1
|
using Core.Entities;
|
2
|
using Core.GraphUtils;
|
3
|
using Ganss.XSS;
|
4
|
using HtmlAgilityPack;
|
5
|
using Models.Tags;
|
6
|
using System;
|
7
|
using System.Collections.Generic;
|
8
|
using System.Linq;
|
9
|
using System.Text;
|
10
|
using System.Threading.Tasks;
|
11
|
|
12
|
namespace Core.Services
|
13
|
{
|
14
|
public class HTMLService : IHTMLService
|
15
|
{
|
16
|
/// <summary>
|
17
|
/// Attribute name for uniquely identifying each tag in the HTML document on the frontend
|
18
|
/// </summary>
|
19
|
private const string TAG_ID_ATTRIBUTE_NAME = "aswi-tag-id";
|
20
|
|
21
|
/// <summary>
|
22
|
/// Attribute name for tag instance so that each marked out tag can be given a style
|
23
|
/// </summary>
|
24
|
private const string TAG_INSTANCE_ATTRIBUTE_NAME = "aswi-tag-instance";
|
25
|
|
26
|
/// <summary>
|
27
|
/// Attribute name for EF Tag IDs so that the frontend can reference them to the backend for deletion
|
28
|
/// </summary>
|
29
|
private const string TAG_EF_ID_ATTRIBUTE_NAME = "aswi-tag-ef-id";
|
30
|
|
31
|
/// <summary>
|
32
|
/// Info that is cached in tha database
|
33
|
/// </summary>
|
34
|
public class CachedInfo
|
35
|
{
|
36
|
/// <summary>
|
37
|
/// Positions of opening tags in the original HTML document
|
38
|
/// </summary>
|
39
|
public List<int> TagStartPositions = new();
|
40
|
/// <summary>
|
41
|
/// Lengths of opening tags in the original HTML document
|
42
|
/// </summary>
|
43
|
public List<int> TagStartLengths = new();
|
44
|
/// <summary>
|
45
|
/// Positions of closing tags in the original HTML document
|
46
|
/// </summary>
|
47
|
public List<int> TagClosingPositions = new();
|
48
|
/// <summary>
|
49
|
/// Lengths for closing tags in the original HTML document
|
50
|
/// </summary>
|
51
|
public List<int> TagClosingLengths = new();
|
52
|
/// <summary>
|
53
|
/// Unused
|
54
|
/// </summary>
|
55
|
public Dictionary<HtmlNode, HtmlNode> NodeDict = new();
|
56
|
/// <summary>
|
57
|
/// Cached CSS for each tag
|
58
|
/// </summary>
|
59
|
public List<TagInstanceCSSInfo> TagInstanceCSS = new();
|
60
|
}
|
61
|
|
62
|
private List<int> TagStartPositions = new();
|
63
|
private List<int> TagStartLengths = new();
|
64
|
private List<int> TagClosingPositions = new();
|
65
|
private List<int> TagClosingLengths = new();
|
66
|
private Dictionary<HtmlNode, HtmlNode> NodeDict = new();
|
67
|
private List<TagInstanceCSSInfo> TagInstanceCSS = new();
|
68
|
|
69
|
/// <summary>
|
70
|
/// Unpack cached info into local (request-scoped) variables
|
71
|
/// </summary>
|
72
|
/// <param name="cachedInfo"></param>
|
73
|
private void UnpackCachedInfo(CachedInfo cachedInfo)
|
74
|
{
|
75
|
TagStartPositions = cachedInfo.TagStartPositions;
|
76
|
TagStartLengths = cachedInfo.TagStartLengths;
|
77
|
TagClosingPositions = cachedInfo.TagClosingPositions;
|
78
|
TagClosingLengths = cachedInfo.TagClosingLengths;
|
79
|
NodeDict = cachedInfo.NodeDict;
|
80
|
TagInstanceCSS = cachedInfo.TagInstanceCSS;
|
81
|
}
|
82
|
|
83
|
|
84
|
/*
|
85
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
86
|
*/
|
87
|
|
88
|
public (string, CachedInfo) FullPreprocessHTML(string htmlSource, List<AnnotationTagGeneric> tags)
|
89
|
{
|
90
|
var docOriginal = new HtmlDocument();
|
91
|
docOriginal.LoadHtml(htmlSource);
|
92
|
var docToEdit = new HtmlDocument();
|
93
|
docToEdit.LoadHtml(htmlSource);
|
94
