public Task <bool> Process(ICrawler crawler, PropertyBag propertyBag)
{
AspectF.Define
.NotNull(crawler, nameof(crawler))
.NotNull(propertyBag, nameof(propertyBag));
if (propertyBag.StatusCode != HttpStatusCode.OK)
{
return(Task.FromResult(true));
}
if (!IsHtmlContent(propertyBag.ContentType))
{
return(Task.FromResult(true));
}
HtmlDocument htmlDoc = new HtmlDocument
{
OptionAddDebuggingAttributes = false,
OptionAutoCloseOnEnd = true,
OptionFixNestedTags = true,
OptionReadEncoding = true
};
using (MemoryStream ms = new MemoryStream(propertyBag.Response))
{
Encoding documentEncoding = htmlDoc.DetectEncoding(ms);
ms.Seek(0, SeekOrigin.Begin);
if (!documentEncoding.IsNull())
{
htmlDoc.Load(ms, documentEncoding, true);
}
else
{
htmlDoc.Load(ms, true);
}
}
string originalContent = htmlDoc.DocumentNode.OuterHtml;
if (HasTextStripRules || HasSubstitutionRules)
{
string content = StripText(originalContent);
content = Substitute(content, propertyBag.Step);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
propertyBag["HtmlDoc"].Value = htmlDoc;
HtmlNodeCollection nodes = htmlDoc.DocumentNode.SelectNodes("//title");
// Extract Title
if (!nodes.IsNull())
{
propertyBag.Title = string.Join(";", nodes.
Select(n => n.InnerText).
ToArray()).Trim();
}
// Extract Meta Data
nodes = htmlDoc.DocumentNode.SelectNodes("//meta[@content and @name]");
if (!nodes.IsNull())
{
propertyBag["Meta"].Value = (
from entry in nodes
let name = entry.Attributes["name"]
let content = entry.Attributes["content"]
where !name.IsNull() && !name.Value.IsNullOrEmpty() && !content.IsNull() && !content.Value.IsNullOrEmpty()
select $"{name.Value}: {content.Value}").ToArray();
}
// Extract text
propertyBag.Text = htmlDoc.ExtractText().Trim();
if (HasLinkStripRules || HasTextStripRules)
{
string content = StripLinks(originalContent);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
string baseUrl = propertyBag.ResponseUri.GetLeftPart(UriPartial.Path);
// Extract Head Base
nodes = htmlDoc.DocumentNode.SelectNodes("//head/base[@href]");
if (!nodes.IsNull())
{
baseUrl = nodes
.Select(entry => new { entry, href = entry.Attributes["href"] })
.Where(arg => !arg.href.IsNull() &&
!arg.href.Value.IsNullOrEmpty() &&
Uri.IsWellFormedUriString(arg.href.Value, UriKind.RelativeOrAbsolute))
.Select(t =>
{
if (Uri.IsWellFormedUriString(t.href.Value, UriKind.Relative))
{
return(propertyBag.ResponseUri.GetComponents(UriComponents.SchemeAndServer, UriFormat.Unescaped) + t.href.Value);
}
return(t.href.Value);
})
.AddToEnd(baseUrl)
.FirstOrDefault();
}
// Extract Links
DocumentWithLinks links = htmlDoc.GetLinks();
foreach (string link in links.Links.Union(links.References))
{
if (link.IsNullOrEmpty())
{
continue;
}
string decodedLink = ExtendedHtmlUtility.HtmlEntityDecode(link);
string normalizedLink = NormalizeLink(baseUrl, decodedLink);
if (normalizedLink.IsNullOrEmpty())
{
continue;
}
crawler.Crawl(new Uri(normalizedLink), propertyBag);
}
return(Task.FromResult(true));
}
public override void Process(Crawler crawler, PropertyBag propertyBag)
{
AspectF.Define.
NotNull(crawler, "crawler").
