volta: parser.rl@51eb85ae4de4 (annotated)

0 eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	1	/* vim: set noet nosta sw=4 ts=4 ft=ragel : */
2 8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	2	/*
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	3	Copyright (c) 2011, Mahlon E. Smith <mahlon@martini.nu>
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	4	All rights reserved.
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	5	Redistribution and use in source and binary forms, with or without
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	6	modification, are permitted provided that the following conditions are met:
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	7
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	8	* Redistributions of source code must retain the above copyright
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	9	notice, this list of conditions and the following disclaimer.
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	10
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	11	* Redistributions in binary form must reproduce the above copyright
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	12	notice, this list of conditions and the following disclaimer in the
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	13	documentation and/or other materials provided with the distribution.
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	14
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	15	* Neither the name of Mahlon E. Smith nor the names of his
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	16	contributors may be used to endorse or promote products derived
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	17	from this software without specific prior written permission.
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	18
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	19	THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND ANY
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	20	EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	21	WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	22	DISCLAIMED. IN NO EVENT SHALL THE REGENTS AND CONTRIBUTORS BE LIABLE FOR ANY
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	23	DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	24	(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	25	LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	26	ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	27	(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	28	SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	29	*/
8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	30
0 eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	31	#include "volta.h"
eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	32
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	33	#define MARK_S( LBL ) p_parsed->tokens.LBL ## _start = p;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	34	#define MARK_E( LBL ) p_parsed->tokens.LBL ## _length = p - ( *pe + p_parsed->tokens.LBL ## _start );
1 823d42546cea Dial in the Makefile and command line option parsing. Better debug Mahlon E. Smith <mahlon@laika.com> parents: 0 diff changeset	35
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	36	/*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	37	* Tokenize an incoming line from squid, returning a parsed and populated
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	38	* structure to make redirection decisions against. This pointer should
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	39	* be freed using finish_parsed() after use.
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	40	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	41	* Squid documentation about redirectors:
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	42	* ---------------------------------------------------------------------------
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	43	* TAG: url_rewrite_program
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	44	* Specify the location of the executable for the URL rewriter.
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	45	* Since they can perform almost any function there isn't one included.
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	46	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	47	* For each requested URL rewriter will receive on line with the format
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	48	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	49	* URL <SP> client_ip "/" fqdn <SP> user <SP> method [<SP> kvpairs]<NL>
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	50	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	51	* In the future, the rewriter interface will be extended with
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	52	* key=value pairs ("kvpairs" shown above). Rewriter programs
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	53	* should be prepared to receive and possibly ignore additional
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	54	* whitespace-separated tokens on each input line.
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	55	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	56	* And the rewriter may return a rewritten URL. The other components of
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	57	* the request line does not need to be returned (ignored if they are).
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	58	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	59	* The rewriter can also indicate that a client-side redirect should
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	60	* be performed to the new URL. This is done by prefixing the returned
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	61	* URL with "301:" (moved permanently) or 302: (moved temporarily).
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	62	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	63	* By default, a URL rewriter is not used.
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	64	* ---------------------------------------------------------------------------
1 823d42546cea Dial in the Makefile and command line option parsing. Better debug Mahlon E. Smith <mahlon@laika.com> parents: 0 diff changeset	65	*/
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	66	parsed *
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	67	parse_request( char *line )
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	68	{
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	69	/* machine required vars */
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	70	unsigned short int cs = 1;
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	71	char *p = line;
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	72	char *pe = p + strlen(p);
13 23a242d7b7fa 1st iteration of volta actually doing something. Process the request, Mahlon E. Smith <mahlon@laika.com> parents: 12 diff changeset	73	char *eof = pe;
2 8c88756f81b0 Ensure that parsing can't be subverted by requests larger than the Mahlon E. Smith <mahlon@martini.nu> parents: 1 diff changeset	74
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	75	/* the client request pointer */
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	76	parsed *p_parsed = init_parsed();
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	77	p_parsed->type = REQUEST;
0 eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	78
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	79	%%{
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	80	machine request_parser;
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	81
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	82	action channel_id_found {
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	83	debug( 1, LOC, "Channel ID found in redirector input. Set 'url_rewrite_concurrency' to '0' in squid.\n" );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	84	fbreak;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	85	}
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	86
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	87	action scheme_start { MARK_S(scheme) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	88	action scheme_finish { MARK_E(scheme) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	89	action host_start { MARK_S(host) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	90	action host_finish { MARK_E(host) }
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	91	action port_start { p_parsed->tokens.port_start = p+1; } # strip leading colon