|
|
95
|
var descendantsOriginal = docOriginal.DocumentNode.DescendantsAndSelf();
|
96
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
97
|
|
98
|
int currentId = 0;
|
99
|
|
100
|
// Create mapping original-document-node -> to-edit-document-node
|
101
|
FillNodeDict(descendantsOriginal, descendantsToEdit);
|
102
|
|
103
|
// Assign IDs to all original rtags
|
104
|
AssignIdsToOriginalDocument(descendantsOriginal, ref currentId);
|
105
|
|
106
|
// Wrap all text that is not in a tag or where the parent node contains tags as children also in spans
|
107
|
WrapTextInSpan(descendantsOriginal, docToEdit);
|
108
|
|
109
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
110
|
int descCount = descendantsToEdit.Count;
|
111
|
|
112
|
foreach (var tag in tags)
|
113
|
{
|
114
|
int i = 0;
|
115
|
List<HtmlNode> addedForSelection = new();
|
116
|
while (i < descCount)
|
117
|
{
|
118
|
for (; i < descCount; i++)
|
119
|
{
|
120
|
var node = descendantsToEdit.ElementAt(i);
|
121
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) || node.ParentNode.Name == "style")
|
122
|
{
|
123
|
continue;
|
124
|
}
|
125
|
|
126
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
127
|
|
128
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
129
|
var end = TagClosingPositions[nodeId];
|
130
|
|
131
|
int selectionStart = tag.Position;
|
132
|
int selectionEnd = tag.Position + tag.Length;
|
133
|
|
134
|
if (selectionStart < end && selectionEnd > start)
|
135
|
{
|
136
|
if (selectionStart <= start && selectionEnd >= end)
|
137
|
{
|
138
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
139
|
}
|
140
|
else if (selectionStart <= start)
|
141
|
{
|
142
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
143
|
}
|
144
|
else if (selectionEnd >= end)
|
145
|
{
|
146
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
147
|
}
|
148
|
else
|
149
|
{
|
150
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
151
|
}
|
152
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
153
|
descCount = descendantsToEdit.Count;
|
154
|
break;
|
155
|
}
|
156
|
}
|
157
|
}
|
158
|
|
159
|
}
|
160
|
|
161
|
ModifyLinks(descendantsToEdit);
|
162
|
string docToRender = docToEdit.DocumentNode.OuterHtml;
|
163
|
HtmlSanitizer sanitizer = new HtmlSanitizer();
|
164
|
sanitizer.AllowedAttributes.Clear();
|
165
|
sanitizer.AllowedAttributes.Add(TAG_ID_ATTRIBUTE_NAME);
|
166
|
sanitizer.AllowedAttributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME);
|
167
|
sanitizer.AllowedAttributes.Add(TAG_EF_ID_ATTRIBUTE_NAME);
|
168
|
sanitizer.AllowedAttributes.Add("href");
|
169
|
sanitizer.AllowedAttributes.Add("end");
|
170
|
sanitizer.AllowedAttributes.Add("start");
|
171
|
sanitizer.AllowedAttributes.Add("class");
|
172
|
sanitizer.AllowedAttributes.Add("target");
|
173
|
sanitizer.AllowedAttributes.Add("id");
|
174
|
if (sanitizer.AllowedTags.Contains("script"))
|
175
|
{
|
176
|
sanitizer.AllowedTags.Remove("script");
|
177
|
}
|
178
|
if (!sanitizer.AllowedTags.Contains("style"))
|
179
|
{
|
180
|
sanitizer.AllowedTags.Add("style");
|
181
|
}
|
182
|
sanitizer.AllowedTags.Add("a");
|
183
|
docToRender = sanitizer.Sanitize(docToRender);
|
184
|
docToRender = docToRender.Replace(" ", " ");
|
185
|
GenerateCSS(tags);
|
186
|
|
187
|
return (docToRender, new CachedInfo()
|
188
|
{
|
189
|
TagStartPositions = TagStartPositions,
|
190
|
