1 |
2c9afc72
|
Vojtěch Bartička
|
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 |
|
|
|
17 |
|
|
private const string TAG_ID_ATTRIBUTE_NAME = "aswi-tag-id";
|
18 |
|
|
private const string TAG_INSTANCE_ATTRIBUTE_NAME = "aswi-tag-instance";
|
19 |
|
|
private const string TAG_EF_ID_ATTRIBUTE_NAME = "aswi-tag-ef-id";
|
20 |
|
|
|
21 |
|
|
public class CachedInfo
|
22 |
|
|
{
|
23 |
|
|
public List<int> TagStartPositions = new();
|
24 |
|
|
public List<int> TagStartLengths = new();
|
25 |
|
|
public List<int> TagClosingPositions = new();
|
26 |
|
|
public List<int> TagClosingLengths = new();
|
27 |
|
|
public Dictionary<HtmlNode, HtmlNode> NodeDict = new();
|
28 |
|
|
public List<TagInstanceCSSInfo> TagInstanceCSS = new();
|
29 |
|
|
}
|
30 |
|
|
|
31 |
|
|
private List<int> TagStartPositions = new();
|
32 |
|
|
private List<int> TagStartLengths = new();
|
33 |
|
|
private List<int> TagClosingPositions = new();
|
34 |
|
|
private List<int> TagClosingLengths = new();
|
35 |
|
|
private Dictionary<HtmlNode, HtmlNode> NodeDict = new();
|
36 |
|
|
private List<TagInstanceCSSInfo> TagInstanceCSS = new();
|
37 |
|
|
|
38 |
|
|
private void UnpackCachedInfo(CachedInfo cachedInfo)
|
39 |
|
|
{
|
40 |
|
|
TagStartPositions = cachedInfo.TagStartPositions;
|
41 |
|
|
TagStartLengths = cachedInfo.TagStartLengths;
|
42 |
|
|
TagClosingPositions = cachedInfo.TagClosingPositions;
|
43 |
|
|
TagClosingLengths = cachedInfo.TagClosingLengths;
|
44 |
|
|
NodeDict = cachedInfo.NodeDict;
|
45 |
|
|
TagInstanceCSS = cachedInfo.TagInstanceCSS;
|
46 |
|
|
}
|
47 |
|
|
|
48 |
|
|
|
49 |
|
|
/*
|
50 |
|
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
51 |
|
|
*/
|
52 |
|
|
|
53 |
|
|
public (string, CachedInfo) FullPreprocessHTML(string htmlSource, List<AnnotationTagGeneric> tags)
|
54 |
|
|
{
|
55 |
|
|
var docOriginal = new HtmlDocument();
|
56 |
|
|
docOriginal.LoadHtml(htmlSource);
|
57 |
|
|
var docToEdit = new HtmlDocument();
|
58 |
|
|
docToEdit.LoadHtml(htmlSource);
|
59 |
|
|
|
60 |
|
|
var descendantsOriginal = docOriginal.DocumentNode.DescendantsAndSelf();
|
61 |
|
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
62 |
|
|
|
63 |
|
|
int currentId = 0;
|
64 |
|
|
|
65 |
|
|
FillNodeDict(descendantsOriginal, descendantsToEdit);
|
66 |
|
|
AssignIdsToOriginalDocument(descendantsOriginal, ref currentId);
|
67 |
|
|
|
68 |
|
|
WrapTextInSpan(descendantsOriginal, docToEdit);
|
69 |
ceadebfe
|
Vojtěch Bartička
|
|
70 |
|
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
71 |
2c9afc72
|
Vojtěch Bartička
|
int descCount = descendantsToEdit.Count;
|
72 |
|
|
|
73 |
|
|
foreach (var tag in tags)
|
74 |
|
|
{
|
75 |
|
|
int i = 0;
|
76 |
|
|
List<HtmlNode> addedForSelection = new();
|
77 |
|
|
while (i < descCount)
|
78 |
|
|
{
|
79 |
|
|
for (; i < descCount; i++)
|
80 |
|
|
{
|
81 |
|
|
var node = descendantsToEdit.ElementAt(i);
