5 * Permission is hereby granted, free of charge, to any person obtaining a copy of
6 * this software and associated documentation files (the "Software"), to deal in
7 * the Software without restriction, including without limitation the rights to
8 * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
9 * of the Software, and to permit persons to whom the Software is furnished to do
10 * so, subject to the following conditions:
12 * The above copyright notice and this permission notice shall be included in all
13 * copies or substantial portions of the Software.
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
36 static const char *keywords_qc[] = {
43 static size_t num_keywords_qc = sizeof(keywords_qc) / sizeof(keywords_qc[0]);
46 static const char *keywords_fg[] = {
47 "switch", "case", "default",
53 "__builtin_debug_printtype"
55 static size_t num_keywords_fg = sizeof(keywords_fg) / sizeof(keywords_fg[0]);
63 void lexerror(lex_file *lex, const char *fmt, ...)
69 con_vprintmsg(LVL_ERROR, lex->name, lex->sline, "parse error", fmt, ap);
71 con_vprintmsg(LVL_ERROR, "", 0, "parse error", fmt, ap);
75 bool lexwarn(lex_file *lex, int warntype, const char *fmt, ...)
82 ctx.line = lex->sline;
85 r = vcompile_warning(ctx, warntype, fmt, ap);
94 token *tok = (token*)mem_a(sizeof(token));
97 memset(tok, 0, sizeof(*tok));
101 void token_delete(token *self)
103 if (self->next && self->next->prev == self)
104 self->next->prev = self->prev;
105 if (self->prev && self->prev->next == self)
106 self->prev->next = self->next;
107 MEM_VECTOR_CLEAR(self, value);
111 token* token_copy(const token *cp)
113 token* self = token_new();
117 self->value_alloc = cp->value_count + 1;
118 self->value_count = cp->value_count;
119 self->value = (char*)mem_a(self->value_alloc);
124 memcpy(self->value, cp->value, cp->value_count);
125 self->value[self->value_alloc-1] = 0;
129 self->ttype = cp->ttype;
130 memcpy(&self->constval, &cp->constval, sizeof(self->constval));
134 void token_delete_all(token *t)
145 token* token_copy_all(const token *cp)
150 out = cur = token_copy(cp);
156 cur->next = token_copy(cp);
158 token_delete_all(out);
161 cur->next->prev = cur;
168 static void lex_token_new(lex_file *lex)
172 token_delete(lex->tok);
173 lex->tok = token_new();
176 vec_shrinkto(lex->tok.value, 0);
177 lex->tok.constval.t = 0;
178 lex->tok.ctx.line = lex->sline;
179 lex->tok.ctx.file = lex->name;
184 lex_file* lex_open(const char *file)
187 FILE *in = file_open(file, "rb");
190 lexerror(NULL, "open failed: '%s'\n", file);
194 lex = (lex_file*)mem_a(sizeof(*lex));
197 lexerror(NULL, "out of memory\n");
201 memset(lex, 0, sizeof(*lex));
204 lex->name = util_strdup(file);
205 lex->line = 1; /* we start counting at 1 */
210 vec_push(lex_filenames, lex->name);
214 lex_file* lex_open_string(const char *str, size_t len, const char *name)
218 lex = (lex_file*)mem_a(sizeof(*lex));
220 lexerror(NULL, "out of memory\n");
224 memset(lex, 0, sizeof(*lex));
227 lex->open_string = str;
228 lex->open_string_length = len;
229 lex->open_string_pos = 0;
231 lex->name = util_strdup(name ? name : "<string-source>");
232 lex->line = 1; /* we start counting at 1 */
237 vec_push(lex_filenames, lex->name);
242 void lex_cleanup(void)
245 for (i = 0; i < vec_size(lex_filenames); ++i)
246 mem_d(lex_filenames[i]);
247 vec_free(lex_filenames);
250 void lex_close(lex_file *lex)
253 for (i = 0; i < vec_size(lex->frames); ++i)
254 mem_d(lex->frames[i].name);
255 vec_free(lex->frames);
258 vec_free(lex->modelname);
261 file_close(lex->file);
264 token_delete(lex->tok);
266 vec_free(lex->tok.value);
268 /* mem_d(lex->name); collected in lex_filenames */
272 static int lex_fgetc(lex_file *lex)
275 return fgetc(lex->file);
276 if (lex->open_string) {
277 if (lex->open_string_pos >= lex->open_string_length)
279 return lex->open_string[lex->open_string_pos++];
284 /* Get or put-back data
285 * The following to functions do NOT understand what kind of data they
287 * The are merely wrapping get/put in order to count line numbers.
