/// <summary>
/// Parse application page
/// </summary>
/// <param name="entry"></param>
static GooglePlayEntry ParseGooglePlayApplicationPage(GooglePlayEntry entry)
{
using (var wc = new WebClient())
{
WebHeaderCollection headers = new WebHeaderCollection
{
[HttpRequestHeader.UserAgent] = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/534.24 (KHTML, like Gecko) Ubuntu/10.04 Chromium/11.0.696.0 Chrome/11.0.696.0 Safari/534.24"
};
wc.Encoding = Encoding.UTF8;
wc.Headers = headers;
var response = wc.DownloadString($"https://play.google.com/{entry.AppId}");
Console.WriteLine($"Parsing - {entry.AppId}");
var doc = new HtmlDocument();
doc.LoadHtml(response);
var descrHtmlNode = doc.DocumentNode.SelectSingleNode(@"//div[@class='show-more-content text-body']");
if (descrHtmlNode == null)
{
return(entry);
}
var descriptionFull = descrHtmlNode.InnerText;
var ratingHtmlNode = doc.DocumentNode.SelectSingleNode(@"//div[@class='rating-box']/div/meta");
if (ratingHtmlNode == null)
{
return(entry);
}
var rating = ratingHtmlNode.GetAttributeValue("content", "0.0");
foreach (var itemprop in new string[] { "datePublished", "numDownloads", "softwareVersion" })
{
var node = doc.DocumentNode.SelectSingleNode($"//div[@itemprop='{itemprop}']");
if (node == null)
{
continue;
}
switch (itemprop)
{
case "datePublished":
entry.Updated = node.InnerText;
break;
case "numDownloads":
entry.Installations = node.InnerText;
break;
case "softwareVersion":
entry.CurrentVersion = node.InnerText;
break;
}
}
}
return(entry);
}
static IEnumerable <GooglePlayEntry> ParseGooglePlaySearchResults(string response, string q)
{
var doc = new HtmlDocument();
doc.LoadHtml(response);
//var nodes = doc.DocumentNode.SelectNodes(@"//div[@class='card no-rationale square-cover apps small']");
var nodes = doc.DocumentNode.SelectNodes(@"//div[contains(@class, 'card no-rationale square-cover apps')]");
var entries = new List <GooglePlayEntry>();
if (nodes != null)
{
foreach (HtmlNode node in nodes)
{
try
{
var entry = new GooglePlayEntry();
entry.SearchQuery = q;
// get application
var detailsNode = node.SelectSingleNode(@"div/div[@class='details']");
var hrefNode = detailsNode.SelectSingleNode("a[@class='title']");
entry.AppId = hrefNode.GetAttributeValue("href", string.Empty);
entry.AppName = hrefNode.InnerText.Trim();
Console.WriteLine(entry.AppId);
// get description from details
var description = detailsNode.SelectSingleNode(@"div[@class='description']");
//Console.WriteLine(description.InnerText.Trim());
entry.Desc = description.InnerText.Trim();
// get developer from description
var ahref = detailsNode.SelectSingleNode(@"div[@class='subtitle-container']/a");
var href = ahref.GetAttributeValue("href", string.Empty);
entry.DevId = href;
entries.Add(entry);
//Console.WriteLine(href);
}
catch (Exception ex)
{
Console.WriteLine($"{ex.Message} while parsing {q}");
}
}
}
return(entries);
}