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	92	action port_finish { MARK_E(port) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	93	action path_start { MARK_S(path) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	94	action path_finish { MARK_E(path) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	95	action meth_start { MARK_S(meth) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	96	action meth_finish { MARK_E(meth) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	97	action c_ip_start { MARK_S(c_ip) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	98	action c_ip_finish { MARK_E(c_ip) }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	99
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	100	action host_error { debug( 3, LOC, "Unable to parse the request hostname.\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	101	action scheme_error { debug( 3, LOC, "Unable to parse the request scheme.\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	102	action meth_error { debug( 3, LOC, "Unable to parse the request method.\n" ); }
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	103	action c_ip_error { debug( 3, LOC, "Unable to parse the client IP address.\n" ); }
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	104
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	105	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	106	# Squid line: URL <SP> client_ip "/" fqdn <SP> user <SP> method [<SP> kvpairs]<NL>
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	107	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	108	# URI Syntax (RFC 3986) misc notes:
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	109	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	110	# - Scheme isn't passed to redirectors on CONNECT method requests
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	111	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	112	# - Hostname segments aren't supposed to be individually greater than 63 chars,
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	113	# and the hostname in total shouldn't exceed 255. They also shouldn't be entirely
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	114	# made up of digits, or contain underscores. In practice, these rules appear to
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	115	# be violated constantly by some pretty big sites.
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	116	# (( alnum ) \| ( alnum . [a-zA-Z0-9\-]{0,63} . alnum )) & !( digit+ );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	117	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	118	# - ipv6 has some utterly insane rules (RFC 5952) in the name of "shortcuts", which
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	119	# only seem like shortcuts to someone writing IP addresses by hand. Anyone that
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	120	# has to parse (or even just read) them has a bunch of seemingly arbitrary work
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	121	# dumped in their lap. Heck, it's impossible to even search for an ipv6 address
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	122	# that contains zeros in a text editor, because you have no idea what how it might
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	123	# be represented. Rad!
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	124	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	125	# The parser just trusts any ipv6 squid hands us as being valid, without
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	126	# any real parsing/validation, other than it consists of hex digits and colons.
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	127	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	128	# - This parser originally validated path/query/fragment as well, but there were
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	129	# enough inconsistencies with unescaped chars and other real-life RFC deviations
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	130	# that I opted to just accept what we get from squid.
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	131	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	132	# - Redirectors aren't handed any userinfo (http://mahlon:password@example.com),
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	133	# so no need to check for that.
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	134	#
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	135
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	136	host_component = alnum \| ( alnum [a-zA-Z0-9\-_]* alnum );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	137	pchar = ( alnum \| [\-._~!$%&'()*+,;=] );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	138	path_segment = '/' ( any - space )*;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	139
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	140	hostname = host_component ( '.' host_component )* '.'?;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	141	ipv4 = digit{1,3} '.' digit{1,3} '.' digit{1,3} '.' digit{1,3};
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	142	ipv6 = ( xdigit \| ':' )+;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	143
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	144	channel_id = ( digit+ space ) %channel_id_found;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	145	scheme = ( alpha{3,5} '://' ) >scheme_start %scheme_finish @!scheme_error;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	146	host = ( hostname \| ipv4 ) >host_start %host_finish @!host_error;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	147	port = ( ':' digit{1,5} ) >port_start %port_finish;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	148	path = path_segment* >path_start %path_finish;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	149	client_ip = ipv4 >c_ip_start %c_ip_finish @!c_ip_error;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	150	method = upper+ >meth_start %meth_finish @!meth_error;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	151
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	152	SquidLine = (
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	153	start: channel_id? -> Url,
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	154	Url: scheme? host port? path? -> Client,
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	155	Client: space client_ip '/' ( hostname \| '-' ) -> User,
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	156	User: space pchar+ -> Method,
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	157	Method: space method -> KVPairs,
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	158	KVPairs: ( space any+ )? -> final
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	159	);
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	160
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	161	main := SquidLine '\n';
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	162	}%%
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	163
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	164	/* state machine */
0 eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	165	%% write exec;
eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	166
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	167	/* If we were given an invalid line, bail early */
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	168	if ( cs < %%{ write first_final; }%% ) {
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	169	free( p_parsed ), p_parsed = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	170	debug( 3, LOC, "Invalid request line (%d), skipped\n", v.timer.lines + 1 );
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	171	debug( 4, LOC, "%s", line );
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	172	return( NULL );
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	173	}
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	174
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	175	debug( 6, LOC, "%s", line );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	176	(void)populate_parsed( p_parsed );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	177	return( p_parsed );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	178	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	179
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	180
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	181	/*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	182	* Tokenize a value string from a successful database lookup, returning a parsed
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	183	* and populated structure. This pointer should be freed using finish_parsed() after use.
