1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-9, 2000 Free Software Foundation, Inc.
3 Written by Ben Pfaff <blp@gnu.org>.
5 This program is free software; you can redistribute it and/or
6 modify it under the terms of the GNU General Public License as
7 published by the Free Software Foundation; either version 2 of the
8 License, or (at your option) any later version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 #include <libpspp/message.h>
29 #include <libpspp/alloc.h>
30 #include <libpspp/assertion.h>
31 #include <language/command.h>
32 #include <libpspp/message.h>
33 #include <libpspp/magic.h>
34 #include <data/settings.h>
35 #include <libpspp/getl.h>
36 #include <libpspp/str.h>
41 #define _(msgid) gettext (msgid)
42 #define N_(msgid) msgid
50 struct string line_buffer;
52 bool (*read_line) (struct string *, enum getl_syntax *);
54 int token; /* Current token. */
55 double tokval; /* T_POS_NUM, T_NEG_NUM: the token's value. */
57 char tokid [LONG_NAME_LEN + 1]; /* T_ID: the identifier. */
59 struct string tokstr; /* T_ID, T_STRING: token string value.
60 For T_ID, this is not truncated as is
63 char *prog; /* Pointer to next token in line_buffer. */
64 bool dot; /* True only if this line ends with a terminal dot. */
65 bool eof; /* True only if the last token returned was T_STOP. */
67 int put_token ; /* If nonzero, next token returned by lex_get().
68 Used only in exceptional circumstances. */
70 struct string put_tokstr;
75 static int parse_id (struct lexer *);
77 /* How a string represents its contents. */
80 CHARACTER_STRING, /* Characters. */
81 BINARY_STRING, /* Binary digits. */
82 OCTAL_STRING, /* Octal digits. */
83 HEX_STRING /* Hexadecimal digits. */
86 static int parse_string (struct lexer *, enum string_type);
89 static void dump_token (void);
94 /* Initializes the lexer. */
96 lex_create (bool (*read_line_func) (struct string *, enum getl_syntax *))
98 struct lexer *lexer = xzalloc (sizeof (*lexer));
100 ds_init_empty (&lexer->tokstr);
101 ds_init_empty (&lexer->put_tokstr);
102 ds_init_empty (&lexer->line_buffer);
103 lexer->read_line = read_line_func;
105 if (!lex_get_line (lexer))
112 lex_destroy (struct lexer *lexer)
116 ds_destroy (&lexer->put_tokstr);
117 ds_destroy (&lexer->tokstr);
118 ds_destroy (&lexer->line_buffer);
125 /* Common functions. */
127 /* Copies put_token, lexer->put_tokstr, put_tokval into token, tokstr,
128 tokval, respectively, and sets tokid appropriately. */
130 restore_token (struct lexer *lexer)
132 assert (lexer->put_token != 0);
133 lexer->token = lexer->put_token;
134 ds_assign_string (&lexer->tokstr, &lexer->put_tokstr);
135 str_copy_trunc (lexer->tokid, sizeof lexer->tokid, ds_cstr (&lexer->tokstr));
136 lexer->tokval = lexer->put_tokval;
137 lexer->put_token = 0;
140 /* Copies token, tokstr, lexer->tokval into lexer->put_token, put_tokstr,
141 put_lexer->tokval respectively. */
143 save_token (struct lexer *lexer)
145 lexer->put_token = lexer->token;
146 ds_assign_string (&lexer->put_tokstr, &lexer->tokstr);
147 lexer->put_tokval = lexer->tokval;
150 /* Parses a single token, setting appropriate global variables to
151 indicate the token's attributes. */
153 lex_get (struct lexer *lexer)
155 /* If a token was pushed ahead, return it. */
156 if (lexer->put_token)
158 restore_token (lexer);
168 /* Skip whitespace. */
171 lexer->token = T_STOP;
177 while (isspace ((unsigned char) *lexer->prog))
191 else if (!lex_get_line (lexer))
194 lexer->token = T_STOP;
201 if (lexer->put_token)
203 restore_token (lexer);
212 /* Actually parse the token. */
213 ds_clear (&lexer->tokstr);
215 switch (*lexer->prog)
218 case '0': case '1': case '2': case '3': case '4':
219 case '5': case '6': case '7': case '8': case '9':
223 /* `-' can introduce a negative number, or it can be a
224 token by itself. If it is not followed by a digit or a
225 decimal point, it is definitely not a number.