|
82 |
|
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) || node.ParentNode.Name == "style")
|
83 |
|
|
{
|
84 |
|
|
continue;
|
85 |
|
|
}
|
86 |
|
|
|
87 |
|
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
88 |
|
|
|
89 |
|
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
90 |
|
|
var end = TagClosingPositions[nodeId];
|
91 |
|
|
|
92 |
|
|
int selectionStart = tag.Position;
|
93 |
|
|
int selectionEnd = tag.Position + tag.Length;
|
94 |
|
|
|
95 |
|
|
if (selectionStart < end && selectionEnd > start)
|
96 |
|
|
{
|
97 |
|
|
if (selectionStart <= start && selectionEnd >= end)
|
98 |
|
|
{
|
99 |
|
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
100 |
|
|
}
|
101 |
|
|
else if (selectionStart <= start)
|
102 |
|
|
{
|
103 |
|
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
104 |
|
|
}
|
105 |
|
|
else if (selectionEnd >= end)
|
106 |
|
|
{
|
107 |
|
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
108 |
|
|
}
|
109 |
|
|
else
|
110 |
|
|
{
|
111 |
|
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tag));
|
112 |
|
|
}
|
113 |
|
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
114 |
|
|
descCount = descendantsToEdit.Count;
|
115 |
|
|
break;
|
116 |
|
|
}
|
117 |
|
|
}
|
118 |
|
|
}
|
119 |
|
|
|
120 |
|
|
}
|
121 |
|
|
|
122 |
|
|
ModifyLinks(descendantsToEdit);
|
123 |
|
|
string docToRender = docToEdit.DocumentNode.OuterHtml;
|
124 |
|
|
HtmlSanitizer sanitizer = new HtmlSanitizer();
|
125 |
|
|
sanitizer.AllowedAttributes.Clear();
|
126 |
|
|
sanitizer.AllowedAttributes.Add(TAG_ID_ATTRIBUTE_NAME);
|
127 |
|
|
sanitizer.AllowedAttributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME);
|
128 |
aa847b20
|
Vojtěch Bartička
|
sanitizer.AllowedAttributes.Add(TAG_EF_ID_ATTRIBUTE_NAME);
|
129 |
2c9afc72
|
Vojtěch Bartička
|
sanitizer.AllowedAttributes.Add("href");
|
130 |
|
|
sanitizer.AllowedAttributes.Add("end");
|
131 |
|
|
sanitizer.AllowedAttributes.Add("start");
|
132 |
|
|
sanitizer.AllowedAttributes.Add("class");
|
133 |
|
|
sanitizer.AllowedAttributes.Add("target");
|
134 |
|
|
if (sanitizer.AllowedTags.Contains("script"))
|
135 |
|
|
{
|
136 |
|
|
sanitizer.AllowedTags.Remove("script");
|
137 |
|
|
}
|
138 |
|
|
if (!sanitizer.AllowedTags.Contains("style"))
|
139 |
|
|
{
|
140 |
|
|
sanitizer.AllowedTags.Add("style");
|
141 |
|
|
}
|
142 |
|
|
sanitizer.AllowedTags.Add("a");
|
143 |
|
|
docToRender = sanitizer.Sanitize(docToRender);
|
144 |
|
|
GenerateCSS(tags);
|
145 |
|
|
|
146 |
|
|
return (docToRender, new CachedInfo()
|
147 |
|
|
{
|
148 |
|
|
TagStartPositions = TagStartPositions,
|
149 |
|
|
TagClosingPositions = TagClosingPositions,
|
150 |
|
|
TagStartLengths = TagStartLengths,
|
151 |
|
|
TagClosingLengths = TagClosingLengths,
|
152 |
|
|
TagInstanceCSS = TagInstanceCSS,
|
153 |
|
|
NodeDict = NodeDict
|
154 |
|
|
});
|
155 |
|
|
}
|
156 |
|
|
|
157 |
