public void scrapData(HtmlNode[] nodes, int dataformat)
{
string urlActual = string.Empty;
cantidad = nodes.Count();
lbTotalNodosIdentificados.Text = nodes.Count().ToString();
cantNodeIndetified = nodes.Count();
progressBarBusqueda.Maximum = Convert.ToInt32(cantidad);
progressBarBusqueda.Step = 1;
foreach (HtmlNode enlace in nodes)
{
itemCount++;
lbTotalNodosEvaluados.Text = itemCount.ToString();
percent = Math.Round((itemCount / cantidad * 100), 2, MidpointRounding.AwayFromZero);
lbPorcent.Text = percent.ToString() + "%";
progressBarBusqueda.PerformStep();
progressBarBusqueda.Refresh();
this.Refresh();
foreach (HtmlAttribute atributo in enlace.Attributes)
{
urlActual = atributo.Value.ToString();
lbEnlaceActual.Text = urlActual;
lbEnlaceActual.Refresh();
if (RegexTool.isUrl(urlActual))
{
if (!isVisited(urlActual))
{
HtmlAgilityPack.HtmlDocument recipe = web.Load(urlActual);
if (hasRecipe(recipe, dataformat))
{
addValueToDic(urlActual, "");
lbTotalRegistroEncontrados.Refresh();
GetMicroData(recipe, dataformat);
}
else
{
var recetas = recipe.DocumentNode.SelectNodes(".//article/a/@href");
addValueToDic(urlActual, "");
if (recetas != null)
{
scrapData(recetas.ToArray(), dataformat);
}
}
}
}
}
}
}
public void scrapData(string Url, int cantidad)
{
browser = new System.Windows.Forms.WebBrowser();
browser.DocumentCompleted += new System.Windows.Forms.WebBrowserDocumentCompletedEventHandler(DocumentCompleted);
HtmlDocument document = null;
int contador = 0;
HtmlWeb web = new HtmlWeb();
while (contador <= cantidad)
{
browser.Navigate(Url);
browser.Visible = true;
browser.Refresh();
while (completed)
{
System.Windows.Forms.Application.DoEvents();
Thread.Sleep(100);
botonClik = browser.Document.GetElementById("btnMoreResults");
botonClik.InvokeMember("click");
document = (HtmlDocument)browser.Document.DomDocument;
HtmlNode[] nodes = document.DocumentNode.SelectNodes(".//a/@href").ToArray();
Console.Write("Buscando Recetas.....");
using (var progress = new ProgressBar())
{
foreach (HtmlNode enlace in nodes)
{
foreach (HtmlAttribute atributo in enlace.Attributes)
{
if (RegexTool.isUrl(atributo.Value))
{
HtmlDocument recipe = web.Load(atributo.Value);
if (hasRecipe(recipe))
{
contador++;
progress.Report((double)contador / cantidad);
Thread.Sleep(20);
getMicroData(recipe);
}
}
}
}
}
}
}
Console.WriteLine("Listo.");
Console.ReadLine();
}
public DataSet getMicroData(HtmlDocument Document)
{
ClDataAcces objDataAccess = new ClDataAcces();
string Name = string.Empty;
string Ingredient = string.Empty;
DataSet recipeData = new DataSet();
DataTable recipe = new DataTable();
DataTable nutritionData = new DataTable();
string calories = string.Empty, fat = string.Empty, saturatefat = string.Empty,
fiber = string.Empty, carbohydrate = string.Empty, protein = string.Empty,
cholesterol = string.Empty, sugar = string.Empty, sodium = string.Empty;
// RECIPE DATA
recipe.Columns.Add("Name");
recipe.Columns.Add("Ingrediente");
// NUTRICION DATA
nutritionData.Columns.Add("Calorias", typeof(decimal));
nutritionData.Columns.Add("fat", typeof(decimal));
