add ipv6 support

README.md

@@ -4,15 +4,15 @@ PAC scripts for proxies
 
 ## 特点
 
-基于 IP 地址白名单设计，位于白名单中的 IP 地址走直连，白名单以外的 IP 地址走代理（暂不支持 IPv6）。
+基于 IP 地址白名单设计，位于白名单中的 IP 地址走直连，白名单以外的 IP 地址走代理。
 
 另有 GFWList 版本从 [gfwlist/gfwlist](https://github.com/gfwlist/gfwlist) 获取域名及 URL 列表，优先匹配列表中的黑白名单，有效防止 DNS 污染。
 
 每周六 12:00 (UTC) 会自动使用 GitHub Actions 运行[生成脚本](build.py)从数据源获取 IP 地址列表并生成 PAC 文件。
 
 ## 使用
 
-获取方式：[本仓库的 Releases](https://github.com/iBug/pac/releases/latest)
+获取方式：[本仓库的 Releases](https://github.com/wits-fe/pac/releases/latest)
 
 - `pac-<name>.txt` 包含从数据源 `<name>` 获取的 IP 地址列表（白名单）
 - `pac-gfwlist-<name>.txt` 在 IP 白名单的基础上添加了 GFWList 的匹配

build.py

@@ -8,11 +8,16 @@
 
 import gfwlist
 
-
-SOURCES = {
+SOURCES_4 = {
     'ipdeny.com': 'http://www.ipdeny.com/ipblocks/data/aggregated/cn-aggregated.zone',
     '17mon': 'https://raw.githubusercontent.com/17mon/china_ip_list/master/china_ip_list.txt',
 }
+SOURCES_6 = {
+    'gaoyifan': 'https://gaoyifan.github.io/china-operator-ip/china6.txt',
+}
+SOURCES_46 = {
+    'maxmind': 'https://github.com/v2fly/geoip/raw/release/text/cn.txt',
+}
 OUT_DIR = "dist"
 
 # Stub content to disable GFWList check
@@ -25,6 +30,8 @@ def fetch_and_convert(src):
     template = "var CHINA = [\n{}\n];\n"
     lines = []
     for iprange in response.text.strip().split("\n"):
+        if iprange.find(":") != -1:
+            break
         ipnet = ipaddress.IPv4Network(iprange)
         netaddr = int(ipnet.network_address)
         netmask = int(ipnet.netmask)
@@ -34,6 +41,101 @@ def fetch_and_convert(src):
     return template.format("\n".join(lines))
 
 
+def fetch_and_convert_ip6(src):
+    response = requests.get(src)
+    response.raise_for_status()
+    text = response.text
+
+    template = "var CHINA6_F = [\n{}\n];\n\n"
+    template2 = "var CHINA6_S = [\n{}\n];\n"
+    lines = []
+    lines2 = []
+    lastnum = 0
+    count = 0
+    begins = False
+    ends = False
+    lower = 0
+    upper = 0
+    networkstr_b = ""
+    iprangestr_b = ""
+    fixlen = len(f"  [0xFFFFFFFF, -1, 0xFFFFFFFF],")
+
+    for iprange in text.strip().split("\n"):
+        if iprange.find(":") == -1:
+            continue
+        ipnet = ipaddress.IPv6Network(iprange)
+        prefixlen = ipnet.prefixlen
+        fulladdr = str(ipnet.exploded).replace(':', '')
+        num1 = int(fulladdr[0:8], 16)
+        num2 = int(fulladdr[8:16], 16)
+        fullmask = f"{ipnet.netmask:X}"
+        mask1 = fullmask[0:8]
+        mask2 = fullmask[8:16]
+
+        if lastnum != num1 and begins:
+            ends = True
+
+        if ends:
+            begins = False
+            ends = False
+            upper = count - 1
+            s2 = networkstr_b
+            if upper == lower:
+                s2 = iprangestr_b
+            s = f"  [0x{lastnum:08X}, {lower}, {upper}],"
+            len1 = len(s)
+            if len1 < fixlen:
+                s = s + " " * (fixlen - len1)
+            s = f"{s} // {s2}"
+            lines.append(s)
+
+        if prefixlen <= 32:
+            if begins:
+                raise NameError(f"Invalid list order: \n{iprange}")
+            s = f"  [0x{num1:08X}, -1, 0x{mask1}], // {iprange}"
+            lines.append(s)
+        else:
+            if not begins:
+                begins = True
+                lower = count
+                networkstr_b = str(ipnet.exploded)[0:10]
+                iprangestr_b = iprange
+
+            s = f"  [0x{num2:08X}, 0x{mask2}], // {count}, {iprange}"
+            lines2.append(s)
+            count = count + 1
+        lastnum = num1
+
+    if begins:
+        begins = False
+        ends = False
+        upper = count - 1
+        s2 = networkstr_b
+        if upper == lower:
+            s2 = iprangestr_b
+        s = f"  [0x{lastnum:08X}, {lower}, {upper}],"
+        len1 = len(s)
+        if len1 < fixlen:
+            s = s + " " * (fixlen - len1)
+        s = f"{s} // {s2}"
+        lines.append(s)
+
+    lines.append("  [0xFFFFFFFF, -1, 0xFFFFFFFF]  // ffff:ffff::/32 placeholder")
+    lines2.append(f"  [0xFFFFFFFF, 0xFFFFFFFF]  // {count}, placeholder, not in use")
+
+    return template.format("\n".join(lines)) + template2.format("\n".join(lines2))
+
+
+def write_pac(filename, code, data, data_6, tail):
+    with open(os.path.join(OUT_DIR, filename), "w") as f:
+        f.write(code)
+        f.write(data)
+        f.write("\n")
+        f.write(data_6)
+        f.write("\n")
+        f.write(tail)
+
+
 def main():
     now = datetime.utcnow()
     date = now.strftime("%Y%m%d")
@@ -45,27 +147,37 @@ def main():
     gfwlist_stub = GFWLIST_STUB
 
