core: refactor http webclient part 7 #8529 (#7689)

Move web-result encoding logic to internal lazy evaluation
This commit is contained in:
Cory
2020-03-16 15:18:15 -05:00
committed by ngosang
parent 0f431471cd
commit 79fa3b9e0d
5 changed files with 43 additions and 143 deletions

View File

@@ -1,22 +1,58 @@
using System;
using System.Collections.Generic; using System.Collections.Generic;
using System.Net; using System.Net;
using System.Text; using System.Text;
using System.Text.RegularExpressions;
namespace Jackett.Common.Utils.Clients namespace Jackett.Common.Utils.Clients
{ {
public abstract class BaseWebResult public abstract class BaseWebResult
{ {
public Encoding Encoding { get; set; } private Encoding _encoding;
public Encoding Encoding
{
get
{
if (_encoding != null)
return _encoding;
if (Request.Encoding != null)
_encoding = Request.Encoding;
else if (Headers.ContainsKey("content-type"))
{
var charsetRegexMatch = Regex.Match(Headers["content-type"][0], @"charset=([\w-]+)", RegexOptions.Compiled);
if (charsetRegexMatch.Success)
{
var charset = charsetRegexMatch.Groups[1].Value;
try
{
_encoding = Encoding.GetEncoding(charset);
}
catch (ArgumentException)
{
// Encoding not found or not enabled on current machine.
}
}
}
_encoding ??= Encoding.UTF8;
return _encoding;
}
set => _encoding = value;
}
public HttpStatusCode Status { get; set; } public HttpStatusCode Status { get; set; }
public string Cookies { get; set; } public string Cookies { get; set; }
public string RedirectingTo { get; set; } public string RedirectingTo { get; set; }
public WebRequest Request { get; set; } public WebRequest Request { get; set; }
public Dictionary<string, string[]> Headers = new Dictionary<string, string[]>(); public Dictionary<string, string[]> Headers { get; protected set; } =
new Dictionary<string, string[]>(StringComparer.OrdinalIgnoreCase);
public bool IsRedirect => Status == System.Net.HttpStatusCode.Redirect || public bool IsRedirect => Status == HttpStatusCode.Redirect ||
Status == System.Net.HttpStatusCode.RedirectKeepVerb || Status == HttpStatusCode.RedirectKeepVerb ||
Status == System.Net.HttpStatusCode.RedirectMethod || Status == HttpStatusCode.RedirectMethod ||
Status == System.Net.HttpStatusCode.Found || Status == HttpStatusCode.Found ||
Status == System.Net.HttpStatusCode.MovedPermanently; Status == HttpStatusCode.MovedPermanently;
} }
} }

View File

@@ -294,40 +294,6 @@ namespace Jackett.Common.Utils.Clients
result.Cookies = cookieBuilder.ToString().Trim(); result.Cookies = cookieBuilder.ToString().Trim();
} }
ServerUtil.ResureRedirectIsFullyQualified(webRequest, result); ServerUtil.ResureRedirectIsFullyQualified(webRequest, result);
Encoding encoding = null;
if (webRequest.Encoding != null)
{
encoding = webRequest.Encoding;
}
else if (result.Headers.ContainsKey("content-type"))
{
var CharsetRegex = new Regex(@"charset=([\w-]+)", RegexOptions.Compiled);
var CharsetRegexMatch = CharsetRegex.Match(result.Headers["content-type"][0]);
if (CharsetRegexMatch.Success)
{
var charset = CharsetRegexMatch.Groups[1].Value;
try
{
encoding = Encoding.GetEncoding(charset);
}
catch (Exception ex)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Error loading encoding {2} based on header {3}: {4}", ClientType, webRequest.Url, charset, result.Headers["content-type"][0], ex));
}
}
else
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Got header without charset: {2}", ClientType, webRequest.Url, result.Headers["content-type"][0]));
}
}
if (encoding == null)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): No encoding detected, defaulting to UTF-8", ClientType, webRequest.Url));
encoding = Encoding.UTF8;
}
result.Encoding = encoding;
return result; return result;
} }
} }

