protected virtual bool IsImage(HTMLTag linkTag)
{
string url = General.GetAbsoluteURL(_url, HttpUtility.HtmlDecode(linkTag.GetAttributeValueOrEmpty("href")));
return url != null && url.IndexOf(ImageURLKeyword, StringComparison.OrdinalIgnoreCase) >= 0;
}
public HTMLTagRange(HTMLTag startTag, HTMLTag endTag)
{
StartTag = startTag;
EndTag = endTag;
}
public string GetHTML(HTMLTag startTag, HTMLTag endTag)
{
return(GetSection(_preprocessedHTML, startTag.Offset, startTag.IsSelfClosing ? startTag.EndOffset : endTag.EndOffset));
}
public static bool ClassAttributeValueHas(HTMLTag tag, string targetClassName)
{
string attributeValue = tag.GetAttributeValue("class");
return(attributeValue != null && ClassAttributeValueHas(attributeValue, targetClassName));
}
public HTMLTag FindCorrespondingEndTag(HTMLTag tag)
{
return(FindCorrespondingEndTag(tag, null));
}
public HTMLTagRange CreateTagRange(HTMLTag tag, HTMLTag stopBeforeTag)
{
HTMLTag endTag = FindCorrespondingEndTag(tag, stopBeforeTag);
return((tag != null && endTag != null) ? new HTMLTagRange(tag, endTag) : null);
}
public IEnumerable<HTMLTag> FindStartTags(HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
return FindTags(false, startAfterTag, stopBeforeTag, names);
}
public IEnumerable <HTMLTag> FindEndTags(HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
return(FindTags(true, startAfterTag, stopBeforeTag, names));
}
public HTMLTag FindCorrespondingEndTag(HTMLTag tag, HTMLTag stopBeforeTag)
{
if (tag == null) {
return null;
}
if (tag.IsEnd) {
throw new ArgumentException("Tag must be a start tag.");
}
if (tag.IsSelfClosing) {
return tag;
}
int startIndex = GetTagIndex(tag) + 1;
int stopIndex = stopBeforeTag != null ? (GetTagIndex(stopBeforeTag) - 1) : (_tags.Count - 1);
int depth = 1;
for (int i = startIndex; i <= stopIndex; i++) {
HTMLTag tag2 = _tags[i];
if (!tag2.IsSelfClosing && tag2.NameEquals(tag.Name)) {
depth += tag2.IsEnd ? -1 : 1;
if (depth == 0) {
return tag2;
}
}
}
return null;
}
public HTMLTag FindStartTag(HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
return FindTag(false, startAfterTag, stopBeforeTag, names);
}
public HTMLTag FindCorrespondingEndTag(HTMLTag tag)
{
return FindCorrespondingEndTag(tag, null);
}
public IEnumerable<HTMLTag> EnumerateTags(HTMLTag startAfterTag, HTMLTag stopBeforeTag)
{
int startIndex = startAfterTag != null ? (GetTagIndex(startAfterTag) + 1) : 0;
int stopIndex = stopBeforeTag != null ? (GetTagIndex(stopBeforeTag) - 1) : (_tags.Count - 1);
for (int i = startIndex; i <= stopIndex; i++) {
yield return _tags[i];
}
}
public HTMLTagRange CreateTagRange(HTMLTag tag, HTMLTag stopBeforeTag)
{
HTMLTag endTag = FindCorrespondingEndTag(tag, stopBeforeTag);
return (tag != null && endTag != null) ? new HTMLTagRange(tag, endTag) : null;
}
public static bool ClassAttributeValueHas(HTMLTag tag, string targetClassName)
{
string attributeValue = tag.GetAttributeValue("class");
return attributeValue != null && ClassAttributeValueHas(attributeValue, targetClassName);
}
public HTMLTag FindTag(bool isEndTag, HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
foreach (HTMLTag tag in FindTags(isEndTag, startAfterTag, stopBeforeTag, names)) {
return tag;
}
return null;
}
public HTMLTagRange CreateTagRange(HTMLTag tag)
{
return CreateTagRange(tag, null);
}
public IEnumerable<HTMLTag> FindTags(bool isEndTag, HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
foreach (HTMLTag tag in EnumerateTags(startAfterTag, stopBeforeTag)) {
if (tag.IsEnd == isEndTag && tag.NameEqualsAny(names)) {
yield return tag;
}
}
}
public HTMLTag FindEndTag(HTMLTag startAfterTag, HTMLTag stopBeforeTag, params string[] names)
{
return(FindTag(true, startAfterTag, stopBeforeTag, names));
}
public string GetHTML(HTMLTag startTag, HTMLTag endTag)