NotNull(propertyBag, "propertyBag");
if (propertyBag.StatusCode != HttpStatusCode.OK)
{
return;
}
if (!IsHtmlContent(propertyBag.ContentType))
{
return;
}
HtmlDocument htmlDoc = new HtmlDocument
{
OptionAddDebuggingAttributes = false,
OptionAutoCloseOnEnd = true,
OptionFixNestedTags = true,
OptionReadEncoding = true
};
using (Stream reader = propertyBag.GetResponse())
{
Encoding documentEncoding = htmlDoc.DetectEncoding(reader);
reader.Seek(0, SeekOrigin.Begin);
if (!documentEncoding.IsNull())
{
htmlDoc.Load(reader, documentEncoding, true);
}
else
{
htmlDoc.Load(reader, true);
}
}
string originalContent = htmlDoc.DocumentNode.OuterHtml;
if (HasTextStripRules || HasSubstitutionRules)
{
string content = StripText(originalContent);
content = Substitute(content, propertyBag.Step);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
propertyBag["HtmlDoc"].Value = htmlDoc;
HtmlNodeCollection nodes = htmlDoc.DocumentNode.SelectNodes("//title");
// Extract Title
if (!nodes.IsNull())
{
propertyBag.Title = string.Join(";", nodes.
Select(n => n.InnerText).
ToArray()).Trim();
}
// Extract Meta Data
nodes = htmlDoc.DocumentNode.SelectNodes("//meta[@content and @name]");
if (!nodes.IsNull())
{
propertyBag["Meta"].Value = (
from entry in nodes
let name = entry.Attributes["name"]
let content = entry.Attributes["content"]
where !name.IsNull() && !name.Value.IsNullOrEmpty() && !content.IsNull() && !content.Value.IsNullOrEmpty()
select name.Value + ": " + content.Value).ToArray();
}
// Extract text
propertyBag.Text = htmlDoc.ExtractText().Trim();
if (HasLinkStripRules || HasTextStripRules)
{
string content = StripLinks(originalContent);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
string baseUrl = propertyBag.ResponseUri.GetLeftPart(UriPartial.Path);
// Extract Head Base
nodes = htmlDoc.DocumentNode.SelectNodes("//head/base[@href]");
if (!nodes.IsNull())
{
baseUrl =
nodes.
Select(entry => new { entry, href = entry.Attributes["href"] }).
Where(@t => [email protected]() && [email protected]() &&
Uri.IsWellFormedUriString(@t.href.Value, UriKind.RelativeOrAbsolute)).
Select(@t => @t.href.Value).
AddToEnd(baseUrl).
FirstOrDefault();
}
// Extract Links
DocumentWithLinks links = htmlDoc.GetLinks();
foreach (string link in links.Links.Union(links.References))
{
if (link.IsNullOrEmpty())
{
continue;
}
string decodedLink = ExtendedHtmlUtility.HtmlEntityDecode(link);
try
{
string normalizedLink = NormalizeLink(baseUrl, decodedLink);
if (normalizedLink.IsNullOrEmpty())
{
continue;
}
AddStepToCrawler(crawler, propertyBag, normalizedLink, link);
}
catch (UriFormatException)
{
//When the link is not propper formatted the link mist be ignored
}
}
}
public void Process(Crawler crawler, PropertyBag propertyBag)
{
AspectF.Define.
NotNull(crawler, "crawler").
NotNull(propertyBag, "propertyBag");
if (propertyBag.StatusCode != HttpStatusCode.OK)
{
return;
}
if (!IsHtmlContent(propertyBag.ContentType))
{
return;
}
HtmlDocument htmlDoc = new HtmlDocument
{
OptionAddDebuggingAttributes = false,
OptionAutoCloseOnEnd = true,
OptionFixNestedTags = true,
OptionReadEncoding = true
};
using (MemoryStream reader = propertyBag.GetResponseStream())
{
Encoding documentEncoding = htmlDoc.DetectEncoding(reader);
reader.Seek(0, SeekOrigin.Begin);
if (!documentEncoding.IsNull())
{
htmlDoc.Load(reader, documentEncoding, true);
}
else
{
htmlDoc.Load(reader, true);
}
}
string originalContent = htmlDoc.DocumentNode.OuterHtml;
if (HasTextStripRules || HasSubstitutionRules)
{
string content = StripText(originalContent);
content = Substitute(content, propertyBag.Step);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
HtmlNodeCollection nodes = htmlDoc.DocumentNode.SelectNodes("//title");
// Extract Title
if (!nodes.IsNull())
{
propertyBag.Title = string.Join(";", nodes.
Select(n => n.InnerText).