|
|
private HtmlNode SolveFullFill(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit, AnnotationTagGeneric tag)
|
158 |
|
|
{
|
159 |
|
|
// full fill
|
160 |
|
|
string textSelected = node.InnerText;
|
161 |
|
|
|
162 |
|
|
var parentNode = node.ParentNode;
|
163 |
|
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
164 |
|
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
165 |
|
|
|
166 |
|
|
EPosition markerPosition = EPosition.MARK_NONE;
|
167 |
|
|
if (selectionEnd == end && selectionStart == start)
|
168 |
|
|
{
|
169 |
|
|
markerPosition = EPosition.MARK_LEFT_RIGHT;
|
170 |
|
|
}
|
171 |
|
|
|
172 |
|
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, start, markerPosition);
|
173 |
|
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
174 |
|
|
|
175 |
|
|
return spanSelected;
|
176 |
|
|
}
|
177 |
|
|
|
178 |
|
|
private List<HtmlNode> SolveRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
179 |
|
|
AnnotationTagGeneric tag)
|
180 |
|
|
{
|
181 |
|
|
// partial fill, end gap
|
182 |
|
|
string text = node.InnerText;
|
183 |
|
|
string textAfter = text.Substring(Math.Min(selectionStart - start + tag.Length, text.Length));
|
184 |
|
|
string textSelected = text.Substring(0, selectionEnd - start);
|
185 |
|
|
|
186 |
|
|
var parentNode = node.ParentNode;
|
187 |
|
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
188 |
|
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
189 |
|
|
|
190 |
|
|
int spanSelectedStart = start;
|
191 |
|
|
int spanAfterStart = start + textSelected.Length;
|
192 |
|
|
|
193 |
|
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_RIGHT);
|
194 |
|
|
parentNode.ChildNodes.Insert(nodeIndex, spanSelected);
|
195 |
|
|
|
196 |
|
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
197 |
|
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanAfter);
|
198 |
|
|
|
199 |
|
|
return new() { spanSelected, spanAfter };
|
200 |
|
|
}
|
201 |
|
|
|
202 |
|
|
private List<HtmlNode> SolveLeftGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
203 |
|
|
AnnotationTagGeneric tag)
|
204 |
|
|
{
|
205 |
|
|
// partial fill, start gap
|
206 |
|
|
string text = node.InnerText;
|
207 |
|
|
string textBefore = text.Substring(0, selectionStart - start);
|
208 |
|
|
string textSelected = text.Substring(selectionStart - start, Math.Min(tag.Length, text.Length - textBefore.Length));
|
209 |
|
|
|
210 |
|
|
var parentNode = node.ParentNode;
|
211 |
|
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
212 |
|
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
213 |
|
|
|
214 |
|
|
int spanBeforeStart = start;
|
215 |
|
|
int spanSelectedStart = start + textBefore.Length;
|
216 |
|
|
|
217 |
|
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
218 |
|
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
219 |
|
|
|
220 |
|
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_LEFT);