{
return GetSection(_preprocessedHTML, startTag.Offset, startTag.IsSelfClosing ? startTag.EndOffset : endTag.EndOffset);
}
public HTMLTagRange CreateTagRange(HTMLTag tag)
{
return(CreateTagRange(tag, null));
}
public string GetInnerHTML(HTMLTag startTag, HTMLTag endTag)
{
return startTag.IsSelfClosing ? String.Empty : GetSection(_preprocessedHTML, startTag.EndOffset, endTag.Offset);
}
private static IEnumerable <HTMLTag> ParseTags(string html, int htmlStart, int htmlEnd)
{
while (htmlStart < htmlEnd)
{
int pos = IndexOf(html, htmlStart, htmlEnd, '<');
if (pos == -1)
{
yield break;
}
HTMLTag tag = new HTMLTag();
tag.Offset = pos;
htmlStart = pos + 1;
tag.IsEnd = StartsWith(html, htmlStart, htmlEnd, '/');
if (StartsWithLetter(html, tag.IsEnd ? (htmlStart + 1) : htmlStart, htmlEnd))
{
// Parse tag name
if (tag.IsEnd)
{
htmlStart += 1;
}
pos = IndexOfAny(html, htmlStart, htmlEnd, true, '/', '>');
if (pos == -1)
{
yield break;
}
tag.Name = GetSectionLower(html, htmlStart, pos);
htmlStart = pos;
// Parse attributes
bool isTagComplete = false;
do
{
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd))
{
htmlStart++;
}
tag.IsSelfClosing = StartsWith(html, htmlStart, htmlEnd, '/');
if (tag.IsSelfClosing)
{
htmlStart += 1;
}
if (StartsWith(html, htmlStart, htmlEnd, '>'))
{
htmlStart += 1;
isTagComplete = true;
}
else if (tag.IsSelfClosing)
{
}
else
{
HTMLAttribute attribute = new HTMLAttribute();
attribute.Offset = htmlStart;
// Parse attribute name
pos = IndexOfAny(html, htmlStart + 1, htmlEnd, true, '=', '/', '>');
if (pos == -1)
{
yield break;
}
attribute.Name = GetSectionLower(html, htmlStart, pos);
htmlStart = pos;
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd))
{
htmlStart++;
}
if (StartsWith(html, htmlStart, htmlEnd, '='))
{
// Parse attribute value
htmlStart += 1;
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd))
{
htmlStart++;
}
if (StartsWithAny(html, htmlStart, htmlEnd, '"', '\''))
{
char quoteChar = html[htmlStart];
htmlStart += 1;
pos = IndexOf(html, htmlStart, htmlEnd, quoteChar);
if (pos == -1)
{
yield break;
}
attribute.Value = GetSection(html, htmlStart, pos);
htmlStart = pos + 1;
}
else
{
pos = IndexOfAny(html, htmlStart, htmlEnd, true, '>');
if (pos == -1)
{
yield break;
}
attribute.Value = GetSection(html, htmlStart, pos);
htmlStart = pos;
}
}
else
{
attribute.Value = String.Empty;
}
attribute.Length = htmlStart - attribute.Offset;
if (tag.GetAttribute(attribute.Name) == null)
{
tag.Attributes.Add(attribute);
}
}
} while (!isTagComplete);
tag.Length = htmlStart - tag.Offset;
// Yield result
yield return(tag);
// Skip contents of special tags whose contents are to be treated as raw text
if (!tag.IsEnd && !tag.IsSelfClosing && tag.NameEqualsAny("script", "style", "title", "textarea"))
{
bool foundEndTag = false;
do
{
pos = IndexOf(html, htmlStart, htmlEnd, '<');
if (pos == -1)
{
yield break;
}
htmlStart = pos + 1;
string endTagText = "/" + tag.Name;
if (StartsWith(html, htmlStart, htmlEnd, endTagText, true) &&
(StartsWithWhiteSpace(html, htmlStart + endTagText.Length, htmlEnd) ||
StartsWithAny(html, htmlStart + endTagText.Length, htmlEnd, '/', '>')))
{
htmlStart -= 1;
foundEndTag = true;
}
} while (!foundEndTag);
}
}
else if (StartsWith(html, htmlStart, htmlEnd, "!--", false) && !StartsWith(html, htmlStart + 3, htmlEnd, '>'))
{
// Skip comment
htmlStart += 3;
bool foundEnd = false;
do
{
pos = IndexOf(html, htmlStart, htmlEnd, '-');
if (pos == -1)
{
yield break;
}
htmlStart = pos + 1;
if (StartsWith(html, htmlStart, htmlEnd, "->", false))
{
htmlStart += 2;
foundEnd = true;
}
else if (StartsWith(html, htmlStart, htmlEnd, "-!>", false))
{
htmlStart += 3;
foundEnd = true;
}
} while (!foundEnd);
}
else if (StartsWithAny(html, htmlStart, htmlEnd, '?', '/', '!'))
