9 MEM_VEC_FUNCTIONS(token, char, value)
10 MEM_VEC_FUNCTIONS(lex_file, frame_macro, frames)
12 VECTOR_MAKE(char*, lex_filenames);
14 void lexerror(lex_file *lex, const char *fmt, ...)
19 vprintmsg(LVL_ERROR, lex->name, lex->sline, "parse error", fmt, ap);
23 bool lexwarn(lex_file *lex, int warntype, const char *fmt, ...)
26 int lvl = LVL_WARNING;
28 if (!OPTS_WARN(warntype))
35 vprintmsg(lvl, lex->name, lex->sline, "warning", fmt, ap);
45 token *tok = (token*)mem_a(sizeof(token));
48 memset(tok, 0, sizeof(*tok));
52 void token_delete(token *self)
54 if (self->next && self->next->prev == self)
55 self->next->prev = self->prev;
56 if (self->prev && self->prev->next == self)
57 self->prev->next = self->next;
58 MEM_VECTOR_CLEAR(self, value);
62 token* token_copy(const token *cp)
64 token* self = token_new();
68 self->value_alloc = cp->value_count + 1;
69 self->value_count = cp->value_count;
70 self->value = (char*)mem_a(self->value_alloc);
75 memcpy(self->value, cp->value, cp->value_count);
76 self->value[self->value_alloc-1] = 0;
80 self->ttype = cp->ttype;
81 memcpy(&self->constval, &cp->constval, sizeof(self->constval));
85 void token_delete_all(token *t)
96 token* token_copy_all(const token *cp)
101 out = cur = token_copy(cp);
107 cur->next = token_copy(cp);
109 token_delete_all(out);
112 cur->next->prev = cur;
119 static void lex_token_new(lex_file *lex)
123 token_delete(lex->tok);
124 lex->tok = token_new();
126 lex->tok.value_count = 0;
127 lex->tok.constval.t = 0;
128 lex->tok.ctx.line = lex->sline;
129 lex->tok.ctx.file = lex->name;
134 lex_file* lex_open(const char *file)
137 FILE *in = util_fopen(file, "rb");
140 lexerror(NULL, "open failed: '%s'\n", file);
144 lex = (lex_file*)mem_a(sizeof(*lex));
147 lexerror(NULL, "out of memory\n");
151 memset(lex, 0, sizeof(*lex));
154 lex->name = util_strdup(file);
155 lex->line = 1; /* we start counting at 1 */
160 lex_filenames_add(lex->name);
165 void lex_cleanup(void)
168 for (i = 0; i < lex_filenames_elements; ++i)
169 mem_d(lex_filenames_data[i]);
170 mem_d(lex_filenames_data);
173 void lex_close(lex_file *lex)
176 for (i = 0; i < lex->frames_count; ++i)
177 mem_d(lex->frames[i].name);
178 MEM_VECTOR_CLEAR(lex, frames);
181 mem_d(lex->modelname);
187 token_delete(lex->tok);
189 MEM_VECTOR_CLEAR(&(lex->tok), value);
191 /* mem_d(lex->name); collected in lex_filenames */
195 /* Get or put-back data
196 * The following to functions do NOT understand what kind of data they
198 * The are merely wrapping get/put in order to count line numbers.
