private static bool prepareItem(FIDCResult result, HtmlNode labelNode, HtmlNode valueTextNode, FIDCItem previousItem)
{
if (labelNode != null && valueTextNode != null)
{
// Key
string key = "";
HtmlAttribute keyAttr = labelNode.Attributes.FirstOrDefault(x => x.Name.ToLower() == "id");
if (keyAttr != null) key = keyAttr.Value;
// Label
string label = labelNode.InnerText;
// Value
string value = "";
value = valueTextNode.InnerText;
// Parent
int level = labelNode.InnerText.CleanString(false).StartsWithCounter(" ");
FIDCItem parent = (previousItem != null && level > 0) ? previousItem.GetParentFromLevel(level - 1) : null;
result.Items.Add(new FIDCItem() { Key = key.CleanString(), Label = label.CleanString(), Value = value.CleanString(), Parent = parent });
return true;
}
return false;
}
private FIDCResult Load()
{
string finalURL = URLSeed;
foreach (ScrapperFilter filter in this.Filters)
{
finalURL += "&" + filter.ToURLParameter();
}
HtmlDocument doc = Scrapper.GetDocument(finalURL);
FIDCResult result = new FIDCResult()
{
Id = 0,
TimeStamp = DateTime.Now,
Items = new List<FIDCItem>()
};
if (doc.DocumentNode != null)
{
// Cabeçalho
HtmlNodeCollection tCabecalho = doc.DocumentNode.SelectNodes("//table[@id='tbCabecalhoInfo']/tr");
if (tCabecalho != null)
{
foreach (HtmlNode row in tCabecalho)
{
HtmlNode labelNode = row.SelectSingleNode("td[1]/span");
HtmlNode valueTextNode = row.SelectSingleNode("td[2]/span");
HtmlNode valueSelectNode = row.SelectSingleNode("td[2]/select/option[@selected]");
if (labelNode != null && (valueTextNode != null || valueSelectNode != null))
{
// Key
string key = "";
HtmlAttribute keyAttr = labelNode.Attributes.FirstOrDefault(x => x.Name.ToLower() == "id");
if (keyAttr != null) key = keyAttr.Value;
// Label
string label = labelNode.InnerText;
// Value
string value = "";
if (valueTextNode != null) value = valueTextNode.InnerText;
else if (valueSelectNode != null) value = valueSelectNode.Attributes.FirstOrDefault(x => x.Name.ToLower() == "value").Value;
result.Items.Add(new FIDCItem() { Key = key.CleanString(), Label = label.CleanString(), Value = value.CleanString() });
}
else
{
System.Diagnostics.Debug.WriteLine("Row Not Processed : " + row.InnerHtml);
}
}
}
// Corpo
HtmlNodeCollection tCorpo = doc.DocumentNode.SelectNodes("//table[@id='tbCorpoInfo']/tr");
if (tCorpo != null)
{
FIDCItem previous = null;
foreach (HtmlNode row in tCorpo)
{
HtmlNode labelNode = row.SelectSingleNode("td[1]/span");
HtmlNode valueTextNode = row.SelectSingleNode("td[2]/span");
if (labelNode != null)
{
// Key
string key = "";
HtmlAttribute keyAttr = labelNode.Attributes.FirstOrDefault(x => x.Name.ToLower() == "id");
if (keyAttr != null) key = keyAttr.Value.CleanString();
// Label
string label = labelNode.InnerText.CleanString();
// Value
string value = (valueTextNode != null) ? valueTextNode.InnerText.CleanString() : null;
// Parent
int level = labelNode.InnerText.CleanString(false).StartsWithCounter(" ");
FIDCItem parent = (previous != null && level > 0) ? previous.GetParentFromLevel(level - 1) : null;
System.Diagnostics.Debug.WriteLine(level + " ::: " + label + " ::: " + labelNode.InnerText);