226 Otherwise, it might be either, but most of the time we
227 want it as a number. When the syntax calls for a `-'
228 token, lex_negative_to_dash() must be used to break
229 negative numbers into two tokens. */
230 if (*lexer->prog == '-')
232 ds_put_char (&lexer->tokstr, *lexer->prog++);
233 while (isspace ((unsigned char) *lexer->prog))
236 if (!isdigit ((unsigned char) *lexer->prog) && *lexer->prog != '.')
241 lexer->token = T_NEG_NUM;
244 lexer->token = T_POS_NUM;
246 /* Parse the number, copying it into tokstr. */
247 while (isdigit ((unsigned char) *lexer->prog))
248 ds_put_char (&lexer->tokstr, *lexer->prog++);
249 if (*lexer->prog == '.')
251 ds_put_char (&lexer->tokstr, *lexer->prog++);
252 while (isdigit ((unsigned char) *lexer->prog))
253 ds_put_char (&lexer->tokstr, *lexer->prog++);
255 if (*lexer->prog == 'e' || *lexer->prog == 'E')
257 ds_put_char (&lexer->tokstr, *lexer->prog++);
258 if (*lexer->prog == '+' || *lexer->prog == '-')
259 ds_put_char (&lexer->tokstr, *lexer->prog++);
260 while (isdigit ((unsigned char) *lexer->prog))
261 ds_put_char (&lexer->tokstr, *lexer->prog++);
264 /* Parse as floating point. */
265 lexer->tokval = strtod (ds_cstr (&lexer->tokstr), &tail);
268 msg (SE, _("%s does not form a valid number."),
269 ds_cstr (&lexer->tokstr));
272 ds_clear (&lexer->tokstr);
273 ds_put_char (&lexer->tokstr, '0');
280 lexer->token = parse_string (lexer, CHARACTER_STRING);
283 case '(': case ')': case ',': case '=': case '+': case '/':
284 lexer->token = *lexer->prog++;
288 if (*++lexer->prog == '*')
291 lexer->token = T_EXP;
298 if (*++lexer->prog == '=')
303 else if (*lexer->prog == '>')
313 if (*++lexer->prog == '=')
323 if (*++lexer->prog == '=')
329 lexer->token = T_NOT;
334 lexer->token = T_AND;
343 if (lexer->prog[1] == '\'' || lexer->prog[1] == '"')
344 lexer->token = parse_string (lexer, BINARY_STRING);
346 lexer->token = parse_id (lexer);
350 if (lexer->prog[1] == '\'' || lexer->prog[1] == '"')
351 lexer->token = parse_string (lexer, OCTAL_STRING);
353 lexer->token = parse_id (lexer);
357 if (lexer->prog[1] == '\'' || lexer->prog[1] == '"')
358 lexer->token = parse_string (lexer, HEX_STRING);
360 lexer->token = parse_id (lexer);
364 if (lex_is_id1 (*lexer->prog))
366 lexer->token = parse_id (lexer);
371 if (isgraph ((unsigned char) *lexer->prog))
372 msg (SE, _("Bad character in input: `%c'."), *lexer->prog++);
374 msg (SE, _("Bad character in input: `\\%o'."), *lexer->prog++);
386 /* Parses an identifier at the current position into tokid and
388 Returns the correct token type. */
390 parse_id (struct lexer *lexer)
392 struct substring rest_of_line
393 = ss_substr (ds_ss (&lexer->line_buffer),
394 ds_pointer_to_position (&lexer->line_buffer, lexer->prog),
396 struct substring id = ss_head (rest_of_line,
397 lex_id_get_length (rest_of_line));
398 lexer->prog += ss_length (id);
400 ds_assign_substring (&lexer->tokstr, id);
401 str_copy_trunc (lexer->tokid, sizeof lexer->tokid, ds_cstr (&lexer->tokstr));
402 return lex_id_to_token (id);
405 /* Reports an error to the effect that subcommand SBC may only be
408 lex_sbc_only_once (const char *sbc)
410 msg (SE, _("Subcommand %s may only be specified once."), sbc);
413 /* Reports an error to the effect that subcommand SBC is
416 lex_sbc_missing (struct lexer *lexer, const char *sbc)
418 lex_error (lexer, _("missing required subcommand %s"), sbc);
421 /* Prints a syntax error message containing the current token and
422 given message MESSAGE (if non-null). */
424 lex_error (struct lexer *lexer, const char *message, ...)