TagClosingPositions = TagClosingPositions,
|
191
|
TagStartLengths = TagStartLengths,
|
192
|
TagClosingLengths = TagClosingLengths,
|
193
|
TagInstanceCSS = TagInstanceCSS,
|
194
|
NodeDict = NodeDict
|
195
|
});
|
196
|
}
|
197
|
|
198
|
/// <summary>
|
199
|
/// Wrap text in span when the target text is the whole text of the target tag
|
200
|
/// </summary>
|
201
|
/// <param name="node">text node to wrap</param>
|
202
|
/// <param name="selectionStart">start of the selection in the edited document</param>
|
203
|
/// <param name="selectionEnd">end of the selection in the edited document</param>
|
204
|
/// <param name="start">start of the node</param>
|
205
|
/// <param name="end">end of the node</param>
|
206
|
/// <param name="docToEdit">document to edit</param>
|
207
|
/// <param name="tag">tag we are adding</param>
|
208
|
/// <returns></returns>
|
209
|
private HtmlNode SolveFullFill(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit, AnnotationTagGeneric tag)
|
210
|
{
|
211
|
// full fill
|
212
|
string textSelected = node.InnerText;
|
213
|
|
214
|
var parentNode = node.ParentNode;
|
215
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
216
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
217
|
|
218
|
EPosition markerPosition = EPosition.MARK_NONE;
|
219
|
if (selectionEnd == end && selectionStart == start)
|
220
|
{
|
221
|
markerPosition = EPosition.MARK_LEFT_RIGHT;
|
222
|
}
|
223
|
else if (selectionEnd == end)
|
224
|
{
|
225
|
markerPosition = EPosition.MARK_RIGHT;
|
226
|
}
|
227
|
else if (selectionStart == start)
|
228
|
{
|
229
|
markerPosition = EPosition.MARK_LEFT;
|
230
|
}
|
231
|
|
232
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, start, markerPosition);
|
233
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
234
|
|
235
|
return spanSelected;
|
236
|
}
|
237
|
|
238
|
/// <summary>
|
239
|
/// Wrap text in span when the target text starts at the start of the tag and leaves space on the right side
|
240
|
/// </summary>
|
241
|
/// <param name="node">text node to wrap</param>
|
242
|
/// <param name="selectionStart">start of the selection in the edited document</param>
|
243
|
/// <param name="selectionEnd">end of the selection in the edited document</param>
|
244
|
/// <param name="start">start of the node</param>
|
245
|
/// <param name="end">end of the node</param>
|
246
|
/// <param name="docToEdit">document to edit</param>
|
247
|
/// <param name="tag">tag we are adding</param>
|
248
|
/// <returns></returns>
|
249
|
private List<HtmlNode> SolveRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
250
|
AnnotationTagGeneric tag)
|
251
|
{
|
252
|
// partial fill, end gap
|
253
|
string text = node.InnerText;
|
254
|
string textAfter = text.Substring(Math.Min(selectionStart - start + tag.Length, text.Length));
|
255
|
string textSelected = text.Substring(0, selectionEnd - start);
|
256
|
|
257
|
var parentNode = node.ParentNode;
|
258
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
259
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
260
|
|
261
|
int spanSelectedStart = start;
|
262
|
int spanAfterStart = start + textSelected.Length;
|
263
|
|
264
|
var position = EPosition.MARK_RIGHT;
|
265
|
if (selectionStart == start)
|
266
|
{
|
267
|
position = EPosition.MARK_LEFT_RIGHT;
|
268
|
}
|
269
|
|
270
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, position);
