X-Git-Url: https://git.xonotic.org/?p=xonotic%2Fgmqcc.git;a=blobdiff_plain;f=lexer.c;h=db0a8f4d9bf541bdd25597c5047e040d7dc36fc5;hp=17af7b6f4b92ad7ca30c2da8bd4f5a2161afbb7b;hb=160e7cf7eebd7fa173fb739aca00143097a3518b;hpb=f65dbda6595f5fe33f263767c37e78871770e06c diff --git a/lexer.c b/lexer.c index 17af7b6..db0a8f4 100644 --- a/lexer.c +++ b/lexer.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2012 + * Copyright (C) 2012, 2013 * Wolfgang Bumiller * * Permission is hereby granted, free of charge, to any person obtaining a copy of @@ -47,7 +47,10 @@ static const char *keywords_fg[] = { "switch", "case", "default", "struct", "union", "break", "continue", - "typedef" + "typedef", + "goto", + + "__builtin_debug_printtype" }; static size_t num_keywords_fg = sizeof(keywords_fg) / sizeof(keywords_fg[0]); @@ -71,20 +74,17 @@ void lexerror(lex_file *lex, const char *fmt, ...) bool lexwarn(lex_file *lex, int warntype, const char *fmt, ...) { + bool r; + lex_ctx ctx; va_list ap; - int lvl = LVL_WARNING; - - if (!OPTS_WARN(warntype)) - return false; - if (opts_werror) - lvl = LVL_ERROR; + ctx.file = lex->name; + ctx.line = lex->sline; va_start(ap, fmt); - con_vprintmsg(lvl, lex->name, lex->sline, "warning", fmt, ap); + r = vcompile_warning(ctx, warntype, fmt, ap); va_end(ap); - - return opts_werror; + return r; } @@ -184,7 +184,7 @@ static void lex_token_new(lex_file *lex) lex_file* lex_open(const char *file) { lex_file *lex; - FILE *in = util_fopen(file, "rb"); + FILE *in = fs_file_open(file, "rb"); if (!in) { lexerror(NULL, "open failed: '%s'\n", file); @@ -193,7 +193,7 @@ lex_file* lex_open(const char *file) lex = (lex_file*)mem_a(sizeof(*lex)); if (!lex) { - fclose(in); + fs_file_close(in); lexerror(NULL, "out of memory\n"); return NULL; } @@ -258,7 +258,7 @@ void lex_close(lex_file *lex) vec_free(lex->modelname); if (lex->file) - fclose(lex->file); + fs_file_close(lex->file); #if 0 if (lex->tok) token_delete(lex->tok); @@ -272,7 +272,7 @@ void lex_close(lex_file *lex) static int lex_fgetc(lex_file *lex) { if (lex->file) - return fgetc(lex->file); + return fs_file_getc(lex->file); if (lex->open_string) { if (lex->open_string_pos >= lex->open_string_length) return EOF; @@ -291,12 +291,16 @@ static int lex_try_trigraph(lex_file *lex, int old) { int c2, c3; c2 = lex_fgetc(lex); + if (!lex->push_line && c2 == '\n') + lex->line++; if (c2 != '?') { lex_ungetch(lex, c2); return old; } c3 = lex_fgetc(lex); + if (!lex->push_line && c3 == '\n') + lex->line++; switch (c3) { case '=': return '#'; case '/': return '\\'; @@ -318,6 +322,11 @@ static int lex_try_digraph(lex_file *lex, int ch) { int c2; c2 = lex_fgetc(lex); + /* we just used fgetc() so count lines + * need to offset a \n the ungetch would recognize + */ + if (!lex->push_line && c2 == '\n') + lex->line++; if (ch == '<' && c2 == ':') return '['; else if (ch == ':' && c2 == '>') @@ -433,7 +442,7 @@ static bool lex_try_pragma(lex_file *lex) goto unroll; } - for (ch = lex_getch(lex); vec_size(param) < 32 && ch != ')' && ch != '\n'; ch = lex_getch(lex)) + for (ch = lex_getch(lex); vec_size(param) < 1024 && ch != ')' && ch != '\n'; ch = lex_getch(lex)) vec_push(param, ch); vec_push(param, 0); @@ -474,29 +483,34 @@ static bool lex_try_pragma(lex_file *lex) lex->line = line; while (ch != '\n' && ch != EOF) ch = lex_getch(lex); + vec_free(command); + vec_free(param); + vec_free(pragma); return true; unroll: if (command) { vec_pop(command); while (vec_size(command)) { - lex_ungetch(lex, vec_last(command)); + lex_ungetch(lex, (unsigned char)vec_last(command)); vec_pop(command); } vec_free(command); + lex_ungetch(lex, ' '); } - if (command) { - vec_pop(command); - while (vec_size(command)) { - lex_ungetch(lex, vec_last(command)); - vec_pop(command); + if (param) { + vec_pop(param); + while (vec_size(param)) { + lex_ungetch(lex, (unsigned char)vec_last(param)); + vec_pop(param); } - vec_free(command); + vec_free(param); + lex_ungetch(lex, ' '); } if (pragma) { vec_pop(pragma); while (vec_size(pragma)) { - lex_ungetch(lex, vec_last(pragma)); + lex_ungetch(lex, (unsigned char)vec_last(pragma)); vec_pop(pragma); } vec_free(pragma); @@ -539,10 +553,10 @@ printf( "line one\n" * here is to store the line of the first character after skipping * the initial whitespace in lex->sline, this happens in lex_do. */ -static int lex_skipwhite(lex_file *lex) +static int lex_skipwhite(lex_file *lex, bool hadwhite) { int ch = 0; - bool haswhite = false; + bool haswhite = hadwhite; do { @@ -729,11 +743,18 @@ static bool lex_finish_frames(lex_file *lex) vec_shrinkto(lex->tok.value, 0); vec_push(lex->frames, m); } while (true); + + return false; } static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote) { + uchar_t chr; int ch = 0; + int nextch; + bool hex; + char u8buf[8]; /* way more than enough */ + int u8len, uc; while (ch != EOF) { @@ -770,6 +791,100 @@ static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote) case 't': ch = '\t'; break; case 'f': ch = '\f'; break; case 'v': ch = '\v'; break; + case 'x': + case 'X': + /* same procedure as in fteqcc */ + ch = 0; + nextch = lex_getch(lex); + if (nextch >= '0' && nextch <= '9') + ch += nextch - '0'; + else if (nextch >= 'a' && nextch <= 'f') + ch += nextch - 'a' + 10; + else if (nextch >= 'A' && nextch <= 'F') + ch += nextch - 'A' + 10; + else { + lexerror(lex, "bad character code"); + lex_ungetch(lex, nextch); + return (lex->tok.ttype = TOKEN_ERROR); + } + + ch *= 0x10; + nextch = lex_getch(lex); + if (nextch >= '0' && nextch <= '9') + ch += nextch - '0'; + else if (nextch >= 'a' && nextch <= 'f') + ch += nextch - 'a' + 10; + else if (nextch >= 'A' && nextch <= 'F') + ch += nextch - 'A' + 10; + else { + lexerror(lex, "bad character code"); + lex_ungetch(lex, nextch); + return (lex->tok.ttype = TOKEN_ERROR); + } + break; + + /* fteqcc support */ + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': + case '8': case '9': + ch = 18 + ch - '0'; + break; + case '<': ch = 29; break; + case '-': ch = 30; break; + case '>': ch = 31; break; + case '[': ch = 16; break; + case ']': ch = 17; break; + case '{': + chr = 0; + nextch = lex_getch(lex); + hex = (nextch == 'x'); + if (!hex) + lex_ungetch(lex, nextch); + for (nextch = lex_getch(lex); nextch != '}'; nextch = lex_getch(lex)) { + if (!hex) { + if (nextch >= '0' && nextch <= '9') + chr = chr * 10 + nextch - '0'; + else { + lexerror(lex, "bad character code"); + return (lex->tok.ttype = TOKEN_ERROR); + } + } else { + if (nextch >= '0' && nextch <= '9') + chr = chr * 0x10 + nextch - '0'; + else if (nextch >= 'a' && nextch <= 'f') + chr = chr * 0x10 + nextch - 'a' + 10; + else if (nextch >= 'A' && nextch <= 'F') + chr = chr * 0x10 + nextch - 'A' + 10; + else { + lexerror(lex, "bad character code"); + return (lex->tok.ttype = TOKEN_ERROR); + } + } + if (chr > 0x10FFFF || (!OPTS_FLAG(UTF8) && chr > 255)) + { + lexerror(lex, "character code out of range"); + return (lex->tok.ttype = TOKEN_ERROR); + } + } + if (OPTS_FLAG(UTF8) && chr >= 128) { + u8len = u8_fromchar(chr, u8buf, sizeof(u8buf)); + if (!u8len) + ch = 0; + else { + --u8len; + for (uc = 0; uc < u8len; ++uc) + lex_tokench(lex, u8buf[uc]); + /* the last character will be inserted with the tokench() call + * below the switch + */ + ch = u8buf[uc]; + } + } + else + ch = chr; + break; + case '\n': ch = '\n'; break; + default: lexwarn(lex, WARN_UNKNOWN_CONTROL_SEQUENCE, "unrecognized control sequence: \\%c", ch); /* so we just add the character plus backslash no matter what it actually is */ @@ -793,7 +908,10 @@ static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch) int ch = lastch; /* parse a number... */ - lex->tok.ttype = TOKEN_INTCONST; + if (ch == '.') + lex->tok.ttype = TOKEN_FLOATCONST; + else + lex->tok.ttype = TOKEN_INTCONST; lex_tokench(lex, ch); @@ -826,7 +944,7 @@ static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch) } } /* NOT else, '.' can come from above as well */ - if (ch == '.' && !ishex) + if (lex->tok.ttype != TOKEN_FLOATCONST && ch == '.' && !ishex) { /* Allow floating comma in non-hex mode */ lex->tok.ttype = TOKEN_FLOATCONST; @@ -863,6 +981,7 @@ static int GMQCC_WARN lex_finish_digit(lex_file *lex, int lastch) int lex_do(lex_file *lex) { int ch, nextch, thirdch; + bool hadwhite = false; lex_token_new(lex); #if 0 @@ -871,10 +990,13 @@ int lex_do(lex_file *lex) #endif while (true) { - ch = lex_skipwhite(lex); + ch = lex_skipwhite(lex, hadwhite); + hadwhite = true; if (!lex->flags.mergelines || ch != '\\') break; ch = lex_getch(lex); + if (ch == '\r') + ch = lex_getch(lex); if (ch != '\n') { lex_ungetch(lex, ch); ch = '\\'; @@ -902,7 +1024,7 @@ int lex_do(lex_file *lex) } /* modelgen / spiritgen commands */ - if (ch == '$') { + if (ch == '$' && !lex->flags.preprocessing) { const char *v; size_t frame; @@ -993,7 +1115,6 @@ int lex_do(lex_file *lex) if (rc < 0) return (lex->tok.ttype = TOKEN_FATAL); - v = lex->tok.value; if (lex->modelname) { frame_macro m; m.value = lex->framevalue; @@ -1048,6 +1169,15 @@ int lex_do(lex_file *lex) switch (ch) { case '[': + nextch = lex_getch(lex); + if (nextch == '[') { + lex_tokench(lex, ch); + lex_tokench(lex, nextch); + lex_endtoken(lex); + return (lex->tok.ttype = TOKEN_ATTRIBUTE_OPEN); + } + lex_ungetch(lex, nextch); + /* FALL THROUGH */ case '(': case ':': case '?': @@ -1057,11 +1187,23 @@ int lex_do(lex_file *lex) return (lex->tok.ttype = ch); else return (lex->tok.ttype = TOKEN_OPERATOR); + + case ']': + if (lex->flags.noops) { + nextch = lex_getch(lex); + if (nextch == ']') { + lex_tokench(lex, ch); + lex_tokench(lex, nextch); + lex_endtoken(lex); + return (lex->tok.ttype = TOKEN_ATTRIBUTE_CLOSE); + } + lex_ungetch(lex, nextch); + } + /* FALL THROUGH */ case ')': case ';': case '{': case '}': - case ']': case '#': lex_tokench(lex, ch); @@ -1071,6 +1213,18 @@ int lex_do(lex_file *lex) break; } + if (ch == '.') { + nextch = lex_getch(lex); + /* digits starting with a dot */ + if (isdigit(nextch)) { + lex_ungetch(lex, nextch); + lex->tok.ttype = lex_finish_digit(lex, ch); + lex_endtoken(lex); + return lex->tok.ttype; + } + lex_ungetch(lex, nextch); + } + if (lex->flags.noops) { /* Detect characters early which are normally @@ -1099,31 +1253,37 @@ int lex_do(lex_file *lex) default: break; } + } - if (ch == '.') - { - lex_tokench(lex, ch); - /* peak ahead once */ - nextch = lex_getch(lex); - if (nextch != '.') { - lex_ungetch(lex, nextch); - lex_endtoken(lex); - return (lex->tok.ttype = ch); - } - /* peak ahead again */ - nextch = lex_getch(lex); - if (nextch != '.') { - lex_ungetch(lex, nextch); - lex_ungetch(lex, '.'); - lex_endtoken(lex); + if (ch == '.') + { + lex_tokench(lex, ch); + /* peak ahead once */ + nextch = lex_getch(lex); + if (nextch != '.') { + lex_ungetch(lex, nextch); + lex_endtoken(lex); + if (lex->flags.noops) return (lex->tok.ttype = ch); - } - /* fill the token to be "..." */ - lex_tokench(lex, ch); - lex_tokench(lex, ch); + else + return (lex->tok.ttype = TOKEN_OPERATOR); + } + /* peak ahead again */ + nextch = lex_getch(lex); + if (nextch != '.') { + lex_ungetch(lex, nextch); + lex_ungetch(lex, '.'); lex_endtoken(lex); - return (lex->tok.ttype = TOKEN_DOTS); + if (lex->flags.noops) + return (lex->tok.ttype = ch); + else + return (lex->tok.ttype = TOKEN_OPERATOR); } + /* fill the token to be "..." */ + lex_tokench(lex, ch); + lex_tokench(lex, ch); + lex_endtoken(lex); + return (lex->tok.ttype = TOKEN_DOTS); } if (ch == ',' || ch == '.') { @@ -1133,15 +1293,23 @@ int lex_do(lex_file *lex) } if (ch == '+' || ch == '-' || /* ++, --, +=, -= and -> as well! */ - ch == '>' || ch == '<' || /* <<, >>, <=, >= */ - ch == '=' || ch == '!' || /* ==, != */ - ch == '&' || ch == '|') /* &&, ||, &=, |= */ - { + ch == '>' || ch == '<' || /* <<, >>, <=, >= */ + ch == '=' || ch == '!' || /* <=>, ==, != */ + ch == '&' || ch == '|' || /* &&, ||, &=, |= */ + ch == '~' /* ~=, ~ */ + ) { lex_tokench(lex, ch); nextch = lex_getch(lex); - if (nextch == ch || nextch == '=') { + if ((nextch == '=' && ch != '<') || (nextch == ch && ch != '!')) { lex_tokench(lex, nextch); + } else if (ch == '<' && nextch == '=') { + lex_tokench(lex, nextch); + if ((thirdch = lex_getch(lex)) == '>') + lex_tokench(lex, thirdch); + else + lex_ungetch(lex, thirdch); + } else if (ch == '-' && nextch == '>') { lex_tokench(lex, nextch); } else if (ch == '&' && nextch == '~') { @@ -1154,8 +1322,20 @@ int lex_do(lex_file *lex) lex_tokench(lex, nextch); lex_tokench(lex, thirdch); } - } else + } + else if (lex->flags.preprocessing && + ch == '-' && isdigit(nextch)) + { + lex->tok.ttype = lex_finish_digit(lex, nextch); + if (lex->tok.ttype == TOKEN_INTCONST) + lex->tok.constval.i = -lex->tok.constval.i; + else + lex->tok.constval.f = -lex->tok.constval.f; + lex_endtoken(lex); + return lex->tok.ttype; + } else { lex_ungetch(lex, nextch); + } lex_endtoken(lex); return (lex->tok.ttype = TOKEN_OPERATOR); @@ -1175,7 +1355,7 @@ int lex_do(lex_file *lex) lex_tokench(lex, ch); nextch = lex_getch(lex); - if (nextch == '=') { + if (nextch == '=' || nextch == '*') { lex_tokench(lex, nextch); } else lex_ungetch(lex, nextch); @@ -1184,6 +1364,12 @@ int lex_do(lex_file *lex) return (lex->tok.ttype = TOKEN_OPERATOR); } + if (ch == '%') { + lex_tokench(lex, ch); + lex_endtoken(lex); + return (lex->tok.ttype = TOKEN_OPERATOR); + } + if (isident_start(ch)) { const char *v; @@ -1221,7 +1407,7 @@ int lex_do(lex_file *lex) if (!strcmp(v, keywords_qc[kw])) return (lex->tok.ttype = TOKEN_KEYWORD); } - if (opts_standard != COMPILER_QCC) { + if (OPTS_OPTION_U32(OPTION_STANDARD) != COMPILER_QCC) { for (kw = 0; kw < num_keywords_fg; ++kw) { if (!strcmp(v, keywords_fg[kw])) return (lex->tok.ttype = TOKEN_KEYWORD); @@ -1243,7 +1429,7 @@ int lex_do(lex_file *lex) while (!lex->flags.preprocessing && lex->tok.ttype == TOKEN_STRINGCONST) { /* Allow c style "string" "continuation" */ - ch = lex_skipwhite(lex); + ch = lex_skipwhite(lex, false); if (ch != '"') { lex_ungetch(lex, ch); break; @@ -1270,8 +1456,9 @@ int lex_do(lex_file *lex) lex_tokench(lex, ch); lex_endtoken(lex); + lex->tok.ttype = TOKEN_CHARCONST; /* It's a vector if we can successfully scan 3 floats */ -#ifdef WIN32 +#ifdef _MSC_VER if (sscanf_s(lex->tok.value, " %f %f %f ", &lex->tok.constval.v.x, &lex->tok.constval.v.y, &lex->tok.constval.v.z) == 3) #else @@ -1282,6 +1469,24 @@ int lex_do(lex_file *lex) { lex->tok.ttype = TOKEN_VECTORCONST; } + else + { + if (!lex->flags.preprocessing && strlen(lex->tok.value) > 1) { + uchar_t u8char; + /* check for a valid utf8 character */ + if (!OPTS_FLAG(UTF8) || !u8_analyze(lex->tok.value, NULL, NULL, &u8char, 8)) { + if (lexwarn(lex, WARN_MULTIBYTE_CHARACTER, + ( OPTS_FLAG(UTF8) ? "invalid multibyte character sequence `%s`" + : "multibyte character: `%s`" ), + lex->tok.value)) + return (lex->tok.ttype = TOKEN_ERROR); + } + else + lex->tok.constval.i = u8char; + } + else + lex->tok.constval.i = lex->tok.value[0]; + } return lex->tok.ttype; } @@ -1293,6 +1498,12 @@ int lex_do(lex_file *lex) return lex->tok.ttype; } - lexerror(lex, "unknown token"); + if (lex->flags.preprocessing) { + lex_tokench(lex, ch); + lex_endtoken(lex); + return (lex->tok.ttype = ch); + } + + lexerror(lex, "unknown token: `%s`", lex->tok.value); return (lex->tok.ttype = TOKEN_ERROR); }