public static ParseRule GetRule(HtmlNodeWithUrl[] nodes, string label, System.Drawing.Size minSize, bool collectIMGTags, bool collectLINKTags, bool collectMETATags)
{
ParseRule result =
GetRuleByLink(nodes, label, collectIMGTags, collectLINKTags, collectMETATags, null)
?? GetRuleByXPath(nodes, label, collectIMGTags, collectLINKTags, collectMETATags, null)
?? GetRuleByLink(nodes, label, collectIMGTags, collectLINKTags, collectMETATags, minSize)
?? GetRuleByXPath(nodes, label, collectIMGTags, collectLINKTags, collectMETATags, minSize)
?? new ParseRule();
return result;
}
private static ParseRule GetRuleByXPath(HtmlNodeWithUrl[] nodesarr, string label, bool collectIMGTags, bool collectLINKTags, bool collectMETATags, System.Drawing.Size? minSize = null)
{
ParseRule result = null;
if (nodesarr != null && nodesarr.Length > 0)
{
#region ByXPath
string mask1 = string.Empty;
string mask2 = string.Empty;
string[] xpaths = nodesarr.Select(n => n.Node.XPath).ToArray();
mask1 = LongestMaskedPathBetween(xpaths);
mask2 = LongestMaskedStringBetween(xpaths);
if (
nodesarr
.Select(n => new { Node = n.Node.OwnerDocument.DocumentNode, Url = n.Url })
.Select(nodeItem =>
{
var links = Helper
.GetAllImagesUrlsFromUrl(nodeItem.Node.OwnerDocument, nodeItem.Url.AbsoluteUri, collectIMGTags, collectLINKTags, collectMETATags, null)
.Where(n => Helper.StringLikes(n.Node.XPath, mask1));
return
minSize == null
? links.Count()
: Helper.GetAllImagesUrlsWithMinSize(links.ToArray(), minSize.Value).Count();
}
)
.Where(c => c != 1)
.Count() == 0)
result = new ParseRule()
{
Label = label,
Condition = ParseFindRuleCondition.ByXPath,
Parameter = mask1
};
if (
nodesarr
.Select(n => new { Node = n.Node.OwnerDocument.DocumentNode, Url = n.Url })
.Select(nodeItem =>
{
var links = Helper
.GetAllImagesUrlsFromUrl(nodeItem.Node.OwnerDocument, nodeItem.Url.AbsoluteUri, collectIMGTags, collectLINKTags, collectMETATags, null)
.Where(n => Helper.StringLikes(n.Node.XPath, mask2));
return
minSize == null
? links.Count()
: Helper.GetAllImagesUrlsWithMinSize(links.ToArray(), minSize.Value).Count();
}
)
.Where(c => c != 1)
.Count() == 0)
result = new ParseRule()
{
Label = label,
Condition = ParseFindRuleCondition.ByXPath,
Parameter = mask2
};
#endregion
#region ByXPathAndIndex
if (result == null)
{
string betterMask = mask2;
int index =
nodesarr
.Select(n => new { Doc = n.Node.OwnerDocument, Node = n.Node.OwnerDocument.DocumentNode, Url = n.Url })
.Select(n =>
{
var links = Helper
.GetAllImagesUrlsFromUrl(n.Node.OwnerDocument, n.Url.AbsoluteUri, collectIMGTags, collectLINKTags, collectMETATags, null)
.Where(i => Helper.StringLikes(i.Node.XPath, betterMask));
string[] images =
(minSize == null
? links.ToArray()
: Helper.GetAllImagesUrlsWithMinSize(links.ToArray(), minSize.Value)
)
.Select(i => i.Url.AbsoluteUri)
.ToArray();
for (int i = 0; i < images.Length; i++)
if (images[i].ToLower() == n.Url.AbsoluteUri.ToLower())
return i;
return -1;
}
).Distinct().OrderBy( i => i).FirstOrDefault();
if (index != -1)
result = new ParseRule()
{
Label = label,
Condition = ParseFindRuleCondition.ByXPathAndIndex,
Parameter = betterMask + ";" + index.ToString()
};
}
#endregion
}
if (result != null)
{
result.CheckImageSize = minSize != null ? true : false;
if (minSize != null)
result.MinImageSize = minSize.Value;
result.CollectIMGTags = collectIMGTags;
result.CollectLINKTags = collectLINKTags;
result.CollectMETATags = collectMETATags;
}
if (minSize != null && result == null)
{
System.Drawing.Size minCalcedSize = new System.Drawing.Size();
foreach (var sz in Helper.GetImageSizes(nodesarr.Select(n => new SomeNodeElement() { Node = n.Node, Url = n.Url }).ToArray()).Select(n => n.Value))
{
if (minCalcedSize.Width > sz.Width)
minCalcedSize.Width = sz.Width;
if (minCalcedSize.Height > sz.Height)
minCalcedSize.Height = sz.Height;
}
if (minSize.Value.Height < minCalcedSize.Height || minSize.Value.Width < minCalcedSize.Width)
result = GetRuleByLink(nodesarr, label, collectIMGTags, collectLINKTags, collectMETATags, minCalcedSize);
}
return result;
}