200 static void lex_ungetch(lex_file *lex, int ch);
201 static int lex_try_trigraph(lex_file *lex, int old)
204 c2 = fgetc(lex->file);
206 lex_ungetch(lex, c2);
210 c3 = fgetc(lex->file);
212 case '=': return '#';
213 case '/': return '\\';
214 case '\'': return '^';
215 case '(': return '[';
216 case ')': return ']';
217 case '!': return '|';
218 case '<': return '{';
219 case '>': return '}';
220 case '-': return '~';
222 lex_ungetch(lex, c3);
223 lex_ungetch(lex, c2);
228 static int lex_try_digraph(lex_file *lex, int ch)
231 c2 = fgetc(lex->file);
232 if (ch == '<' && c2 == ':')
234 else if (ch == ':' && c2 == '>')
236 else if (ch == '<' && c2 == '%')
238 else if (ch == '%' && c2 == '>')
240 else if (ch == '%' && c2 == ':')
242 lex_ungetch(lex, c2);
246 static int lex_getch(lex_file *lex)
252 if (lex->peek[lex->peekpos] == '\n')
254 return lex->peek[lex->peekpos];
257 ch = fgetc(lex->file);
261 return lex_try_trigraph(lex, ch);
262 else if (ch == '<' || ch == ':' || ch == '%')
263 return lex_try_digraph(lex, ch);
267 static void lex_ungetch(lex_file *lex, int ch)
269 lex->peek[lex->peekpos++] = ch;
274 /* classify characters
275 * some additions to the is*() functions of ctype.h
278 /* Idents are alphanumberic, but they start with alpha or _ */
279 static bool isident_start(int ch)
281 return isalpha(ch) || ch == '_';
284 static bool isident(int ch)
286 return isident_start(ch) || isdigit(ch);
289 /* isxdigit_only is used when we already know it's not a digit
290 * and want to see if it's a hex digit anyway.
292 static bool isxdigit_only(int ch)
294 return (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F');
297 /* Skip whitespace and comments and return the first
298 * non-white character.
299 * As this makes use of the above getch() ungetch() functions,
300 * we don't need to care at all about line numbering anymore.
302 * In theory, this function should only be used at the beginning
303 * of lexing, or when we *know* the next character is part of the token.
304 * Otherwise, if the parser throws an error, the linenumber may not be
305 * the line of the error, but the line of the next token AFTER the error.
307 * This is currently only problematic when using c-like string-continuation,
308 * since comments and whitespaces are allowed between 2 such strings.
312 "A continuation of the previous string"
313 // This line is skipped
316 * In this case, if the parse decides it didn't actually want a string,
317 * and uses lex->line to print an error, it will show the ', foo);' line's
320 * On the other hand, the parser is supposed to remember the line of the next
321 * token's beginning. In this case we would want skipwhite() to be called
322 * AFTER reading a token, so that the parser, before reading the NEXT token,
323 * doesn't store teh *comment's* linenumber, but the actual token's linenumber.
326 * here is to store the line of the first character after skipping
327 * the initial whitespace in lex->sline, this happens in lex_do.
329 static int lex_skipwhite(lex_file *lex)
336 while (ch != EOF && isspace(ch)) ch = lex_getch(lex);
342 /* one line comment */
345 /* check for special: '/', '/', '*', '/' */
354 while (ch != EOF && ch != '\n') {
361 /* multiline comment */
373 if (ch == '/') /* allow *//* direct following comment */
375 lex_ungetch(lex, ch);
376 ch = ' '; /* cause TRUE in the isspace check */
380 /* Otherwise roll back to the slash and break out of the loop */
381 lex_ungetch(lex, ch);
385 } while (ch != EOF && isspace(ch));
390 /* Append a character to the token buffer */
391 static bool GMQCC_WARN lex_tokench(lex_file *lex, int ch)
393 if (!token_value_add(&lex->tok, ch)) {
394 lexerror(lex, "out of memory");
400 /* Append a trailing null-byte */
401 static bool GMQCC_WARN lex_endtoken(lex_file *lex)
