7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "lib.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "net.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "str.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "strescape.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "http-url.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch#include "http-parser.h"

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_token_char_mask = (1<<0)|(1<<1);

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_value_char_mask = (1<<0)|(1<<1)|(1<<2);

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_text_char_mask = (1<<0)|(1<<1)|(1<<2)|(1<<6);

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_qdtext_char_mask = (1<<3)|(1<<4)|(1<<6);

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_ctext_char_mask = (1<<3)|(1<<5)|(1<<6);

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschconst unsigned char _http_token68_char_mask = (1<<0)|(1<<7);

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Boschconst unsigned char _http_char_lookup[256] = {

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 0, 0, 0, 0, 0, 0, 0, 0, 0, 64, 0, 0, 0, 0, 0, 0, // 00

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 10

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 10, 36, 50, 50, 50, 50, 50, 20, 20, 10, 9, 12, 9, 9, 140, // 20

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 12, 12, 12, 12, 12, 12, // 30

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 12, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, // 40

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 12, 4, 12, 10, 9, // 50

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 10, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, // 60

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 12, 10, 12, 9, 0, // 70

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // 80

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // 90

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // A0

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // B0

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // C0

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // D0

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // E0

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, // F0

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch};

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Boschvoid http_parser_init(struct http_parser *parser,

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch const unsigned char *data, size_t size)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch{

efe78d3ba24fc866af1c79b9223dc0809ba26cadStephan Bosch i_zero(parser);

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->begin = data;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->cur = data;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->end = data + size;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch}

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Boschvoid http_parse_ows(struct http_parser *parser)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch{

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch /* OWS = *( SP / HTAB ) */

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch if (parser->cur >= parser->end)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch return;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch while (parser->cur < parser->end &&

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch (parser->cur[0] == ' ' || parser->cur[0] == '\t')) {

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->cur++;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch }

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch}

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschint http_parser_skip_token(struct http_parser *parser)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch{

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch /* token = 1*tchar */

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch if (parser->cur >= parser->end || !http_char_is_token(*parser->cur))

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch return 0;

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch parser->cur++;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch while (parser->cur < parser->end && http_char_is_token(*parser->cur))

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->cur++;

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch return 1;

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch}

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Boschint http_parse_token(struct http_parser *parser, const char **token_r)

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch{

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch const unsigned char *first = parser->cur;

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch int ret;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch if ((ret=http_parser_skip_token(parser)) <= 0)

0b5bdb5ba6c8cb928b6d2dcc4636148bfa1a7ec7Stephan Bosch return ret;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch *token_r = t_strndup(first, parser->cur - first);

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch return 1;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch}

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Boschint http_parse_token_list_next(struct http_parser *parser,

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch const char **token_r)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch{

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch /* http://tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-21;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch for (;;) {

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch if (http_parse_token(parser, token_r) > 0)

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch break;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch http_parse_ows(parser);

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch if (parser->cur >= parser->end || parser->cur[0] != ',')

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch return 0;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch parser->cur++;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch http_parse_ows(parser);

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch }

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch return 1;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch}

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Boschint http_parse_quoted_string(struct http_parser *parser, const char **str_r)

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch{

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch string_t *str;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* quoted-string = DQUOTE *( qdtext / quoted-pair ) DQUOTE

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* DQUOTE */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (parser->cur >= parser->end || parser->cur[0] != '"')

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return 0;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch parser->cur++;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* *( qdtext / quoted-pair ) */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch str = t_str_new(256);

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch for (;;) {

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch const unsigned char *first;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* *qdtext */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch first = parser->cur;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch while (parser->cur < parser->end && http_char_is_qdtext(*parser->cur))

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch parser->cur++;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (parser->cur >= parser->end)

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return -1;

7384b4e78eaab44693c985192276e31322155e32Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch str_append_n(str, first, parser->cur - first);

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* DQUOTE */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (*parser->cur == '"') {

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch parser->cur++;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch break;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* "\" */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch } else if (*parser->cur == '\\') {

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch parser->cur++;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (parser->cur >= parser->end || !http_char_is_text(*parser->cur))

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return -1;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch str_append_c(str, *parser->cur);

aefccddf68a92e43209aea013a36d0d7d374f410Stephan Bosch parser->cur++;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch /* ERROR */

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch } else {

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return -1;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch }

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch }

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch *str_r = str_c(str);

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return 1;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch}

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch

6ee9ce5ed955a1283dc22ad28980bf9cc23d4c4eStephan Boschint http_parse_token_or_qstring(struct http_parser *parser,

6ee9ce5ed955a1283dc22ad28980bf9cc23d4c4eStephan Bosch const char **word_r)

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch{

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (parser->cur >= parser->end)

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return 0;

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch if (parser->cur[0] == '"')

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return http_parse_quoted_string(parser, word_r);

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch return http_parse_token(parser, word_r);

233a96794faa7734dfc3aea19503c14b29cccfb7Stephan Bosch}