abnf.c

/* Parser for ABNF, per RFC 5234. */

#include <hammer/hammer.h>
#include <hammer/glue.h>

const HParser* init_parser() {
  static const HParser *ret = NULL;
  if (ret)
    return ret;

  /* Core rules from Appendix B.1. Some rules appear out
   * of order because of precedence. */
  H_RULE(ALPHA,   h_choice(h_ch_range('A', 'Z'),
                           h_ch_range('a', 'z'), NULL));
  H_RULE(BIT,     h_choice(h_ch('0'), h_ch('1'), NULL));
  H_RULE(CHAR,    h_ch_range('\x01', '\x7f'));
  H_RULE(CR,      h_ch('\x0d'));
  H_RULE(LF,      h_ch('\x0a'));
  H_RULE(CRLF,    h_sequence(CR, LF, NULL));
  H_RULE(CTL,     h_choice(h_ch_range('\x00', '\x1f'),
                           h_ch('\x7f'), NULL));
  H_RULE(DIGIT,   h_ch_range('0', '9'));
  H_RULE(DQUOTE,  h_ch('"'));
  H_RULE(HEXDIG,  h_choice(DIGIT, 
                           h_ch_range('A', 'F'), NULL));
  H_RULE(HTAB,    h_ch('\x09'));
  H_RULE(OCTET,   h_ch_range('\x00', '\xff'));
  H_RULE(SP,      h_ch(' '));
  H_RULE(VCHAR,   h_ch_range('\x21', '\x7e'));
  H_RULE(WSP,     h_choice(SP, HTAB, NULL));
  H_RULE(LWSP,    h_many(h_choice(WSP,
                                  h_sequence(CRLF, WSP, NULL),
                                  NULL)));

  /* ABNF definition of ABNF from section 4.2, mostly from the bottom up. */
  H_RULE(prose_val,     h_sequence(h_ch('<'), h_many(h_choice(h_ch_range('\x20', '\x3d'),
                                                              h_ch_range('\x3f', '\x7e'), NULL)),
                                   h_ch('>'), NULL));
  H_RULE(hex_val,       h_sequence(h_ch('x'), 
                                   h_many1(HEXDIG),
                                   h_optional(h_choice(h_many1(h_sequence(h_ch('.'), 
                                                                          h_many1(HEXDIG), NULL)),
                                                       h_sequence(h_ch('-'), h_many1(HEXDIG), NULL),
                                                       NULL)),
                                   NULL));
  H_RULE(dec_val,       h_sequence(h_ch('d'),
                                   h_many1(DIGIT),
                                   h_optional(h_choice(h_many1(h_sequence(h_ch('.'),
                                                                          h_many1(DIGIT), NULL)),
                                                       h_sequence(h_ch('-'), h_many1(DIGIT), NULL),
                                                       NULL)),
                                   NULL));
  H_RULE(bin_val,       h_sequence(h_ch('b'),
                                   h_many1(BIT),
                                   h_optional(h_choice(h_many1(h_sequence(h_ch('.'),
                                                                          h_many1(BIT), NULL)),
                                                       h_sequence(h_ch('-'), h_many1(BIT), NULL),
                                                       NULL)),
                                   NULL));
  H_RULE(num_val,       h_sequence(h_ch('%'), h_choice(bin_val, dec_val, hex_val, NULL), NULL));
  
  H_RULE(char_val,      h_sequence(DQUOTE, h_many(h_choice(h_ch_range('\x20', '\x21'),
                                                           h_ch_range('\x23', '\x7e'), NULL)),
                                   DQUOTE, NULL));
  H_RULE(comment,       h_sequence(h_ch(';'), h_many(h_choice(WSP, VCHAR, NULL)), CRLF, NULL));
  H_RULE(c_nl,          h_choice(comment, CRLF, NULL));
  H_RULE(c_wsp,         h_choice(WSP, h_sequence(c_nl, WSP, NULL), NULL));
  H_RULE(alternation,   h_indirect());
  H_RULE(option,        h_sequence(h_ch('['), h_many(c_wsp), alternation, h_many(c_wsp), h_ch(']'), NULL));
  H_RULE(group,         h_sequence(h_ch('('), h_many(c_wsp), alternation, h_many(c_wsp), h_ch(')'), NULL));
  H_RULE(rulename,      h_sequence(ALPHA, h_many(h_choice(ALPHA, DIGIT, h_ch('-'), NULL)), NULL));
  H_RULE(element,       h_choice(rulename, group, option, char_val, num_val, prose_val, NULL));
  H_RULE(repeat,        h_choice(h_many1(DIGIT), 
                                 h_sequence(h_many(DIGIT), h_ch('*'), h_many(DIGIT), NULL), NULL));
  H_RULE(repetition,    h_sequence(h_optional(repeat), element, NULL));
  H_RULE(concatenation, h_sequence(repetition, h_many(h_sequence(h_many1(c_wsp), repetition, NULL)), NULL));
  h_bind_indirect(alternation, h_sequence(concatenation, h_many(h_sequence(h_many(c_wsp), h_ch('/'),
                                                                           h_many(c_wsp), concatenation,
                                                                           NULL)), NULL));
  H_RULE(elements,      h_sequence(alternation, h_many(c_wsp), NULL));
  H_RULE(defined_as,    h_sequence(h_many(c_wsp), h_choice(h_ch('='), h_token("=/", 2), NULL), 
                                   h_many(c_wsp), NULL));
  H_RULE(rule,          h_sequence(rulename, defined_as, elements, c_nl, NULL));
  H_RULE(rulelist,      h_many1(h_choice(rule, h_sequence(h_many(c_wsp), c_nl, NULL), NULL)));

  ret = rulelist;
  return ret;
}