|
221 |
|
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
222 |
|
|
|
223 |
|
|
return new() { spanSelected, spanBefore };
|
224 |
|
|
}
|
225 |
|
|
|
226 |
|
|
private List<HtmlNode> SolveLeftRightGap(HtmlNode node, int selectionStart, int selectionEnd, int start, int end, HtmlDocument docToEdit,
|
227 |
|
|
AnnotationTagGeneric tag)
|
228 |
|
|
{
|
229 |
|
|
// partial fill, start gap end gap
|
230 |
|
|
string text = node.InnerText;
|
231 |
|
|
string textBefore = text.Substring(0, selectionStart - start);
|
232 |
|
|
string textAfter = text.Substring(selectionStart - start + tag.Length);
|
233 |
|
|
string textSelected = text.Substring(selectionStart - start, tag.Length);
|
234 |
|
|
|
235 |
|
|
var parentNode = node.ParentNode;
|
236 |
|
|
int nodeIndex = parentNode.ChildNodes.IndexOf(node);
|
237 |
|
|
parentNode.ChildNodes.RemoveAt(nodeIndex);
|
238 |
|
|
|
239 |
|
|
int spanBeforeStart = start;
|
240 |
|
|
int spanSelectedStart = start + textBefore.Length;
|
241 |
|
|
int spanAfterStart = start + textBefore.Length + textSelected.Length;
|
242 |
|
|
|
243 |
|
|
HtmlNode spanBefore = CreateSpan(docToEdit, textBefore, TagStartPositions.Count, null, null, spanBeforeStart);
|
244 |
|
|
parentNode.ChildNodes.Insert(nodeIndex, spanBefore);
|
245 |
|
|
|
246 |
|
|
HtmlNode spanSelected = CreateSpan(docToEdit, textSelected, TagStartPositions.Count, tag.Instance, tag.Id, spanSelectedStart, EPosition.MARK_LEFT_RIGHT);
|
247 |
|
|
parentNode.ChildNodes.Insert(nodeIndex + 1, spanSelected);
|
248 |
|
|
|
249 |
|
|
HtmlNode spanAfter = CreateSpan(docToEdit, textAfter, TagStartPositions.Count, null, null, spanAfterStart);
|
250 |
|
|
parentNode.ChildNodes.Insert(nodeIndex + 2, spanAfter);
|
251 |
|
|
|
252 |
|
|
return new() { spanSelected, spanBefore, spanAfter };
|
253 |
|
|
}
|
254 |
|
|
|
255 |
|
|
private HtmlNode CreateSpan(HtmlDocument doc, string text, int tagId, Guid? instanceId, Guid? entityId, int startPosition, EPosition position = EPosition.MARK_NONE)
|
256 |
|
|
{
|
257 |
|
|
HtmlNode span = doc.CreateElement("span");
|
258 |
|
|
span.InnerHtml = text;
|
259 |
|
|
TagStartPositions.Add(startPosition);
|
260 |
|
|
TagStartLengths.Add(0);
|
261 |
|
|
TagClosingPositions.Add(startPosition + text.Length);
|
262 |
|
|
TagClosingLengths.Add(0);
|
263 |
|
|
span.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, tagId.ToString());
|
264 |
|
|
|
265 |
|
|
if (instanceId != null)
|
266 |
|
|
{
|
267 |
|
|
span.AddClass("annotation");
|
268 |
|
|
span.Attributes.Add(TAG_INSTANCE_ATTRIBUTE_NAME, instanceId.Value.ToString());
|
269 |
|
|
span.Attributes.Add(TAG_EF_ID_ATTRIBUTE_NAME, entityId.ToString());
|
270 |
|
|
|
271 |
|
|
if (position == EPosition.MARK_LEFT || position == EPosition.MARK_LEFT_RIGHT)
|
272 |
|
|
{
|
273 |
|
|
span.Attributes.Add("start", "1");
|
274 |
|
|
}
|
275 |
|
|