429 token_rep = lex_token_representation (lexer);
430 if (lexer->token == T_STOP)
431 strcpy (where, "end of file");
432 else if (lexer->token == '.')
433 strcpy (where, "end of command");
435 snprintf (where, sizeof where, "`%s'", token_rep);
443 va_start (args, message);
444 vsnprintf (buf, 1024, message, args);
447 msg (SE, _("Syntax error %s at %s."), buf, where);
450 msg (SE, _("Syntax error at %s."), where);
453 /* Checks that we're at end of command.
454 If so, returns a successful command completion code.
455 If not, flags a syntax error and returns an error command
458 lex_end_of_command (struct lexer *lexer)
460 if (lexer->token != '.')
462 lex_error (lexer, _("expecting end of command"));
469 /* Token testing functions. */
471 /* Returns true if the current token is a number. */
473 lex_is_number (struct lexer *lexer)
475 return lexer->token == T_POS_NUM || lexer->token == T_NEG_NUM;
478 /* Returns the value of the current token, which must be a
479 floating point number. */
481 lex_number (struct lexer *lexer)
483 assert (lex_is_number (lexer));
484 return lexer->tokval;
487 /* Returns true iff the current token is an integer. */
489 lex_is_integer (struct lexer *lexer)
491 return (lex_is_number (lexer)
492 && lexer->tokval != NOT_LONG
493 && lexer->tokval >= LONG_MIN
494 && lexer->tokval <= LONG_MAX
495 && floor (lexer->tokval) == lexer->tokval);
498 /* Returns the value of the current token, which must be an
501 lex_integer (struct lexer *lexer)
503 assert (lex_is_integer (lexer));
504 return lexer->tokval;
507 /* Token matching functions. */
509 /* If TOK is the current token, skips it and returns true
510 Otherwise, returns false. */
512 lex_match (struct lexer *lexer, int t)
514 if (lexer->token == t)
523 /* If the current token is the identifier S, skips it and returns
524 true. The identifier may be abbreviated to its first three
526 Otherwise, returns false. */
528 lex_match_id (struct lexer *lexer, const char *s)
530 if (lexer->token == T_ID
531 && lex_id_match (ss_cstr (s), ss_cstr (lexer->tokid)))
540 /* If the current token is integer N, skips it and returns true.
541 Otherwise, returns false. */
543 lex_match_int (struct lexer *lexer, int x)
545 if (lex_is_integer (lexer) && lex_integer (lexer) == x)
554 /* Forced matches. */
556 /* If this token is identifier S, fetches the next token and returns
558 Otherwise, reports an error and returns zero. */
560 lex_force_match_id (struct lexer *lexer, const char *s)
562 if (lex_match_id (lexer, s))
566 lex_error (lexer, _("expecting `%s'"), s);
571 /* If the current token is T, skips the token. Otherwise, reports an
572 error and returns from the current function with return value false. */
574 lex_force_match (struct lexer *lexer, int t)
576 if (lexer->token == t)
583 lex_error (lexer, _("expecting `%s'"), lex_token_name (t));
588 /* If this token is a string, does nothing and returns true.
589 Otherwise, reports an error and returns false. */
591 lex_force_string (struct lexer *lexer)
593 if (lexer->token == T_STRING)
597 lex_error (lexer, _("expecting string"));
602 /* If this token is an integer, does nothing and returns true.
603 Otherwise, reports an error and returns false. */
605 lex_force_int (struct lexer *lexer)
607 if (lex_is_integer (lexer))
611 lex_error (lexer, _("expecting integer"));
616 /* If this token is a number, does nothing and returns true.
617 Otherwise, reports an error and returns false. */
619 lex_force_num (struct lexer *lexer)
621 if (lex_is_number (lexer))
624 lex_error (lexer, _("expecting number"));
628 /* If this token is an identifier, does nothing and returns true.