nutritionData.Columns.Add("saturatefat", typeof(decimal));
nutritionData.Columns.Add("fiber", typeof(decimal));
nutritionData.Columns.Add("carbohydrate", typeof(decimal));
nutritionData.Columns.Add("protein", typeof(decimal));
nutritionData.Columns.Add("cholesterol", typeof(decimal));
nutritionData.Columns.Add("sugar", typeof(decimal));
nutritionData.Columns.Add("sodium", typeof(decimal));
HtmlNode[] nodes = Document.DocumentNode.SelectNodes(".//*[@itemtype='http://schema.org/Recipe']").ToArray();
HtmlNode[] child = null;
HtmlNode NameNode = Document.DocumentNode.SelectSingleNode("/html/body/div[2]/div/div[1]/div[3]/section[1]/section[2]/div/h3");
Name = NameNode.InnerText;
foreach (HtmlNode item in nodes)
{
child = item.SelectNodes(".//*[@itemprop='ingredients']").ToArray();
Console.WriteLine("INGREDIENTES\r");
Console.WriteLine("-----------------------------------------------------------------------");
foreach (HtmlNode chldnode in child)
{
for (int ind = 0; ind < chldnode.Attributes.Count; ind++)
{
Console.WriteLine(chldnode.Attributes[ind].Name + ": " +
chldnode.Attributes[ind].Value.ToString() + " ---> " +
chldnode.InnerText.Trim() + "\r");
Ingredient = chldnode.InnerText.Trim();
recipe.NewRow();
recipe.Rows.Add(Name, Ingredient);
}
}
nodes = Document.DocumentNode.SelectNodes(".//*[@itemtype='http://schema.org/NutritionInformation']").ToArray();
Console.WriteLine("INFORMACION DE NUTRICION\r");
Console.WriteLine("-----------------------------------------------------------------------");
foreach (HtmlNode chldnode in nodes)
{
child = chldnode.SelectNodes(".//*[@itemprop]").ToArray();
foreach (HtmlNode ntNodo in child)
{
for (int ind = 0; ind < ntNodo.Attributes.Count; ind++)
{
Console.WriteLine(ntNodo.Attributes[ind].Value + ": " +
RegexTool.GetNumber(ntNodo.InnerText) + "\r");
switch (ntNodo.Attributes[ind].Value)
{
case "calories":
calories = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "fatContent":
fat = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "saturatedFatContent":
saturatefat = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "fiberContent":
fiber = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "carbohydrateContent":
carbohydrate = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "proteinContent":
protein = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "cholesterolContent":
cholesterol = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "sugarContent":
sugar = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "sodiumContent":
sodium = RegexTool.GetNumber(ntNodo.InnerText);
break;
}
}
}
}
nutritionData.Rows.Add(Convert.ToDecimal(calories), Convert.ToDecimal(fat), Convert.ToDecimal(saturatefat),
Convert.ToDecimal(fiber), Convert.ToDecimal(carbohydrate), Convert.ToDecimal(protein), Convert.ToDecimal(cholesterol)
, Convert.ToDecimal(sugar), Convert.ToDecimal(sodium));
}
recipeData.Tables.Add(recipe);
recipeData.Tables.Add(nutritionData);
objDataAccess.insertRecipeData(recipeData, false);
Console.WriteLine("Datos Registrados con exito");
