/// <summary>
/// 净化富文本
/// </summary>
/// <param name="html">HTML 代码</param>
/// <returns>净化后的 HTML 代码</returns>
public static string SanitizeRichText(string html)
{
Config.HtmlEncoder = new HtmlEncoderNone();
Config.OutputFormatter = OutputFormatters.HtmlEncodingNone;
var sanitizer = new HtmlSanitizer(
new[]
{
"br", "span", "a", "img", "b", "strong", "i", "strike", "u", "p", "blockquote", "h1", "hr",
"comment", "spoiler", "table", "colgroup", "col", "thead", "tr", "th", "tbody", "td"
},
null,
new[] { "src", "alt", "width", "height", "data-non-image", "href", "style" },
null,
new[] { "text-align" });
var dom = CQ.Create(sanitizer.Sanitize(html));
foreach (var img in dom["img"])
{
if (string.IsNullOrWhiteSpace(img.Attributes["src"]))
{
img.Remove();
}
else
{
var fileName = UpyunProvider.ExtractFileName(img.Attributes["src"]);
if (string.IsNullOrWhiteSpace(fileName))
{
continue;
}
img.Attributes["article-image-src"] = $"keylol://{fileName}";
img.RemoveAttribute("src");
}
}
return(dom.Render());
}
/// <summary>
/// 如果图片属于自有存储，则提取出来替换 Schema，否则返回原图
/// </summary>
/// <param name="coverImage">封面图</param>
/// <returns>净化后的封面图</returns>
public static string SanitizeCoverImage(string coverImage)
{
var fileName = UpyunProvider.ExtractFileName(coverImage);
return(string.IsNullOrWhiteSpace(fileName) ? coverImage : $"keylol://{fileName}");
}
private async Task <string> UploadFromUrlAsync([NotNull] string url)
{
try
{
url = Regex.Replace(url, @"^\/\/", "http://");
if (!Regex.IsMatch(url, @"^https?:\/\/"))
{
url = $"http://{url}";
}
var request = WebRequest.CreateHttp(url);
request.Referer = url;
request.UserAgent =
"Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/48.0.2564.116 Safari/537.36";
request.Accept = "image/webp,image/*,*/*;q=0.8";
request.Headers["Accept-Language"] = "en-US,en;q=0.8,zh-CN;q=0.6,zh;q=0.4";
request.Timeout = 10000;
request.ReadWriteTimeout = 60000;
using (var response = await request.GetResponseAsync())
using (var ms = new MemoryStream(response.ContentLength > 0 ? (int)response.ContentLength : 0))
{
var extension = MimeTypeToFileExtension(response.ContentType?.Split(';')[0]);
if (extension == null) // 不支持的类型
{
_logger.Warn($"Unsupported MIME type: {url}");
return(null);
}
if (response.ContentLength > UpyunProvider.MaxImageSize)
{
_logger.Warn($"Image (Content-Length) is too large: {url}");
return(null);
}
var responseStream = response.GetResponseStream();
if (responseStream == null)
{
_logger.Warn($"Null response stream: {url}");
return(null);
}
await responseStream.CopyToAsync(ms);
var fileData = ms.ToArray();
if (fileData.Length <= 0)
{
_logger.Warn($"Empty response stream: {url}");
return(null);
}
if (fileData.Length > UpyunProvider.MaxImageSize)
{
_logger.Warn($"Image (response stream length) is too large: {url}");
return(null);
}
var name = await UpyunProvider.UploadFile(fileData, extension);
if (string.IsNullOrWhiteSpace(name))
{
_logger.Warn($"Upload failed: {url}");
return(null);
}
return($"keylol://{name}");
}
}
catch (Exception e)
{
_logger.Warn($"Download failed: {url}", e);
return(null);
}
}
