using AutoMapper;
|
13
|
using Models.Tags;
|
14
|
using Ganss.XSS;
|
15
|
using HtmlAgilityPack;
|
16
|
using System.Text.RegularExpressions;
|
17
|
using Newtonsoft.Json;
|
18
|
using Core.GraphUtils;
|
19
|
|
20
|
namespace Core.Services.AnnotationService
|
21
|
{
|
22
|
public class AnnotationServiceEF : IAnnotationService
|
23
|
{
|
24
|
private readonly DatabaseContext context;
|
25
|
private readonly ILogger logger;
|
26
|
private readonly IMapper mapper;
|
27
|
|
28
|
private const string TAG_ID_ATTRIBUTE_NAME = "aswi-tag-id";
|
29
|
private const string TAG_INSTANCE_ATTRIBUTE_NAME = "aswi-tag-instance";
|
30
|
private const string TAG_EF_ID_ATTRIBUTE_NAME = "aswi-tag-ef-id";
|
31
|
|
32
|
public AnnotationServiceEF(DatabaseContext context, ILogger logger, IMapper mapper)
|
33
|
{
|
34
|
this.context = context;
|
35
|
this.logger = logger;
|
36
|
this.mapper = mapper;
|
37
|
}
|
38
|
|
39
|
public void CreateDocumentAnnotations(AnnotationsAddRequest request, Guid clientUserId)
|
40
|
{
|
41
|
User addingUser = context.Users.Single(u => u.Id == clientUserId);
|
42
|
|
43
|
// Check the documents exist
|
44
|
var documents = context.Documents.Where(d => request.DocumentIdList.Contains(d.Id)).ToList();
|
45
|
if (documents.Count() != request.DocumentIdList.Count)
|
46
|
{
|
47
|
logger.Information($"Received a non-existent Document ID when assigning documents to users");
|
48
|
throw new InvalidOperationException($"{request.DocumentIdList.Count - documents.Count()} of the received documents do not exist");
|
49
|
}
|
50
|
|
51
|
var users = context.Users.Where(u => request.UserIdList.Contains(u.Id)).ToList();
|
52
|
foreach (var user in users)
|
53
|
{
|
54
|
var userAnnotatedDocuments = context.Annotations.Where(a => a.User == user).Select(a => a.Document).ToList();
|
55
|
foreach (var doc in documents)
|
56
|
{
|
57
|
if (userAnnotatedDocuments.Contains(doc))
|
58
|
{
|
59
|
logger.Information($"User {user.Username} has already been assigned the document {doc.Id}, ignoring");
|
60
|
continue;
|
61
|
}
|
62
|
|
63
|
context.Annotations.Add(new Annotation()
|
64
|
{
|
65
|
User = user,
|
66
|
UserAssigned = addingUser,
|
67
|
DateAssigned = DateTime.Now,
|
68
|
DateLastChanged = DateTime.Now,
|
69
|
Document = doc,
|
70
|
State = EState.NEW,
|
71
|
Note = ""
|
72
|
});
|
73
|
}
|
74
|
}
|
75
|
|
76
|
context.SaveChanges();
|
77
|
}
|
78
|
|
79
|
public AnnotationListResponse GetUserAnnotations(Guid userId)
|
80
|
{
|
81
|
var annotations = context.Annotations.Where(a => a.User.Id == userId).Include(a => a.Document).ToList();
|
82
|
var documentIds = annotations.Select(a => a.Document.Id).ToList();
|
83
|
var documents = context.Documents.Where(d => documentIds.Contains(d.Id));
|
84
|
var infos = new List<AnnotationListInfo>();
|
85
|
|
86
|
var annotationsDocuments = annotations.Zip(documents, (a, d) => new { Annotation = a, Document = d });
|
87
|
foreach (var ad in annotationsDocuments)
|
88
|
{
|
89
|
infos.Add(new AnnotationListInfo()
|
90
|
{
|
91
|
AnnotationId = ad.Annotation.Id,
|
92
|
DocumentName = ad.Document.Name,
|
93
|
State = ad.Annotation.State
|
94
|
});
|
95
|
}
|
96
|
|
97
|
return new AnnotationListResponse()
|
98
|
{
|
99
|
Annotations = infos
|
100
|
};
|
101
|
}
|
102
|
|
103
|
public void AddNoteToAnnotation(Guid annotationId, Guid userId, ERole userRole, AddNoteToAnnotationRequest request)
|
104
|
{
|
105
|
Annotation annotation = null;
|
106
|
try
|
107
|
{
|
108
|
annotation = context.Annotations.Include(a => a.User).First(a => a.Id == annotationId);
|
109
|
}
|
110
|
catch (Exception)
|
111
|
{
|
112
|
throw new InvalidOperationException("Annotation not found");
|
113
|
}
|
114
|
|
115
|
if (userRole < ERole.ADMINISTRATOR && annotation.User.Id != userId)
|
116
|
{
|
117
|
throw new UnauthorizedAccessException("User does not have access to this annotation");
|
118
|
}
|
119
|
|
120
|
annotation.Note = request.Note;
|
121
|
context.SaveChanges();
|
122
|
}
|
123
|
|
124
|
|
125
|
public AnnotationInfo GetAnnotation(Guid annotationId, Guid userId, ERole userRole)
