支持用户自定义规则

genpac.py

@@ -19,7 +19,7 @@
 # 如果你可以正常访问gfwlistUrl，可以设置为不使用代理
 # gfwProxyType 0 不使用代理; 1 SOCKS4; 2 SOCKS5; 3 HTTP
 gfwProxyType = 2
-gfwProxyHost = '127.0.0.1'
+gfwProxyHost = '127.0.0.2'
 gfwProxyPort = 9527
 gfwProxyUsr  = None
 gfwProxyPwd  = None
@@ -40,11 +40,15 @@
 # ********************************************************************** #
 
 VERSION = '0.1a'
-DEBUGMODE = False
+DEBUGMODE = True
+
+
+gfwlistContent = ''
 
 import sys, os, base64, re
 
 def fetchGFWList():
+    global gfwlistContent
     import socks, socket, urllib2
     if (gfwProxyType == socks.PROXY_TYPE_SOCKS4) or (gfwProxyType == socks.PROXY_TYPE_SOCKS5) or (gfwProxyType == socks.PROXY_TYPE_HTTP):
         socks.setdefaultproxy(gfwProxyType, gfwProxyHost, gfwProxyPort, True, gfwProxyUsr, gfwProxyPwd)
@@ -58,10 +62,10 @@ def fetchGFWList():
 
     try:
         response = urllib2.urlopen(gfwlistUrl)
-        content = response.read()
+        gfwlistContent = response.read()
     except Exception, e:
         return False, e
-    return True, content
+    return True, None
 
 def wildcardToRegexp(pattern):
     pattern = re.sub(r"([\\\+\|\{\}\[\]\(\)\^\$\.\#])", r"\\\1", pattern);
@@ -155,26 +159,57 @@ def convertListToJSArray(list):
         array = "\n" + array + "\n" + indent;
     return "[" + array + "]"
 
-def generatePacRules(gfwlist):
-    gfwlist = base64.decodestring(gfwlist)
+def parseGFWListRules():
+    global gfwlistContent
+    gfwlist = base64.decodestring(gfwlistContent)
     if DEBUGMODE:
         with open('tmp/gfwlist.txt', 'w') as f:
             f.write(gfwlist)
 
-    directRegexpList, directWildcardList, proxyRegexpList, proxyWildcardList = parseRuleList(gfwlist)
+    return parseRuleList(gfwlist)
 
-    rules = '''    var directRegexpList   = %s;
+def parseUserRules():
+    directUserRegexpList = []
+    directUserWildcardList = []
+    proxyUserRegexpList = []
+    proxyUserWildcardList = []
+    try:
+        with open('user-rules.txt') as f:
+            directUserRegexpList, directUserWildcardList, proxyUserRegexpList, proxyUserWildcardList = parseRuleList(f.read())
+    except Exception, e:
+        pass
+
+    return directUserRegexpList, directUserWildcardList, proxyUserRegexpList, proxyUserWildcardList
+
+def generatePACRuls(userRules, gfwListRules):
+    directRegexpList, directWildcardList, proxyRegexpList, proxyWildcardList = parseGFWListRules()
+    directUserRegexpList, directUserWildcardList, proxyUserRegexpList, proxyUserWildcardList = parseUserRules()
+
+    rules = '''    //User Rules
+    var directUserRegexpList   = %s;
+    var directUserWildcardList = %s;
+    var proxyUserRegexpList    = %s;
+    var proxyUserWildcardList  = %s;
+
+    //gfwlist Rules
+    var directRegexpList   = %s;
     var directWildcardList = %s;
     var proxyRegexpList    = %s;
     var proxyWildcardList  = %s;
-    ''' % ( convertListToJSArray(directRegexpList), 
+
+    ''' % ( convertListToJSArray(directUserRegexpList), 
+            convertListToJSArray(directUserWildcardList), 
+            convertListToJSArray(proxyUserRegexpList), 
+            convertListToJSArray(proxyUserWildcardList),
+            convertListToJSArray(directRegexpList), 
             convertListToJSArray(directWildcardList), 
             convertListToJSArray(proxyRegexpList), 
             convertListToJSArray(proxyWildcardList)
           )
     return rules
 