289 static void lex_ungetch(lex_file *lex, int ch);
290 static int lex_try_trigraph(lex_file *lex, int old)
295 lex_ungetch(lex, c2);
301 case '=': return '#';
302 case '/': return '\\';
303 case '\'': return '^';
304 case '(': return '[';
305 case ')': return ']';
306 case '!': return '|';
307 case '<': return '{';
308 case '>': return '}';
309 case '-': return '~';
311 lex_ungetch(lex, c3);
312 lex_ungetch(lex, c2);
317 static int lex_try_digraph(lex_file *lex, int ch)
321 /* we just used fgetc() so count lines
322 * need to offset a \n the ungetch would recognize
324 if (!lex->push_line && c2 == '\n')
326 if (ch == '<' && c2 == ':')
328 else if (ch == ':' && c2 == '>')
330 else if (ch == '<' && c2 == '%')
332 else if (ch == '%' && c2 == '>')
334 else if (ch == '%' && c2 == ':')
336 lex_ungetch(lex, c2);
340 static int lex_getch(lex_file *lex)
346 if (!lex->push_line && lex->peek[lex->peekpos] == '\n')
348 return lex->peek[lex->peekpos];
352 if (!lex->push_line && ch == '\n')
355 return lex_try_trigraph(lex, ch);
356 else if (!lex->flags.nodigraphs && (ch == '<' || ch == ':' || ch == '%'))
357 return lex_try_digraph(lex, ch);
361 static void lex_ungetch(lex_file *lex, int ch)
363 lex->peek[lex->peekpos++] = ch;
364 if (!lex->push_line && ch == '\n')
368 /* classify characters
369 * some additions to the is*() functions of ctype.h
372 /* Idents are alphanumberic, but they start with alpha or _ */
373 static bool isident_start(int ch)
375 return isalpha(ch) || ch == '_';
378 static bool isident(int ch)
380 return isident_start(ch) || isdigit(ch);
383 /* isxdigit_only is used when we already know it's not a digit
384 * and want to see if it's a hex digit anyway.
386 static bool isxdigit_only(int ch)
388 return (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F');
391 /* Append a character to the token buffer */
392 static void lex_tokench(lex_file *lex, int ch)
394 vec_push(lex->tok.value, ch);
397 /* Append a trailing null-byte */
398 static void lex_endtoken(lex_file *lex)
400 vec_push(lex->tok.value, 0);
401 vec_shrinkby(lex->tok.value, 1);
404 static bool lex_try_pragma(lex_file *lex)
408 char *command = NULL;
412 if (lex->flags.preprocessing)
419 lex_ungetch(lex, ch);
423 for (ch = lex_getch(lex); vec_size(pragma) < 8 && ch >= 'a' && ch <= 'z'; ch = lex_getch(lex))
424 vec_push(pragma, ch);
427 if (ch != ' ' || strcmp(pragma, "pragma")) {
428 lex_ungetch(lex, ch);
432 for (ch = lex_getch(lex); vec_size(command) < 32 && ch >= 'a' && ch <= 'z'; ch = lex_getch(lex))
433 vec_push(command, ch);
434 vec_push(command, 0);
437 lex_ungetch(lex, ch);
441 for (ch = lex_getch(lex); vec_size(param) < 1024 && ch != ')' && ch != '\n'; ch = lex_getch(lex))
446 lex_ungetch(lex, ch);
450 if (!strcmp(command, "push")) {
451 if (!strcmp(param, "line")) {
453 if (lex->push_line == 1)
459 else if (!strcmp(command, "pop")) {
460 if (!strcmp(param, "line")) {
463 if (lex->push_line == 0)
469 else if (!strcmp(command, "file")) {
470 lex->name = util_strdup(param);
471 vec_push(lex_filenames, lex->name);
473 else if (!strcmp(command, "line")) {
474 line = strtol(param, NULL, 0)-1;
480 while (ch != '\n' && ch != EOF)
487 while (vec_size(command)) {
488 lex_ungetch(lex, vec_last(command));
492 lex_ungetch(lex, ' ');
496 while (vec_size(command)) {
497 lex_ungetch(lex, vec_last(command));
501 lex_ungetch(lex, ' ');
505 while (vec_size(pragma)) {
506 lex_ungetch(lex, vec_last(pragma));
511 lex_ungetch(lex, '#');
517 /* Skip whitespace and comments and return the first
518 * non-white character.
