/// <summary>
/// Get the source by index.
/// </summary>
/// <param name="index">The index.</param>
/// <returns>source section</returns>
public string GetSource(int index)
{
string source = string.Empty;
if (_matches != null && index < _matches.Count)
{
Match sub = _matches[index];
if (sub.Length != 0)
{
MatchTag start = (MatchTag)_page.tags[sub.Index];
int tagIndex = sub.Index + sub.Length;
int sourceLength;
if (_page.tags.Count > tagIndex)
{
// get source data until the start of the next tag
MatchTag end = (MatchTag)_page.tags[tagIndex];
sourceLength = end.Index - start.Index;
}
else
{
// If last tag, get source data until the end of the current tag
MatchTag end = (MatchTag)_page.tags[tagIndex - 1];
sourceLength = end.Index + end.Length - start.Index;
}
source = _pageSource.Substring(start.Index, sourceLength);
}
}
return(source);
}
/// <summary>
/// Tags the list.
/// </summary>
/// <param name="source">The source.</param>
/// <returns></returns>
public static MatchTagCollection TagList(string source)
{
MatchTagCollection list = new MatchTagCollection();
for (int i = 0; i < source.Length; i++)
{
if (source[i] == '<')
{
int length = GetTagLength(source, i);
// check if tag is too short or comment
// min lenght for a tag is 3 <x>
// comments start <!--
if (length >= 3 &&
(source[i + 1] == '/' || source[i + 1] == '!' ||
(source[i + 1] >= 'A' && source[i + 1] <= 'Z') ||
(source[i + 1] >= 'a' && source[i + 1] <= 'z')))
{
MatchTag tag = new MatchTag(source, i, length);
list.Add(tag);
i = i + length - 1;
}
}
}
return list;
}
/// <summary>
/// Tags the list.
/// </summary>
/// <param name="source">The source.</param>
/// <returns></returns>
public static MatchTagCollection TagList(string source)
{
MatchTagCollection list = new MatchTagCollection();
for (int i = 0; i < source.Length; i++)
{
if (source[i] == '<')
{
int length = GetTagLength(source, i);
// check if tag is too short or comment
// min lenght for a tag is 3 <x>
// comments start <!--
if (length >= 3 &&
(source[i + 1] == '/' || source[i + 1] == '!' ||
(source[i + 1] >= 'A' && source[i + 1] <= 'Z') ||
(source[i + 1] >= 'a' && source[i + 1] <= 'z')))
{
MatchTag tag = new MatchTag(source, i, length);
list.Add(tag);
i = i + length - 1;
}
}
}
return(list);
}
/// <summary>
/// Builds the profile.
/// </summary>
/// <param name="source">The source.</param>
/// <returns>the profile</returns>
private Profile BuildProfile(string source)
{
Profile build = new Profile();
if (source == null || source.Length == 0)
{
return(build);
}
MatchTagCollection tags = HtmlString.TagList(source);
build.tags = new MatchTagCollection();
build.tagMap = string.Empty;
for (int i = 0; i < tags.Count; i++)
{
MatchTag tag = tags[i];
char tagStart = char.ToUpper(tag.TagName[0]);
// Test if special tag -- used for regex searchs
// these tags are only copied into the tagMap no position index is stored
if (tagStart == 'Z')
{
build.tagMap += tag.TagName.Substring(1);
}
if (_sectionTemplate.Tags.IndexOf(tagStart) != -1 &&
tag.TagName != "br")
{
if (tagStart == 'T')
{
if (char.ToUpper(tag.TagName[1]) != 'A')
{
tagStart = char.ToUpper(tag.TagName[1]);
}
}
if (tag.IsClose)
{
build.tagMap += tagStart;
}
else
{
build.tagMap += char.ToLower(tagStart);
}
build.tags.Add(tag);
}
}
return(build);
}
/// <summary>
/// Strips the unwanted HTML tags.
/// </summary>
/// <param name="source">The source.</param>
/// <returns>stripped source</returns>
private string StripTags(string source)
{
string stripped = string.Empty;
source = HtmlString.NewLines(source);
MatchTagCollection tags = HtmlString.TagList(source);
for (int i = 0; i < tags.Count; i++)
{
MatchTag tag = tags[i];
if (_template.Tags.IndexOf(char.ToUpper(tag.TagName[0])) != -1)
{
stripped += source.Substring(tag.Index, tag.Length);
}
int start = tag.Index + tag.Length;
if (start < source.Length)
{
int end;
if (i + 1 < tags.Count)
{
tag = tags[i + 1];
end = tag.Index;
}
else
{
end = source.Length;
}
stripped += source.Substring(start, end - start);
}
}
return(stripped);
}
public virtual void Add(MatchTag value)
{
this.List.Add(value);
}
public virtual void Add(MatchTag value)
{
this.List.Add(value);
}
/// <summary>
/// Gets the sections.
/// </summary>
/// <param name="source">The source.</param>
/// <returns></returns>
private Sections GetSections(string source)
{
source = StripTags(source);
Sections data = new Sections();
data.dataFields = new ArrayList();
data.minFields = 0;
data.optionalData = false;
data.dataTags = 0;
MatchTagCollection tags = HtmlString.TagList(source);
bool isOptionalTag = false;
bool zTag = false;
for (int i = 0; i < tags.Count; i++)
{
MatchTag tag = tags[i];
// check if tag + data is optional / regex
zTag = false;
if (char.ToUpper(tag.TagName[0]) == 'Z')
{
zTag = true;
isOptionalTag = true;
if (tag.IsClose)
{
isOptionalTag = false;
}
//i++;
//if (i < tags.Count)
// tag = tags[i];
//else
// break;
}
// Check if tag is one of interest
if (_template.Tags.IndexOf(char.ToUpper(tag.TagName[0])) != -1)
{
DataField section;
if (!zTag)
{
// Add tag to array of fields
section = new DataField();
section.optional = isOptionalTag;
section.htmlTag = tag;
section.hasData = false;
section.source = tag.FullTag;
if (section.source.IndexOf("<#") != -1 || section.source.IndexOf("<*") != -1)
{
section.hasData = true;
if (isOptionalTag)
{
data.optionalData = true;
}
section.dataElements = GetElements(section.source);
data.dataTags += section.dataElements.Count;
}
data.dataFields.Add(section);
if (!isOptionalTag)
{
data.minFields++;
}
}
// Add data between this tag and the next to field array
int start = tag.Index + tag.Length;
int end;
if (i + 1 < tags.Count)
{
tag = tags[i + 1];
zTag = false;
if (char.ToUpper(tag.TagName[0]) == 'Z')
{
zTag = true;
isOptionalTag = true;
if (tag.IsClose)
{
isOptionalTag = false;
}
}
// start = tag.Index + tag.Length;
// i++;
// if (i + 1 < tags.Count)
// tag = tags[i + 1];
// else
// break;
//}
end = tag.Index;
}
else
{
end = source.Length;
}
if (!zTag)
{
section = new DataField();
section.optional = isOptionalTag;
section.htmlTag = null;
section.source = HtmlString.Decode(source.Substring(start, end - start));
section.hasData = false;
if (section.source.IndexOf("<#") != -1 || section.source.IndexOf("<*") != -1)
{
section.hasData = true;
if (isOptionalTag)
{
data.optionalData = true;
}
section.dataElements = GetElements(section.source);
data.dataTags += section.dataElements.Count;
}
data.dataFields.Add(section);
if (!isOptionalTag)
{
data.minFields++;
}
}
}
}
return(data);
}