|
271
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
272
|
|
273
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
274
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanAfter);
|
275
|
|
276
|
return new() { spanSelected, spanAfter };
|
277
|
}
|
278
|
|
279
|
/// <summary>
|
280
|
/// Wrap text in span when the target text ends at the end of the tag but leaves space on the left
|
281
|
/// </summary>
|
282
|
/// <param name="node">text node to wrap</param>
|
283
|
/// <param name="selectionStart">start of the selection in the edited document</param>
|
284
|
/// <param name="selectionEnd">end of the selection in the edited document</param>
|
285
|
/// <param name="start">start of the node</param>
|
286
|
/// <param name="end">end of the node</param>
|
287
|
/// <param name="docToEdit">document to edit</param>
|
288
|
/// <param name="tag">tag we are adding</param>
|
289
|
/// <returns></returns>
|
290
|
private List<HtmlNode> SolveLeftGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
291
|
AnnotationTagGeneric tag)
|
292
|
{
|
293
|
// partial fill, start gap
|
294
|
string text = node.InnerText;
|
295
|
string textBefore = text.Substring(0, selectionStart - start);
|
296
|
string textSelected = text.Substring(selectionStart - start, Math.Min(tag.Length, text.Length - textBefore.Length));
|
297
|
|
298
|
var parentNode = node.ParentNode;
|
299
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
300
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
301
|
|
302
|
int spanBeforeStart = start;
|
303
|
int spanSelectedStart = start + textBefore.Length;
|
304
|
|
305
|
var position = EPosition.MARK_LEFT;
|
306
|
if (selectionEnd == end)
|
307
|
{
|
308
|
position = EPosition.MARK_LEFT_RIGHT;
|
309
|
}
|
310
|
|
311
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
312
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
313
|
|
314
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, position);
|
315
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
316
|
|
317
|
return new() { spanSelected, spanBefore };
|
318
|
}
|
319
|
|
320
|
/// <summary>
|
321
|
/// Wrap text in span when the target text is contained in the node but neither starts or ends match
|
322
|
/// </summary>
|
323
|
/// <param name="node">text node to wrap</param>
|
324
|
/// <param name="selectionStart">start of the selection in the edited document</param>
|
325
|
/// <param name="selectionEnd">end of the selection in the edited document</param>
|
326
|
/// <param name="start">start of the node</param>
|
327
|
/// <param name="end">end of the node</param>
|
328
|
/// <param name="docToEdit">document to edit</param>
|
329
|
/// <param name="tag">tag we are adding</param>
|
330
|
/// <returns></returns>
|
331
|
private List<HtmlNode> SolveLeftRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
332
|
AnnotationTagGeneric tag)
|
333
|
{
|
334
|
// partial fill, start gap end gap
|
335
|
string text = node.InnerText;
|
336
|
string textBefore = text.Substring(0, selectionStart - start);
|
337
|
string textAfter = text.Substring(selectionStart - start + tag.Length);
|
338
|
string textSelected = text.Substring(selectionStart - start, tag.Length);
|
339
|
|
340
|
var parentNode = node.ParentNode;
|
341
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
342
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