519 * As this makes use of the above getch() ungetch() functions,
520 * we don't need to care at all about line numbering anymore.
522 * In theory, this function should only be used at the beginning
523 * of lexing, or when we *know* the next character is part of the token.
524 * Otherwise, if the parser throws an error, the linenumber may not be
525 * the line of the error, but the line of the next token AFTER the error.
527 * This is currently only problematic when using c-like string-continuation,
528 * since comments and whitespaces are allowed between 2 such strings.
532 "A continuation of the previous string"
533 // This line is skipped
536 * In this case, if the parse decides it didn't actually want a string,
537 * and uses lex->line to print an error, it will show the ', foo);' line's
540 * On the other hand, the parser is supposed to remember the line of the next
541 * token's beginning. In this case we would want skipwhite() to be called
542 * AFTER reading a token, so that the parser, before reading the NEXT token,
543 * doesn't store teh *comment's* linenumber, but the actual token's linenumber.
546 * here is to store the line of the first character after skipping
547 * the initial whitespace in lex->sline, this happens in lex_do.
549 static int lex_skipwhite(lex_file *lex, bool hadwhite)
552 bool haswhite = hadwhite;
557 while (ch != EOF && isspace(ch)) {
559 if (lex_try_pragma(lex))
562 if (lex->flags.preprocessing) {
565 /* see if there was whitespace first */
566 if (haswhite) { /* (vec_size(lex->tok.value)) { */
567 lex_ungetch(lex, ch);
571 /* otherwise return EOL */
575 lex_tokench(lex, ch);
584 /* one line comment */
587 if (lex->flags.preprocessing) {
590 lex_tokench(lex, '/');
591 lex_tokench(lex, '/');
593 lex_tokench(lex, ' ');
594 lex_tokench(lex, ' ');
597 while (ch != EOF && ch != '\n') {
598 if (lex->flags.preprocessing)
599 lex_tokench(lex, ' '); /* ch); */
602 if (lex->flags.preprocessing) {
603 lex_ungetch(lex, '\n');
611 /* multiline comment */
612 if (lex->flags.preprocessing) {
615 lex_tokench(lex, '/');
616 lex_tokench(lex, '*');
618 lex_tokench(lex, ' ');
619 lex_tokench(lex, ' ');
628 if (lex->flags.preprocessing) {
630 lex_tokench(lex, '*');
631 lex_tokench(lex, '/');
633 lex_tokench(lex, ' ');
634 lex_tokench(lex, ' ');
638 lex_ungetch(lex, ch);
640 if (lex->flags.preprocessing) {
642 lex_tokench(lex, '\n');
644 lex_tokench(lex, ' '); /* ch); */
647 ch = ' '; /* cause TRUE in the isspace check */
650 /* Otherwise roll back to the slash and break out of the loop */
651 lex_ungetch(lex, ch);
655 } while (ch != EOF && isspace(ch));
659 lex_ungetch(lex, ch);
666 static bool GMQCC_WARN lex_finish_ident(lex_file *lex)
671 while (ch != EOF && isident(ch))
673 lex_tokench(lex, ch);
677 /* last ch was not an ident ch: */
678 lex_ungetch(lex, ch);
683 /* read one ident for the frame list */
684 static int lex_parse_frame(lex_file *lex)
691 while (ch != EOF && ch != '\n' && isspace(ch))
697 if (!isident_start(ch)) {
698 lexerror(lex, "invalid framename, must start with one of a-z or _, got %c", ch);
702 lex_tokench(lex, ch);
703 if (!lex_finish_ident(lex))
709 /* read a list of $frames */
