internal static Section[] GetSections(IElement elem, Section[] sections = null)
{
if (sections == null)
{
return(GetSections(elem, new Section[0]));
}
if (FiFiPageEntry.IsEnd(elem))
{
return(sections);
}
var sectionHeadTags = new string[2] {
"H4", "H5"
};
if (sectionHeadTags.Contains(elem.TagName))
{
// some sections have no text content
var sectionElems = FiFiPage.GetElementsUntil(sectionHeadTags, elem.NextElementSibling);
bool hasContent = sectionElems.Count() > 0 && sectionElems.Select(e => e.TextContent.Trim()).Aggregate((x, y) => x + y).Length > 0;
if (hasContent)
{
return(GetSections(elem.NextElementSibling, AddSection(sections, new Section(elem))));
}
}
return(GetSections(elem.NextElementSibling, sections));
}
public static TermObject GetTerm(string term)
{
FiFiPage fifiPage = FiFiPage.GetPage(term);
var fifiObject = (!fifiPage.IsEmpty() && fifiPage.IsFinnish()) ? (TermObject)fifiPage.GetJsonObject() : new TermObject()
{
term = term, entries = Enumerable.Empty <EntryObject>()
};
EnFiPage enfiPage = EnFiPage.GetPage(term);
var enfiObject = (!enfiPage.IsEmpty() && enfiPage.IsFinnish()) ? (TermObject)enfiPage.GetJsonObject() : new TermObject()
{
term = term, entries = Enumerable.Empty <EntryObject>()
};
var combinedTerms = new TermObject()
{
term = fifiObject.term,
entries = fifiObject.entries.Concat(enfiObject.entries)
};
return(combinedTerms);
}
internal static FiFiPageEntry[] GetEntries(IElement[] elements, FiFiPageEntry[] entries)
{
// cycle through Finnish elements on HtmlPage
IElement elem = elements.FirstOrDefault();
if (FiFiPage.IsEndOfPage(elem))
{
return(entries);
}
// page entries are headed by h3 tags
// if element is not an h3 then
if (elem.TagName != "H3")
{
return(GetEntries(elements.Skip(1).ToArray(), entries));
}
// otherwise, create a new PageEntry, add it to entries, and continue.
FiFiPageEntry entry = new FiFiPageEntry(elements);
return(GetEntries(elements.Skip(1).ToArray(), FiFiPageEntry.AddEntry(entries, entry)));
}
internal static Item[] GetItems(IElement elem, Item[] items = null)
{
if (items == null)
{
return(GetItems(elem, new Item[0]));
}
if (FiFiPage.IsEndOfPage(elem))
{
return(items);
}
if (String.IsNullOrEmpty(elem.TextContent.Trim()))
{
return(GetItems(elem.NextElementSibling, items));
}
switch (elem.TagName)
{
case "H3":
case "H4":
case "H5":
return(items);
case "UL":
case "OL":
case "DIV":
case "DD":
var lItems = elem.QuerySelectorAll("li,dl").Select(li => new Item(li)).ToArray();
return(GetItems(elem.NextElementSibling, items.Concat(lItems).ToArray()));
default:
return(GetItems(elem.NextElementSibling, items.Concat(new Item[1] {
new Item(elem)
}).ToArray()));
}
}
internal static bool IsEnd(IElement elem)
{ // this should be moved to WiktionaryPage
return(FiFiPage.IsEndOfPage(elem) || elem.TagName == "H3");
}
// end of section / entry
internal static bool IsEnd(IElement elem)
{
return(FiFiPage.IsEndOfPage(elem) || elem.TagName == "H3");
}