return(recipeData);
}
public DataSet GetMicroData(HtmlAgilityPack.HtmlDocument Document, int dataFormat)
{
string dataError = string.Empty, pictureRoute = string.Empty, appRoute = string.Empty, filename = string.Empty;
ClDataAcces objDataAccess = new ClDataAcces();
string Name = string.Empty;
string Ingredient = string.Empty;
DataSet recipeData = new DataSet();
DataTable recipeTable = new DataTable();
DataTable nutritionData = new DataTable();
HtmlNode[] nodes = null;
HtmlNode[] child = null;
HtmlNode NameNode = null;
WebClient cliente = new WebClient();
IList <string> listIngredient = null;
Uri Url = null;
string calories = "0", fat = "0", saturatefat = "0",
fiber = "0", carbohydrate = "0", protein = "0",
cholesterol = "0", sugar = "0", sodium = "0", imagenUrl = string.Empty;
// RECIPE DATA
recipeTable.Columns.Add("Name", typeof(string));
recipeTable.Columns.Add("Ingrediente", typeof(string));
recipeTable.Columns.Add("recipeTipoPlatoData", typeof(string));
recipeTable.Columns.Add("recipeCulturaData", typeof(string));
recipeTable.Columns.Add("recipeNacionalidadData", typeof(string));
recipeTable.Columns.Add("recipeMomentoData", typeof(string));
recipeTable.Columns.Add("recipeTemporadaData", typeof(string));
recipeTable.Columns.Add("Picture", typeof(string));
recipeTable.Columns.Add("Origen", typeof(string));
recipeTable.Columns.Add("esSopa", typeof(Boolean));
recipeTable.Columns.Add("esPasta", typeof(Boolean));
recipeTable.Columns.Add("esMarisco", typeof(Boolean));
recipeTable.Columns.Add("esEnsalada", typeof(Boolean));
recipeTable.Columns.Add("esBebida", typeof(Boolean));
recipeTable.Columns.Add("esBajoColesterol", typeof(Boolean));
recipeTable.Columns.Add("esBajoEnCalorias", typeof(Boolean));
recipeTable.Columns.Add("esLibreGluten", typeof(Boolean));
// NUTRICION DATA
nutritionData.Columns.Add("Calorias", typeof(decimal));
nutritionData.Columns.Add("fat", typeof(decimal));
nutritionData.Columns.Add("saturatefat", typeof(decimal));
nutritionData.Columns.Add("fiber", typeof(decimal));
nutritionData.Columns.Add("carbohydrate", typeof(decimal));
nutritionData.Columns.Add("protein", typeof(decimal));
nutritionData.Columns.Add("cholesterol", typeof(decimal));
nutritionData.Columns.Add("sugar", typeof(decimal));
nutritionData.Columns.Add("sodium", typeof(decimal));
appRoute = Application.StartupPath;
switch (dataFormat)
{
case 0:
nodes = Document.DocumentNode.SelectNodes(".//*[@itemtype='http://schema.org/Recipe']").ToArray();
NameNode = Document.DocumentNode.SelectSingleNode(".//*[@class='recipeDetailHeader showOnTabletToDesktop']");
imagenUrl = Document.DocumentNode.Descendants("img")
.Where(node => node.Attributes["class"] != null && node.Attributes["class"].Value == "recipeDetailSummaryImageMain")
.Select(node => node.Attributes["src"].Value)
.DefaultIfEmpty(string.Empty)
.FirstOrDefault()
.ToString();
cliente.DownloadFileCompleted += new AsyncCompletedEventHandler(Cliente_DownloadFileCompleted);
Url = new Uri(imagenUrl);
filename = System.IO.Path.GetFileName(Url.LocalPath);