result.Items.Add(new FIDCItem() { Parent = parent, Key = key, Label = label, Value = value });
previous = result.Items.Last();
}
else
{
System.Diagnostics.Debug.WriteLine("Row Not Processed : " + row.InnerHtml);
}
}
}
// CotaSenior
HtmlNodeCollection tHorizontais = doc.DocumentNode.SelectNodes("//table[@id='dgInfoCotaSenior'] | //table[@id='dgInfoCotaSubrd']");
if (tHorizontais != null && tHorizontais.Count() == 2)
{
foreach (HtmlNode tCotaSenior in tHorizontais)
{
HtmlNode parentNode = tCotaSenior.SelectSingleNode("preceding-sibling::*[1]/tr[last()]");
FIDCItem parent = new FIDCItem() { Key = "Anexo I - " + parentNode.InnerText.ToKey(), Label = "Anexo I - " + parentNode.InnerText.CleanString() };
result.Items.Add(parent);
if (tCotaSenior != null)
{
HtmlNodeCollection labelCol = tCotaSenior.SelectNodes("tr[1]/td");
HtmlNodeCollection valuesCol = tCotaSenior.SelectNodes("tr[2]/td");
if (labelCol != null && valuesCol != null)
{
int itemNumber = System.Math.Min(labelCol.Count(), valuesCol.Count());
string key = "";
string label = "";
string value = "";
for (int i = 0; i < itemNumber; i++)
{
label = labelCol[i].InnerText;
value = valuesCol[i].InnerText;
result.Items.Add(new FIDCItem() { Parent = parent, Key = key.CleanString(), Label = label.CleanString(), Value = value.CleanString() });
}
}
}
}
}
else
{
System.Diagnostics.Debug.WriteLine("TABELAS HORIZONTAIS: ERROR!!!");
}
}
return result;
}
private FIDCResult Load()
{
FIDCResult result = new FIDCResult()
{
Id = 0,
TimeStamp = DateTime.Now,
Items = new List<FIDCItem>()
};
string finalURL = URLSeed;
foreach (ScrapperFilter filter in this.Filters)
{
finalURL += "&" + filter.ToURLParameter();
}
HtmlDocument doc = Scrapper.GetDocument(finalURL);
//HtmlDocument doc = Scrapper.GetDocumentFromFile(AppDomain.CurrentDomain.BaseDirectory + "Services\\TestData\\Mensal489.html");
if (doc.DocumentNode != null)
{
FIDCItem previous = null;
foreach (HtmlNode table in doc.DocumentNode.SelectNodes("//table"))
{
TableStyle style = TableStyle.NotDefined;
// Try to find Table Style
if (style == TableStyle.NotDefined)
{
int columns = 0;
foreach (HtmlNode row in table.SelectNodes("tr"))
{
HtmlNodeCollection cols = row.SelectNodes("td");
if (cols != null && cols.Count > columns)
{
columns = cols.Count;
}
}
if (columns == 2) style = TableStyle.Vertical;
if (columns > 3) style = TableStyle.Horizontal;
}
if (style == TableStyle.Vertical)
{
foreach (HtmlNode row in table.SelectNodes("tr"))
{
HtmlNodeCollection labelNodes = row.SelectNodes("td[1]/span");
HtmlNodeCollection valueTextNodes = row.SelectNodes("td[2]/span");
if (labelNodes != null && valueTextNodes != null)
{
if (labelNodes.Count == 1 && valueTextNodes.Count == 1)
{
prepareItem(result, labelNodes[0], valueTextNodes[0], previous);
previous = result.Items.Last();
}
if (labelNodes.Count == 2)
{
prepareItem(result, labelNodes[0], labelNodes[1], previous);
previous = result.Items.Last();
prepareItem(result, valueTextNodes[0], valueTextNodes[1], previous);
previous = result.Items.Last();
}
}
}
}
else if (style == TableStyle.Horizontal)
{
}
}
}
return result;
}
