public void DoSearch(string query, LoadingContext<BookSummary> context)
{
context.Progress.Progress = null;
Http.RequestHtml(GetSearchUrl(query), doc => {
if (doc.DocumentNode.Descendants("title").First().InnerText.StartsWith("The Library Catalog : Item #")) {
context.AddResult(new DetailsPageSummary(doc));
context.SetCompleted();
return;
}
var serverInput = doc.DocumentNode.Descendants("input").First(n => n.GetAttributeValue("name", "") == "server");
var countString = serverInput.PreviousSibling.InnerText;
var count = int.Parse(countParser.Match(countString).Groups[1].Value);
if (count == 0)
context.SetCompleted();
else {
context.Progress.Maximum = count;
ParseResultList(doc, context);
}
});
}
static void ParseResultList(HtmlDocument doc, LoadingContext<BookSummary> context)
{
var mainTable = doc.DocumentNode.FirstChild.Element("body").Element("center").Element("table");
foreach (var elem in mainTable.Elements("tr")) {
if (context.Progress.WasCanceled) {
context.SetCompleted();
return;
}
if (elem.Elements("th").Count() != 2)
continue; //These are <hr> rows; ignore them
context.AddResult(new SearchResult(elem));
context.Progress.Progress++;
}
//Their next link is supposed to be in a <form> tag,
//but the <form> tag is malformed. I do not want to
//rely entirely on text matching for this, so I look
//in the <form>'s parent.
var nextLink = doc.DocumentNode.Descendants("form").SelectMany(f => f.ParentNode.Descendants("a"))
.FirstOrDefault(a => a.InnerText.Contains("Next"));
if (nextLink == null) {
Debug.Assert(context.Progress.Progress == context.Progress.Maximum, "Book count mismatch");
context.SetCompleted();
} else {
Http.RequestHtml(new Uri(baseUri, nextLink.GetAttributeValue("href", "")),
newDoc => ParseResultList(newDoc, context)
);
}
}