629 Otherwise, reports an error and returns false. */
631 lex_force_id (struct lexer *lexer)
633 if (lexer->token == T_ID)
636 lex_error (lexer, _("expecting identifier"));
640 /* Weird token functions. */
642 /* Returns the first character of the next token, except that if the
643 next token is not an identifier, the character returned will not be
644 a character that can begin an identifier. Specifically, the
645 hexstring lead-in X' causes lookahead() to return '. Note that an
646 alphanumeric return value doesn't guarantee an ID token, it could
647 also be a reserved-word token. */
649 lex_look_ahead (struct lexer *lexer)
651 if (lexer->put_token)
652 return lexer->put_token;
661 while (isspace ((unsigned char) *lexer->prog))
668 else if (!lex_get_line (lexer))
671 if (lexer->put_token)
672 return lexer->put_token;
675 if ((toupper ((unsigned char) *lexer->prog) == 'X'
676 || toupper ((unsigned char) *lexer->prog) == 'B'
677 || toupper ((unsigned char) *lexer->prog) == 'O')
678 && (lexer->prog[1] == '\'' || lexer->prog[1] == '"'))
685 /* Makes the current token become the next token to be read; the
686 current token is set to T. */
688 lex_put_back (struct lexer *lexer, int t)
694 /* Makes the current token become the next token to be read; the
695 current token is set to the identifier ID. */
697 lex_put_back_id (struct lexer *lexer, const char *id)
699 assert (lex_id_to_token (ss_cstr (id)) == T_ID);
702 ds_assign_cstr (&lexer->tokstr, id);
703 str_copy_trunc (lexer->tokid, sizeof lexer->tokid, ds_cstr (&lexer->tokstr));
706 /* Weird line processing functions. */
708 /* Returns the entire contents of the current line. */
710 lex_entire_line (struct lexer *lexer)
712 return ds_cstr (&lexer->line_buffer);
715 const struct string *
716 lex_entire_line_ds (struct lexer *lexer)
718 return &lexer->line_buffer;
721 /* As lex_entire_line(), but only returns the part of the current line
722 that hasn't already been tokenized.
723 If END_DOT is non-null, stores nonzero into *END_DOT if the line
724 ends with a terminal dot, or zero if it doesn't. */
726 lex_rest_of_line (struct lexer *lexer, int *end_dot)
729 *end_dot = lexer->dot;
733 /* Causes the rest of the current input line to be ignored for
734 tokenization purposes. */
736 lex_discard_line (struct lexer *lexer)
738 ds_cstr (&lexer->line_buffer); /* Ensures ds_end points to something valid */
739 lexer->prog = ds_end (&lexer->line_buffer);
741 lexer->put_token = 0;
745 /* Discards the rest of the current command.
746 When we're reading commands from a file, we skip tokens until
747 a terminal dot or EOF.
748 When we're reading commands interactively from the user,
749 that's just discarding the current line, because presumably
750 the user doesn't want to finish typing a command that will be
753 lex_discard_rest_of_command (struct lexer *lexer)
755 if (!getl_is_interactive ())
757 while (lexer->token != T_STOP && lexer->token != '.')
761 lex_discard_line (lexer);
764 /* Weird line reading functions. */
766 /* Remove C-style comments in STRING, begun by slash-star and
767 terminated by star-slash or newline. */
769 strip_comments (struct string *string)
777 for (cp = ds_cstr (string); *cp; )
779 /* If we're not in a comment, check for quote marks. */
784 else if (*cp == '\'' || *cp == '"')
788 /* If we're not inside a quotation, check for comment. */
791 if (cp[0] == '/' && cp[1] == '*')
798 else if (in_comment && cp[0] == '*' && cp[1] == '/')
807 /* Check commenting. */
814 /* Prepares LINE, which is subject to the given SYNTAX rules, for
815 tokenization by stripping comments and determining whether it
816 is the beginning or end of a command and storing into
817 *LINE_STARTS_COMMAND and *LINE_ENDS_COMMAND appropriately. */
819 lex_preprocess_line (struct string *line,
820 enum getl_syntax syntax,
821 bool *line_starts_command,
822 bool *line_ends_command)
824 strip_comments (line);
825 ds_rtrim (line, ss_cstr (CC_SPACES));
826 *line_ends_command = (ds_chomp (line, get_endcmd ())
827 || (ds_is_empty (line) && get_nulline ()));
828 *line_starts_command = false;
829 if (syntax == GETL_BATCH)
831 int first = ds_first (line);
832 *line_starts_command = !isspace (first);
833 if (first == '+' || first == '-')
834 *ds_data (line) = ' ';
838 /* Reads a line, without performing any preprocessing.