710 static bool lex_finish_frames(lex_file *lex)
717 rc = lex_parse_frame(lex);
718 if (rc > 0) /* end of line */
720 if (rc < 0) /* error */
723 for (i = 0; i < vec_size(lex->frames); ++i) {
724 if (!strcmp(lex->tok.value, lex->frames[i].name)) {
725 lex->frames[i].value = lex->framevalue++;
726 if (lexwarn(lex, WARN_FRAME_MACROS, "duplicate frame macro defined: `%s`", lex->tok.value))
731 if (i < vec_size(lex->frames))
734 m.value = lex->framevalue++;
735 m.name = util_strdup(lex->tok.value);
736 vec_shrinkto(lex->tok.value, 0);
737 vec_push(lex->frames, m);
743 static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote)
748 char u8buf[8]; /* way more than enough */
755 return TOKEN_STRINGCONST;
757 if (lex->flags.preprocessing && ch == '\\') {
758 lex_tokench(lex, ch);
761 lexerror(lex, "unexpected end of file");
762 lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
763 return (lex->tok.ttype = TOKEN_ERROR);
765 lex_tokench(lex, ch);
767 else if (ch == '\\') {
770 lexerror(lex, "unexpected end of file");
771 lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
772 return (lex->tok.ttype = TOKEN_ERROR);
779 case 'a': ch = '\a'; break;
780 case 'b': ch = '\b'; break;
781 case 'r': ch = '\r'; break;
782 case 'n': ch = '\n'; break;
783 case 't': ch = '\t'; break;
784 case 'f': ch = '\f'; break;
785 case 'v': ch = '\v'; break;
788 /* same procedure as in fteqcc */
790 nextch = lex_getch(lex);
791 if (nextch >= '0' && nextch <= '9')
793 else if (nextch >= 'a' && nextch <= 'f')
794 ch += nextch - 'a' + 10;
795 else if (nextch >= 'A' && nextch <= 'F')
796 ch += nextch - 'A' + 10;
798 lexerror(lex, "bad character code");
799 lex_ungetch(lex, nextch);
800 return (lex->tok.ttype = TOKEN_ERROR);
804 nextch = lex_getch(lex);
805 if (nextch >= '0' && nextch <= '9')
807 else if (nextch >= 'a' && nextch <= 'f')
808 ch += nextch - 'a' + 10;
809 else if (nextch >= 'A' && nextch <= 'F')
810 ch += nextch - 'A' + 10;
812 lexerror(lex, "bad character code");
813 lex_ungetch(lex, nextch);
814 return (lex->tok.ttype = TOKEN_ERROR);
819 case '0': case '1': case '2': case '3':
820 case '4': case '5': case '6': case '7':
824 case '<': ch = 29; break;
825 case '-': ch = 30; break;
826 case '>': ch = 31; break;
827 case '[': ch = 16; break;
828 case ']': ch = 17; break;
831 nextch = lex_getch(lex);
832 hex = (nextch == 'x');
834 lex_ungetch(lex, nextch);
835 for (nextch = lex_getch(lex); nextch != '}'; nextch = lex_getch(lex)) {
837 if (nextch >= '0' && nextch <= '9')
838 ch = ch * 10 + nextch - '0';
840 lexerror(lex, "bad character code");
841 return (lex->tok.ttype = TOKEN_ERROR);
844 if (nextch >= '0' || nextch <= '9')
845 ch = ch * 16 + nextch - '0';
846 else if (nextch >= 'a' && nextch <= 'f')
847 ch = ch * 16 + nextch - 'a' + 10;
848 else if (nextch >= 'A' && nextch <= 'F')
849 ch = ch * 16 + nextch - 'A' + 10;
851 lexerror(lex, "bad character code");
852 return (lex->tok.ttype = TOKEN_ERROR);
855 if ( (!OPTS_FLAG(UTF8) && ch > 255) ||
856 ( OPTS_FLAG(UTF8) && ch > 0x10FFFF) )
858 lexerror(lex, "character code out of range");
859 return (lex->tok.ttype = TOKEN_ERROR);
862 if (OPTS_FLAG(UTF8) && ch >= 128) {
