/// <summary>
/// Extracts the raw data from craigslist by hitting
/// http://vancouver.en.craigslist.ca/search/sss?query=bike+kona+dawg&srchType=A&format=rss
/// then breaks down the individual links and hits those concurrently parsing out the relevant details
/// I cannot find a paging paramter for this RSS feed
/// The start at parameter is &s=75
/// You only get 25 items on a page and have no clue in the RSS feed of how many matching
/// items there are in total. The non rss version of the page has a page count / total
/// matching items count
///
/// See also another piece of software that does this stuff too :
/// http://www.zentastic.com/blog/zencash/
///
/// </summary>
/// <param name="cityName"></param>
/// <param name="stolenDate"></param>
/// <param name="itemDescription"></param>
/// <returns></returns>
public async Task<CraigslistInfo[]> GetCraigslistInfoByRSS(string cityName, string stolenDate, int pageNumber, string itemDescription)
{
XmlDocument xmlDoc = new XmlDocument(); //* create an xml document object.
xmlDoc.Load(String.Format("http://{0}.en.craigslist.ca/search/sss?query={1}&srchType=A&format=rss&s={2}", cityName, itemDescription, pageNumber * RSS_RESULTS_PAGE_FEED_PAGE_SIZE));
//
// Again a bit of fragile code here, but whaddaya gonna do!
//
if (xmlDoc.ChildNodes[1].ChildNodes[0].ChildNodes[13].ChildNodes[0].ChildNodes.Count > 0)
{
if (xmlDoc.ChildNodes[1].ChildNodes[0].ChildNodes[13].ChildNodes[0].ChildNodes.Count < RSS_RESULTS_PAGE_FEED_PAGE_SIZE)
{
//* Get elements.
XmlNodeList items = xmlDoc.ChildNodes[1].ChildNodes[0].ChildNodes[13].ChildNodes[0].ChildNodes;
//
// This multitaking stuff is a bit of a basard
// used this link http://stackoverflow.com/questions/10806951/how-to-limit-the-amount-of-concurrent-async-i-o-operations and also
// See http://msdn.microsoft.com/en-gb/library/vstudio/hh300224.aspx
// and especially http://msdn.microsoft.com/en-us/library/hh556530.aspx
return await GetAllUrlsAsync(items);
}
else
{
//
// Break all the rest rules - which say the error message should be in the statusText header
// see: http://stackoverflow.com/questions/1077340/best-way-to-return-error-messages-on-rest-services
// instead do it wrong and return an error disguised as the object
//
CraigslistInfo cockupTooManyresults = new CraigslistInfo();
cockupTooManyresults.DescriptionHTML = "Error";
cockupTooManyresults.Title = String.Format("More than {0} results returned, please be more selective", RSS_RESULTS_PAGE_FEED_PAGE_SIZE);
cockupTooManyresults.DodgyScore = 0;
CraigslistInfo[] results = new CraigslistInfo[1];
results[0] = cockupTooManyresults;
return results;
}
}
else
{
//
// Break all the rest rules - which say the error message should be in the statusText header
// see: http://stackoverflow.com/questions/1077340/best-way-to-return-error-messages-on-rest-services
// instead do it wrong and return an error disguised as the object
// Oh yeah and breech the Dont reeat Yoursefl principle too!