if (position == EPosition.MARK_RIGHT || position == EPosition.MARK_LEFT_RIGHT)
|
276 |
|
|
{
|
277 |
|
|
span.Attributes.Add("end", "1");
|
278 |
|
|
}
|
279 |
|
|
}
|
280 |
|
|
|
281 |
|
|
return span;
|
282 |
|
|
}
|
283 |
|
|
|
284 |
|
|
private enum EPosition
|
285 |
|
|
{
|
286 |
|
|
MARK_LEFT = 5,
|
287 |
|
|
MARK_RIGHT = 3,
|
288 |
|
|
MARK_LEFT_RIGHT = 2,
|
289 |
|
|
MARK_NONE = 0
|
290 |
|
|
}
|
291 |
|
|
|
292 |
|
|
private void ModifyLinks(IEnumerable<HtmlNode> descendantsOriginal)
|
293 |
|
|
{
|
294 |
|
|
foreach (var descendant in descendantsOriginal)
|
295 |
|
|
{
|
296 |
|
|
if (descendant.Name == "a")
|
297 |
|
|
{
|
298 |
|
|
if (descendant.Attributes.Contains("href"))
|
299 |
|
|
{
|
300 |
|
|
descendant.SetAttributeValue("href", "/link?url=" + descendant.Attributes["href"].Value);
|
301 |
|
|
descendant.SetAttributeValue("target", "_blank");
|
302 |
|
|
}
|
303 |
|
|
}
|
304 |
|
|
}
|
305 |
|
|
}
|
306 |
|
|
|
307 |
|
|
private void WrapTextInSpan(IEnumerable<HtmlNode> descendantsOriginal, HtmlDocument docToEdit)
|
308 |
|
|
{
|
309 |
|
|
// Special case for non-html documents
|
310 |
|
|
if (descendantsOriginal.Count() == 2)
|
311 |
|
|
{
|
312 |
|
|
var documentNode = descendantsOriginal.ElementAt(0);
|
313 |
|
|
var childNode = descendantsOriginal.ElementAt(1);
|
314 |
|
|
if (documentNode.Name == "#document" && childNode.Name == "#text")
|
315 |
|
|
{
|
316 |
|
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
317 |
|
|
coveringSpan.InnerHtml = childNode.InnerHtml;
|
318 |
|
|
TagStartPositions.Add(childNode.InnerStartIndex);
|
319 |
|
|
TagStartLengths.Add(0);
|
320 |
|
|
TagClosingPositions.Add(childNode.InnerStartIndex + childNode.InnerLength);
|
321 |
|
|
TagClosingLengths.Add(0);
|
322 |
|
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
323 |
|
|
|
324 |
|
|
var parent = NodeDict[documentNode];
|
325 |
|
|
|
326 |
|
|
parent.ChildNodes.RemoveAt(0);
|
327 |
|
|
parent.ChildNodes.Add(coveringSpan);
|
328 |
|
|
|
329 |
|
|
return;
|
330 |
|
|
}
|
331 |
|
|
}
|
332 |
|
|
|
333 |
|
|
foreach (var node in descendantsOriginal)
|
334 |
|
|
{
|
335 |
|
|
var originalNode = node;
|
336 |
|
|
var toEditNode = NodeDict[node];
|
337 |
|
|
|
338 |
|
|
if (originalNode.Name.Contains("#"))
|
339 |
|
|
{
|
340 |
|
|
continue;
|
341 |
|
|
}
|
342 |
|
|
else
|
343 |
|
|
{
|
344 |
|
|
bool onlyText = true;
|
345 |
|
|
bool onlySubtags = true;
|
346 |
|
|
|
347 |
|
|
foreach (var child in node.ChildNodes)
|
348 |
|
|
{
|
349 |
|
|
if (child.Name.Contains("#"))
|
350 |
|
|
{
|
351 |
|
|
onlySubtags = false;
|
352 |
|
|
}
|
353 |
|
|
else
|
354 |
|
|
{
|
355 |
|
|
onlyText = false;
|
356 |
|
|
}
|
357 |
|
|
}
|
358 |
|
|
|
359 |
|
|
if (onlyText || onlySubtags)
|
360 |
|
|
{
|
361 |
|
|
continue;
|
362 |
|
|
}
|
363 |
|
|
else
|
364 |
|
|
{
|
365 |
|
|
|
366 |
|
|
foreach (var child in node.ChildNodes)
|
367 |
|
|
{