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	184	*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	185	*/
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	186	parsed *
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	187	parse_rule( char *rewrite )
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	188	{
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	189	/* machine required vars */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	190	unsigned short int cs = 1;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	191	char *p = rewrite;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	192	char *pe = p + strlen(p);
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	193	char *eof = pe;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	194
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	195	/* the client rule pointer */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	196	parsed *p_parsed = init_parsed();
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	197	p_parsed->type = RULE;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	198
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	199	%%{
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	200	machine rule_parser;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	201
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	202	action match_start { MARK_S(path_re) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	203	action match_finish { MARK_E(path_re) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	204	action redir_start { MARK_S(redir) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	205	action redir_finish { p_parsed->tokens.redir_length = 3; } # strip trailing colon
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	206
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	207	action scheme_start { MARK_S(scheme) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	208	action scheme_finish { MARK_E(scheme) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	209	action host_start { MARK_S(host) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	210	action host_finish { MARK_E(host) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	211	action port_start { p_parsed->tokens.port_start = p+1; } # strip leading colon
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	212	action port_finish { MARK_E(port) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	213	action path_start { MARK_S(path) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	214	action path_finish { MARK_E(path) }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	215
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	216	action match_error { debug( 3, LOC, "Unable to parse the rule path matcher.\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	217	action host_error { debug( 3, LOC, "Unable to parse the rule hostname.\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	218
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	219	host_component = alnum \| ( alnum [a-zA-Z0-9\-_]* alnum );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	220	path_segment = '/' ( any - space )*;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	221
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	222	sep = space+;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	223	hostname = host_component ( '.' host_component )* '.'?;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	224	ipv4 = digit{1,3} '.' digit{1,3} '.' digit{1,3} '.' digit{1,3};
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	225	ipv6 = ( xdigit \| ':' )+;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	226
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	227	path_re = ( any - space )+ >match_start %match_finish @!match_error;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	228	redir = ( digit{3} ':' ) >redir_start %redir_finish;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	229
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	230	scheme = ( alpha{3,5} '://' ) >scheme_start %scheme_finish;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	231	host = ( hostname \| ipv4 ) >host_start %host_finish @!host_error;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	232	port = ( ':' digit{1,5} ) >port_start %port_finish;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	233	path = path_segment* >path_start %path_finish;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	234
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	235	main := path_re sep ( redir? scheme? host port? path? );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	236	}%%
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	237
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	238	/* state machine */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	239	%% write exec;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	240
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	241	/* If we were given an invalid rule, bail early */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	242	if ( cs < %%{ write first_final; }%% ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	243	free( p_parsed ), p_parsed = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	244	debug( 3, LOC, "Invalid rule\n" );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	245	debug( 4, LOC, "%s\n", rewrite );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	246	return( NULL );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	247	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	248
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	249	(void)populate_parsed( p_parsed );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	250	return( p_parsed );
0 eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	251	}
eac7211fe522 Initial commit of an experimental little Squid redirector. Mahlon E. Smith <mahlon@martini.nu> parents: diff changeset	252
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	253
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	254	/*
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	255	* Tokenize a line from an ascii representation of the database, returning
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	256	* a pointer to a parsed struct. Used for creation of a new cdb file,
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	257	* validating data prior to use.
