/// <summary>
/// Returns the content of specified tag.
/// </summary>
/// <param name="pageHtml">html code</param>
/// <param name="justTagName">The tag name like TABLE</param>
public static string GetTagContent(ref string pageHtml, string justTagName)
{
string startTag = '<' + justTagName;
string endTag = justTagName + '>';
int start, end;
start = StringCompare.IndexOfIgnoreCase(ref pageHtml, startTag);
if (start == -1)
{
return("");
}
start = StringCompare.IndexOfMatchCase(ref pageHtml, '>', start);
if (start == -1)
{
return("");
}
start++;
end = StringCompare.IndexOfIgnoreCase(ref pageHtml, endTag, start);
if (end == -1)
{
return("");
}
end = StringCompare.LastIndexOfMatchCase(ref pageHtml, '<', end);
if (end == -1 || start > end)
{
return("");
}
return(pageHtml.Substring(start, end - start).Trim());
}
public static TextRange FindCSSClassStyleUrlValuePosition(ref string pageHtml, int startindex)
{
int valueStart, valueEnd;
const string strCSSUrlValue = "url(";
TextRange result; // = new TextRange(-1, -1);
result.End = -1;
result.Start = -1;
//==============================
if (startindex >= pageHtml.Length)
{
return(result);
}
// Find first position
valueStart = StringCompare.IndexOfIgnoreCase(ref pageHtml, strCSSUrlValue, startindex);
if (valueStart == -1)
{
return(result);
}
valueStart += strCSSUrlValue.Length;
valueEnd = StringCompare.IndexOfMatchCase(ref pageHtml, ")", valueStart);
if (valueEnd == -1)
{
return(result);
}
if (valueEnd > StringCompare.IndexOfMatchCase(ref pageHtml, ";", valueStart))
{
return(result);
}
result.Start = valueStart;
result.End = valueEnd;
return(result);
}
/// <summary>
/// Removes specified tag
/// </summary>
/// <param name="pageHtml"></param>
/// <param name="tagName"></param>
public static void RemoveTagOnly(ref string pageHtml, string tagName)
{
string tagEnd = "</" + tagName + ">";
string tagStart = "<" + tagName;
string tagStartEnd = ">";
int tagStartPos = 0;
int tagStartEndPos = 0;
int tagEndPos = 0;
do
{
tagStartPos = StringCompare.IndexOfIgnoreCase(ref pageHtml, tagStart, tagStartPos);
if (tagStartPos == -1)
{
return;
}
tagStartEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagStartEnd, tagStartPos);
if (tagStartEndPos == -1)
{
return;
}
tagStartEndPos += tagStartEnd.Length;
// removes tag start
pageHtml = pageHtml.Remove(tagStartPos, tagStartEndPos - tagStartPos);
// find tag end pos
tagEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagEnd, tagStartPos);
if (tagEndPos == -1)
{
return;
}
pageHtml = pageHtml.Remove(tagEndPos, tagEnd.Length);
} while (tagEndPos != -1);
}
/// <summary>
/// Locates @import rule value
/// </summary>
/// <param name="cssCodes"></param>
/// <param name="startindex"></param>
/// <param name="onlyWithoutUrlOption">Specifies that operator should not catch rule with URL attribute.</param>
/// <param name="captureExactInUrl">Specifies that operator should strip the URL attribute if there is any</param>
/// <returns></returns>
public static TextRange FindImportRuleUrlPosition(ref string cssCodes, int startindex, bool onlyWithoutUrlOption, bool captureExactlyInUrl)
{
int valueStart, valueEnd;
TextRange result = new TextRange(-1, -1);
const string strCSSImportRule = "@import";
const string strCSSUrlValue = "url(";
//==============================
if (startindex >= cssCodes.Length)
{
return(result);
}
// Find first position
valueStart = StringCompare.IndexOfIgnoreCase(ref cssCodes, strCSSImportRule, startindex);
if (valueStart == -1)
{
return(result);
}
valueStart += strCSSImportRule.Length;
//
if (cssCodes.Substring(valueStart, 1).Trim().Length > 0)