863 u8len = u8_fromchar((uchar_t)ch, u8buf, sizeof(u8buf));
868 for (uc = 0; uc < u8len; ++uc)
869 lex_tokench(lex, u8buf[uc]);
870 /* the last character will be inserted with the tokench() call
877 case '\n': ch = '\n'; break;
880 lexwarn(lex, WARN_UNKNOWN_CONTROL_SEQUENCE, "unrecognized control sequence: \\%c", ch);
881 /* so we just add the character plus backslash no matter what it actually is */
882 lex_tokench(lex, '\\');
884 /* add the character finally */
885 lex_tokench(lex, ch);
888 lex_tokench(lex, ch);
890 lexerror(lex, "unexpected end of file within string constant");
891 lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
892 return (lex->tok.ttype = TOKEN_ERROR);
895 static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch)
901 /* parse a number... */
903 lex->tok.ttype = TOKEN_FLOATCONST;
905 lex->tok.ttype = TOKEN_INTCONST;
907 lex_tokench(lex, ch);
910 if (ch != '.' && !isdigit(ch))
912 if (lastch != '0' || ch != 'x')
914 /* end of the number or EOF */
915 lex_ungetch(lex, ch);
918 lex->tok.constval.i = lastch - '0';
919 return lex->tok.ttype;
925 /* EOF would have been caught above */
929 lex_tokench(lex, ch);
931 while (isdigit(ch) || (ishex && isxdigit_only(ch)))
933 lex_tokench(lex, ch);
937 /* NOT else, '.' can come from above as well */
938 if (lex->tok.ttype != TOKEN_FLOATCONST && ch == '.' && !ishex)
940 /* Allow floating comma in non-hex mode */
941 lex->tok.ttype = TOKEN_FLOATCONST;
942 lex_tokench(lex, ch);
944 /* continue digits-only */
948 lex_tokench(lex, ch);
952 /* put back the last character */
953 /* but do not put back the trailing 'f' or a float */
954 if (lex->tok.ttype == TOKEN_FLOATCONST && ch == 'f')
957 /* generally we don't want words to follow numbers: */
959 lexerror(lex, "unexpected trailing characters after number");
960 return (lex->tok.ttype = TOKEN_ERROR);
962 lex_ungetch(lex, ch);
965 if (lex->tok.ttype == TOKEN_FLOATCONST)
966 lex->tok.constval.f = strtod(lex->tok.value, NULL);
968 lex->tok.constval.i = strtol(lex->tok.value, NULL, 0);
969 return lex->tok.ttype;
972 int lex_do(lex_file *lex)
974 int ch, nextch, thirdch;
975 bool hadwhite = false;
984 ch = lex_skipwhite(lex, hadwhite);
986 if (!lex->flags.mergelines || ch != '\\')
990 lex_ungetch(lex, ch);
994 /* we reached a linemerge */
995 lex_tokench(lex, '\n');
999 if (lex->flags.preprocessing && (ch == TOKEN_WHITE || ch == TOKEN_EOL || ch == TOKEN_FATAL)) {
1000 return (lex->tok.ttype = ch);
1003 lex->sline = lex->line;
1004 lex->tok.ctx.line = lex->sline;
1005 lex->tok.ctx.file = lex->name;
1008 return (lex->tok.ttype = TOKEN_FATAL);
1012 return (lex->tok.ttype = TOKEN_EOF);
1015 /* modelgen / spiritgen commands */
1016 if (ch == '$' && !lex->flags.preprocessing) {
1020 ch = lex_getch(lex);
1021 if (!isident_start(ch)) {
1022 lexerror(lex, "hanging '$' modelgen/spritegen command line");
1025 lex_tokench(lex, ch);
1026 if (!lex_finish_ident(lex))
1027 return (lex->tok.ttype = TOKEN_ERROR);
1029 /* skip the known commands */
1032 if (!strcmp(v, "frame") || !strcmp(v, "framesave"))
1034 /* frame/framesave command works like an enum
1035 * similar to fteqcc we handle this in the lexer.