|
126
|
{
|
127
|
var annotation = context.Annotations
|
128
|
.Where(a => a.Id == annotationId)
|
129
|
.Include(a => a.User)
|
130
|
.Include(a => a.Document).ThenInclude(d => d.Content)
|
131
|
.First();
|
132
|
|
133
|
if (userRole < ERole.ADMINISTRATOR)
|
134
|
{
|
135
|
if (annotation.User.Id != userId)
|
136
|
{
|
137
|
throw new UnauthorizedAccessException($"User {userId} does not have assigned annotation {annotationId}");
|
138
|
}
|
139
|
}
|
140
|
|
141
|
var documentContent = context.Documents.Where(d => d.Id == annotation.Document.Id).Select(d => d.Content).First();
|
142
|
|
143
|
var tags = context.AnnotationTags.Where(at => at.Annotation.Id == annotationId)
|
144
|
.Include(at => at.Tag)
|
145
|
.ThenInclude(t => t.Category)
|
146
|
.Include(at => at.SubTag)
|
147
|
.OrderBy(at => at.Position)
|
148
|
.ToList();
|
149
|
|
150
|
List<TagInstanceInfo> tagInstanceInfos = new();
|
151
|
foreach (var tag in tags)
|
152
|
{
|
153
|
var tagInstance = mapper.Map<TagInstanceInfo>(tag);
|
154
|
tagInstanceInfos.Add(tagInstance);
|
155
|
}
|
156
|
|
157
|
var docToRender = "";
|
158
|
if (annotation.CachedDocumentHTML == "")
|
159
|
{
|
160
|
docToRender = FullPreprocessHTML(documentContent.Content, tags);
|
161
|
annotation.CachedStartPositions = JsonConvert.SerializeObject(TagStartPositions);
|
162
|
annotation.CachedLengths = JsonConvert.SerializeObject(TagStartLengths);
|
163
|
annotation.CachedClosingPositions = JsonConvert.SerializeObject(TagClosingPositions);
|
164
|
annotation.CachedClosingLengths = JsonConvert.SerializeObject(TagClosingLengths);
|
165
|
annotation.CachedCSS = JsonConvert.SerializeObject(TagInstanceCSS);
|
166
|
annotation.ModifiedType = EModified.NONE;
|
167
|
annotation.CachedDocumentHTML = docToRender;
|
168
|
context.SaveChanges();
|
169
|
}
|
170
|
else
|
171
|
{
|
172
|
docToRender = annotation.CachedDocumentHTML;
|
173
|
TagStartPositions = JsonConvert.DeserializeObject<List<int>>(annotation.CachedStartPositions);
|
174
|
TagStartLengths = JsonConvert.DeserializeObject<List<int>>(annotation.CachedLengths);
|
175
|
TagClosingPositions = JsonConvert.DeserializeObject<List<int>>(annotation.CachedClosingPositions);
|
176
|
TagClosingLengths = JsonConvert.DeserializeObject<List<int>>(annotation.CachedClosingLengths);
|
177
|
TagInstanceCSS = JsonConvert.DeserializeObject<List<TagInstanceCSSInfo>>(annotation.CachedCSS);
|
178
|
|
179
|
// The annotation has been modified and we need to either add the new tag or remove the tag
|
180
|
if (annotation.ModifiedType != EModified.NONE)
|
181
|
{
|
182
|
if (annotation.ModifiedType == EModified.ADDED)
|
183
|
{
|
184
|
var lastModifiedTag = context.AnnotationTags.Where(at => at.Id == annotation.LastModifiedTagId).First();
|
185
|
docToRender = PartialPreprocessHTMLAddTag(docToRender, documentContent.Content, lastModifiedTag, tags);
|
186
|
}
|
187
|
else if (annotation.ModifiedType == EModified.REMOVED)
|
188
|
{
|
189
|
docToRender = PartialPreprocessHTMLRemoveTag(docToRender, documentContent.Content, new AnnotationTag() { Id = annotation.LastModifiedTagId.Value }, tags);
|
190
|
}
|
191
|
|
192
|
annotation.ModifiedType = EModified.NONE;
|
193
|
annotation.CachedStartPositions = JsonConvert.SerializeObject(TagStartPositions);
|
194
|
annotation.CachedLengths = JsonConvert.SerializeObject(TagStartLengths);
|
195
|
annotation.CachedClosingPositions = JsonConvert.SerializeObject(TagClosingPositions);
|
196
|
annotation.CachedClosingLengths = JsonConvert.SerializeObject(TagClosingLengths);
|
197
|
annotation.CachedDocumentHTML = docToRender;
|
198
|
annotation.CachedCSS = JsonConvert.SerializeObject(TagInstanceCSS);
|
199
|
context.SaveChanges();
|
200
|
}
|
201
|
}
|
202
|
|
203
|
// We probably cannot use AutoMapper since we are dealing with too many different entities
|
204
|
AnnotationInfo annotationInfo = new()
|
205
|
{
|
206
|
SourceDocumentContent = documentContent.Content,
|
207
|
DocumentToRender = docToRender,
|
208
|
TagStartPositions = TagStartPositions.ToArray(),