{
return(result);
}
else
{
valueStart++;
}
valueEnd = StringCompare.IndexOfMatchCase(ref cssCodes, ";", valueStart);
if (valueEnd == -1)
{
return(result);
}
else
{
int urlPos = StringCompare.IndexOfIgnoreCase(ref cssCodes, strCSSUrlValue, valueStart);
if (urlPos != -1 && urlPos < valueEnd)
{
if (onlyWithoutUrlOption)
{
result.Start = valueEnd;
result.End = -1;
return(result);
}
if (captureExactlyInUrl)
{
valueStart = urlPos + strCSSUrlValue.Length;
urlPos = StringCompare.IndexOfMatchCase(ref cssCodes, ")", valueStart);
if (urlPos != -1 && urlPos < valueEnd)
{
valueEnd = urlPos;
}
}
}
}
result.Start = valueStart;
result.End = valueEnd;
return(result);
}
public static void ReplaceFormsSources(ref string htmlCodes,
string pageUrlNoQuery,
string newPageFormat,
string pagePath,
string siteRootUrl,
bool encodeUrl,
bool changeMethod,
string extraAttributeFormat)
{
TextRange methodResult; // = new TextRange();
TextRange actionResult; // = new TextRange();
int cursorPos = 0;
string newAttribute = "";
string formMethod = "";
string tmp, actionSrc = "";
string orgValue = "";
bool addNewAttribute = false;
bool hasNewAttribute = false;
if (!string.IsNullOrEmpty(extraAttributeFormat))
{
addNewAttribute = true;
hasNewAttribute = true;
}
do
{
addNewAttribute = hasNewAttribute;
if (changeMethod)
{
methodResult = HtmlParser.GetTagAttributeValuePos(ref htmlCodes, "<form", "method", cursorPos);
if (methodResult.Start > -1 && methodResult.End > -1)
{
// get the method
formMethod = htmlCodes.Substring(methodResult.Start, 2);
// validate the method
formMethod = WebMethods.DetectMethod(formMethod, WebMethods.DefaultMethods.GET);
htmlCodes = htmlCodes.Remove(methodResult.Start, methodResult.End - methodResult.Start);
htmlCodes = htmlCodes.Insert(methodResult.Start, "POST");
}
else
{
int formPos = StringCompare.IndexOfIgnoreCase(ref htmlCodes, "<form", cursorPos);
int tagEnd;
if (formPos != -1)
{
tagEnd = StringCompare.IndexOfMatchCase(ref htmlCodes, '>', formPos);
if (tagEnd != -1)
{
htmlCodes = htmlCodes.Insert(tagEnd, " method=POST ");
}
}
formMethod = WebMethods.GET;
}
}
actionResult = HtmlParser.GetTagAttributeValuePos(ref htmlCodes, "<form", "action", cursorPos);
if (actionResult.Start == -1)
{
break;
}
if (actionResult.Start > -1 && actionResult.End > -1)
{
cursorPos = actionResult.Start;
//====== Correct value position according to quotes existence=======
// actionResult = ASProxyFunctions.CorrectValueIfQuoteExists(ref pageHtml, actionResult);
// Get the value
actionSrc = htmlCodes.Substring(actionResult.Start, actionResult.End - actionResult.Start);
// BUG fixed in v5 beta 2
// now supports forms with javascript
if (UrlProvider.IsClientSitdeUrl(actionSrc) == false)
{
//====== Convert virtual url to absolute ======
actionSrc = UrlProvider.JoinUrl(actionSrc, pageUrlNoQuery, pagePath, siteRootUrl);
//====== Delete invalid character such as tab and line feed ======
actionSrc = UrlProvider.IgnoreInvalidUrlCharctersInHtml(actionSrc);
orgValue = actionSrc;
//===== If another site url, has bookmark=====
if (actionSrc.IndexOf('#') != -1)
{
actionSrc = UrlBuilder.RemoveUrlBookmark(actionSrc, out tmp);
}
//=====Get desired address=======
actionSrc = HttpUtility.HtmlDecode(actionSrc);
//====== Encode url to make unknown it ======
if (encodeUrl)
{
actionSrc = UrlProvider.EncodeUrl(actionSrc);
}
else
{
// just url safe
actionSrc = UrlProvider.EscapeUrlQuery(actionSrc);
}
//====== Add it to our url ======
actionSrc = string.Format(newPageFormat, actionSrc);
if (changeMethod)
{