-def CreatePacFile(gfwlist):
+
+def CreatePacFile(gfwlistRules, userRules):
     pacContent = '''/**
  * Generated by GenPAC %(ver)s
  * Author: JinnLynn http://jeeker.net
@@ -188,10 +223,33 @@ def CreatePacFile(gfwlist):
 function FindProxyForURL(url, host) {
     var P = "%(proxy)s";
     var D = "DIRECT";
+%(rules)s
     var i = 0;
     var length = 0;
-%(rules)s
-    // gfwlist Rules
+
+    length = directUserRegexpList.length;
+    for (i = 0; i < length; i++)
+    {
+        if(regExpMatch(url, directUserRegexpList[i])) return D;
+    }
+
+    length = directUserWildcardList.length;
+    for (i = 0; i < length; i++)
+    {
+        if (shExpMatch(url, directUserWildcardList[i])) return D;
+    }
+
+    length = proxyUserRegexpList.length;
+    for (i = 0; i < length; i++)
+    {
+        if(regExpMatch(url, proxyUserRegexpList[i])) return P;
+    }
+
+    length = proxyUserWildcardList.length;
+    for (i = 0; i < length; i++)
+    {
+        if(shExpMatch(url, proxyUserWildcardList[i])) return P;
+    }
     length = directRegexpList.length;
     for (i = 0; i < length; i++)
     {
@@ -219,9 +277,9 @@ def CreatePacFile(gfwlist):
     return D;
 }
 '''
-    result = { 'ver': VERSION,
-               'proxy': generateProxyVar(),
-               'rules': generatePacRules(gfwlist),
+    result = { 'ver':       VERSION,
+               'proxy':     generateProxyVar(),
+               'rules':     generatePACRuls(userRules, gfwlistRules)
               }
     with open(pacFile, 'w') as handle:
         handle.write(pacContent % result)
@@ -238,11 +296,20 @@ def CreatePacFile(gfwlist):
     os.system("rm -rf tmp/*")
 
     print "正在获取GFWList %s ..." % gfwlistUrl
-    res, content = fetchGFWList()
+    res, errorInfo = fetchGFWList()
     if res == False:
         print "GFWList获取失败，请检查相关内容是否配置正确。"
-        print "错误信息: %s" % content
+        print "错误信息: %s" % errorInfo
     else:
-        print "正在生成 %s ..." % pacFile
-        CreatePacFile(content)
-        print "一切就绪。"
+        print '正在解析 GFWList Rules ...'
+
+    # 无论gfwlist是否获取成功，都要解析，否则PAC文件有错，只是获取失败时解析的是空数据
+    gfwlistRules = parseGFWListRules()
+
+    print '正在解析 User Rules ...'
+    userRules = parseUserRules()
+
+    print "正在生成 %s ..." % pacFile
+    CreatePacFile(userRules, gfwlistRules)
+
+    print "一切就绪。"

user-rules.txt

@@ -0,0 +1,13 @@
+! 用户自定义的代理规则
+! 语法与gfwlist相同，即AdBlock Plus过滤规则
+! 简单说明如下：
+! 通配符支持，如 *.example.com/* 实际书写时可省略* 如.example.com/ 意即*.example.com/* 
+! 正则表达式支持，以\开始和结束， 如 \[\w]+:\/\/example.com\
+! 例外规则 @@，如 @@*.example.com/* 满足@@后规则的地址不使用代理
+! 匹配地址开始和结尾 |，如 |http://example.com、example.com|分别表示以http://example.com开始和以example.com结束的地址
+! || 标记，如 ||example.com 则http://example.com、https://example.com、ftp://example.com等地址均满足条件
+! 注释 ! 如 ! Comment
+!
+! 更详细说明 请访问 http://adblockplus.org/en/filters
+!
+!