|
209
|
TagLengths = TagStartLengths.ToArray(),
|
210
|
Note = annotation.Note,
|
211
|
State = annotation.State,
|
212
|
Type = IsHtml(documentContent.Content) ? EDocumentType.HTML : EDocumentType.TEXT,
|
213
|
TagInstances = tagInstanceInfos,
|
214
|
CSSInfo = TagInstanceCSS
|
215
|
};
|
216
|
|
217
|
NodeDict.Clear();
|
218
|
|
219
|
return annotationInfo;
|
220
|
}
|
221
|
|
222
|
private List<int> TagStartPositions = new();
|
223
|
private List<int> TagStartLengths = new();
|
224
|
private List<int> TagClosingPositions = new();
|
225
|
private List<int> TagClosingLengths = new();
|
226
|
private Dictionary<HtmlNode, HtmlNode> NodeDict = new();
|
227
|
private List<TagInstanceCSSInfo> TagInstanceCSS = new();
|
228
|
|
229
|
/*
|
230
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
231
|
*/
|
232
|
|
233
|
|
234
|
private string FullPreprocessHTML(string htmlSource, List<AnnotationTag> tags)
|
235
|
{
|
236
|
var docOriginal = new HtmlDocument();
|
237
|
docOriginal.LoadHtml(htmlSource);
|
238
|
var docToEdit = new HtmlDocument();
|
239
|
docToEdit.LoadHtml(htmlSource);
|
240
|
|
241
|
var descendantsOriginal = docOriginal.DocumentNode.DescendantsAndSelf();
|
242
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
243
|
|
244
|
int currentId = 0;
|
245
|
|
246
|
FillNodeDict(descendantsOriginal, descendantsToEdit);
|
247
|
AssignIdsToOriginalDocument(descendantsOriginal, ref currentId);
|
248
|
|
249
|
WrapTextInSpan(descendantsOriginal, docToEdit);
|
250
|
|
251
|
int descCount = descendantsToEdit.Count;
|
252
|
|
253
|
foreach (var tag in tags)
|
254
|
{
|
255
|
int i = 0;
|
256
|
List<HtmlNode> addedForSelection = new();
|
257
|
while (i < descCount)
|
258
|
{
|
259
|
for (; i < descCount; i++)
|
260
|
{
|
261
|
var node = descendantsToEdit.ElementAt(i);
|
262
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) ||
|
263
|
node.ParentNode.Name == "style")
|
264
|
{
|
265
|
continue;
|
266
|
}
|
267
|
|
268
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
269
|
|
270
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
271
|
var end = TagClosingPositions[nodeId];
|
272
|
|
273
|
int selectionStart = tag.Position;
|
274
|
int selectionEnd = tag.Position + tag.Length;
|
275
|
|
276
|
if (selectionStart < end && selectionEnd > start)
|
277
|
{
|
278
|
if (selectionStart <= start && selectionEnd >= end)
|
279
|
{
|
280
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
281
|
}
|
282
|
else if (selectionStart <= start)
|
283
|
{
|
284
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
285
|
}
|
286
|
else if (selectionEnd >= end)
|
287
|
{
|
288
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
289
|
}
|
290
|
else
|
291
|
{
|
292
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
293
|
}
|
294
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
295
|
descCount = descendantsToEdit.Count;
|
296
|
break;
|
297
|
}
|
298
|
}
|
299
|
}
|
300
|
|
301
|
}
|
302
|
|
303
|
string docToRender = docToEdit.DocumentNode.OuterHtml;
|
304
|
HtmlSanitizer sanitizer = new HtmlSanitizer();
|
305
|
sanitizer.AllowedAttributes.Clear();
|
306
|
sanitizer.AllowedAttributes.Add(TAG_ID_ATTRIBUTE_NAME);
|
307
|
sanitizer.AllowedAttributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME);
|
308
|
sanitizer.AllowedAttributes.Add("end");
|
309
|
sanitizer.AllowedAttributes.Add("start");
|
310
|
sanitizer.AllowedAttributes.Add("class");
|
311
|
if (sanitizer.AllowedTags.Contains("script"))
|
312
|
{
|
313
|
sanitizer.AllowedTags.Remove("script");
|
314
|
}
|
315
|
if (!sanitizer.AllowedTags.Contains("style"))
|
316
|
{
|
317
|
sanitizer.AllowedTags.Add("style");
|
318
|
}
|
319
|
docToRender = sanitizer.Sanitize(docToRender);
|
320
|
GenerateCSS(tags);
|
321
|
return docToRender;
|
322
|
}
|
323
|
|
324
|
private HtmlNode SolveFullFill(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit, AnnotationTag tag)
|
325
|
{
|
326
|
// full fill
|
327
|
string textSelected = node.InnerText;
|
328
|
|
329
|
var parentNode = node.ParentNode;
|
330