//actionSrc = UrlBuilder.AddUrlQuery(actionSrc, Consts.qIsPostForm, ((int)method).ToString());
actionSrc = UrlBuilder.AddUrlQueryToEnd(actionSrc, Consts.Query.WebMethod, formMethod);
}
// Make it html safe
actionSrc = HttpUtility.HtmlEncode(actionSrc);
//====== Replace it with old url ======
htmlCodes = htmlCodes.Remove(actionResult.Start, actionResult.End - actionResult.Start);
htmlCodes = htmlCodes.Insert(actionResult.Start, actionSrc);
}
else
{
// this is client side url
addNewAttribute = false;
}
if (addNewAttribute)
{
// Apply orginal value and encoded value to format
newAttribute = string.Format(extraAttributeFormat, orgValue, actionSrc, "POST");
// Locate end of tag
cursorPos = StringCompare.IndexOfMatchCase(ref htmlCodes, '>', actionResult.Start);
if (htmlCodes[cursorPos - 1] == '/')
{
cursorPos--;
}
// Insert to it
htmlCodes = htmlCodes.Insert(cursorPos, newAttribute);
}
}
else
{
if (actionResult.Start != -1)
{
cursorPos = actionResult.Start;
}
cursorPos = StringCompare.IndexOfMatchCase(ref htmlCodes, ">", cursorPos);
}
}while (actionResult.Start != -1);
}
/// <summary>
/// Locate and remove a tag from html codes
/// </summary>
/// <param name="pageHtml">Html codes</param>
/// <param name="tagName">Name of tags</param>
/// <param name="removeTag">Remove the tag also</param>
public static void RemoveTagContent(ref string pageHtml, string tagName, bool removeTag)
{
string tagEnd = "</" + tagName + ">";
string tagStart = "<" + tagName;
string tagStartEnd = ">";
int tagStartPos = 0;
int tagStartEndPos = 0;
int tagEndPos = 0;
if (removeTag)
{
do
{
tagStartPos = StringCompare.IndexOfIgnoreCase(ref pageHtml, tagStart, tagStartPos);
if (tagStartPos == -1)
{
return;
}
tagStartEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagStartEnd, tagStartPos);
if (tagStartEndPos == -1)
{
return;
}
tagStartEndPos += tagStartEnd.Length;
tagEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagEnd, tagStartEndPos);
if (tagEndPos == -1)
{
return;
}
tagEndPos += tagEnd.Length;
pageHtml = pageHtml.Remove(tagStartPos, tagEndPos - tagStartPos);
} while (tagEndPos != -1);
}
else
{
do
{
tagStartPos = StringCompare.IndexOfIgnoreCase(ref pageHtml, tagStart, tagStartPos);
if (tagStartPos == -1)
{
return;
}
tagStartEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagStartEnd, tagStartPos);
if (tagStartEndPos == -1)
{
return;
}
tagStartEndPos += tagStartEnd.Length;
tagEndPos = StringCompare.IndexOfMatchCase(ref pageHtml, tagEnd, tagStartEndPos);
if (tagEndPos == -1)
{
return;
}
pageHtml = pageHtml.Remove(tagStartEndPos, tagEndPos - tagStartEndPos);
} while (tagEndPos != -1);
}
}
/// <summary>
/// Locate value of attribute position for specified html tag
/// </summary>
public static TextRange GetTagAttributeValuePos(ref string pageHtml, string tagName, string attributeName, int start)
{
int tagStart, tagEnd;
int attrStart;
int valueStart;
// Allocate result with default values
TextRange result = new TextRange(-1, -1);
// Incorrect input
if (start >= pageHtml.Length)
{
return(result);
}
// Correct tag name
if (tagName[0] != '<')
{
tagName = '<' + tagName;
}
//=============================================================//
// Find tag first position
tagStart = StringCompare.IndexOfIgnoreCase(ref pageHtml, tagName, start);
if (tagStart == -1)
{
return(result);
}
// Locate end of tag
tagEnd = StringCompare.IndexOfMatchCase(ref pageHtml, ">", tagStart);
// Wrong html code
if (tagEnd == -1)
{
return(result);
}
// Find the Attribute position
attrStart = StringCompare.IndexOfIgnoreCase(ref pageHtml, attributeName, tagStart);
// There is no attribute, so go away!