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	258	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	259	*/
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	260	struct db_input *
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	261	parse_dbinput( char *line )
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	262	{
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	263	/* machine required vars */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	264	unsigned short int cs = 1;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	265	char *p = line;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	266	char *pe = p + strlen(p);
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	267	char *eof = pe;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	268
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	269	/* the db line input pointer */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	270	struct db_input *dbline = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	271	if ( (dbline = malloc( sizeof(struct db_input) )) == NULL ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	272	debug( 5, LOC, "Unable to allocate memory for db input: %s\n", strerror(errno) );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	273	return( NULL );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	274	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	275	dbline->klen = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	276	dbline->vlen = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	277	dbline->kstart = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	278	dbline->key = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	279	dbline->vstart = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	280	dbline->val = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	281
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	282	%%{
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	283	machine dbinput_parser;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	284
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	285	action key_start { dbline->kstart = p; }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	286	action key_finish { dbline->klen = p - ( *pe + dbline->kstart ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	287	action key_error { debug( 0, LOC, "Invalid key format\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	288	action val_start { dbline->vstart = p; }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	289	action val_finish { dbline->vlen = p - ( *pe + dbline->vstart ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	290	action val_error { debug( 0, LOC, "Invalid rewrite value\n" ); }
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	291
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	292	sep = space+;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	293	host_component = alnum \| ( alnum [a-zA-Z0-9\-_]* alnum );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	294	hostname = host_component ( '.' host_component )* '.'?;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	295	ipv4 = digit{1,3} '.' digit{1,3} '.' digit{1,3} '.' digit{1,3};
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	296	token = ( any - space )+;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	297	redir = ( digit{3} ':' );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	298	host = ( hostname \| ipv4 );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	299
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	300	key = ( host \| '*' ) >key_start %key_finish @!key_error;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	301	val = ( token sep token ) >val_start %val_finish @!val_error;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	302
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	303	main:= key sep val '\n';
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	304	}%%
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	305
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	306	/* state machine */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	307	%% write exec;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	308
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	309	/* If the input line was invalid, bail early */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	310	if ( cs < %%{ write first_final; }%% ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	311	free( dbline ), dbline = NULL;
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	312	return( NULL );
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	313	}
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	314
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	315	/* populate struct */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	316	dbline->key = copy_string_token( dbline->kstart, dbline->klen );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	317	dbline->val = copy_string_token( dbline->vstart, dbline->vlen );
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	318
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	319	/* check the val to make sure it is a valid rewrite rule */
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	320	parsed *valstr = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	321	valstr = parse_rule( dbline->val );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	322	if ( valstr == NULL ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	323	free( dbline->key );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	324	free( dbline->val );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	325	free( dbline );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	326	return( NULL );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	327	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	328	finish_parsed( valstr );
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	329
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	330	return( dbline );
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	331	}
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	332
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	333
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	334
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	335	/*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	336	* Initialize and return a pointer to a new parser object.
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	337	*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	338	*/
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	339	parsed *
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	340	init_parsed( void )
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	341	{
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	342	parsed *p_parsed = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	343	if ( (p_parsed = malloc( sizeof(parsed) )) == NULL ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	344	debug( 5, LOC, "Unable to allocate memory for parsed struct: %s\n", strerror(errno) );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	345	return( NULL );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	346	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	347
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	348	p_parsed->type = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	349	p_parsed->path_re = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	350	p_parsed->redir = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	351	p_parsed->scheme = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	352	p_parsed->host = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	353	p_parsed->tld = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	354	p_parsed->port = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	355	p_parsed->path = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	356	p_parsed->user = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	357	p_parsed->method = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	358	p_parsed->client_ip = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	359
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	360	p_parsed->tokens.path_re_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	361	p_parsed->tokens.redir_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	362	p_parsed->tokens.scheme_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	363	p_parsed->tokens.host_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	364	p_parsed->tokens.port_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	365	p_parsed->tokens.path_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	366	p_parsed->tokens.meth_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	367	p_parsed->tokens.c_ip_start = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	368	p_parsed->tokens.path_re_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	369	p_parsed->tokens.redir_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	370	p_parsed->tokens.scheme_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	371	p_parsed->tokens.host_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	372	p_parsed->tokens.port_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	373	p_parsed->tokens.path_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	374	p_parsed->tokens.meth_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	375	p_parsed->tokens.c_ip_length = 0;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	376
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	377	return p_parsed;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	378	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	379
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	380
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	381	/*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	382	* Release memory used by the parsed struct.
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	383	*
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	384	*/
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	385	void
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	386	finish_parsed( parsed *p_parsed )
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	387	{
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	388	if ( p_parsed == NULL ) return;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	389
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	390	free( p_parsed->scheme );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	391	free( p_parsed->host );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	392	free( p_parsed->path );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	393	free( p_parsed->port );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	394
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	395	if ( p_parsed->type == REQUEST ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	396	free( p_parsed->tld );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	397	free( p_parsed->method );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	398	free( p_parsed->client_ip );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	399	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	400
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	401	if ( p_parsed->type == RULE ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	402	free( p_parsed->path_re );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	403	free( p_parsed->redir );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	404	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	405
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	406	free( p_parsed ), p_parsed = NULL;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	407
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	408	return;
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	409	}
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	410
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	411
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	412	#define COPY_STR( LBL ) copy_string_token( p_parsed->tokens.LBL ## _start, p_parsed->tokens.LBL ## _length )
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	413	/* #define COPY_IP4( LBL ) copy_ipv4_token( p_request->tokens.LBL ## _start, p_request->tokens.LBL ## _length ) */
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	414
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	415	/*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	416	* Take the previously parsed token locations and copy them into the request struct.