343
|
|
344
|
int spanBeforeStart = start;
|
345
|
int spanSelectedStart = start + textBefore.Length;
|
346
|
int spanAfterStart = start + textBefore.Length + textSelected.Length;
|
347
|
|
348
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
349
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
350
|
|
351
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_LEFT_RIGHT);
|
352
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
353
|
|
354
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
355
|
parentNode.ChildNodes.Insert(nodeIndex + 2, spanAfter);
|
356
|
|
357
|
return new() { spanSelected, spanBefore, spanAfter };
|
358
|
}
|
359
|
|
360
|
/// <summary>
|
361
|
/// Create span utility method
|
362
|
/// </summary>
|
363
|
/// <param name="doc">document</param>
|
364
|
/// <param name="text">inner text of the span</param>
|
365
|
/// <param name="tagId">ID of the tag (TAG_ID_ATTRIBUTE_NAME)</param>
|
366
|
/// <param name="instanceId">ID of the instance (TAG_INSTANCE_ATTRIBUTE_NAME)</param>
|
367
|
/// <param name="entityId">Id of the EF entity (TAG_EF_ID_NAME)</param>
|
368
|
/// <param name="startPosition"></param>
|
369
|
/// <param name="position"></param>
|
370
|
/// <returns></returns>
|
371
|
private HtmlNode CreateSpan(HtmlDocument doc, string text, int tagId, Guid? instanceId, Guid? entityId, int startPosition, EPosition position = EPosition.MARK_NONE)
|
372
|
{
|
373
|
HtmlNode span = doc.CreateElement("span");
|
374
|
span.InnerHtml = text;
|
375
|
TagStartPositions.Add(startPosition);
|
376
|
TagStartLengths.Add(0);
|
377
|
TagClosingPositions.Add(startPosition + text.Length);
|
378
|
TagClosingLengths.Add(0);
|
379
|
span.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, tagId.ToString());
|
380
|
|
381
|
if (instanceId != null)
|
382
|
{
|
383
|
span.AddClass("annotation");
|
384
|
span.Attributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME, instanceId.Value.ToString());
|
385
|
span.Attributes.Add(TAG_EF_ID_ATTRIBUTE_NAME, entityId.ToString());
|
386
|
span.Attributes.Add("id", entityId.ToString());
|
387
|
|
388
|
if (position == EPosition.MARK_LEFT || position == EPosition.MARK_LEFT_RIGHT)
|
389
|
{
|
390
|
span.Attributes.Add("start", "1");
|
391
|
}
|
392
|
if (position == EPosition.MARK_RIGHT || position == EPosition.MARK_LEFT_RIGHT)
|
393
|
{
|
394
|
span.Attributes.Add("end", "1");
|
395
|
}
|
396
|
}
|
397
|
|
398
|
return span;
|
399
|
}
|
400
|
|
401
|
private enum EPosition
|
402
|
{
|
403
|
MARK_LEFT = 5,
|
404
|
MARK_RIGHT = 3,
|
405
|
MARK_LEFT_RIGHT = 2,
|
406
|
MARK_NONE = 0
|
407
|
}
|
408
|
|
409
|
/// <summary>
|
410
|
/// Redirect all links
|
411
|
/// </summary>
|
412
|
/// <param name="descendantsOriginal"></param>
|
413
|
private void ModifyLinks(IEnumerable<HtmlNode> descendantsOriginal)
|
414
|
{
|
415
|
foreach (var descendant in descendantsOriginal)
|
416
|
{
|
417
|
if (descendant.Name == "a")
|
418
|
{
|
419
|
if (descendant.Attributes.Contains("href"))
|
420
|
{
|
421
|
descendant.SetAttributeValue("href", "/link?url=" + descendant.Attributes["href"].Value);
|
422
|
descendant.SetAttributeValue("target", "_blank");
|
423
|
}
|
424
|
}
|
425
|
}
|
426
|
}
|
427
|
|
428
|
/// <summary>
|
429
|
/// Wrap all #text nodes in spans
|
430
|
/// </summary>
|
431
|
/// <param name="descendantsOriginal"></param>
|
432
|
/// <param name="docToEdit"></param>
