private static async Task <string> PostIt(Uri uri, FormSearchSos formSearchSos)
{
var formContent = new FormUrlEncodedContent(formSearchSos.FormDataList());
// PrintKeyValuePairs(formDataList);
var request = new HttpRequestMessage {
RequestUri = uri, Method = HttpMethod.Post, Content = formContent
};
SetRequestHeaders(request);
_httpRespMsg = await NetHttpClient.Client.SendAsync(request);
if (!_httpRespMsg.IsSuccessStatusCode)
{
CurrentStatus.LastOpMessage = $"PostIt could not retrieve URL, StatusCode: {_httpRespMsg.StatusCode}";
CurrentStatus.ScrapeComplete = true;
return(string.Empty);
}
var stringContent = await _httpRespMsg.Content.ReadAsStringAsync();
BytesReceived += stringContent.Length;
return(stringContent);
}
public static ScrapeStatus GetElections(FormSearchSos formSearchSos)
{
// TODO: put exception catcher here for network problems.
var contentString = PostIt(new Uri(OfficeSearchResultsUrl), formSearchSos).Result;
if (!_httpRespMsg.IsSuccessStatusCode)
{
CurrentStatus.LastOpMessage = $"GetElections call returned Status Code: {_httpRespMsg.StatusCode}";
CurrentStatus.ScrapeComplete = true;
CurrentStatus.LastPageCompleted++;
return(CurrentStatus);
}
if (string.IsNullOrEmpty(contentString))
{
CurrentStatus.LastOpMessage = "GetElections received null content";
CurrentStatus.ScrapeComplete = true;
CurrentStatus.LastPageCompleted++;
return(CurrentStatus);
}
CurrentStatus.LastOpMessage = "GetElections received document length = " + contentString.Length;
var htmlDoc = new HtmlDocument();
htmlDoc.LoadHtml(contentString);
const string tgtSelect = "//select[@id='id_election']/option";
var electNodes = htmlDoc.DocumentNode.SelectNodes(tgtSelect);
if (electNodes == null)
{
CurrentStatus.ScrapeComplete = true;
CurrentStatus.LastOpMessage = "ElectNodes search returned null.";
return(CurrentStatus);
}
var elections = electNodes.Select(node => new Election(node.Attributes[0].Value, node.InnerText)).ToList();
CurrentStatus.ScrapeComplete = true;
CurrentStatus.ScrapeSuccess = true;
CurrentStatus.Elections = elections;
return(CurrentStatus);
}
public static ScrapeStatus GetCandidates(FormSearchSos formSearchSos)
{
var contentString = PostIt(new Uri(OfficeSearchResultsUrl), formSearchSos).Result;
if (!_httpRespMsg.IsSuccessStatusCode)
{
CurrentStatus.LastOpMessage = $"GetCandidates call returned Status Code: {_httpRespMsg.StatusCode}";
CurrentStatus.ScrapeComplete = true;
CurrentStatus.LastPageCompleted++;
return(CurrentStatus);
}
if (string.IsNullOrEmpty(contentString))
{
CurrentStatus.LastOpMessage = "GetCandidates received null content";
CurrentStatus.ScrapeComplete = true;
CurrentStatus.LastPageCompleted++;
return(CurrentStatus);
}
CurrentStatus.LastOpMessage = "GetCandidates received document length = " + contentString.Length;
var htmlDoc = new HtmlDocument();
htmlDoc.LoadHtml(contentString);
var year = formSearchSos.ElectionYear;
var candList = new List <CandidateSos>();
const string tgtDiv = "//*[@class=\"col1Inner\"]/table/tr";
var nodes = htmlDoc.DocumentNode.SelectNodes(tgtDiv);
if (nodes == null)
{
CurrentStatus.ScrapeComplete = true;
CurrentStatus.ScrapeSuccess = false;
CurrentStatus.LastOpMessage = "Data table search returned null.";
return(CurrentStatus);
}
var candDesc = string.Empty;
foreach (var nodetr in nodes)
{
var tdObj = nodetr.ChildNodes[0]; // td 1
if (tdObj.Attributes.Count > 0 && tdObj.Attributes[0].Name == "colspan") // Account for #text children here
{
// Start of candidate section
candDesc = CleanUpWhiteSpace(tdObj.InnerText);
}
else
{
// Get candidates
if (nodetr.InnerHtml.Contains("Qualified - Signatures Required"))
{
// We're at the end of the list, asterisk
break;
}
var subdoc = new HtmlDocument();
subdoc.LoadHtml(nodetr.InnerHtml);
var rowcntr = 1;
// November fails here
var candTrRows = from table in subdoc.DocumentNode.SelectNodes("//table")
from row in table.SelectNodes("tr")
from cell in row.SelectNodes("td")
select new CellData {
RowNum = rowcntr++, CellText = CleanUpWhiteSpace(cell.InnerText)
};
candList.Add(FillCandidate(candTrRows.ToList(), candDesc, year.VarValue));
}
}
CurrentStatus.ScrapeComplete = true;
CurrentStatus.ScrapeSuccess = true;
CurrentStatus.Candidates = candList;
return(CurrentStatus);
}