1036 * The reason for this is that it is sensitive to newlines,
1037 * which the parser is unaware of
1039 if (!lex_finish_frames(lex))
1040 return (lex->tok.ttype = TOKEN_ERROR);
1044 if (!strcmp(v, "framevalue"))
1046 ch = lex_getch(lex);
1047 while (ch != EOF && isspace(ch) && ch != '\n')
1048 ch = lex_getch(lex);
1051 lexerror(lex, "$framevalue requires an integer parameter");
1056 lex->tok.ttype = lex_finish_digit(lex, ch);
1058 if (lex->tok.ttype != TOKEN_INTCONST) {
1059 lexerror(lex, "$framevalue requires an integer parameter");
1062 lex->framevalue = lex->tok.constval.i;
1066 if (!strcmp(v, "framerestore"))
1072 rc = lex_parse_frame(lex);
1075 lexerror(lex, "$framerestore requires a framename parameter");
1079 return (lex->tok.ttype = TOKEN_FATAL);
1082 for (frame = 0; frame < vec_size(lex->frames); ++frame) {
1083 if (!strcmp(v, lex->frames[frame].name)) {
1084 lex->framevalue = lex->frames[frame].value;
1088 lexerror(lex, "unknown framename `%s`", v);
1092 if (!strcmp(v, "modelname"))
1098 rc = lex_parse_frame(lex);
1101 lexerror(lex, "$modelname requires a parameter");
1105 return (lex->tok.ttype = TOKEN_FATAL);
1107 if (lex->modelname) {
1109 m.value = lex->framevalue;
1110 m.name = lex->modelname;
1111 lex->modelname = NULL;
1112 vec_push(lex->frames, m);
1114 lex->modelname = lex->tok.value;
1115 lex->tok.value = NULL;
1119 if (!strcmp(v, "flush"))
1122 for (fi = 0; fi < vec_size(lex->frames); ++fi)
1123 mem_d(lex->frames[fi].name);
1124 vec_free(lex->frames);
1125 /* skip line (fteqcc does it too) */
1126 ch = lex_getch(lex);
1127 while (ch != EOF && ch != '\n')
1128 ch = lex_getch(lex);
1132 if (!strcmp(v, "cd") ||
1133 !strcmp(v, "origin") ||
1134 !strcmp(v, "base") ||
1135 !strcmp(v, "flags") ||
1136 !strcmp(v, "scale") ||
1140 ch = lex_getch(lex);
1141 while (ch != EOF && ch != '\n')
1142 ch = lex_getch(lex);
1146 for (frame = 0; frame < vec_size(lex->frames); ++frame) {
1147 if (!strcmp(v, lex->frames[frame].name)) {
1148 lex->tok.constval.i = lex->frames[frame].value;
1149 return (lex->tok.ttype = TOKEN_INTCONST);
1153 lexerror(lex, "invalid frame macro");
1157 /* single-character tokens */
1161 nextch = lex_getch(lex);
1162 if (nextch == '[') {
1163 lex_tokench(lex, ch);
1164 lex_tokench(lex, nextch);
1166 return (lex->tok.ttype = TOKEN_ATTRIBUTE_OPEN);
1168 lex_ungetch(lex, nextch);
1173 lex_tokench(lex, ch);
1175 if (lex->flags.noops)
1176 return (lex->tok.ttype = ch);
1178 return (lex->tok.ttype = TOKEN_OPERATOR);
1181 if (lex->flags.noops) {
1182 nextch = lex_getch(lex);
1183 if (nextch == ']') {
1184 lex_tokench(lex, ch);
1185 lex_tokench(lex, nextch);
1187 return (lex->tok.ttype = TOKEN_ATTRIBUTE_CLOSE);
1189 lex_ungetch(lex, nextch);
1198 lex_tokench(lex, ch);
1200 return (lex->tok.ttype = ch);
1206 nextch = lex_getch(lex);
1207 /* digits starting with a dot */
1208 if (isdigit(nextch)) {
1209 lex_ungetch(lex, nextch);
1210 lex->tok.ttype = lex_finish_digit(lex, ch);
1212 return lex->tok.ttype;
1214 lex_ungetch(lex, nextch);
1217 if (lex->flags.noops)
1219 /* Detect characters early which are normally
1220 * operators OR PART of an operator.