|
368 |
|
|
if (child.Name.Contains("#text"))
|
369 |
|
|
{
|
370 |
|
|
HtmlNode coveringSpan = docToEdit.CreateElement("span");
|
371 |
|
|
coveringSpan.InnerHtml = child.InnerHtml;
|
372 |
|
|
TagStartPositions.Add(child.InnerStartIndex);
|
373 |
|
|
TagStartLengths.Add(0);
|
374 |
|
|
TagClosingPositions.Add(child.InnerStartIndex + child.InnerLength);
|
375 |
|
|
TagClosingLengths.Add(0);
|
376 |
|
|
coveringSpan.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, (TagStartPositions.Count - 1).ToString());
|
377 |
|
|
|
378 |
|
|
var parent = NodeDict[node];
|
379 |
|
|
var index = parent.ChildNodes.IndexOf(NodeDict[child]);
|
380 |
|
|
|
381 |
|
|
parent.ChildNodes.RemoveAt(index);
|
382 |
|
|
parent.ChildNodes.Insert(index, coveringSpan);
|
383 |
|
|
}
|
384 |
|
|
}
|
385 |
|
|
}
|
386 |
|
|
}
|
387 |
|
|
}
|
388 |
|
|
}
|
389 |
|
|
|
390 |
|
|
private void GenerateCSS(List<AnnotationTagGeneric> tags)
|
391 |
|
|
{
|
392 |
|
|
/*string inner = "span.annotation {border-bottom: 2px solid;}";
|
393 |
|
|
inner += "span {line-height: 30px}\n";*/
|
394 |
|
|
|
395 |
|
|
var tagPaddingDict = Intersections.ColorGraph(Intersections.FindIntersections(tags));
|
396 |
|
|
foreach (var tag in tags.DistinctBy(t => t.Instance))
|
397 |
|
|
{
|
398 |
|
|
var padding = (tagPaddingDict[tag] + 1) * 2;
|
399 |
|
|
TagInstanceCSS.Add(new()
|
400 |
|
|
{
|
401 |
|
|
InstanceId = tag.Instance,
|
402 |
|
|
Color = tag.Tag.Color,
|
403 |
|
|
Padding = padding
|
404 |
|
|
});
|
405 |
|
|
}
|
406 |
|
|
}
|
407 |
|
|
|
408 |
|
|
private void AssignIdsToOriginalDocument(IEnumerable<HtmlNode> descendantsOriginal, ref int currentId)
|
409 |
|
|
{
|
410 |
|
|
foreach (var node in descendantsOriginal)
|
411 |
|
|
{
|
412 |
|
|
var originalNode = node;
|
413 |
|
|
var toEditNode = NodeDict[node];
|
414 |
|
|
|
415 |
|
|
if (originalNode.Name.Contains("#"))
|
416 |
|
|
{
|
417 |
|
|
continue;
|
418 |
|
|
}
|
419 |
|
|
else
|
420 |
|
|
{
|
421 |
|
|
TagStartPositions.Add(originalNode.OuterStartIndex);
|
422 |
|
|
TagStartLengths.Add(originalNode.InnerStartIndex - originalNode.OuterStartIndex);
|
423 |
|
|
currentId = TagStartPositions.Count - 1;
|
424 |
|
|
toEditNode.Attributes.Add(TAG_ID_ATTRIBUTE_NAME, currentId.ToString());
|
425 |
|
|
|
426 |
|
|
TagClosingPositions.Add(originalNode.InnerStartIndex + originalNode.InnerLength);
|
427 |
|
|
TagClosingLengths.Add((originalNode.OuterStartIndex + originalNode.OuterLength) - (originalNode.InnerStartIndex + originalNode.InnerLength));
|
428 |
|
|
}
|
429 |
|
|
}
|
430 |
|
|
}
|
431 |
|
|
|
432 |
|
|
private void FillNodeDict(IEnumerable<HtmlNode> descendantsOriginal, IEnumerable<HtmlNode> descendantsToEdit)
|
433 |
|
|
{
|
434 |
|
|
var zipped = descendantsOriginal.Zip(descendantsToEdit, (orig, toEdit) => new
|
435 |
|
|
{
|
436 |
|
|
Original = orig,
|
437 |
|
|
ToEdit = toEdit
|
438 |
|
|
});
|
