private void ConfigureItemSections(IEnumerable <string> contentInLowerCase, ListPageHandler handler)
{
string[] testedDishNames = new string[] { "kurczak", "gulasz", "makaron", "filet" };
string[] bannedNames = new string[] { };
var possibleSections = FilterForExcept(contentInLowerCase, testedDishNames, bannedNames);
var morePossibleSections = FilterFor(possibleSections, new string[]
{ "<a href" });
ApplyItemSections(morePossibleSections, testedDishNames, handler);
}
public IListPageHandler CreateForList(Uri firstListPage)
{
_currentPage = firstListPage;
_pageContent = GetContent(firstListPage);
ListPageHandler handler = new ListPageHandler();
ConfigureItemSections(_pageContent.Select(l => l.ToLower()), handler);
ConfigureNextPageSwitch(handler);
return(handler);
}
private void ConfigureNextPageSwitch(ListPageHandler handler)
{
var pageSwitch = new InceremntPageStrategy(_currentPage);
var page = pageSwitch.NextPage();
try
{
GetContent(page);
handler.NextPageStrategy = pageSwitch;
}
catch (WebException)
{
handler.NextPageStrategy = new SearchContentForUriStrategy();
}
}
private void ApplyItemSections(IEnumerable <string> sections, string[] testedDishNames, ListPageHandler handler)
{
foreach (var section in sections)
{
int begin = section.IndexOf("<a href=");
int betweenBegin = section.IndexOf('>', begin + 5);
int betweenEnd = section.IndexOf('<', betweenBegin);
int end = section.IndexOf("</a>", betweenEnd - 1) + 4;
var dishName = section.Substring(betweenBegin + 1, betweenEnd - betweenBegin - 1);
if (testedDishNames.Any(w => dishName.Contains(w)))
{
var preceedingSectionBegin = section.LastIndexOf('<', begin - 1);
string sectionBegin = section.Substring(0, begin);
var preceedingSectionEnd = section.IndexOf('>', end + 1);
string sectionEnd = section.Substring(end);
handler.ItemSection = new KeyValuePair <string, string>(sectionBegin, sectionEnd);
handler.ItemUriExtractor = GetItemExtractStrategyFor(section.Substring(begin, betweenBegin - begin));
// So far the engine sucks a little and misses cases, can't put all the stake on one miss, so so far all hrefs are acceptable
// when engine improves add return; after first match
}
}
}