View File

@@ -314,40 +314,6 @@ namespace Jackett.Common.Utils.Clients
result.Cookies = cookieBuilder.ToString().Trim(); result.Cookies = cookieBuilder.ToString().Trim();
} }
ServerUtil.ResureRedirectIsFullyQualified(webRequest, result); ServerUtil.ResureRedirectIsFullyQualified(webRequest, result);
Encoding encoding = null;
if (webRequest.Encoding != null)
{
encoding = webRequest.Encoding;
}
else if (result.Headers.ContainsKey("content-type"))
{
var CharsetRegex = new Regex(@"charset=([\w-]+)", RegexOptions.Compiled);
var CharsetRegexMatch = CharsetRegex.Match(result.Headers["content-type"][0]);
if (CharsetRegexMatch.Success)
{
var charset = CharsetRegexMatch.Groups[1].Value;
try
{
encoding = Encoding.GetEncoding(charset);
}
catch (Exception ex)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Error loading encoding {2} based on header {3}: {4}", ClientType, webRequest.Url, charset, result.Headers["content-type"][0], ex));
}
}
else
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Got header without charset: {2}", ClientType, webRequest.Url, result.Headers["content-type"][0]));
}
}
if (encoding == null)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): No encoding detected, defaulting to UTF-8", ClientType, webRequest.Url));
encoding = Encoding.UTF8;
}
result.Encoding = encoding;
return result; return result;
} }

View File

@@ -314,40 +314,6 @@ namespace Jackett.Common.Utils.Clients
result.Cookies = cookieBuilder.ToString().Trim(); result.Cookies = cookieBuilder.ToString().Trim();
} }
ServerUtil.ResureRedirectIsFullyQualified(webRequest, result); ServerUtil.ResureRedirectIsFullyQualified(webRequest, result);
Encoding encoding = null;
if (webRequest.Encoding != null)
{
encoding = webRequest.Encoding;
}
else if (result.Headers.ContainsKey("content-type"))
{
var CharsetRegex = new Regex(@"charset=([\w-]+)", RegexOptions.Compiled);
var CharsetRegexMatch = CharsetRegex.Match(result.Headers["content-type"][0]);
if (CharsetRegexMatch.Success)
{
var charset = CharsetRegexMatch.Groups[1].Value;
try
{
encoding = Encoding.GetEncoding(charset);
}
catch (Exception ex)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Error loading encoding {2} based on header {3}: {4}", ClientType, webRequest.Url, charset, result.Headers["content-type"][0], ex));
}
}
else
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Got header without charset: {2}", ClientType, webRequest.Url, result.Headers["content-type"][0]));
}
}
if (encoding == null)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): No encoding detected, defaulting to UTF-8", ClientType, webRequest.Url));
encoding = Encoding.UTF8;
}
result.Encoding = encoding;
return result; return result;
} }

View File

@@ -297,40 +297,6 @@ namespace Jackett.Common.Utils.Clients
result.Cookies = cookieBuilder.ToString().Trim(); result.Cookies = cookieBuilder.ToString().Trim();
} }
ServerUtil.ResureRedirectIsFullyQualified(webRequest, result); ServerUtil.ResureRedirectIsFullyQualified(webRequest, result);
Encoding encoding = null;
if (webRequest.Encoding != null)
{
encoding = webRequest.Encoding;
}
else if (result.Headers.ContainsKey("content-type"))
{
var CharsetRegex = new Regex(@"charset=([\w-]+)", RegexOptions.Compiled);
var CharsetRegexMatch = CharsetRegex.Match(result.Headers["content-type"][0]);
if (CharsetRegexMatch.Success)
{
var charset = CharsetRegexMatch.Groups[1].Value;
try
{
encoding = Encoding.GetEncoding(charset);
}
catch (Exception ex)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Error loading encoding {2} based on header {3}: {4}", ClientType, webRequest.Url, charset, result.Headers["content-type"][0], ex));
}
}
else
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): Got header without charset: {2}", ClientType, webRequest.Url, result.Headers["content-type"][0]));
}
}
if (encoding == null)
{
logger.Error(string.Format("WebClient({0}).GetString(Url:{1}): No encoding detected, defaulting to UTF-8", ClientType, webRequest.Url));
encoding = Encoding.UTF8;
}
result.Encoding = encoding;
return result; return result;
} }
} }