     os.makedirs(OUT_DIR, mode=0o755, exist_ok=True)
-    for key in SOURCES:
-        print(f"Generating PAC script from source {key}")
+    for key in SOURCES_4:
+        key_6 = list(SOURCES_6)[0]
+        print(f"Generating PAC script from source {key}(IPv4) & {key_6}(IPv6)")
+        try:
+            data = fetch_and_convert(SOURCES_4[key])
+            data_6 = fetch_and_convert_ip6(SOURCES_6[key_6])
+        except RequestException:
+            continue
+        except HTTPError:
+            continue
+
+        filename = f"pac-IPv4_{key}--IPv6_{key_6}.txt"
+        filename_gfwlist = f"pac-gfwlist-IPv4_{key}--IPv6_{key_6}.txt"
+        write_pac(filename, code, data, data_6, gfwlist_stub)
+        write_pac(filename_gfwlist, code, data, data_6, gfwlist_part)
+
+    for key in SOURCES_46:
+        print(f"Generating PAC script from source {key}(IPv4v6)")
         try:
-            data = fetch_and_convert(SOURCES[key])
+            data = fetch_and_convert(SOURCES_46[key])
+            data_6 = fetch_and_convert_ip6(SOURCES_46[key])
         except RequestException:
             continue
         except HTTPError:
             continue
 
-        filename = f"pac-{key}.txt"
-        filename_gfwlist = f"pac-gfwlist-{key}.txt"
-        with open(os.path.join(OUT_DIR, filename), "w") as f:
-            f.write(code)
-            f.write(data)
-            f.write("\n")
-            f.write(gfwlist_stub)
-        with open(os.path.join(OUT_DIR, filename_gfwlist), "w") as f:
-            f.write(code)
-            f.write(data)
-            f.write("\n")
-            f.write(gfwlist_part)
+        filename = f"pac-IPv4v6_{key}.txt"
+        filename_gfwlist = f"pac-gfwlist-IPv4v6_{key}.txt"
+        write_pac(filename, code, data, data_6, gfwlist_stub)
+        write_pac(filename_gfwlist, code, data, data_6, gfwlist_part)
+
 
 
 if __name__ == '__main__':

code.js

@@ -5,8 +5,100 @@
 var proxy = __PROXY__;
 var direct = "DIRECT";
 