1239 lex_tokench(lex, ch);
1241 return (lex->tok.ttype = ch);
1248 lex_tokench(lex, ch);
1249 /* peak ahead once */
1250 nextch = lex_getch(lex);
1251 if (nextch != '.') {
1252 lex_ungetch(lex, nextch);
1254 return (lex->tok.ttype = ch);
1256 /* peak ahead again */
1257 nextch = lex_getch(lex);
1258 if (nextch != '.') {
1259 lex_ungetch(lex, nextch);
1260 lex_ungetch(lex, '.');
1262 return (lex->tok.ttype = ch);
1264 /* fill the token to be "..." */
1265 lex_tokench(lex, ch);
1266 lex_tokench(lex, ch);
1268 return (lex->tok.ttype = TOKEN_DOTS);
1272 if (ch == ',' || ch == '.') {
1273 lex_tokench(lex, ch);
1275 return (lex->tok.ttype = TOKEN_OPERATOR);
1278 if (ch == '+' || ch == '-' || /* ++, --, +=, -= and -> as well! */
1279 ch == '>' || ch == '<' || /* <<, >>, <=, >= */
1280 ch == '=' || ch == '!' || /* ==, != */
1281 ch == '&' || ch == '|') /* &&, ||, &=, |= */
1283 lex_tokench(lex, ch);
1285 nextch = lex_getch(lex);
1286 if (nextch == '=' || (nextch == ch && ch != '!')) {
1287 lex_tokench(lex, nextch);
1288 } else if (ch == '-' && nextch == '>') {
1289 lex_tokench(lex, nextch);
1290 } else if (ch == '&' && nextch == '~') {
1291 thirdch = lex_getch(lex);
1292 if (thirdch != '=') {
1293 lex_ungetch(lex, thirdch);
1294 lex_ungetch(lex, nextch);
1297 lex_tokench(lex, nextch);
1298 lex_tokench(lex, thirdch);
1301 lex_ungetch(lex, nextch);
1304 return (lex->tok.ttype = TOKEN_OPERATOR);
1308 if (ch == '^' || ch == '~' || ch == '!')
