followup: minor fixes according to pr comments

This includes:
* Add set_proxy_options to README TOC.
* Make a copy of proxy options gotten from the user.
* Rewrite no_proxy matching to use ngx.re.gmatch and friends instead
  of Lua's built-in gmatch functions. Since the regular expression
  formats are completely different, the no_proxy matching process had
  to be modified a bit.
* Fix 14-host-header.t on systems with ipv6 support by disabling ipv6
  address resolution.
* Make 16-http-proxy.t more reliable by ignoring the order of the
  headers in the CONNECT request (the order is not guaranteed). The
  new check uses a regular expression to check that the CONNECT line
  is correct and that the headers include correct Host in some position.

Author: sjakthol

README.md

@@ -23,6 +23,7 @@ Production ready.
 * [new](#new)
 * [connect](#connect)
 * [connect_proxy](#connect_proxy)
+* [set_proxy_options](#set_proxy_options)
 * [set_timeout](#set_timeout)
 * [set_timeouts](#set_timeouts)
 * [ssl_handshake](#ssl_handshake)

lib/resty/http.lua

@@ -15,6 +15,8 @@ local tbl_concat = table.concat
 local tbl_insert = table.insert
 local ngx_encode_args = ngx.encode_args
 local ngx_re_match = ngx.re.match
+local ngx_re_gmatch = ngx.re.gmatch
+local ngx_re_sub = ngx.re.sub
 local ngx_re_gsub = ngx.re.gsub
 local ngx_re_find = ngx.re.find
 local ngx_log = ngx.log
@@ -958,7 +960,7 @@ function _M.proxy_response(self, response, chunksize)
 end
 
 function _M.set_proxy_options(self, opts)
-    self.proxy_opts = opts
+    self.proxy_opts = tbl_copy(opts)  -- Take by value
 end
 
 function _M.get_proxy_uri(self, scheme, host)
@@ -977,20 +979,27 @@ function _M.get_proxy_uri(self, scheme, host)
         local no_proxy_set = {}
         -- wget allows domains in no_proxy list to be prefixed by "."
         -- e.g. no_proxy=.mit.edu
-        for host_suffix in self.proxy_opts.no_proxy:gmatch("%.?([^,]+)") do
-            no_proxy_set[host_suffix] = true
+        for host_suffix in ngx_re_gmatch(self.proxy_opts.no_proxy, "\\.?([^,]+)") do
+            no_proxy_set[host_suffix[1]] = true
         end
 
 		-- From curl docs:
 		-- matched as either a domain which contains the hostname, or the
 		-- hostname itself. For example local.com would match local.com,
-		-- local.com:80, and www.local.com, but not www.notlocal.com.
-		for pos in host:gmatch("%f[^%z%.]()") do
-			local host_suffix = host:sub(pos, -1)
-			if no_proxy_set[host_suffix] then
-				return nil
-			end
-		end
+        -- local.com:80, and www.local.com, but not www.notlocal.com.
+        --
+        -- Therefore, we keep stripping subdomains from the host, compare
+        -- them to the ones in the no_proxy list and continue until we find
+        -- a match or until there's only the TLD left
+        repeat
+            if no_proxy_set[host] then
+                return nil
+            end
+
+            -- Strip the next level from the domain and check if that one
+            -- is on the list
+            host = ngx_re_sub(host, "^[^.]+\\.", "")
+        until not ngx_re_find(host, "\\.")
     end
 
     if scheme == "http" and self.proxy_opts.http_proxy then

t/14-host-header.t

@@ -12,7 +12,7 @@ $ENV{TEST_COVERAGE} ||= 0;
 our $HttpConfig = qq{
     lua_package_path "$pwd/lib/?.lua;/usr/local/share/lua/5.1/?.lua;;";
     error_log logs/error.log debug;
-    resolver 8.8.8.8;
+    resolver 8.8.8.8 ipv6=off;
 
     init_by_lua_block {
         if $ENV{TEST_COVERAGE} == 1 then

t/16-http-proxy.t

@@ -279,10 +279,8 @@ GET /lua
         }
     }
 --- tcp_listen: 12345
---- tcp_query eval_stdout
-# Note: The incoming request contains CRLF line endings and print needs to
-# be used here to get the same line breaks to the expected request
-print "CONNECT 127.0.0.1:443 HTTP/1.1\r\nUser-Agent: test_ua\r\nHost: 127.0.0.1:443\r\n\r\n"
+--- tcp_query eval
+qr/CONNECT 127.0.0.1:443 HTTP\/1.1\r\n.*Host: 127.0.0.1:443\r\n.*/s
 
 # The reply cannot be successful or otherwise the client would start
 # to do a TLS handshake with the proxied host and that we cannot