403 if (!token_value_add(&lex->tok, 0)) {
404 lexerror(lex, "out of memory");
407 lex->tok.value_count--;
412 static bool GMQCC_WARN lex_finish_ident(lex_file *lex)
417 while (ch != EOF && isident(ch))
419 if (!lex_tokench(lex, ch))
420 return (lex->tok.ttype = TOKEN_FATAL);
424 /* last ch was not an ident ch: */
425 lex_ungetch(lex, ch);
430 /* read one ident for the frame list */
431 static int lex_parse_frame(lex_file *lex)
438 while (ch != EOF && ch != '\n' && isspace(ch))
444 if (!isident_start(ch)) {
445 lexerror(lex, "invalid framename, must start with one of a-z or _, got %c", ch);
449 if (!lex_tokench(lex, ch))
451 if (!lex_finish_ident(lex))
453 if (!lex_endtoken(lex))
458 /* read a list of $frames */
459 static bool lex_finish_frames(lex_file *lex)
466 rc = lex_parse_frame(lex);
467 if (rc > 0) /* end of line */
469 if (rc < 0) /* error */
472 for (i = 0; i < lex->frames_count; ++i) {
473 if (!strcmp(lex->tok.value, lex->frames[i].name)) {
474 lex->frames[i].value = lex->framevalue++;
475 if (lexwarn(lex, WARN_FRAME_MACROS, "duplicate frame macro defined: `%s`", lex->tok.value))
480 if (i < lex->frames_count)
483 m.value = lex->framevalue++;
484 m.name = lex->tok.value;
485 lex->tok.value = NULL;
486 lex->tok.value_alloc = lex->tok.value_count = 0;
487 if (!lex_file_frames_add(lex, m)) {
488 lexerror(lex, "out of memory");
494 static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote)
502 return TOKEN_STRINGCONST;
507 lexerror(lex, "unexpected end of file");
508 lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
509 return (lex->tok.ttype = TOKEN_ERROR);
514 case 'a': ch = '\a'; break;
515 case 'b': ch = '\b'; break;
516 case 'r': ch = '\r'; break;
517 case 'n': ch = '\n'; break;
518 case 't': ch = '\t'; break;
519 case 'f': ch = '\f'; break;
520 case 'v': ch = '\v'; break;
522 lexwarn(lex, WARN_UNKNOWN_CONTROL_SEQUENCE, "unrecognized control sequence: \\%c", ch);
523 /* so we just add the character plus backslash no matter what it actually is */
524 if (!lex_tokench(lex, '\\'))
525 return (lex->tok.ttype = TOKEN_FATAL);
527 /* add the character finally */
528 if (!lex_tokench(lex, ch))
529 return (lex->tok.ttype = TOKEN_FATAL);
531 else if (!lex_tokench(lex, ch))
532 return (lex->tok.ttype = TOKEN_FATAL);
534 lexerror(lex, "unexpected end of file within string constant");
535 lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
536 return (lex->tok.ttype = TOKEN_ERROR);
539 static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch)
545 /* parse a number... */
546 lex->tok.ttype = TOKEN_INTCONST;
548 if (!lex_tokench(lex, ch))
549 return (lex->tok.ttype = TOKEN_FATAL);
552 if (ch != '.' && !isdigit(ch))
554 if (lastch != '0' || ch != 'x')
556 /* end of the number or EOF */
557 lex_ungetch(lex, ch);
558 if (!lex_endtoken(lex))
559 return (lex->tok.ttype = TOKEN_FATAL);
561 lex->tok.constval.i = lastch - '0';
562 return lex->tok.ttype;
568 /* EOF would have been caught above */
572 if (!lex_tokench(lex, ch))
573 return (lex->tok.ttype = TOKEN_FATAL);
575 while (isdigit(ch) || (ishex && isxdigit_only(ch)))
577 if (!lex_tokench(lex, ch))
578 return (lex->tok.ttype = TOKEN_FATAL);
582 /* NOT else, '.' can come from above as well */
583 if (ch == '.' && !ishex)
585 /* Allow floating comma in non-hex mode */
586 lex->tok.ttype = TOKEN_FLOATCONST;
587 if (!lex_tokench(lex, ch))
588 return (lex->tok.ttype = TOKEN_FATAL);
590 /* continue digits-only */
594 if (!lex_tokench(lex, ch))
595 return (lex->tok.ttype = TOKEN_FATAL);
599 /* put back the last character */
600 /* but do not put back the trailing 'f' or a float */
601 if (lex->tok.ttype == TOKEN_FLOATCONST && ch == 'f')
604 /* generally we don't want words to follow numbers: */
606 lexerror(lex, "unexpected trailing characters after number");
607 return (lex->tok.ttype = TOKEN_ERROR);