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
331
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
332
|
|
333
|
EPosition markerPosition = EPosition.MARK_NONE;
|
334
|
if (selectionEnd == end && selectionStart == start)
|
335
|
{
|
336
|
markerPosition = EPosition.MARK_LEFT_RIGHT;
|
337
|
}
|
338
|
|
339
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, start, markerPosition);
|
340
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
341
|
|
342
|
return spanSelected;
|
343
|
}
|
344
|
|
345
|
private List<HtmlNode> SolveRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
346
|
AnnotationTag tag)
|
347
|
{
|
348
|
// partial fill, end gap
|
349
|
string text = node.InnerText;
|
350
|
string textAfter = text.Substring(Math.Min(selectionStart - start + tag.Length, text.Length));
|
351
|
string textSelected = text.Substring(0, selectionEnd - start);
|
352
|
|
353
|
var parentNode = node.ParentNode;
|
354
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
355
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
356
|
|
357
|
int spanSelectedStart = start;
|
358
|
int spanAfterStart = start + textSelected.Length;
|
359
|
|
360
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_RIGHT);
|
361
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
362
|
|
363
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
364
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanAfter);
|
365
|
|
366
|
return new() { spanSelected, spanAfter };
|
367
|
}
|
368
|
|
369
|
private List<HtmlNode> SolveLeftGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
370
|
AnnotationTag tag)
|
371
|
{
|
372
|
// partial fill, start gap
|
373
|
string text = node.InnerText;
|
374
|
string textBefore = text.Substring(0, selectionStart - start);
|
375
|
string textSelected = text.Substring(selectionStart - start, Math.Min(tag.Length, text.Length - textBefore.Length));
|
376
|
|
377
|
var parentNode = node.ParentNode;
|
378
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
379
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
380
|
|
381
|
int spanBeforeStart = start;
|
382
|
int spanSelectedStart = start + textBefore.Length;
|
383
|
|
384
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
385
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
386
|
|
387
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_LEFT);
|
388
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
389
|
|
390
|
return new() { spanSelected, spanBefore };
|
391
|
}
|
392
|
|
393
|
private List<HtmlNode> SolveLeftRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
394
|
AnnotationTag tag)
|
395
|
{
|
396
|
// partial fill, start gap end gap
|
397
|
string text = node.InnerText;
|
398
|
string textBefore = text.Substring(0, selectionStart - start);
|
399
|
string textAfter = text.Substring(selectionStart - start + tag.Length);
|
400
|
string textSelected = text.Substring(selectionStart - start, tag.Length);
|
401
|
|
402
|
var parentNode = node.ParentNode;
|
403
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
404
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
405
|
|
406
|
int spanBeforeStart = start;
|
407
|
int spanSelectedStart = start + textBefore.Length;
|
408
|
int spanAfterStart = start + textBefore.Length + textSelected.Length;
|
409
|
|
410
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
411
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
412
|
|
413
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_LEFT_RIGHT);
|
414
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
415
|
|
416
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
417
|
parentNode.ChildNodes.Insert(nodeIndex + 2, spanAfter);
|
418
|
|
419
|
return new() { spanSelected, spanBefore, spanAfter };
|
420
|
}
|
421
|
|
422
|
private HtmlNode CreateSpan(HtmlDocument doc, string text, int tagId, Guid? instanceId, Guid? entityId, int startPosition, EPosition position = EPosition.MARK_NONE)
|
423
|
{
|
424
|
HtmlNode span = doc.CreateElement("span");
|
425
|
span.InnerHtml = text;
|
426
|
