public static MIMEPart ParseContent(string content)
{
MIMEPart part = ParseHeader(content);
if (content.Length > part.HeaderLength + 4)
{
part.Content = content.Substring(part.HeaderLength + 4); // 4=\r\n\r\n
}
return(part);
}
public static MHTAsset FromPart(MIMEPart part)
{
var asset = new MHTAsset();
try
{
asset.Raw = part.Content;
asset.Type = part.MediaType;
asset.Location = part.Header["Content-Location"];
asset.Encoding = part.Header["Content-Transfer-Encoding"];
} catch (Exception ex)
{
throw new Exception("不能将 MIMEPart 解析为 Asset", ex);
}
return(asset);
}
public static MHTContainer ParseContent(string content)
{
var mht = new MHTContainer();
// 解析 Header 部分
mht.HeaderPart = MIMEPart.ParseHeader(content);
content = content.Substring(mht.HeaderPart.HeaderLength + 4);
mht.Parts.Add(mht.HeaderPart);
// 获取 boundary
string boundary = mht.HeaderPart.ContentType.Boundary;
//if (boundary[0] == '"')
// boundary = boundary.Trim('"');
// 分割
//var contentParts = content.Split(new[] { boundary }, StringSplitOptions.RemoveEmptyEntries);
string[] contentParts = Regex.Split(content, "^.*?" + Regex.Escape(boundary) + ".*"
, RegexOptions.Compiled
| RegexOptions.Multiline);
contentParts = contentParts.Skip(0).ToArray();
// 解析每一部分
foreach (string strPart in contentParts)
{
string str = strPart.Trim();
if (string.IsNullOrWhiteSpace(str)) // 忽略开头和结尾的两部分
{
continue;
}
MIMEPart part = MIMEPart.ParseContent(str);
mht.Parts.Add(part);
string location;
if (part.Header.TryGetValue("Content-Location", out location))
{
mht.Locations[location] = part;
}
}
return(mht);
}
public static MIMEPart ParseHeader(string content)
{
var part = new MIMEPart();
// 获取 Header 长度
part.HeaderLength = content.IndexOf("\r\n\r\n", StringComparison.Ordinal);
if (part.HeaderLength < 0)
{
part.HeaderLength = content.Length;
}
// 获取 HeaderRaw
part.HeaderRaw = content.Substring(0, part.HeaderLength);
// 解析 Header
{
Match match = HeaderRegex.Match(part.HeaderRaw);
while (match != Match.Empty)
{
string k = match.Groups["key"].Value.Trim();
string v = match.Groups["value"].Value.Trim(); // remove white space
part.Header.Add(k, v);
match = match.NextMatch();
}
}
// 解析 Type 参数
{
string ct = part.Header["Content-Type"];
// 不知道为什么,这里不能解析 Type=text/html 这段参数
ct = Regex.Replace(ct, "type=[^;]+", "");
// 也不能包含有换行
ct = ct.Replace("\r\n\t", "");
part.ContentType = new ContentType(ct);
}
return(part);
}
public static bool IsAsset(MIMEPart part)
{
return(true);
}