609 lex_ungetch(lex, ch);
611 if (!lex_endtoken(lex))
612 return (lex->tok.ttype = TOKEN_FATAL);
613 if (lex->tok.ttype == TOKEN_FLOATCONST)
614 lex->tok.constval.f = strtod(lex->tok.value, NULL);
616 lex->tok.constval.i = strtol(lex->tok.value, NULL, 0);
617 return lex->tok.ttype;
620 int lex_do(lex_file *lex)
630 ch = lex_skipwhite(lex);
631 lex->sline = lex->line;
632 lex->tok.ctx.line = lex->sline;
633 lex->tok.ctx.file = lex->name;
636 return (lex->tok.ttype = TOKEN_FATAL);
640 return (lex->tok.ttype = TOKEN_EOF);
643 /* modelgen / spiritgen commands */
649 if (!isident_start(ch)) {
650 lexerror(lex, "hanging '$' modelgen/spritegen command line");
653 if (!lex_tokench(lex, ch))
654 return (lex->tok.ttype = TOKEN_FATAL);
655 if (!lex_finish_ident(lex))
656 return (lex->tok.ttype = TOKEN_ERROR);
657 if (!lex_endtoken(lex))
658 return (lex->tok.ttype = TOKEN_FATAL);
659 /* skip the known commands */
662 if (!strcmp(v, "frame") || !strcmp(v, "framesave"))
664 /* frame/framesave command works like an enum
665 * similar to fteqcc we handle this in the lexer.
666 * The reason for this is that it is sensitive to newlines,
667 * which the parser is unaware of
669 if (!lex_finish_frames(lex))
670 return (lex->tok.ttype = TOKEN_ERROR);
674 if (!strcmp(v, "framevalue"))
677 while (ch != EOF && isspace(ch) && ch != '\n')
681 lexerror(lex, "$framevalue requires an integer parameter");
686 lex->tok.ttype = lex_finish_digit(lex, ch);
687 if (!lex_endtoken(lex))
688 return (lex->tok.ttype = TOKEN_FATAL);
689 if (lex->tok.ttype != TOKEN_INTCONST) {
690 lexerror(lex, "$framevalue requires an integer parameter");
693 lex->framevalue = lex->tok.constval.i;
697 if (!strcmp(v, "framerestore"))
703 rc = lex_parse_frame(lex);
706 lexerror(lex, "$framerestore requires a framename parameter");
710 return (lex->tok.ttype = TOKEN_FATAL);
713 for (frame = 0; frame < lex->frames_count; ++frame) {
714 if (!strcmp(v, lex->frames[frame].name)) {
715 lex->framevalue = lex->frames[frame].value;
719 lexerror(lex, "unknown framename `%s`", v);
723 if (!strcmp(v, "modelname"))
729 rc = lex_parse_frame(lex);
732 lexerror(lex, "$framerestore requires a framename parameter");
736 return (lex->tok.ttype = TOKEN_FATAL);
739 if (lex->modelname) {
741 m.value = lex->framevalue;
742 m.name = lex->modelname;
743 lex->modelname = NULL;
744 if (!lex_file_frames_add(lex, m)) {
745 lexerror(lex, "out of memory");
746 return (lex->tok.ttype = TOKEN_FATAL);
749 lex->modelname = lex->tok.value;
750 lex->tok.value = NULL;
751 lex->tok.value_alloc = lex->tok.value_count = 0;
752 for (frame = 0; frame < lex->frames_count; ++frame) {
753 if (!strcmp(v, lex->frames[frame].name)) {
754 lex->framevalue = lex->frames[frame].value;
761 if (!strcmp(v, "flush"))
764 for (frame = 0; frame < lex->frames_count; ++frame)
765 mem_d(lex->frames[frame].name);
766 MEM_VECTOR_CLEAR(lex, frames);
767 /* skip line (fteqcc does it too) */
769 while (ch != EOF && ch != '\n')
774 if (!strcmp(v, "cd") ||
775 !strcmp(v, "origin") ||
776 !strcmp(v, "base") ||
777 !strcmp(v, "flags") ||
778 !strcmp(v, "scale") ||
783 while (ch != EOF && ch != '\n')
788 for (frame = 0; frame < lex->frames_count; ++frame) {
789 if (!strcmp(v, lex->frames[frame].name)) {
790 lex->tok.constval.i = lex->frames[frame].value;
791 return (lex->tok.ttype = TOKEN_INTCONST);
795 lexerror(lex, "invalid frame macro");
799 /* single-character tokens */
803 if (!lex_tokench(lex, ch) ||
806 return (lex->tok.ttype = TOKEN_FATAL);
808 if (lex->flags.noops)
809 return (lex->tok.ttype = ch);
811 return (lex->tok.ttype = TOKEN_OPERATOR);
820 if (!lex_tokench(lex, ch) ||
823 return (lex->tok.ttype = TOKEN_FATAL);
825 return (lex->tok.ttype = ch);
830 if (lex->flags.noops)
832 /* Detect characters early which are normally
833 * operators OR PART of an operator.