|
433
|
private void WrapTextInSpan(IEnumerable<HtmlNode> descendantsOriginal, HtmlDocument docToEdit)
|
434
|
{
|
435
|
// Special case for non-html documents
|
436
|
if (descendantsOriginal.Count() == 2)
|
437
|
{
|
438
|
var documentNode = descendantsOriginal.ElementAt(0);
|
439
|
var childNode = descendantsOriginal.ElementAt(1);
|
440
|
if (documentNode.Name == "#document" && childNode.Name == "#text")
|
441
|
{
|
442
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
443
|
coveringSpan.InnerHtml = childNode.InnerHtml;
|
444
|
TagStartPositions.Add(childNode.InnerStartIndex);
|
445
|
TagStartLengths.Add(0);
|
446
|
TagClosingPositions.Add(childNode.InnerStartIndex + childNode.InnerLength);
|
447
|
TagClosingLengths.Add(0);
|
448
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
449
|
|
450
|
var parent = NodeDict[documentNode];
|
451
|
|
452
|
parent.ChildNodes.RemoveAt(0);
|
453
|
parent.ChildNodes.Add(coveringSpan);
|
454
|
|
455
|
return;
|
456
|
}
|
457
|
}
|
458
|
|
459
|
foreach (var node in descendantsOriginal)
|
460
|
{
|
461
|
var originalNode = node;
|
462
|
var toEditNode = NodeDict[node];
|
463
|
|
464
|
if (originalNode.Name.Contains("#"))
|
465
|
{
|
466
|
continue;
|
467
|
}
|
468
|
else
|
469
|
{
|
470
|
bool onlyText = true;
|
471
|
bool onlySubtags = true;
|
472
|
|
473
|
foreach (var child in node.ChildNodes)
|
474
|
{
|
475
|
if (child.Name.Contains("#"))
|
476
|
{
|
477
|
onlySubtags = false;
|
478
|
}
|
479
|
else
|
480
|
{
|
481
|
onlyText = false;
|
482
|
}
|
483
|
}
|
484
|
|
485
|
if (onlyText || onlySubtags)
|
486
|
{
|
487
|
continue;
|
488
|
}
|
489
|
else
|
490
|
{
|
491
|
|
492
|
foreach (var child in node.ChildNodes)
|
493
|
{
|
494
|
if (child.Name.Contains("#text"))
|
495
|
{
|
496
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
497
|
coveringSpan.InnerHtml = child.InnerHtml;
|
498
|
|
499
|
// Add length of 0 - in they were not in the original document
|
500
|
TagStartPositions.Add(child.InnerStartIndex);
|
501
|
TagStartLengths.Add(0);
|
502
|
TagClosingPositions.Add(child.InnerStartIndex + child.InnerLength);
|
503
|
TagClosingLengths.Add(0);
|
504
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
505
|
|
506
|
var parent = NodeDict[node];
|
507
|
var index = parent.ChildNodes.IndexOf(NodeDict[child]);
|
508
|
|
509
|
parent.ChildNodes.RemoveAt(index);
|
510
|
parent.ChildNodes.Insert(index, coveringSpan);
|
511
|
}
|
512
|
}
|
513
|
}
|
514
|
}
|
515
|
}
|
516
|
}
|
517
|
|
518
|
/// <summary>
|
519
|
/// Generate CSS for all tag instances
|
520
|
/// </summary>
|
521
|
/// <param name="tags"></param>
|
522
|
private void GenerateCSS(List<AnnotationTagGeneric> tags)
|
523
|
{
|
524
|
/*string inner = "span.annotation {border-bottom: 2px solid;}";
|
525
|
inner += "span {line-height: 30px}\n";*/
|
526
|
|
527
|
var tagPaddingDict = Intersections.ColorGraph(Intersections.FindIntersections(tags));
|
528
|
foreach (var tag in tags.DistinctBy(t => t.Instance))
|
529
|
{
|
530
|
var padding = (tagPaddingDict[tag] + 2) * 2;
|
531
|
TagInstanceCSS.Add(new()
|
532
|
{
|
533
|
InstanceId = tag.Instance,
|
534
|
Color = tag.Tag.Color,
|
535
|
Padding = padding
|
536
|
});
|
537
|
}
|
538
|
}
|
539
|
|
540
|
/// <summary>
|
541
|
/// Add IDs to all tags in the original document
|
542
|
/// </summary>
|
543
|
/// <param name="descendantsOriginal"></param>
|
544
|