TagStartPositions.Add(startPosition);
|
427
|
TagStartLengths.Add(0);
|
428
|
TagClosingPositions.Add(startPosition + text.Length);
|
429
|
TagClosingLengths.Add(0);
|
430
|
span.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, tagId.ToString());
|
431
|
|
432
|
if (instanceId != null)
|
433
|
{
|
434
|
span.AddClass("annotation");
|
435
|
span.Attributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME, instanceId.Value.ToString());
|
436
|
span.Attributes.Add(TAG_EF_ID_ATTRIBUTE_NAME, entityId.ToString());
|
437
|
|
438
|
if (position == EPosition.MARK_LEFT || position == EPosition.MARK_LEFT_RIGHT)
|
439
|
{
|
440
|
span.Attributes.Add("start", "1");
|
441
|
}
|
442
|
if (position == EPosition.MARK_RIGHT || position == EPosition.MARK_LEFT_RIGHT)
|
443
|
{
|
444
|
span.Attributes.Add("end", "1");
|
445
|
}
|
446
|
}
|
447
|
|
448
|
return span;
|
449
|
}
|
450
|
|
451
|
private enum EPosition
|
452
|
{
|
453
|
MARK_LEFT = 5,
|
454
|
MARK_RIGHT = 3,
|
455
|
MARK_LEFT_RIGHT = 2,
|
456
|
MARK_NONE = 0
|
457
|
}
|
458
|
|
459
|
private void WrapTextInSpan(IEnumerable<HtmlNode> descendantsOriginal, HtmlDocument docToEdit)
|
460
|
{
|
461
|
// Special case for non-html documents
|
462
|
if (descendantsOriginal.Count() == 2)
|
463
|
{
|
464
|
var documentNode = descendantsOriginal.ElementAt(0);
|
465
|
var childNode = descendantsOriginal.ElementAt(1);
|
466
|
if (documentNode.Name == "#document" && childNode.Name == "#text")
|
467
|
{
|
468
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
469
|
coveringSpan.InnerHtml = childNode.InnerHtml;
|
470
|
TagStartPositions.Add(childNode.InnerStartIndex);
|
471
|
TagStartLengths.Add(0);
|
472
|
TagClosingPositions.Add(childNode.InnerStartIndex + childNode.InnerLength);
|
473
|
TagClosingLengths.Add(0);
|
474
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
475
|
|
476
|
var parent = NodeDict[documentNode];
|
477
|
|
478
|
parent.ChildNodes.RemoveAt(0);
|
479
|
parent.ChildNodes.Add(coveringSpan);
|
480
|
|
481
|
return;
|
482
|
}
|
483
|
}
|
484
|
|
485
|
foreach (var node in descendantsOriginal)
|
486
|
{
|
487
|
var originalNode = node;
|
488
|
var toEditNode = NodeDict[node];
|
489
|
|
490
|
if (originalNode.Name.Contains("#"))
|
491
|
{
|
492
|
continue;
|
493
|
}
|
494
|
else
|
495
|
{
|
496
|
bool onlyText = true;
|
497
|
bool onlySubtags = true;
|
498
|
|
499
|
foreach (var child in node.ChildNodes)
|
500
|
{
|
501
|
if (child.Name.Contains("#"))
|
502
|
{
|
503
|
onlySubtags = false;
|
504
|
}
|
505
|
else
|
506
|
{
|
507
|
onlyText = false;
|
508
|
}
|
509
|
}
|
510
|
|
511
|
if (onlyText || onlySubtags)
|
512
|
{
|
513
|
continue;
|
514
|
}
|
515
|
else
|
516
|
{
|
517
|
|
518
|
foreach (var child in node.ChildNodes)
|
519
|
{
|
520
|
if (child.Name.Contains("#text"))
|
521
|
{
|
522
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
523
|
coveringSpan.InnerHtml = child.InnerHtml;
|
524
|
TagStartPositions.Add(child.InnerStartIndex);
|
525
|
TagStartLengths.Add(0);
|
526
|
TagClosingPositions.Add(child.InnerStartIndex + child.InnerLength);
|
527
|
TagClosingLengths.Add(0);
|
528
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
529
|
|
530
|
var parent = NodeDict[node];
|
531
|
var index = parent.ChildNodes.IndexOf(NodeDict[child]);
|
532
|
|
533
|
parent.ChildNodes.RemoveAt(index);
|
534
|
parent.ChildNodes.Insert(index, coveringSpan);
|
535
|
}
|
536
|
}
|
537
|
}
|
538
|
}
|
539
|
}
|
540
|
}
|
541
|
|
542
|
private void GenerateCSS(List<AnnotationTag> tags)
|
543
|
{
|
544
|
/*string inner = "span.annotation {border-bottom: 2px solid;}";
|
545
|
inner += "span {line-height: 30px}\n";*/
|
546
|
|
547
|
var tagPaddingDict = Intersections.ColorGraph(Intersections.FindIntersections(tags));
|
548
|
foreach (var tag in tags)
|
549
|
{
|
550
|
var padding = (tagPaddingDict[tag] + 1) * 2;
|
551
|
TagInstanceCSS.Add(new()
|
552
|
{
|
553
|
InstanceId = tag.Instance,
|
554
|
Color = tag.Tag.Color,
|
555
|
Padding = padding
|
556
|
});
|
557
|
}
|
558
|
}
|
559
|
|
560
|