850 if (!lex_tokench(lex, ch) ||
853 return (lex->tok.ttype = TOKEN_FATAL);
855 return (lex->tok.ttype = ch);
862 if (!lex_tokench(lex, ch))
863 return (lex->tok.ttype = TOKEN_FATAL);
864 /* peak ahead once */
865 nextch = lex_getch(lex);
867 lex_ungetch(lex, nextch);
868 if (!lex_endtoken(lex))
869 return (lex->tok.ttype = TOKEN_FATAL);
870 return (lex->tok.ttype = ch);
872 /* peak ahead again */
873 nextch = lex_getch(lex);
875 lex_ungetch(lex, nextch);
876 lex_ungetch(lex, nextch);
877 if (!lex_endtoken(lex))
878 return (lex->tok.ttype = TOKEN_FATAL);
879 return (lex->tok.ttype = ch);
881 /* fill the token to be "..." */
882 if (!lex_tokench(lex, ch) ||
883 !lex_tokench(lex, ch) ||
886 return (lex->tok.ttype = TOKEN_FATAL);
888 return (lex->tok.ttype = TOKEN_DOTS);
892 if (ch == ',' || ch == '.') {
893 if (!lex_tokench(lex, ch) ||
896 return (lex->tok.ttype = TOKEN_FATAL);
898 return (lex->tok.ttype = TOKEN_OPERATOR);
901 if (ch == '+' || ch == '-' || /* ++, --, +=, -= and -> as well! */
902 ch == '>' || ch == '<' || /* <<, >>, <=, >= */
903 ch == '=' || ch == '!' || /* ==, != */
904 ch == '&' || ch == '|') /* &&, ||, &=, |= */
906 if (!lex_tokench(lex, ch))
907 return (lex->tok.ttype = TOKEN_FATAL);
909 nextch = lex_getch(lex);
910 if (nextch == ch || nextch == '=') {
911 if (!lex_tokench(lex, nextch))
912 return (lex->tok.ttype = TOKEN_FATAL);
913 } else if (ch == '-' && nextch == '>') {
914 if (!lex_tokench(lex, nextch))
915 return (lex->tok.ttype = TOKEN_FATAL);
917 lex_ungetch(lex, nextch);
919 if (!lex_endtoken(lex))
920 return (lex->tok.ttype = TOKEN_FATAL);
921 return (lex->tok.ttype = TOKEN_OPERATOR);
925 if (ch == '^' || ch == '~' || ch == '!')