/// <param name="currentId"></param>
|
545
|
private void AssignIdsToOriginalDocument(IEnumerable<HtmlNode> descendantsOriginal, ref int currentId)
|
546
|
{
|
547
|
foreach (var node in descendantsOriginal)
|
548
|
{
|
549
|
var originalNode = node;
|
550
|
var toEditNode = NodeDict[node];
|
551
|
|
552
|
if (originalNode.Name.Contains("#"))
|
553
|
{
|
554
|
continue;
|
555
|
}
|
556
|
else
|
557
|
{
|
558
|
TagStartPositions.Add(originalNode.OuterStartIndex);
|
559
|
TagStartLengths.Add(originalNode.InnerStartIndex - originalNode.OuterStartIndex);
|
560
|
currentId = TagStartPositions.Count - 1;
|
561
|
toEditNode.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, currentId.ToString());
|
562
|
|
563
|
// Cross-referencing of the original document to the edited document
|
564
|
TagClosingPositions.Add(originalNode.InnerStartIndex + originalNode.InnerLength);
|
565
|
TagClosingLengths.Add((originalNode.OuterStartIndex + originalNode.OuterLength) - (originalNode.InnerStartIndex + originalNode.InnerLength));
|
566
|
}
|
567
|
}
|
568
|
}
|
569
|
|
570
|
private void FillNodeDict(IEnumerable<HtmlNode> descendantsOriginal, IEnumerable<HtmlNode> descendantsToEdit)
|
571
|
{
|
572
|
var zipped = descendantsOriginal.Zip(descendantsToEdit, (orig, toEdit) => new
|
573
|
{
|
574
|
Original = orig,
|
575
|
ToEdit = toEdit
|
576
|
});
|
577
|
foreach (var node in zipped)
|
578
|
{
|
579
|
var originalNode = node.Original;
|
580
|
var toEditNode = node.ToEdit;
|
581
|
NodeDict.Add(originalNode, toEditNode);
|
582
|
}
|
583
|
}
|
584
|
|
585
|
|
586
|
/*
|
587
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
588
|
*/
|
589
|
|
590
|
/*
|
591
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
592
|
*/
|
593
|
|
594
|
/// <summary>
|
595
|
/// Incrementally modifies the html to render and cached info by adding a new tag
|
596
|
/// </summary>
|
597
|
/// <param name="htmlToEdit"></param>
|
598
|
/// <param name="htmlOriginal"></param>
|
599
|
/// <param name="tagToAdd"></param>
|
600
|
/// <param name="tags"></param>
|
601
|
/// <param name="cachedInfo"></param>
|
602
|
/// <returns></returns>
|
603
|
public (string, CachedInfo) PartialPreprocessHTMLAddTag(string htmlToEdit, string htmlOriginal, AnnotationTagGeneric tagToAdd, List<AnnotationTagGeneric> tags, CachedInfo cachedInfo)
|
604
|
{
|
605
|
UnpackCachedInfo(cachedInfo);
|
606
|
|
607
|
var docOriginal = new HtmlDocument();
|
608
|
docOriginal.LoadHtml(htmlOriginal);
|
609
|
var docToEdit = new HtmlDocument();
|
610
|
docToEdit.LoadHtml(htmlToEdit);
|
611
|
|
612
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
613
|
|
614
|
int i = 0;
|
615
|
List<HtmlNode> addedForSelection = new();
|
616
|
|
617
|
int descendantsCount = descendantsToEdit.Count();
|
618
|
while (i < descendantsCount)
|
619
|
{
|
620
|
for (; i < descendantsCount; i++)
|
621
|
{
|
622
|
var node = descendantsToEdit.ElementAt(i);
|
623
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) ||
|
624
|
node.ParentNode.Name == "style" || node.ParentNode.Name.StartsWith("#"))
|
625
|
{
|
626
|
continue;
|
627
|
}
|
628
|
|
629
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
630
|
|
631
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
632
|
var end = TagClosingPositions[nodeId];
|
633
|
|
634
|
int selectionStart = tagToAdd.Position;
|
635
|
int selectionEnd = tagToAdd.Position + tagToAdd.Length;