pictureRoute = appRoute.Remove(appRoute.Length - 10) + "\\Picture\\" + filename;
cliente.DownloadFileAsync(Url, pictureRoute);
Name = NameNode.InnerText;
foreach (HtmlNode item in nodes)
{
child = item.SelectNodes(".//*[@itemprop='ingredients']").ToArray();
foreach (HtmlNode chldnode in child)
{
for (int ind = 0; ind < chldnode.Attributes.Count; ind++)
{
Ingredient = chldnode.InnerText.Trim();
recipeTable.NewRow();
recipeTable.Rows.Add(Name, Ingredient, CboTipoPlato.Text, cboCultura.Text,
cboNacionalidad.Text, cboMomentoComida.Text, cboTemporada.Text, pictureRoute,
txtOrigenRegistro.Text, chkSopa.Checked,
chkPasta.Checked, chkPescadoMarisco.Checked, chkEnsalada.Checked,
chkBebida.Checked, chkBajoColesterol.Checked, chkBajoCalorias.Checked,
chkLibreGluten.Checked);
}
}
if (Document.DocumentNode.SelectNodes(".//*[@itemtype='http://schema.org/NutritionInformation']") != null)
{
nodes = Document.DocumentNode.SelectNodes(".//*[@itemtype='http://schema.org/NutritionInformation']").ToArray();
foreach (HtmlNode chldnode in nodes)
{
child = chldnode.SelectNodes(".//*[@itemprop]").ToArray();
foreach (HtmlNode ntNodo in child)
{
for (int ind = 0; ind < ntNodo.Attributes.Count; ind++)
{
Console.WriteLine(ntNodo.Attributes[ind].Value + ": " +
RegexTool.GetNumber(ntNodo.InnerText) + "\r");
if (ntNodo.InnerText != "")
{
switch (ntNodo.Attributes[ind].Value)
{
case "calories":
calories = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "fatContent":
fat = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "saturatedFatContent":
saturatefat = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "fiberContent":
fiber = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "carbohydrateContent":
carbohydrate = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "proteinContent":
protein = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "cholesterolContent":
cholesterol = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "sugarContent":
sugar = RegexTool.GetNumber(ntNodo.InnerText);
break;
case "sodiumContent":
sodium = RegexTool.GetNumber(ntNodo.InnerText);
break;
}
}
}
}
}
try
{
nutritionData.Rows.Add(Convert.ToDecimal(calories), Convert.ToDecimal(fat), Convert.ToDecimal(saturatefat),
Convert.ToDecimal(fiber), Convert.ToDecimal(carbohydrate), Convert.ToDecimal(protein), Convert.ToDecimal(cholesterol)
, Convert.ToDecimal(sugar), Convert.ToDecimal(sodium));
}
catch (Exception error)
{
dataError = error.Message;
}
}
}
if (nutritionData.Rows.Count > 0)
{
recipeData.Tables.Add(recipeTable);
recipeData.Tables.Add(nutritionData);
dataError = objDataAccess.insertRecipeData(recipeData, chkClasificacion.Checked);
if (dataError != "")
{
errorCount++;
lbErrorCount.Text = errorCount.ToString() + "\r MicroData scraper ";
Ftool.WriteLogFile(DateTime.Now.ToShortDateString() + " " + DateTime.Now.ToShortTimeString() + " " + dataError);
}
else
{
contRegistroEncontrados++;
lbTotalRegistroEncontrados.Text = contRegistroEncontrados.ToString();
}
}
break;
case 1:
// RECUPERANDO EN VALOR JSON EN LA PAGINA
string jsonValue = Document.DocumentNode.