private void AssignIdsToOriginalDocument(IEnumerable<HtmlNode> descendantsOriginal, ref int currentId)
|
561
|
{
|
562
|
foreach (var node in descendantsOriginal)
|
563
|
{
|
564
|
var originalNode = node;
|
565
|
var toEditNode = NodeDict[node];
|
566
|
|
567
|
if (originalNode.Name.Contains("#"))
|
568
|
{
|
569
|
continue;
|
570
|
}
|
571
|
else
|
572
|
{
|
573
|
TagStartPositions.Add(originalNode.OuterStartIndex);
|
574
|
TagStartLengths.Add(originalNode.InnerStartIndex - originalNode.OuterStartIndex);
|
575
|
currentId = TagStartPositions.Count - 1;
|
576
|
toEditNode.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, currentId.ToString());
|
577
|
|
578
|
TagClosingPositions.Add(originalNode.InnerStartIndex + originalNode.InnerLength);
|
579
|
TagClosingLengths.Add((originalNode.OuterStartIndex + originalNode.OuterLength) - (originalNode.InnerStartIndex + originalNode.InnerLength));
|
580
|
}
|
581
|
}
|
582
|
}
|
583
|
|
584
|
private void FillNodeDict(IEnumerable<HtmlNode> descendantsOriginal, IEnumerable<HtmlNode> descendantsToEdit)
|
585
|
{
|
586
|
var zipped = descendantsOriginal.Zip(descendantsToEdit, (orig, toEdit) => new
|
587
|
{
|
588
|
Original = orig,
|
589
|
ToEdit = toEdit
|
590
|
});
|
591
|
foreach (var node in zipped)
|
592
|
{
|
593
|
var originalNode = node.Original;
|
594
|
var toEditNode = node.ToEdit;
|
595
|
NodeDict.Add(originalNode, toEditNode);
|
596
|
}
|
597
|
}
|
598
|
|
599
|
|
600
|
/*
|
601
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
602
|
*/
|
603
|
|
604
|
/*
|
605
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
606
|
*/
|
607
|
|
608
|
private string PartialPreprocessHTMLAddTag(string htmlToEdit, string htmlOriginal, AnnotationTag tagToAdd, List<AnnotationTag> tags)
|
609
|
{
|
610
|
var docOriginal = new HtmlDocument();
|
611
|
docOriginal.LoadHtml(htmlOriginal);
|
612
|
var docToEdit = new HtmlDocument();
|
613
|
docToEdit.LoadHtml(htmlToEdit);
|
614
|
|
615
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
616
|
|
617
|
int i = 0;
|
618
|
List<HtmlNode> addedForSelection = new();
|
619
|
|
620
|
int descendantsCount = descendantsToEdit.Count();
|
621
|
while (i < descendantsCount)
|
622
|
{
|
623
|
for (; i < descendantsCount; i++)
|
624
|
{
|
625
|
var node = descendantsToEdit.ElementAt(i);
|
626
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) ||
|
627
|
node.ParentNode.Name == "style" || node.ParentNode.Name.StartsWith("#"))
|
628
|
{
|
629
|
continue;
|
630
|
}
|
631
|
|
632
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
633
|
|
634
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
635
|
var end = TagClosingPositions[nodeId];
|
636
|
|
637
|
int selectionStart = tagToAdd.Position;
|
638
|
int selectionEnd = tagToAdd.Position + tagToAdd.Length;
|
639
|
|
640
|
if (selectionStart < end && selectionEnd > start)
|
641
|
{
|
642
|
if (selectionStart <= start && selectionEnd >= end)
|
643
|
{
|
644
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
645
|
}
|
646
|
else if (selectionStart <= start)
|
647
|
{
|
648
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
649
|
}
|
650
|
else if (selectionEnd >= end)
|
651
|
{
|
652
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
653
|
}
|
654
|
else
|
655
|
{
|
656
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
657
|
}
|
658
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
659
|
descendantsCount = descendantsToEdit.Count();
|
660
|
break;
|
661
|
}
|
662
|
}
|
663
|
}
|
664
|
|
665
|
GenerateCSS(tags);
|
666
|
return docToEdit.DocumentNode.OuterHtml;
|
667
|
}
|
668
|
|
669
|
|
670
|
private string PartialPreprocessHTMLRemoveTag(string htmlToEdit, string htmlOriginal, AnnotationTag tagToRemove, List<AnnotationTag> tags)
|
671
|
{
|
672
|
var docOriginal = new HtmlDocument();
|
673
|
docOriginal.LoadHtml(htmlOriginal);
|
674
|
var docToEdit = new HtmlDocument();
|
675
|
docToEdit.LoadHtml(htmlToEdit);
|
676
|
|
677
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
678
|
|
679
|
int i = 0;
|
680
|
int descendantsCount = descendantsToEdit.Count();
|
681
|