{
// Skip bogus comment or DOCTYPE
htmlStart += 1;
pos = IndexOf(html, htmlStart, htmlEnd, '>');
if (pos == -1)
{
yield break;
}
htmlStart = pos + 1;
}
}
}
private static IEnumerable<HTMLTag> ParseTags(string html, int htmlStart, int htmlEnd)
{
while (htmlStart < htmlEnd) {
int pos = IndexOf(html, htmlStart, htmlEnd, '<');
if (pos == -1) yield break;
HTMLTag tag = new HTMLTag();
tag.Offset = pos;
htmlStart = pos + 1;
tag.IsEnd = StartsWith(html, htmlStart, htmlEnd, '/');
if (StartsWithLetter(html, tag.IsEnd ? (htmlStart + 1) : htmlStart, htmlEnd)) {
// Parse tag name
if (tag.IsEnd) htmlStart += 1;
pos = IndexOfAny(html, htmlStart, htmlEnd, true, '/', '>');
if (pos == -1) yield break;
tag.Name = GetSectionLower(html, htmlStart, pos);
htmlStart = pos;
// Parse attributes
bool isTagComplete = false;
do {
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd)) htmlStart++;
tag.IsSelfClosing = StartsWith(html, htmlStart, htmlEnd, '/');
if (tag.IsSelfClosing) htmlStart += 1;
if (StartsWith(html, htmlStart, htmlEnd, '>')) {
htmlStart += 1;
isTagComplete = true;
}
else if (tag.IsSelfClosing) { }
else {
HTMLAttribute attribute = new HTMLAttribute();
attribute.Offset = htmlStart;
// Parse attribute name
pos = IndexOfAny(html, htmlStart + 1, htmlEnd, true, '=', '/', '>');
if (pos == -1) yield break;
attribute.Name = GetSectionLower(html, htmlStart, pos);
htmlStart = pos;
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd)) htmlStart++;
if (StartsWith(html, htmlStart, htmlEnd, '=')) {
// Parse attribute value
htmlStart += 1;
while (StartsWithWhiteSpace(html, htmlStart, htmlEnd)) htmlStart++;
if (StartsWithAny(html, htmlStart, htmlEnd, '"', '\'')) {
char quoteChar = html[htmlStart];
htmlStart += 1;
pos = IndexOf(html, htmlStart, htmlEnd, quoteChar);
if (pos == -1) yield break;
attribute.Value = GetSection(html, htmlStart, pos);
htmlStart = pos + 1;
}
else {
pos = IndexOfAny(html, htmlStart, htmlEnd, true, '>');
if (pos == -1) yield break;
attribute.Value = GetSection(html, htmlStart, pos);
htmlStart = pos;
}
}
else {
attribute.Value = String.Empty;
}
attribute.Length = htmlStart - attribute.Offset;
if (tag.GetAttribute(attribute.Name) == null) {
tag.Attributes.Add(attribute);
}
}
} while (!isTagComplete);
tag.Length = htmlStart - tag.Offset;
// Yield result
yield return tag;
// Skip contents of special tags whose contents are to be treated as raw text
if (!tag.IsEnd && !tag.IsSelfClosing && tag.NameEqualsAny("script", "style", "title", "textarea")) {
bool foundEndTag = false;
do {
pos = IndexOf(html, htmlStart, htmlEnd, '<');
if (pos == -1) yield break;
htmlStart = pos + 1;
string endTagText = "/" + tag.Name;
if (StartsWith(html, htmlStart, htmlEnd, endTagText, true) &&
(StartsWithWhiteSpace(html, htmlStart + endTagText.Length, htmlEnd) ||
StartsWithAny(html, htmlStart + endTagText.Length, htmlEnd, '/', '>')))
{
htmlStart -= 1;
foundEndTag = true;
}
} while (!foundEndTag);
}
}
else if (StartsWith(html, htmlStart, htmlEnd, "!--", false) && !StartsWith(html, htmlStart + 3, htmlEnd, '>')) {
// Skip comment
htmlStart += 3;
bool foundEnd = false;
do {
pos = IndexOf(html, htmlStart, htmlEnd, '-');
if (pos == -1) yield break;
htmlStart = pos + 1;
if (StartsWith(html, htmlStart, htmlEnd, "->", false)) {
htmlStart += 2;
foundEnd = true;
}
else if (StartsWith(html, htmlStart, htmlEnd, "-!>", false)) {
htmlStart += 3;
foundEnd = true;
}
} while (!foundEnd);
}
else if (StartsWithAny(html, htmlStart, htmlEnd, '?', '/', '!')) {
// Skip bogus comment or DOCTYPE
htmlStart += 1;
pos = IndexOf(html, htmlStart, htmlEnd, '>');
if (pos == -1) yield break;
htmlStart = pos + 1;
}
}
}
public string GetInnerHTML(HTMLTag startTag, HTMLTag endTag)
{
return(startTag.IsSelfClosing ? String.Empty : GetSection(_preprocessedHTML, startTag.EndOffset, endTag.Offset));
}
private int GetTagIndex(HTMLTag tag)
{
int i;
if (!_offsetToIndex.TryGetValue(tag.Offset, out i)) {
throw new Exception("Unable to locate the specified tag.");
}
return i;
}
public HTMLTagRange(HTMLTag startTag, HTMLTag endTag)
{
StartTag = startTag;
EndTag = endTag;
}
protected override bool IsImage(HTMLTag linkTag)
{
return Enumerable.FirstOrDefault(Enumerable.Where(_htmlParser.FindStartTags(_htmlParser.CreateTagRange(linkTag), "img"), t => HTMLParser.ClassAttributeValueHas(t, "thumb"))) != null;
}