SelectSingleNode("//script[@type='application/ld+json']").InnerText;
//SERIALIZANDO EL OBJETO JSON A UN OBJETO DINAMICO
dynamic jsonData = JsonConvert.DeserializeObject(jsonValue);
try
{
jsonData.nutrition.Remove("@type");
Dictionary <string, string> values = jsonData.nutrition.ToObject <Dictionary <string, string> >();
listIngredient = jsonData.recipeIngredient.ToObject <IList <string> >();
// RECUPERANDO VALORES NUTRICIONALES
foreach (KeyValuePair <string, string> valor in values)
{
switch (valor.Key)
{
case "calories":
calories = RegexTool.GetNumber(valor.Value);
break;
case "fatContent":
fat = RegexTool.GetNumber(valor.Value);
break;
case "saturatedFatContent":
saturatefat = RegexTool.GetNumber(valor.Value);
break;
case "proteinContent":
protein = RegexTool.GetNumber(valor.Value);
break;
case "carbohydrateContent":
carbohydrate = RegexTool.GetNumber(valor.Value);
break;
case "sugarContent":
sugar = RegexTool.GetNumber(valor.Value);
break;
case "sodiumContent":
sodium = RegexTool.GetNumber(valor.Value);
break;
case "fiberContent":
fiber = RegexTool.GetNumber(valor.Value);
break;
}
}
}
catch (Exception e)
{
dataError = e.Message;
lbErrorCount.Text = errorCount.ToString() + "\r Json-Ld Scraper ";
Ftool.WriteLogFile(DateTime.Now.ToShortDateString() + " " + DateTime.Now.ToShortTimeString() + " " + dataError);
}
try
{
nutritionData.Rows.Clear();
nutritionData.Rows.Add(Convert.ToDecimal(calories), Convert.ToDecimal(fat), Convert.ToDecimal(saturatefat),
Convert.ToDecimal(fiber), Convert.ToDecimal(carbohydrate), Convert.ToDecimal(protein), Convert.ToDecimal(cholesterol)
, Convert.ToDecimal(sugar), Convert.ToDecimal(sodium));
//RECUPERANDO INGREDIENTES
foreach (string ingrediente in listIngredient)
{
Ingredient = ingrediente;
Name = jsonData.name;
recipeTable.NewRow();
recipeTable.Rows.Add(Name, Ingredient, CboTipoPlato.Text, cboCultura.Text,
cboNacionalidad.Text, cboMomentoComida.Text, cboTemporada.Text, pictureRoute, txtOrigenRegistro.Text, chkSopa.Checked,
chkPasta.Checked, chkPescadoMarisco.Checked, chkEnsalada.Checked, chkBebida.Checked, chkBajoColesterol.Checked,
chkBajoCalorias.Checked, chkLibreGluten.Checked);
}
}
catch (Exception error)
{
errorCount++;
dataError = error.Message;
lbErrorCount.Text = errorCount.ToString() + "\r Json-Ld Scraper ";
Ftool.WriteLogFile(DateTime.Now.ToShortDateString() + " " + DateTime.Now.ToShortTimeString() + " " + dataError);
}
if (nutritionData.Rows.Count > 0)
{
if (nutritionData.Rows[0]["Calorias"].ToString() != "0" || nutritionData.Rows[0]["fat"].ToString() != "0" ||
nutritionData.Rows[0]["saturatefat"].ToString() != "0" || nutritionData.Rows[0]["fiber"].ToString() != "0" ||
nutritionData.Rows[0]["protein"].ToString() != "0" || nutritionData.Rows[0]["carbohydrate"].ToString() != "0" ||
nutritionData.Rows[0]["cholesterol"].ToString() != "0" || nutritionData.Rows[0]["sugar"].ToString() != "0" ||
nutritionData.Rows[0]["sodium"].ToString() != "0"
)
{
//RECUPERANDO IMAGEN DEL LA RECETA
imagenUrl = jsonData.image;
cliente.DownloadFileCompleted += new AsyncCompletedEventHandler(Cliente_DownloadFileCompleted);
if (RegexTool.isUrl(imagenUrl))
{
Url = new Uri(imagenUrl);
filename = System.IO.Path.GetFileName(Url.LocalPath);
pictureRoute = appRoute.Remove(appRoute.Length - 10) + "\\Picture\\" + filename;
cliente.DownloadFileAsync(Url, pictureRoute);
}
recipeData.Tables.Add(recipeTable);
recipeData.Tables.Add(nutritionData);
dataError = objDataAccess.insertRecipeData(recipeData, chkClasificacion.Checked);
if (dataError != "")
{
errorCount++;
lbErrorCount.Text = errorCount.ToString();
Ftool.WriteLogFile(DateTime.Now.ToShortDateString() + " " + DateTime.Now.ToShortTimeString() + " " + dataError);
}
else
{
contRegistroEncontrados++;
lbTotalRegistroEncontrados.Text = contRegistroEncontrados.ToString();
}
}
}
break;
}
return(recipeData);
}