public List <Dictionary <string, string> > ProcessHtml(string html, List <RegexPattern> regexPatterns)
{
var results = new List <Dictionary <string, string> >();
RegexInfo pageTypeMatch = GetPageType(html, regexPatterns, 0);
if (pageTypeMatch == null)
{
return(null);
}
// can use reflection
if (pageTypeMatch.RegexType.Equals(RegexType.List))
{
var selectionRegexs = regexPatterns.Where(r => r.Type.Equals(RegexType.Selection) && r.ParentId.Equals(pageTypeMatch.RegexId)).ToList();
var selections = GetMatchedResults(html, selectionRegexs);
foreach (var s in selections)
{
var rows = ListPageProcessor(s.Value, regexPatterns, pageTypeMatch);
results.AddRange(rows);
}
}
return(results);
}
public List <Dictionary <string, string> > ListPageProcessor(string html, List <RegexPattern> regexPatterns, RegexInfo pageTypeMatch)
{
var results = new List <Dictionary <string, string> >();
var globalResults = new List <KeyValuePair <string, string> >();
var globalRegexes = regexPatterns.Where(r => r.Type.Equals(RegexType.Global) && r.ParentId.Equals(pageTypeMatch.RegexId)).ToList();
if (globalRegexes != null && globalRegexes.Count > 0)
{
globalResults = GetMatchedResults(html, globalRegexes);
}
var regexes = regexPatterns.Where(r => r.ParentId.Equals(pageTypeMatch.RegexId) && r.Type.Equals(RegexType.Item)).ToList();
if (regexes == null)
{
return(null);
}
var items = GetMatchedResults(html, regexes);
foreach (var regex in regexes)
{
var detailRegexes = regexPatterns.Where(r => r.Type.Equals(RegexType.Detail) && r.ParentId.Equals(regex.Id)).ToList();
foreach (var item in items)
{
var itemDetails = GetMatchedResults(item.Value, detailRegexes);
if (itemDetails != null && itemDetails.Count > 0)
{
if (globalResults.Count > 0)
{
itemDetails.AddRange(globalResults);
}
results.Add(itemDetails.GroupBy(f => f.Key).Select(g => g.First()).ToList().ToDictionary(x => x.Key, x => x.Value));
}
}
}
return(results);
}