927 if (!lex_tokench(lex, ch) ||
930 return (lex->tok.ttype = TOKEN_FATAL);
932 return (lex->tok.ttype = TOKEN_OPERATOR);
936 if (ch == '*' || ch == '/') /* *=, /= */
938 if (!lex_tokench(lex, ch))
939 return (lex->tok.ttype = TOKEN_FATAL);
941 nextch = lex_getch(lex);
943 if (!lex_tokench(lex, nextch))
944 return (lex->tok.ttype = TOKEN_FATAL);
946 lex_ungetch(lex, nextch);
948 if (!lex_endtoken(lex))
949 return (lex->tok.ttype = TOKEN_FATAL);
950 return (lex->tok.ttype = TOKEN_OPERATOR);
953 if (isident_start(ch))
957 if (!lex_tokench(lex, ch))
958 return (lex->tok.ttype = TOKEN_FATAL);
959 if (!lex_finish_ident(lex)) {
961 return (lex->tok.ttype = TOKEN_ERROR);
963 if (!lex_endtoken(lex))
964 return (lex->tok.ttype = TOKEN_FATAL);
965 lex->tok.ttype = TOKEN_IDENT;
968 if (!strcmp(v, "void")) {
969 lex->tok.ttype = TOKEN_TYPENAME;
970 lex->tok.constval.t = TYPE_VOID;
971 } else if (!strcmp(v, "int")) {
972 lex->tok.ttype = TOKEN_TYPENAME;
973 lex->tok.constval.t = TYPE_INTEGER;
974 } else if (!strcmp(v, "float")) {
975 lex->tok.ttype = TOKEN_TYPENAME;
976 lex->tok.constval.t = TYPE_FLOAT;
977 } else if (!strcmp(v, "string")) {
978 lex->tok.ttype = TOKEN_TYPENAME;
979 lex->tok.constval.t = TYPE_STRING;
980 } else if (!strcmp(v, "entity")) {
981 lex->tok.ttype = TOKEN_TYPENAME;
982 lex->tok.constval.t = TYPE_ENTITY;
983 } else if (!strcmp(v, "vector")) {
984 lex->tok.ttype = TOKEN_TYPENAME;
985 lex->tok.constval.t = TYPE_VECTOR;
986 } else if (!strcmp(v, "for") ||
987 !strcmp(v, "while") ||
990 !strcmp(v, "else") ||
991 !strcmp(v, "local") ||
992 !strcmp(v, "return") ||
994 lex->tok.ttype = TOKEN_KEYWORD;
996 return lex->tok.ttype;
1001 lex->tok.ttype = lex_finish_string(lex, '"');
1002 while (lex->tok.ttype == TOKEN_STRINGCONST)
1004 /* Allow c style "string" "continuation" */
1005 ch = lex_skipwhite(lex);
1007 lex_ungetch(lex, ch);
1011 lex->tok.ttype = lex_finish_string(lex, '"');
1013 if (!lex_endtoken(lex))
1014 return (lex->tok.ttype = TOKEN_FATAL);
1015 return lex->tok.ttype;
1020 /* we parse character constants like string,
1021 * but return TOKEN_CHARCONST, or a vector type if it fits...
1022 * Likewise actual unescaping has to be done by the parser.
1023 * The difference is we don't allow 'char' 'continuation'.
1025 lex->tok.ttype = lex_finish_string(lex, '\'');
1026 if (!lex_endtoken(lex))
1027 return (lex->tok.ttype = TOKEN_FATAL);
1029 /* It's a vector if we can successfully scan 3 floats */
1031 if (sscanf_s(lex->tok.value, " %f %f %f ",
1032 &lex->tok.constval.v.x, &lex->tok.constval.v.y, &lex->tok.constval.v.z) == 3)
1034 if (sscanf(lex->tok.value, " %f %f %f ",
1035 &lex->tok.constval.v.x, &lex->tok.constval.v.y, &lex->tok.constval.v.z) == 3)
1038 lex->tok.ttype = TOKEN_VECTORCONST;
1041 return lex->tok.ttype;
1046 lex->tok.ttype = lex_finish_digit(lex, ch);
1047 if (!lex_endtoken(lex))
1048 return (lex->tok.ttype = TOKEN_FATAL);
1049 return lex->tok.ttype;
1052 lexerror(lex, "unknown token");
1053 return (lex->tok.ttype = TOKEN_ERROR);