//
CraigslistInfo cockupTooManyresults = new CraigslistInfo();
cockupTooManyresults.DescriptionHTML = "Error";
cockupTooManyresults.Title = "No results returned, alter your search terms";
cockupTooManyresults.DodgyScore = 0;
CraigslistInfo[] results = new CraigslistInfo[1];
results[0] = cockupTooManyresults;
return results;
}
}
/// <summary>
/// Parses data from an HTML node for a listing items to populate most of the data into a craigslistInfo object
/// </summary>
/// <param name="item"></param>
/// <returns></returns>
public static CraigslistInfo transformSummaryHTMLNodeIntoCraigslistInfo(HtmlAgilityPack.HtmlNode item)
{
CraigslistInfo craigslistInfo = new CraigslistInfo();
//
// the node is a block of HTml that looks like this
//<p class="row" data-latitude="" data-longitude="">
// <span class="ih" id="images:3Ee3K43m25Ne5K95Hecbmc8b000ac1a0e1376.jpg"> </span>
// <span class="itemdate"> Dec 11</span>
// <span class="itemsep"> - </span>
// <a href="http://vancouver.en.craigslist.ca/pml/ctd/3427888027.html">2003 HYUNDAI TIBURON GT</a>
// <span class="itemsep"> - </span>
// <span class="itemph"></span>
// <span class="itempp"> $6595</span>
// <span class="itempn"><font size="-1"> (MISSION)</font></span>
// <span class="itemcg" title="ctd"> <small class="gc"><a href="/ctd/">cars & trucks - by dealer</a></small></span>
// <span class="itempx"> <span class="p"> pic</span></span>
// <br class="c">
// </p>
//
// First of all parse the date : Dec 11
//
// .:TODO:. this is not working and throwing the exception every time - and using the output in jscript breaks the page!
//
string[] formats = { " MMM dd" };
try
{
craigslistInfo.DatePostedReal = DateTime.ParseExact(item.SelectSingleNode("//span[@class='itemdate']").InnerText.Substring(0, formats[0].Length), formats, null, System.Globalization.DateTimeStyles.AssumeUniversal);
}
catch (FormatException)
{
craigslistInfo.DatePostedReal = DateTime.MinValue;
Console.WriteLine("Unable to convert '{0}' to a date.", item.SelectSingleNode("//span[@class='itemdate']").InnerText);
}
//
// Next the Price
//
craigslistInfo.Price = item.SelectSingleNode("//span[@class='itempp']").InnerText.Trim();
craigslistInfo.SaleArea = item.SelectSingleNode("//font").InnerText;
craigslistInfo.LinkURL = item.ChildNodes[7].Attributes["href"].Value;
craigslistInfo.ShortTitle = item.ChildNodes[7].InnerText;
craigslistInfo.Category = item.ChildNodes[17].SelectSingleNode("//a").InnerText;
craigslistInfo.DatePostedParsed = craigslistInfo.DatePostedReal.ToString("dd MMM yyyy HH:mm:ss");
return craigslistInfo;
}
/// <summary>
/// The actions from the foreach loop are moved to this async method.
/// </summary>
/// <param name="url"></param>
/// <param name="client"></param>
/// <returns></returns>
public static async Task<CraigslistInfo> ProcessURL(string url, HttpClient client)
{
//
// The detail page is not available in RSS, so have to parse the HTML for it (much harder than the nice XML you get frmo RSS feed)
// to kae this easier use a special gizmo called the HTMl agility pack that parses the
// HTMl into a document tree (and fixes errors etc in the HTMl) - this allows us to traverse the page safely
//
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
//
// There are various options, set as needed
// Here I just want any errors tidied up to make a valid Docuemtn Tree so we can parse with confidence
//
htmlDoc.OptionFixNestedTags = true;
htmlDoc.Load((await (client.GetStreamAsync(url))), true);
//
// Now ots of magic to parse the data from all the HTML
//
CraigslistInfo currentListing = new CraigslistInfo();
currentListing.updateCraigslistInfoFromFullItemDetailsPage(htmlDoc, url);
return currentListing;
}
/// <summary>
/// <summary>
/// Extracts the raw data from craigslist by hitting
/// http://vancouver.en.craigslist.ca/search/sss?query=bike+kona+dawg&srchType=A
/// then breaks down the individual links and hits those concurrently parsing out the relevant details
/// We use the HTmll page not the RSS feed since the RSS feed doe not do paging correctly
