public static IEnumerable <T> Links <T>(this ParsedHtml self, Func <string, HtmlObject, T> selector)
{
return
(from a in self.QuerySelectorAll("a[href]")
let href = a.GetAttributeValue("href")
where !string.IsNullOrWhiteSpace(href)
select selector(self.TryBaseHref(href), a));
}
public static IQuery <HtmlObject> Tables(ParsedHtml html) => html.Tables(null).ToQuery();
public static IQuery <T> Links <T>(ParsedHtml html, Func <string, string, T> selector) => html.Links((href, ho) => selector(href, ho.InnerHtml)) .Select(link => link) .ToQuery();
public static IQuery <string> Links(ParsedHtml html) => Links(html, (href, _) => href);
public static IQuery <HtmlForm> Forms(ParsedHtml html) => Query.Return(html.Forms);
public static IQuery <DataTable> FormsAsDataTable(ParsedHtml html)
{
var forms =
from f in html.Forms
select f.GetForm((fd, fs) => new
{
f.Name,
Action = new Uri(html.TryBaseHref(f.Action), UriKind.Absolute),
Method = f.Method.ToString().ToUpperInvariant(),
f.EncType,
Data = fd,
Submittables = fs,
});
var dt = new DataTable();
dt.Columns.AddRange(new []
{
new DataColumn("#")
{
AllowDBNull = false
},
new DataColumn("FormName")
{
AllowDBNull = true
},
new DataColumn("FormAction")
{
AllowDBNull = false
},
new DataColumn("FormMethod")
{
AllowDBNull = true
},
new DataColumn("FormEncoding")
{
AllowDBNull = true
},
new DataColumn("Name")
{
AllowDBNull = false
},
new DataColumn("Value")
{
AllowDBNull = false
},
new DataColumn("Submittable", typeof(bool))
{
AllowDBNull = false
},
});
foreach (var form in
from fi in forms.Select((f, i) => (i + 1).AsKeyTo(f))
let form = fi.Value
from controls in new[]
{
from e in form.Data.AsEnumerable()
from v in e.Value
select new { e.Key, Value = v, Submittable = false },
from e in form.Submittables.AsEnumerable()
from v in e.Value
select new { e.Key, Value = v, Submittable = true },
}
from control in controls
select new object[]
{
fi.Key,
form.Name,
form.Action.OriginalString,
form.Method,
form.EncType,
control.Key,
control.Value,
control.Submittable,
})
{
dt.Rows.Add(form);
}
return(Query.Singleton(dt));
}
public static IEnumerable <HtmlObject> Tables(this ParsedHtml self, string selector) =>
from e in self.QuerySelectorAll(selector ?? "table")
where "table".Equals(e.Name, StringComparison.OrdinalIgnoreCase)
select e;
public static IEnumerable <HtmlObject> Tables(this ParsedHtml html) => html.Tables(null);
public static IEnumerable <string> Links(this ParsedHtml html) => html.Links((href, _) => href);