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	417	*
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	418	*/
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	419	void
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	420	populate_parsed( parsed *p_parsed )
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	421	{
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	422	p_parsed->scheme = COPY_STR( scheme );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	423	p_parsed->host = COPY_STR( host );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	424	p_parsed->path = COPY_STR( path );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	425	p_parsed->port = COPY_STR( port );
13 23a242d7b7fa 1st iteration of volta actually doing something. Process the request, Mahlon E. Smith <mahlon@laika.com> parents: 12 diff changeset	426
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	427	if ( p_parsed->type == REQUEST ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	428	p_parsed->method = COPY_STR( meth );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	429	p_parsed->client_ip = COPY_STR( c_ip );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	430	/* p_request->client_ip = COPY_IP4( c_ip ); */
13 23a242d7b7fa 1st iteration of volta actually doing something. Process the request, Mahlon E. Smith <mahlon@laika.com> parents: 12 diff changeset	431
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	432	(void)lowercase_str( p_parsed->host, p_parsed->tokens.host_length );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	433	(void)parse_tld( p_parsed );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	434	}
13 23a242d7b7fa 1st iteration of volta actually doing something. Process the request, Mahlon E. Smith <mahlon@laika.com> parents: 12 diff changeset	435
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	436	if ( p_parsed->type == RULE ) {
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	437	p_parsed->path_re = COPY_STR( path_re );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	438	p_parsed->redir = COPY_STR( redir );
51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	439	}
10 d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	440
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	441	return;
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	442	}
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	443
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	444
d07309450285 Get the ragel line parser properly tokenizing the input lines. Add a Mahlon E. Smith <mahlon@martini.nu> parents: 2 diff changeset	445	/*
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	446	* Pull the top level domain out of the requested hostname.
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	447	*
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	448	*/
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	449	void
14 51eb85ae4de4 There isn't a fast way to look up ( value exists or null ) for every Mahlon E. Smith <mahlon@martini.nu> parents: 13 diff changeset	450	parse_tld( parsed *p_request )
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	451	{
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	452	unsigned short int cs = 5, mark = 0;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	453	char *p = p_request->host;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	454	char *pe = p + p_request->tokens.host_length;
13 23a242d7b7fa 1st iteration of volta actually doing something. Process the request, Mahlon E. Smith <mahlon@laika.com> parents: 12 diff changeset	455	char ts = 0, te = 0, *eof = pe;
12 191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	456
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	457	%%{
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	458	machine tld_parser;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	459
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	460	host_component = alnum \| ( alnum [a-zA-Z0-9\-_]* alnum );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	461	tld = ( host_component '.' host_component );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	462
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	463	main := \|*
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	464	tld => { mark = ( p_request->tokens.host_length - (int)strlen(te) ); };
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	465	*\|;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	466	}%%
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	467
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	468	/* It's far easier (and faster) to scan from left to right rather than
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	469	backtrack, so start by reversing the requested host string. */
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	470	reverse_str( p_request->host );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	471
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	472	/* scanner */
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	473	%% write exec;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	474
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	475	/* If there was a mark, then allocate memory and copy. */
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	476	if ( mark != 0 ) {
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	477	if ( (p_request->tld = calloc( mark + 1, sizeof(char) )) == NULL ) {
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	478	debug( 5, LOC, "Unable to allocate memory for tld token: %s\n", strerror(errno) );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	479	reverse_str( p_request->host );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	480	return;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	481	}
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	482
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	483	memcpy( p_request->tld, p_request->host, mark );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	484	reverse_str( p_request->tld );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	485	}
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	486
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	487	/* restore the hostname. */
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	488	reverse_str( p_request->host );
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	489	return;
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	490	}
191b3c25974a Clean up redundant parser actions via preprocessor macros, add a Mahlon E. Smith <mahlon@martini.nu> parents: 10 diff changeset	491

author	Mahlon E. Smith <mahlon@martini.nu>
	Fri, 04 Nov 2011 20:34:28 -0700
changeset 14	51eb85ae4de4
parent 13	23a242d7b7fa
child 15	2706fc514dea
permissions	-rw-r--r--