1310 lex_tokench(lex, ch);
1312 return (lex->tok.ttype = TOKEN_OPERATOR);
1316 if (ch == '*' || ch == '/') /* *=, /= */
1318 lex_tokench(lex, ch);
1320 nextch = lex_getch(lex);
1321 if (nextch == '=') {
1322 lex_tokench(lex, nextch);
1324 lex_ungetch(lex, nextch);
1327 return (lex->tok.ttype = TOKEN_OPERATOR);
1330 if (isident_start(ch))
1334 lex_tokench(lex, ch);
1335 if (!lex_finish_ident(lex)) {
1337 return (lex->tok.ttype = TOKEN_ERROR);
1340 lex->tok.ttype = TOKEN_IDENT;
1343 if (!strcmp(v, "void")) {
1344 lex->tok.ttype = TOKEN_TYPENAME;
1345 lex->tok.constval.t = TYPE_VOID;
1346 } else if (!strcmp(v, "int")) {
1347 lex->tok.ttype = TOKEN_TYPENAME;
1348 lex->tok.constval.t = TYPE_INTEGER;
1349 } else if (!strcmp(v, "float")) {
1350 lex->tok.ttype = TOKEN_TYPENAME;
1351 lex->tok.constval.t = TYPE_FLOAT;
1352 } else if (!strcmp(v, "string")) {
1353 lex->tok.ttype = TOKEN_TYPENAME;
1354 lex->tok.constval.t = TYPE_STRING;
1355 } else if (!strcmp(v, "entity")) {
1356 lex->tok.ttype = TOKEN_TYPENAME;
1357 lex->tok.constval.t = TYPE_ENTITY;
1358 } else if (!strcmp(v, "vector")) {
1359 lex->tok.ttype = TOKEN_TYPENAME;
1360 lex->tok.constval.t = TYPE_VECTOR;
1363 for (kw = 0; kw < num_keywords_qc; ++kw) {
1364 if (!strcmp(v, keywords_qc[kw]))
1365 return (lex->tok.ttype = TOKEN_KEYWORD);
1367 if (opts.standard != COMPILER_QCC) {
1368 for (kw = 0; kw < num_keywords_fg; ++kw) {
1369 if (!strcmp(v, keywords_fg[kw]))
1370 return (lex->tok.ttype = TOKEN_KEYWORD);
1375 return lex->tok.ttype;
1380 lex->flags.nodigraphs = true;
1381 if (lex->flags.preprocessing)
1382 lex_tokench(lex, ch);
1383 lex->tok.ttype = lex_finish_string(lex, '"');
1384 if (lex->flags.preprocessing)
1385 lex_tokench(lex, ch);
1386 while (!lex->flags.preprocessing && lex->tok.ttype == TOKEN_STRINGCONST)
1388 /* Allow c style "string" "continuation" */
1389 ch = lex_skipwhite(lex, false);
1391 lex_ungetch(lex, ch);
1395 lex->tok.ttype = lex_finish_string(lex, '"');
1397 lex->flags.nodigraphs = false;
1399 return lex->tok.ttype;
1404 /* we parse character constants like string,
1405 * but return TOKEN_CHARCONST, or a vector type if it fits...
1406 * Likewise actual unescaping has to be done by the parser.
1407 * The difference is we don't allow 'char' 'continuation'.
1409 if (lex->flags.preprocessing)
1410 lex_tokench(lex, ch);
1411 lex->tok.ttype = lex_finish_string(lex, '\'');
1412 if (lex->flags.preprocessing)
1413 lex_tokench(lex, ch);
1416 lex->tok.ttype = TOKEN_CHARCONST;
1417 /* It's a vector if we can successfully scan 3 floats */
1419 if (sscanf_s(lex->tok.value, " %f %f %f ",
1420 &lex->tok.constval.v.x, &lex->tok.constval.v.y, &lex->tok.constval.v.z) == 3)
1422 if (sscanf(lex->tok.value, " %f %f %f ",
1423 &lex->tok.constval.v.x, &lex->tok.constval.v.y, &lex->tok.constval.v.z) == 3)
1427 lex->tok.ttype = TOKEN_VECTORCONST;
1431 if (!lex->flags.preprocessing && strlen(lex->tok.value) > 1) {
1433 /* check for a valid utf8 character */
1434 if (!OPTS_FLAG(UTF8) || !u8_analyze(lex->tok.value, NULL, NULL, &u8char, 8)) {
1435 if (lexwarn(lex, WARN_MULTIBYTE_CHARACTER,
1436 ( OPTS_FLAG(UTF8) ? "invalid multibyte character sequence `%s`"
1437 : "multibyte character: `%s`" ),
1439 return (lex->tok.ttype = TOKEN_ERROR);
1442 lex->tok.constval.i = u8char;
1445 lex->tok.constval.i = lex->tok.value[0];
1448 return lex->tok.ttype;
1453 lex->tok.ttype = lex_finish_digit(lex, ch);
1455 return lex->tok.ttype;
1458 if (lex->flags.preprocessing) {
1459 lex_tokench(lex, ch);
1461 return (lex->tok.ttype = ch);
1464 lexerror(lex, "unknown token");
1465 return (lex->tok.ttype = TOKEN_ERROR);