private static void AddNameOwnerInformation(DaycareProvider provider, HtmlDocument doc)
{
var headers = doc.DocumentNode.Descendants("h2").ToList();
provider.name = headers[0].InnerText;
provider.owner = headers[1].InnerText;
}
private static void AddContactInformation(DaycareProvider provider, HtmlDocument doc)
{
var header = doc.DocumentNode.Descendants("h4").Where(d => d.InnerText.Contains("Contact Information"));
if (header.Count() > 0)
{
var headerContent = header.First();
var sibling = headerContent.NextSibling;
while (sibling != null && sibling.Name != "h4")
{
if (sibling.InnerText.Trim().StartsWith("Email address:"))
{
provider.email = sibling.NextSibling.InnerText;
}
if (sibling.InnerText.Trim().StartsWith("Phone:"))
{
provider.phone = sibling.InnerText.Replace("Phone:", "").Trim();
}
if (sibling.InnerText.Trim().StartsWith("Web Site Link:"))
{
provider.website = sibling.NextSibling.Attributes["href"].Value;
}
sibling = sibling.NextSibling;
}
}
}
private static void AddOperationInformation(DaycareProvider provider, HtmlDocument doc)
{
var header = doc.DocumentNode.Descendants("h4").Where(d => d.InnerText.Contains("Days/Hours"));
if (header.Count() > 0)
{
var headerContent = header.First();
var sibling = headerContent.NextSibling;
while (sibling != null && sibling.Name != "h4")
{
if (sibling.InnerText.Trim().StartsWith("Days of the Week:"))
{
provider.daysOfTheWeek = sibling.InnerText.Replace("Days of the Week:", "").Trim();
}
if (sibling.InnerText.Trim().StartsWith("Hours:"))
{
provider.hours = sibling.InnerText.Replace("Hours:", "").Trim();
}
sibling = sibling.NextSibling;
}
}
}
private static void AddLocationInformation(DaycareProvider provider, HtmlDocument doc)
{
var header = doc.DocumentNode.Descendants("h4").Where(d => d.InnerText.Contains("Location:"));
if (header.Count() > 0)
{
var headerContent = header.First();
provider.location = headerContent.InnerText.Split(new char[] { ' ' }, 2).Last();
var sibling = headerContent.NextSibling;
while (sibling != null && sibling.Name != "h4")
{
if (sibling.InnerText.Trim().Length > 0 && sibling.Name != "a")
{
provider.address += sibling.InnerText.Replace("\n", "") + ", ";
}
sibling = sibling.NextSibling;
}
if (provider.address != null)
{
provider.address = provider.address.Trim().TrimEnd(',');
}
}
}
private static DaycareProvider ScrapeDaycareProvider(string url)
{
HtmlWeb hweb = new HtmlWeb();
HtmlDocument hdoc = hweb.Load(url);
var provider = new DaycareProvider();
provider.listingUrl = url;
AddNameOwnerInformation(provider, hdoc);
AddLocationInformation(provider, hdoc);
AddContactInformation(provider, hdoc);
AddOperationInformation(provider, hdoc);
return(provider);
}