+// lower: lower_index
+// upper: (upper_index + 1) / (array_length if upper_index = last)
+function binarySearch(list, num, lower, upper) {
+  var x = lower, y = upper, middle;
+  while (y - x > 1) {
+    middle = Math.floor((x + y) / 2);
+    if (list[middle][0] > num)
+      y = middle;
+    else
+      x = middle;
+  }
+  return x;
+}
+
+function convertToUInt6(high, low) {
+  var num1 = parseInt(high, 16) & 0xFFFF;
+  var num2 = parseInt(low, 16) & 0xFFFF;
+  return (((num1 << 16) | num2) >>> 0);
+}
+
+function isInNet6(parts, list, list2) {
+  var num = convertToUInt6(parts[0], parts[1]);
+  var x = binarySearch(list, num, 0, list.length);
+
+  if (list[x][1] == -1) 
+    return (((num & list[x][2]) ^ list[x][0]) === 0);
+
+  // not in net (/33 - /64)
+  if (num !== list[x][0]) 
+    return false;
+
+  var num2 = convertToUInt6(parts[2], parts[3]);
+  var x2 = binarySearch(list2, num2, list[x][1], list[x][2] + 1);
+
+  return (((num2 & list2[x2][1]) ^ list2[x2][0]) === 0);
+}
+
+function isLanOrChina6(host) {
+  var addr = host;
+  if (addr.indexOf("[") !== -1) {
+    addr = addr.substring(1, addr.length - 1);
+  }
+
+  if (addr.indexOf("::") === -1) {
+    var groups = addr.split(":");
+    if (groups.length != 8)
+      return false; // invalid ipv6 format
+    return isInNet6(groups, CHINA6_F, CHINA6_S) || isInNet6(groups, LAN6_F, LAN6_S);
+  }
+
+  var halfs = addr.split("::");
+  var left = halfs[0];
+  var right = halfs[1];
+  if (left.length < 1) left = "0000";
+  if (right.length < 1) right = "0000";
+
+  var groups1 = left.split(":");
+  if (groups1.length > 3)
+    return isInNet6(groups1, CHINA6_F, CHINA6_S) || isInNet6(groups1, LAN6_F, LAN6_S);
+
+  var groups2 = right.split(":");
+  var zeros = 8 - (groups1.length + groups2.length);
+  if (zeros < 2)
+    return false; // invalid ipv6 format
+
+  var parts = ["0", "0", "0", "0"];
+  parts[0] = groups1[0];
+  var i = 1;
+  for (var j = 1; j < groups1.length; j++) {
+    parts[i] = groups1[j];
+    i = i + 1;
+    if (i == 4) break;
+  }
+
+  if (i < 4) {
+    for (var k = 0; k < zeros; k++) {
+      parts[i] = "0000";
+      i = i + 1;
+      if (i == 4) break;
+    }
+  }
+  if (i == 3) parts[3] = groups2[0];
+
+  return isInNet6(parts, CHINA6_F, CHINA6_S) || isInNet6(parts, LAN6_F, LAN6_S);
+}
+
+function convertToUInt(host) {
+  var bytes = host.split(".");
+  var result = ((bytes[0] & 0xFF) << 24) | ((bytes[1] & 0xFF) << 16) | ((bytes[2] & 0xFF) <<  8) | (bytes[3] & 0xFF);
+  return (result >>> 0);
+}
+
 function belongsToSubnet(host, list) {
-  var ip = convert_addr(host) >>> 0;
+  var ip = convertToUInt(host);
 
   if (list.length === 0 || ip < list[0][0])
     return false;
@@ -83,11 +175,35 @@ function FindProxyForURL(url, host) {
   }
 
   // Fallback to IP whitelist
-  var remote = dnsResolve(host);
-  if (!remote || remote.indexOf(":") !== -1) {
-    // resolution failed or is IPv6 addr
+
+  // if host is IPv6
+  if (host.indexOf(":") !== -1) {
+    if (isLanOrChina6(host)) {
+      return direct;
+    }
+    return proxy;
+  }
+
+  var remote;
+  if(typeof dnsResolveEx == 'function') {
+    remote = dnsResolveEx(host);
+  } else {
+    remote = dnsResolve(host);
+  }
+
+  if (!remote) {
     return proxy;
+  } else {
+    remote = remote.split(";")[0];
   }
+
+  if (remote.indexOf(":") !== -1) {
+    if (isLanOrChina6(remote)) {
+      return direct;
+    }
+    return proxy;
+  }
+
   if (isLan(remote) || isChina(remote)) {
     return direct;
   }
@@ -104,3 +220,21 @@ var LAN = [
   [0xC0A80000, 0xFFFF0000]  // 192.168.0.0/16
 ];
 
+// not support /65 - /128
+var LAN6_F = [
+  [0x00000000, 0, 0],           // ::/64
+  [0x0064FF9B, 1, 2],           // 64:ff9b:
+  [0x01000000, 3, 3],           // 100::/64
+  [0x20010000, -1, 0xFFFFFFFF], // 2001::/32 - teredo, may remove
+  [0xFC000000, -1, 0xFE000000], // fc00::/7
+  [0xFE800000, -1, 0xFFC00000], // fe80::/10
+  [0xFF000000, -1, 0xFF000000]  // ff00::/8
+];
+
+var LAN6_S = [
+  [0x00000000, 0xFFFFFFFF], // 0, ::/64 - catch {::, ::1, ::ffff:0:0/96, ::ffff:0:0:0/96}, may remove
+  [0x00000000, 0xFFFFFFFF], // 1, 64:ff9b::/64 - catch {64:ff9b::/96, NAT64}, may remove
+  [0x00010000, 0xFFFF0000], // 2, 64:ff9b:1::/48 - NAT64, may remove
+  [0x00000000, 0xFFFFFFFF]  // 3, 100::/64
+];
+