if (attrStart == -1 || attrStart >= tagEnd)
{
// Set found start position to result
result.Start = tagStart;
result.End = -1;
return(result);
}
// Find value start position
valueStart = StringCompare.IndexOfMatchCase(ref pageHtml, '=', attrStart);
// There is no value, so go away!
if (valueStart == -1 || valueStart >= tagEnd)
{
// Set found start position to result
result.Start = attrStart;
result.End = -1;
return(result);
}
// Locate value of attribute position
result = FindAttributeValuePosition(ref pageHtml, tagEnd, valueStart);
return(result);
}
/// <summary>
/// Finds and override url addresses within "style" tag.
/// </summary>
public static void ReplaceStyleTagStyleUrl(
ref string htmlCode,
string currentUrlWithoutParameters,
string importRuleUrl,
string othersNewUrl,
string replacmentBasePath,
string siteAddress,
bool encodeUrl)
{
const string styleStart = "<style";
const string styleEnd = "</style>";
const string htmlTagEnd = ">";
TextRange position;
int index = 0;
//string bookmarkPart = "";
do
{
string styleContent = "";
// Find style tag position start
position.Start = StringCompare.IndexOfIgnoreCase(ref htmlCode, styleStart, index);
if (position.Start == -1)
{
break;
}
// locate tag end
position.Start = StringCompare.IndexOfMatchCase(ref htmlCode, htmlTagEnd, position.Start);
if (position.Start == -1)
{
break;
}
position.Start++;
// Locate end tag
position.End = StringCompare.IndexOfIgnoreCase(ref htmlCode, styleEnd, position.Start);
if (position.End == -1)
{
break;
}
// Set next searching start position
index = position.End;
// Get the content
styleContent = htmlCode.Substring(position.Start, position.End - position.Start);
// If there is nothing, go to next tag
if (styleContent.Trim().Length == 0)
{
continue;
}
// Process the style content for "Import Rule"
// We don't need "Import Rule" process with "url" property any more. It will done by bext command. Since v4.0
ReplaceCSSClassStyleUrl(ref styleContent, currentUrlWithoutParameters, importRuleUrl, replacmentBasePath, siteAddress, encodeUrl, true);
// Process the style content for backgrounds.
// So, this breach the backgound option role. Since v4.0
ReplaceCSSClassStyleUrl(ref styleContent, currentUrlWithoutParameters, othersNewUrl, replacmentBasePath, siteAddress, encodeUrl, false);
// Replace the new style
htmlCode = htmlCode.Remove(position.Start, position.End - position.Start);
htmlCode = htmlCode.Insert(position.Start, styleContent);
} while (index != -1);
}
/// <summary>
/// Finds and encode codes within "script" tag.
/// </summary>
public static void ReplaceScriptTagCodes(ref string htmlCode)
{
const string scriptStart = "<script";
const string scriptEnd = "</script>";
const string htmlTagEnd = ">";
TextRange position;
int index = 0;
do
{
string scriptContent = "";
// Find script tag position start
position.Start = StringCompare.IndexOfIgnoreCase(ref htmlCode, scriptStart, index);
if (position.Start == -1)
{
break;
}
// locate tag end
position.Start = StringCompare.IndexOfMatchCase(ref htmlCode, htmlTagEnd, position.Start);
if (position.Start == -1)
{
break;
}
position.Start++;
// Locate end tag
position.End = StringCompare.IndexOfIgnoreCase(ref htmlCode, scriptEnd, position.Start);
if (position.End == -1)
{
break;
}
// Set next searching start position
index = position.End;
// Get the content
scriptContent = htmlCode.Substring(position.Start, position.End - position.Start);
// If there is nothing, go to next tag
if (scriptContent.Trim().Length == 0)
{
continue;
}
// Very bad use of processor, But there is no other way to do!
// Create a new instance of processor
JSProcessor processor = new JSProcessor();
// Process the script content
processor.Execute(ref scriptContent,
null, null, null, null);
// Replace the new style
htmlCode = htmlCode.Remove(position.Start, position.End - position.Start);
htmlCode = htmlCode.Insert(position.Start, scriptContent);
} while (index != -1);
}