439 |
|
|
foreach (var node in zipped)
|
440 |
|
|
{
|
441 |
|
|
var originalNode = node.Original;
|
442 |
|
|
var toEditNode = node.ToEdit;
|
443 |
|
|
NodeDict.Add(originalNode, toEditNode);
|
444 |
|
|
}
|
445 |
|
|
}
|
446 |
|
|
|
447 |
|
|
|
448 |
|
|
/*
|
449 |
|
|
* Full HTML Preprocessing -------------------------------------------------------------------------------
|
450 |
|
|
*/
|
451 |
|
|
|
452 |
|
|
/*
|
453 |
|
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
454 |
|
|
*/
|
455 |
|
|
|
456 |
f260dac8
|
Vojtěch Bartička
|
public (string, CachedInfo) PartialPreprocessHTMLAddTag(string htmlToEdit, string htmlOriginal, AnnotationTagGeneric tagToAdd, List<AnnotationTagGeneric> tags, CachedInfo cachedInfo)
|
457 |
2c9afc72
|
Vojtěch Bartička
|
{
|
458 |
|
|
UnpackCachedInfo(cachedInfo);
|
459 |
|
|
|
460 |
|
|
var docOriginal = new HtmlDocument();
|
461 |
|
|
docOriginal.LoadHtml(htmlOriginal);
|
462 |
|
|
var docToEdit = new HtmlDocument();
|
463 |
|
|
docToEdit.LoadHtml(htmlToEdit);
|
464 |
|
|
|
465 |
|
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
466 |
|
|
|
467 |
|
|
int i = 0;
|
468 |
|
|
List<HtmlNode> addedForSelection = new();
|
469 |
|
|
|
470 |
|
|
int descendantsCount = descendantsToEdit.Count();
|
471 |
|
|
while (i < descendantsCount)
|
472 |
|
|
{
|
473 |
|
|
for (; i < descendantsCount; i++)
|
474 |
|
|
{
|
475 |
|
|
var node = descendantsToEdit.ElementAt(i);
|
476 |
|
|
if (!node.Name.Contains("#text") || addedForSelection.Contains(node) || addedForSelection.Contains(node.ParentNode) ||
|
477 |
|
|
node.ParentNode.Name == "style" || node.ParentNode.Name.StartsWith("#"))
|
478 |
|
|
{
|
479 |
|
|
continue;
|
480 |
|
|
}
|
481 |
|
|
|
482 |
|
|
int nodeId = node.ParentNode.GetAttributeValue(TAG_ID_ATTRIBUTE_NAME, -1);
|
483 |
|
|
|
484 |
|
|
var start = TagStartPositions[nodeId] + TagStartLengths[nodeId];
|
485 |
|
|
var end = TagClosingPositions[nodeId];
|
486 |
|
|
|
487 |
|
|
int selectionStart = tagToAdd.Position;
|
488 |
|
|
int selectionEnd = tagToAdd.Position + tagToAdd.Length;
|
489 |
|
|
|
490 |
|
|
if (selectionStart < end && selectionEnd > start)
|
491 |
|
|
{
|
492 |
|
|
if (selectionStart <= start && selectionEnd >= end)
|
493 |
|
|
{
|
494 |
|
|
addedForSelection.Add(SolveFullFill(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
495 |
|
|
}
|
496 |
|
|
else if (selectionStart <= start)
|
497 |
|
|
{
|
498 |
|
|
addedForSelection.AddRange(SolveRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
499 |
|
|
}
|
500 |
|
|
else if (selectionEnd >= end)
|
501 |
|
|
{
|
502 |
|
|
addedForSelection.AddRange(SolveLeftGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
503 |
|
|
}
|
504 |
|
|
else
|
505 |
|
|
{
|
506 |
|
|
addedForSelection.AddRange(SolveLeftRightGap(node, selectionStart, selectionEnd, start, end, docToEdit, tagToAdd));
|
507 |
|
|
}
|
508 |
|