while (i < descendantsCount)
|
682
|
{
|
683
|
for (; i < descendantsCount; i++)
|
684
|
{
|
685
|
var node = descendantsToEdit.ElementAt(i);
|
686
|
if (!node.Attributes.Contains(TAG_EF_ID_ATTRIBUTE_NAME))
|
687
|
{
|
688
|
continue;
|
689
|
}
|
690
|
else
|
691
|
{
|
692
|
if (node.Attributes[TAG_EF_ID_ATTRIBUTE_NAME].Value != tagToRemove.Id.ToString())
|
693
|
{
|
694
|
continue;
|
695
|
}
|
696
|
|
697
|
node.Attributes.Remove(TAG_EF_ID_ATTRIBUTE_NAME);
|
698
|
node.Attributes.Remove(TAG_INSTANCE_ATTRIBUTE_NAME);
|
699
|
node.Attributes.Remove("class");
|
700
|
node.Attributes.Remove("start");
|
701
|
node.Attributes.Remove("end");
|
702
|
|
703
|
/*var parent = node.ParentNode;
|
704
|
var contents = node.ChildNodes;
|
705
|
int index = parent.ChildNodes.IndexOf(node);
|
706
|
parent.ChildNodes.RemoveAt(index);
|
707
|
|
708
|
List<HtmlNode> newChildren = new();
|
709
|
for (int j = 0; j < index; j++)
|
710
|
{
|
711
|
newChildren.Add(parent.ChildNodes[j]);
|
712
|
}
|
713
|
for (int j = 0; j < contents.Count; j++)
|
714
|
{
|
715
|
newChildren.Add(contents[j]);
|
716
|
}
|
717
|
for (int j = index; j < parent.ChildNodes.Count; j++)
|
718
|
{
|
719
|
newChildren.Add(parent.ChildNodes[j]);
|
720
|
}
|
721
|
|
722
|
parent.ChildNodes.Clear();
|
723
|
foreach (var child in newChildren) { parent.ChildNodes.Add(child); }*/
|
724
|
|
725
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
726
|
descendantsCount = descendantsToEdit.Count();
|
727
|
break;
|
728
|
}
|
729
|
}
|
730
|
}
|
731
|
|
732
|
GenerateCSS(tags);
|
733
|
return docToEdit.DocumentNode.OuterHtml;
|
734
|
}
|
735
|
|
736
|
/*
|
737
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
738
|
*/
|
739
|
|
740
|
|
741
|
// TODO temporary
|
742
|
private bool IsHtml(string text)
|
743
|
{
|
744
|
return text.Contains("<html>");
|
745
|
}
|
746
|
|
747
|
public void AddAnnotationInstance(Guid annotationId, Guid userId, ERole userRole, AnnotationInstanceAddRequest request)
|
748
|
{
|
749
|
var annotation = context.Annotations
|
750
|
.Where(a => a.Id == annotationId)
|
751
|
.Include(a => a.User)
|
752
|
.Include(a => a.Document).ThenInclude(d => d.Content)
|
753
|
.First();
|
754
|
|
755
|
if (userRole < ERole.ADMINISTRATOR)
|
756
|
{
|
757
|
if (annotation.User.Id != userId)
|
758
|
{
|
759
|
throw new UnauthorizedAccessException($"User {userId} does not have assigned annotation {annotationId}");
|
760
|
}
|
761
|
}
|
762
|
|
763
|
if (annotation.State == EState.NEW)
|
764
|
{
|
765
|
annotation.State = EState.IN_PROGRESS;
|
766
|
}
|
767
|
|
768
|
AnnotationTag annotationTag = new()
|
769
|
{
|
770
|
Id = Guid.NewGuid(),
|
771
|
Annotation = annotation,
|
772
|
Instance = request.InstanceId == null ? Guid.NewGuid() : request.InstanceId.Value,
|
773
|
Length = request.Length,
|
774
|
Position = request.Position,
|
775
|
Note = ""
|
776
|
};
|
777
|
|
778
|
if (request.Type == ETagType.TAG)
|
779
|
{
|
780
|
annotationTag.Tag = context.Tags.Where(t => t.Id == request.Id).Single();
|
781
|
annotationTag.SubTag = null;
|
782
|
|
783
|
if (annotationTag.Tag.SentimentEnabled)
|
784
|
{
|
785
|
annotationTag.Sentiment = ETagSentiment.NEUTRAL;
|
786
|
}
|
787
|
|
788
|
// If for the same annotation exists a tag with same position and length and of the same type, ignore
|
789
|
if (context.AnnotationTags.Any(at =>
|
790
|
at.Position == annotationTag.Position &&
|
791
|
at.Length == annotationTag.Length &&
|
792
|
at.Annotation == annotation &&
|
793
|
at.Tag == annotationTag.Tag))
|
794
|
{
|
795
|
throw new InvalidOperationException("Duplicate tag");
|
796
|
}
|
797
|
|
798
|
}
|
799
|
else if (request.Type == ETagType.SUBTAG)
|
800
|
{
|
801
|
var subTag = context.SubTags.Where(st => st.Id == request.Id).Include(st => st.Tag).Single();
|
802
|
annotationTag.SubTag = subTag;
|
803
|
annotationTag.Tag = subTag.Tag;
|
804
|
|
805
|
if (annotationTag.SubTag.SentimentEnabled)
|
806
|
{
|
807
|
annotationTag.Sentiment = ETagSentiment.NEUTRAL;
|
808
|
}
|
809
|
|
810
|
if (context.AnnotationTags.Any(at =>
|
811
|