/// which makes things crap
/// The start at parameter is &s=75
///
/// See also another piece of software that does this stuff too :
/// http://www.zentastic.com/blog/zencash/
/// /// </summary>
/// <param name="cityName"></param>
/// <param name="stolenDate"></param>
/// <param name="pageNumber"></param>
/// <param name="itemDescription"></param>
/// <returns></returns>
public async Task<CraigslistInfo[]> GetCraigslistInfo(string cityName, string stolenDate, int pageNumber, string itemDescription)
{
List<CraigslistInfo> listings = new List<CraigslistInfo>();
//
// The search result page is available in RSS, but that page does not have paging so try to use the HTML
// version instead (much harder to parse than the nice XML you get from RSS feed)
// to make this easier use a special gizmo called the HtmlAgilityPack that parses the
// HTMl into a document tree (and fixes errors etc in the HTML) - this allows us to traverse the
// page strucvture safely
//
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
//
// There are various options, set as needed
// Here I just want any errors tidied up to make a valid Document Tree so we can parse with confidence
// Note: We get this first URl synchronously, but call up the fine detail for each item within it
// by calling it's details URL asynchronoursly (and concurrently) for speed purposes
//
htmlDoc.OptionFixNestedTags = true;
System.Net.WebRequest webReq = System.Net.WebRequest.Create(String.Format("http://{0}.en.craigslist.ca/search/sss?query={1}&srchType=A&s={2}", cityName, itemDescription, pageNumber * HTML_RESULTS_PAGE_FEED_PAGE_SIZE));
System.Net.WebResponse webRes = webReq.GetResponse();
System.IO.Stream mystream = webRes.GetResponseStream();
if (mystream != null)
{
htmlDoc.Load(mystream);
}
//
// Use XPAth queries (see http://www.w3schools.com/xpath/xpath_syntax.asp)
// to parse the DOm the HtmlAgilityPack has made us
//
//HtmlAgilityPack.HtmlNode pageCountNode = htmlDoc.DocumentNode.SelectSingleNode("//h4/b[2]");
//string searchMatchCount = pageCountNode.InnerText.Substring(pageCountNode.InnerText.IndexOf("Found: "),pageCountNode.InnerText.IndexOf(" Displaying:") - pageCountNode.InnerText.IndexOf("Found: "));
//
// Get the inodvidual items for sales
//
HtmlAgilityPack.HtmlNodeCollection itemNodes = htmlDoc.DocumentNode.SelectNodes("//p[@class='row']");
if (itemNodes != null)
{
if (itemNodes.Count < MAX_RESULTS_WARN)
{
foreach (HtmlAgilityPack.HtmlNode item in itemNodes)
{
CraigslistInfo craigslistInfo = CraigslistInfo.transformSummaryHTMLNodeIntoCraigslistInfo(item);
listings.Add(craigslistInfo);
}
}
else
{
//
// Break all the rest rules - which say the error message should be in the statusText header
// see: http://stackoverflow.com/questions/1077340/best-way-to-return-error-messages-on-rest-services
// instead do it wrong and return an error disguised as the object
//
CraigslistInfo cockupTooManyresults = new CraigslistInfo();
cockupTooManyresults.DescriptionHTML = "Error";
cockupTooManyresults.Title = String.Format("More than {0} results returned, please be more selective", MAX_RESULTS_WARN);
cockupTooManyresults.DodgyScore = 0;
CraigslistInfo[] results = new CraigslistInfo[1];
results[0] = cockupTooManyresults;
return results;
}
}
else
{
//
// Break all the rest rules - which say the error message should be in the statusText header
// see: http://stackoverflow.com/questions/1077340/best-way-to-return-error-messages-on-rest-services
// instead do it wrong and return an error disguised as the object
// Oh yeah and breech the Dont reeat Yoursefl principle too!
//
CraigslistInfo cockupTooManyresults = new CraigslistInfo();
cockupTooManyresults.DescriptionHTML = "";
cockupTooManyresults.Title = "No results returned, alter your search terms";
cockupTooManyresults.DodgyScore = 0;
CraigslistInfo[] results = new CraigslistInfo[1];
results[0] = cockupTooManyresults;
return results;
}
return await UpdateAllListItemsWithDetailsFromTheirDetailPageAsync(listings);
}