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
509 |
|
|
descendantsCount = descendantsToEdit.Count();
|
510 |
|
|
break;
|
511 |
|
|
}
|
512 |
|
|
}
|
513 |
|
|
}
|
514 |
|
|
|
515 |
|
|
GenerateCSS(tags);
|
516 |
|
|
return (docToEdit.DocumentNode.OuterHtml, new()
|
517 |
|
|
{
|
518 |
|
|
TagStartPositions = TagStartPositions,
|
519 |
|
|
TagStartLengths = TagStartLengths,
|
520 |
|
|
TagClosingPositions = TagClosingPositions,
|
521 |
|
|
TagClosingLengths = TagClosingLengths,
|
522 |
|
|
TagInstanceCSS = TagInstanceCSS,
|
523 |
|
|
NodeDict = NodeDict
|
524 |
|
|
});
|
525 |
|
|
}
|
526 |
|
|
|
527 |
|
|
|
528 |
f260dac8
|
Vojtěch Bartička
|
public (string, CachedInfo) PartialPreprocessHTMLRemoveTag(string htmlToEdit, string htmlOriginal, AnnotationTagGeneric tagToRemove, List<AnnotationTagGeneric> tags, CachedInfo cachedInfo)
|
529 |
2c9afc72
|
Vojtěch Bartička
|
{
|
530 |
|
|
UnpackCachedInfo(cachedInfo);
|
531 |
|
|
|
532 |
|
|
var docOriginal = new HtmlDocument();
|
533 |
|
|
docOriginal.LoadHtml(htmlOriginal);
|
534 |
|
|
var docToEdit = new HtmlDocument();
|
535 |
|
|
docToEdit.LoadHtml(htmlToEdit);
|
536 |
|
|
|
537 |
|
|
var descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
538 |
|
|
|
539 |
|
|
int i = 0;
|
540 |
|
|
int descendantsCount = descendantsToEdit.Count();
|
541 |
|
|
while (i < descendantsCount)
|
542 |
|
|
{
|
543 |
|
|
for (; i < descendantsCount; i++)
|
544 |
|
|
{
|
545 |
|
|
var node = descendantsToEdit.ElementAt(i);
|
546 |
|
|
if (!node.Attributes.Contains(TAG_EF_ID_ATTRIBUTE_NAME))
|
547 |
|
|
{
|
548 |
|
|
continue;
|
549 |
|
|
}
|
550 |
|
|
else
|
551 |
|
|
{
|
552 |
|
|
if (node.Attributes[TAG_EF_ID_ATTRIBUTE_NAME].Value != tagToRemove.Id.ToString())
|
553 |
|
|
{
|
554 |
|
|
continue;
|
555 |
|
|
}
|
556 |
|
|
|
557 |
|
|
node.Attributes.Remove(TAG_EF_ID_ATTRIBUTE_NAME);
|
558 |
|
|
node.Attributes.Remove(TAG_INSTANCE_ATTRIBUTE_NAME);
|
559 |
|
|
node.Attributes.Remove("class");
|
560 |
|
|
node.Attributes.Remove("start");
|
561 |
|
|
node.Attributes.Remove("end");
|
562 |
|
|
|
563 |
|
|
descendantsToEdit = docToEdit.DocumentNode.DescendantsAndSelf().ToList();
|
564 |
|
|
descendantsCount = descendantsToEdit.Count();
|
565 |
|
|
break;
|
566 |
|
|
}
|
567 |
|
|
}
|
568 |
|
|
}
|
569 |
|
|
|
570 |
|
|
GenerateCSS(tags);
|
571 |
|
|
return (docToEdit.DocumentNode.OuterHtml, new()
|
572 |
|
|
{
|
573 |
|
|
TagStartPositions = TagStartPositions,
|
574 |
|
|
TagStartLengths = TagStartLengths,
|
575 |
|
|
TagClosingPositions = TagClosingPositions,
|
576 |
|
|
TagClosingLengths = TagClosingLengths,
|
577 |
|
|
TagInstanceCSS = TagInstanceCSS,
|
578 |
|
|
NodeDict = NodeDict
|
579 |
|
|
});
|
580 |
|
|
}
|
581 |
|
|
|
582 |
|
|
/*
|
583 |
|
|
* Partial HTML Preprocessing ----------------------------------------------------------------------------
|
584 |
|
|
*/
|
585 |
|
|
}
|
586 |
|
|
}
|