839 Sets *SYNTAX, if SYNTAX is non-null, to the line's syntax
842 lex_get_line_raw (struct lexer *lexer, enum getl_syntax *syntax)
844 enum getl_syntax dummy;
845 bool ok = lexer->read_line (&lexer->line_buffer,
846 syntax != NULL ? syntax : &dummy);
850 /* Reads a line for use by the tokenizer, and preprocesses it by
851 removing comments, stripping trailing whitespace and the
852 terminal dot, and removing leading indentors. */
854 lex_get_line (struct lexer *lexer)
856 bool line_starts_command;
857 enum getl_syntax syntax;
859 if (!lex_get_line_raw (lexer, &syntax))
862 lex_preprocess_line (&lexer->line_buffer, syntax,
863 &line_starts_command, &lexer->dot);
864 if (line_starts_command)
865 lexer->put_token = '.';
867 lexer->prog = ds_cstr (&lexer->line_buffer);
873 /* Returns the name of a token. */
875 lex_token_name (int token)
877 if (lex_is_keyword (token))
878 return lex_id_name (token);
879 else if (token < 256)
881 static char t[256][2];
891 /* Returns an ASCII representation of the current token as a
892 malloc()'d string. */
894 lex_token_representation (struct lexer *lexer)
898 switch (lexer->token)
903 return ds_xstrdup (&lexer->tokstr);
911 for (sp = ds_cstr (&lexer->tokstr); sp < ds_end (&lexer->tokstr); sp++)
912 if (!isprint ((unsigned char) *sp))
918 token_rep = xmalloc (2 + ds_length (&lexer->tokstr) * 2 + 1 + 1);
926 for (sp = ds_cstr (&lexer->tokstr); *sp; )
930 *dp++ = (unsigned char) *sp++;
933 for (sp = ds_cstr (&lexer->tokstr); sp < ds_end (&lexer->tokstr); sp++)
935 *dp++ = (((unsigned char) *sp) >> 4)["0123456789ABCDEF"];
936 *dp++ = (((unsigned char) *sp) & 15)["0123456789ABCDEF"];
946 token_rep = xmalloc (1);
951 return xstrdup ("**");
954 return xstrdup (lex_token_name (lexer->token));
960 /* Really weird functions. */
962 /* Most of the time, a `-' is a lead-in to a negative number. But
963 sometimes it's actually part of the syntax. If a dash can be part
964 of syntax then this function is called to rip it off of a
967 lex_negative_to_dash (struct lexer *lexer)
969 if (lexer->token == T_NEG_NUM)
971 lexer->token = T_POS_NUM;
972 lexer->tokval = -lexer->tokval;
973 ds_assign_substring (&lexer->tokstr, ds_substr (&lexer->tokstr, 1, SIZE_MAX));
979 /* We're not at eof any more. */
981 lex_reset_eof (struct lexer *lexer)
986 /* Skip a COMMENT command. */
988 lex_skip_comment (struct lexer *lexer)
992 if (!lex_get_line (lexer))
994 lexer->put_token = T_STOP;
999 if (lexer->put_token == '.')