ToArray()).Trim();
}
// Extract Meta Data
nodes = htmlDoc.DocumentNode.SelectNodes("//meta[@content and @name]");
if (!nodes.IsNull())
{
propertyBag["Meta"].Value = (
from entry in nodes
let name = entry.Attributes["name"]
let content = entry.Attributes["content"]
where !name.IsNull() && !name.Value.IsNullOrEmpty() && !content.IsNull() && !content.Value.IsNullOrEmpty()
select name.Value + ": " + content.Value).ToArray();
}
propertyBag.Text = htmlDoc.ExtractText().Trim();
if (HasLinkStripRules || HasTextStripRules)
{
string content = StripLinks(originalContent);
using (TextReader tr = new StringReader(content))
{
htmlDoc.Load(tr);
}
}
// Extract Links
DocumentWithLinks links = htmlDoc.GetLinks();
foreach (string link in links.Links.Union(links.References))
{
if (link.IsNullOrEmpty())
{
continue;
}
string baseUrl = propertyBag.ResponseUri.GetLeftPart(UriPartial.Path);
string decodedLink = ExtendedHtmlUtility.HtmlEntityDecode(link);
string normalizedLink = NormalizeLink(baseUrl, decodedLink);
if (normalizedLink.IsNullOrEmpty())
{
continue;
}
crawler.AddStep(new Uri(normalizedLink), propertyBag.Step.Depth + 1,
propertyBag.Step, new Dictionary <string, object>
{
{ Resources.PropertyBagKeyOriginalUrl, link },
{ Resources.PropertyBagKeyOriginalReferrerUrl, propertyBag.ResponseUri }
});
}
}
protected static void SaveUrlToDB(string strReturnPage, SpiderTemplate s, List <DetailLink> list)
{
Dictionary <string, string> m_dicLink2Text = new Dictionary <string, string>();
string strUrlFilterRule = s.ListPattern;
//strUrlFilterRule = ParseUrl(strUrlFilterRule);
HtmlAgilityPack.HtmlDocument htmlDoc = GetHtmlDocument(strReturnPage);
// string baseUrl = new Uri(strVisitUrl).GetLeftPart(UriPartial.Authority);
string baseUrl = GetUrlLeftPart(s.ListUrl);
DocumentWithLinks links = htmlDoc.GetLinks();
bool bNoArticle = true;
List <string> lstRevomeSame = new List <string>();
// int nCountPerPage = 0;
// bool bExistFind = false;
// List<string> lstNeedDownLoad = new List<string>();
foreach (string link in links.Links.Union(links.References))
{
if (string.IsNullOrEmpty(link))
{
continue;
}
//string decodedLink = ExtendedHtmlUtility.HtmlEntityDecode(link);
string decodedLink = link;
//if (decodedLink != link)
//{
// int a = 1;
//}
//Console.WriteLine(decodedLink);
string normalizedLink = GetNormalizedLink(baseUrl, decodedLink);
//Console.WriteLine(normalizedLink);
if (string.IsNullOrEmpty(normalizedLink))
{
continue;
}
MatchCollection matchs = Regex.Matches(normalizedLink, strUrlFilterRule, RegexOptions.Singleline);
if (matchs.Count > 0)
{
string strLinkText = "";
foreach (string strTemp in links.m_dicLink2Text.Keys)
{
if (strTemp.Contains(normalizedLink))
{
strLinkText = links.m_dicLink2Text[strTemp];
break;
}
}
//if (links.m_dicLink2Text.Keys.Contains(normalizedLink))
// strLinkText = links.m_dicLink2Text[normalizedLink];
if (strLinkText == "")
{
if (links.m_dicLink2Text.Keys.Contains(link))
{
strLinkText = links.m_dicLink2Text[link].TrimEnd().TrimStart();
}
if (links.m_dicLink2Text.Keys.Contains(link.ToLower()))
{
strLinkText = links.m_dicLink2Text[link.ToLower()].TrimEnd().TrimStart();
}
}
if (lstRevomeSame.Contains(normalizedLink))
{
continue;
}
else
{
lstRevomeSame.Add(normalizedLink);
}
//bool bRet = AddLayerNodeToSaveUrlToDB(m_strWholeDbName, normalizedLink, ref strLinkText);
DetailLink lnk = new DetailLink();
lnk.Template = s;
lnk.Url = normalizedLink;
lnk.Title = strLinkText;
list.Add(lnk);
}
//Console.WriteLine(" uri is " + normalizedLink.ToString());
}
return;
}