|
636
|
|
637
|
if (selectionStart < end && selectionEnd > start)
|
638
|
{
|
639
|
if (selectionStart <= start && selectionEnd >= end)
|
640
|
{
|
641
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
642
|
}
|
643
|
else if (selectionStart <= start)
|
644
|
{
|
645
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
646
|
}
|
647
|
else if (selectionEnd >= end)
|
648
|
{
|
649
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
650
|
}
|
651
|
else
|
652
|
{
|
653
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
654
|
}
|
655
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
656
|
descendantsCount = descendantsToEdit.Count();
|
657
|
break;
|
658
|
}
|
659
|
}
|
660
|
}
|
661
|
|
662
|
GenerateCSS(tags);
|
663
|
return (docToEdit.DocumentNode.OuterHtml, new()
|
664
|
{
|
665
|
TagStartPositions = TagStartPositions,
|
666
|
TagStartLengths = TagStartLengths,
|
667
|
TagClosingPositions = TagClosingPositions,
|
668
|
TagClosingLengths = TagClosingLengths,
|
669
|
TagInstanceCSS = TagInstanceCSS,
|
670
|
NodeDict = NodeDict
|
671
|
});
|
672
|
}
|
673
|
|
674
|
/// <summary>
|
675
|
/// Removing tag effectively just cleans up the attributes and CSS thus has no effect
|
676
|
/// </summary>
|
677
|
/// <param name="htmlToEdit"></param>
|
678
|
/// <param name="htmlOriginal"></param>
|
679
|
/// <param name="tagToRemove"></param>
|
680
|
/// <param name="tags"></param>
|
681
|
/// <param name="cachedInfo"></param>
|
682
|
/// <returns></returns>
|
683
|
public (string, CachedInfo) PartialPreprocessHTMLRemoveTag(string htmlToEdit, string htmlOriginal, AnnotationTagGeneric tagToRemove, List<AnnotationTagGeneric> tags, CachedInfo cachedInfo)
|
684
|
{
|
685
|
UnpackCachedInfo(cachedInfo);
|
686
|
|
687
|
var docOriginal = new HtmlDocument();
|
688
|
docOriginal.LoadHtml(htmlOriginal);
|
689
|
var docToEdit = new HtmlDocument();
|
690
|
docToEdit.LoadHtml(htmlToEdit);
|
691
|
|
692
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
693
|
|
694
|
int i = 0;
|
695
|
int descendantsCount = descendantsToEdit.Count();
|
696
|
while (i < descendantsCount)
|
697
|
{
|
698
|
for (; i < descendantsCount; i++)
|
699
|
{
|
700
|
var node = descendantsToEdit.ElementAt(i);
|
701
|
if (!node.Attributes.Contains(TAG_EF_ID_ATTRIBUTE_NAME))
|
702
|
{
|
703
|
continue;
|
704
|
}
|
705
|
else
|
706
|
{
|
707
|
if (node.Attributes[TAG_EF_ID_ATTRIBUTE_NAME].Value != tagToRemove.Id.ToString())
|
708
|
{
|
709
|
continue;
|
710
|
}
|
711
|
|
712
|
node.Attributes.Remove(TAG_EF_ID_ATTRIBUTE_NAME);
|
713
|
node.Attributes.Remove(TAG_INSTANCE_ATTRIBUTE_NAME);
|
714
|
node.Attributes.Remove("class");
|
715
|
node.Attributes.Remove("start");
|
716
|
node.Attributes.Remove("end");
|
717
|
|
718
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
719
|
descendantsCount = descendantsToEdit.Count();
|
720
|
break;
|
721
|
}
|
722
|
}
|
723
|
}
|
724
|
|
725
|
GenerateCSS(tags);
|
726
|
return (docToEdit.DocumentNode.OuterHtml, new()
|
727
|
{
|
728
|
TagStartPositions = TagStartPositions,
|
729
|
TagStartLengths = TagStartLengths,
|
730
|
TagClosingPositions = TagClosingPositions,
|
731
|
TagClosingLengths = TagClosingLengths,
|
732
|
TagInstanceCSS = TagInstanceCSS,
|
733
|
NodeDict = NodeDict
|
734
|
});
|
735
|
}
|
736
|
|
737
|
/*
|
738
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
739
|
*/
|
740
|
}
|
741
|
}
|