1002 ds_cstr (&lexer->line_buffer); /* Ensures ds_end will point to a valid char */
1003 lexer->prog = ds_end (&lexer->line_buffer);
1009 /* Private functions. */
1011 /* When invoked, tokstr contains a string of binary, octal, or
1012 hex digits, according to TYPE. The string is converted to
1013 characters having the specified values. */
1015 convert_numeric_string_to_char_string (struct lexer *lexer,
1016 enum string_type type)
1018 const char *base_name;
1028 base_name = _("binary");
1033 base_name = _("octal");
1038 base_name = _("hex");
1046 byte_cnt = ds_length (&lexer->tokstr) / chars_per_byte;
1047 if (ds_length (&lexer->tokstr) % chars_per_byte)
1048 msg (SE, _("String of %s digits has %d characters, which is not a "
1050 base_name, ds_length (&lexer->tokstr), chars_per_byte);
1052 p = ds_cstr (&lexer->tokstr);
1053 for (i = 0; i < byte_cnt; i++)
1059 for (j = 0; j < chars_per_byte; j++, p++)
1063 if (*p >= '0' && *p <= '9')
1067 static const char alpha[] = "abcdef";
1068 const char *q = strchr (alpha, tolower ((unsigned char) *p));
1077 msg (SE, _("`%c' is not a valid %s digit."), *p, base_name);
1079 value = value * base + v;
1082 ds_cstr (&lexer->tokstr)[i] = (unsigned char) value;
1085 ds_truncate (&lexer->tokstr, byte_cnt);
1088 /* Parses a string from the input buffer into tokstr. The input
1089 buffer pointer lexer->prog must point to the initial single or double
1090 quote. TYPE indicates the type of string to be parsed.
1091 Returns token type. */
1093 parse_string (struct lexer *lexer, enum string_type type)
1095 if (type != CHARACTER_STRING)
1098 /* Accumulate the entire string, joining sections indicated by +
1102 /* Single or double quote. */
1103 int c = *lexer->prog++;
1105 /* Accumulate section. */
1108 /* Check end of line. */
1109 if (*lexer->prog == '\0')
1111 msg (SE, _("Unterminated string constant."));
1115 /* Double quote characters to embed them in strings. */
1116 if (*lexer->prog == c)
1118 if (lexer->prog[1] == c)
1124 ds_put_char (&lexer->tokstr, *lexer->prog++);
1128 /* Skip whitespace after final quote mark. */
1133 while (isspace ((unsigned char) *lexer->prog))
1141 if (!lex_get_line (lexer))
1145 /* Skip plus sign. */
1146 if (*lexer->prog != '+')
1150 /* Skip whitespace after plus sign. */
1155 while (isspace ((unsigned char) *lexer->prog))
1163 if (!lex_get_line (lexer))
1165 msg (SE, _("Unexpected end of file in string concatenation."));
1170 /* Ensure that a valid string follows. */
1171 if (*lexer->prog != '\'' && *lexer->prog != '"')
1173 msg (SE, _("String expected following `+'."));
1178 /* We come here when we've finished concatenating all the string sections
1179 into one large string. */
1181 if (type != CHARACTER_STRING)
1182 convert_numeric_string_to_char_string (lexer, type);
1184 if (ds_length (&lexer->tokstr) > 255)
1186 msg (SE, _("String exceeds 255 characters in length (%d characters)."),
1187 ds_length (&lexer->tokstr));
1188 ds_truncate (&lexer->tokstr, 255);
1195 /* Reads one token from the lexer and writes a textual representation
1196 on stdout for debugging purposes. */
1198 dump_token (struct lexer *lexer)
1204 getl_location (&curfn, &curln);
1206 fprintf (stderr, "%s:%d\t", curfn, curln);
1209 switch (lexer->token)
1212 fprintf (stderr, "ID\t%s\n", lexer->tokid);
1217 fprintf (stderr, "NUM\t%f\n", lexer->tokval);
1221 fprintf (stderr, "STRING\t\"%s\"\n", ds_cstr (&lexer->tokstr));
1225 fprintf (stderr, "STOP\n");
1229 fprintf (stderr, "MISC\tEXP\"");
1233 fprintf (stderr, "MISC\tEOF\n");
1237 if (lex_is_keyword (token))
1238 fprintf (stderr, "KEYWORD\t%s\n", lex_token_name (token));
1240 fprintf (stderr, "PUNCT\t%c\n", lexer->token);
1244 #endif /* DUMP_TOKENS */
1247 /* Token Accessor Functions */
1250 lex_token (const struct lexer *lexer)
1252 return lexer->token;
1256 lex_tokval (const struct lexer *lexer)
1258 return lexer->tokval;
1262 lex_tokid (const struct lexer *lexer)
1264 return lexer->tokid;
1267 const struct string *
1268 lex_tokstr (const struct lexer *lexer)
1270 return &lexer->tokstr;