at.Position == annotationTag.Position &&
|
812
|
at.Length == annotationTag.Length &&
|
813
|
at.Annotation == annotation &&
|
814
|
at.Tag == annotationTag.Tag &&
|
815
|
at.SubTag == annotationTag.SubTag))
|
816
|
{
|
817
|
throw new InvalidOperationException("Duplicate tag");
|
818
|
}
|
819
|
}
|
820
|
else
|
821
|
{
|
822
|
throw new ArgumentException($"Unknown tag type {request.Type}");
|
823
|
}
|
824
|
|
825
|
annotation.LastModifiedTagId = annotationTag.Id;
|
826
|
annotation.ModifiedType = EModified.ADDED;
|
827
|
|
828
|
context.AnnotationTags.Add(annotationTag);
|
829
|
context.SaveChanges();
|
830
|
}
|
831
|
|
832
|
public void DeleteAnnotationInstance(Guid annotationId, Guid tagInstanceId, Guid loggedUserId, ERole userRole)
|
833
|
{
|
834
|
Annotation annotation = null;
|
835
|
try
|
836
|
{
|
837
|
annotation = context.Annotations
|
838
|
.Where(a => a.Id == annotationId)
|
839
|
.Include(a => a.User)
|
840
|
.Include(a => a.Document).ThenInclude(d => d.Content)
|
841
|
.First();
|
842
|
|
843
|
}
|
844
|
catch (Exception ex)
|
845
|
{
|
846
|
throw new InvalidOperationException("Could not find annotation");
|
847
|
}
|
848
|
|
849
|
|
850
|
if (userRole < ERole.ADMINISTRATOR)
|
851
|
{
|
852
|
if (annotation.User.Id != loggedUserId)
|
853
|
{
|
854
|
throw new UnauthorizedAccessException($"User {loggedUserId} does not have assigned annotation {annotationId}");
|
855
|
}
|
856
|
}
|
857
|
|
858
|
if (!context.AnnotationTags.Any(at => at.Id == tagInstanceId))
|
859
|
{
|
860
|
throw new InvalidOperationException("Could not find tag instance");
|
861
|
}
|
862
|
|
863
|
|
864
|
var annotationTag = context.AnnotationTags.First(at => at.Id == tagInstanceId);
|
865
|
annotation.LastModifiedTagId = annotationTag.Id;
|
866
|
annotation.ModifiedType = EModified.REMOVED;
|
867
|
|
868
|
context.AnnotationTags.Remove(annotationTag);
|
869
|
|
870
|
context.SaveChanges();
|
871
|
}
|
872
|
|
873
|
public void SetTagInstanceSentiment(Guid annotationId, Guid instanceId, Guid userId, ERole userRole, ETagSentiment sentiment)
|
874
|
{
|
875
|
Annotation annotation = null;
|
876
|
try
|
877
|
{
|
878
|
annotation = context.Annotations
|
879
|
.Where(a => a.Id == annotationId)
|
880
|
.Include(a => a.User)
|
881
|
.Include(a => a.Document).ThenInclude(d => d.Content)
|
882
|
.First();
|
883
|
|
884
|
}
|
885
|
catch (Exception ex)
|
886
|
{
|
887
|
throw new InvalidOperationException("Could not find annotation");
|
888
|
}
|
889
|
|
890
|
|
891
|
if (userRole < ERole.ADMINISTRATOR)
|
892
|
{
|
893
|
if (annotation.User.Id != userId)
|
894
|
{
|
895
|
throw new UnauthorizedAccessException($"User {userId} does not have assigned annotation {annotationId}");
|
896
|
}
|
897
|
}
|
898
|
|
899
|
|
900
|
var tagInstances = context.AnnotationTags.Where(at => at.Instance == instanceId).ToList();
|
901
|
if (tagInstances.Count() == 0)
|
902
|
{
|
903
|
throw new InvalidOperationException("No such instance found");
|
904
|
}
|
905
|
|
906
|
foreach (var tagInstance in tagInstances)
|
907
|
{
|
908
|
tagInstance.Sentiment = sentiment;
|
909
|
}
|
910
|
|
911
|
context.SaveChanges();
|
912
|
}
|
913
|
|
914
|
public void MarkAnnotationAsDone(Guid annotationId, Guid userId, ERole userRole, bool done)
|
915
|
{
|
916
|
Annotation annotation = null;
|
917
|
try
|
918
|
{
|
919
|
annotation = context.Annotations
|
920
|
.Where(a => a.Id == annotationId)
|
921
|
.Include(a => a.User)
|
922
|
.Include(a => a.Document).ThenInclude(d => d.Content)
|
923
|
.First();
|
924
|
|
925
|
}
|
926
|
catch (Exception ex)
|
927
|
{
|
928
|
throw new InvalidOperationException("Could not find annotation");
|
929
|
}
|
930
|
|
931
|
|
932
|
if (userRole < ERole.ADMINISTRATOR)
|
933
|
{
|
934
|
if (annotation.User.Id != userId)
|
935
|
{
|
936
|
throw new UnauthorizedAccessException($"User {userId} does not have assigned annotation {annotationId}");
|
937
|
}
|
938
|
}
|
939
|
|
940
|
annotation.State = done ? EState.DONE : EState.IN_PROGRESS;
|
941
|
context.SaveChanges();
|
942
|
}
|
943
|
}
|
944
|
}
|