From e5ff1d3d4bc01d98b12d05f9cb85457c7f15c424 Mon Sep 17 00:00:00 2001 From: IronClawTrem Date: Tue, 6 Aug 2019 02:17:23 +0100 Subject: first commit --- src/tools/lcc/lburg/gram.c | 677 ++++++++++++++++++++++++++++++++++++++++++++ src/tools/lcc/lburg/gram.y | 202 +++++++++++++ src/tools/lcc/lburg/lburg.1 | 179 ++++++++++++ src/tools/lcc/lburg/lburg.c | 671 +++++++++++++++++++++++++++++++++++++++++++ src/tools/lcc/lburg/lburg.h | 65 +++++ 5 files changed, 1794 insertions(+) create mode 100644 src/tools/lcc/lburg/gram.c create mode 100644 src/tools/lcc/lburg/gram.y create mode 100644 src/tools/lcc/lburg/lburg.1 create mode 100644 src/tools/lcc/lburg/lburg.c create mode 100644 src/tools/lcc/lburg/lburg.h (limited to 'src/tools/lcc/lburg') diff --git a/src/tools/lcc/lburg/gram.c b/src/tools/lcc/lburg/gram.c new file mode 100644 index 0000000..c09f897 --- /dev/null +++ b/src/tools/lcc/lburg/gram.c @@ -0,0 +1,677 @@ +#if defined(__STDC__) || defined(__cplusplus) +#define YYCONST const +#define YYPARAMS(x) x +#define YYDEFUN(name, arglist, args) name(args) +#define YYAND , +#define YYPTR void * +#else +#define YYCONST +#define YYPARAMS(x) () +#define YYDEFUN(name, arglist, args) name arglist args; +#define YYAND ; +#define YYPTR char * +#endif +#define YYBYACC 1 +#ifndef YYDONT_INCLUDE_STDIO +#include +#endif +#include /* for malloc/realloc/free */ +#line 2 "lburg/gram.y" +#include +#include "lburg.h" +/*lint -e616 -e527 -e652 -esym(552,yynerrs) -esym(563,yynewstate,yyerrlab) */ +static int yylineno = 0; +#line 8 "lburg/gram.y" +typedef union { + int n; + char *string; + Tree tree; +} YYSTYPE; +#line 37 "y.tab.c" +#define TERMINAL 257 +#define START 258 +#define PPERCENT 259 +#define ID 260 +#define TEMPLATE 261 +#define CODE 262 +#define INT 263 +#define YYERRCODE 256 +static YYCONST short yylhs[] = { -1, + 0, 0, 4, 4, 6, 6, 6, 6, 7, 7, + 5, 5, 5, 5, 1, 3, 3, 3, 2, +}; +static YYCONST short yylen[] = { 2, + 3, 1, 0, 2, 3, 3, 1, 2, 0, 4, + 0, 7, 2, 3, 1, 1, 4, 6, 1, +}; +static YYCONST short yydefred[] = { 3, + 0, 0, 0, 9, 0, 11, 7, 4, 8, 0, + 15, 0, 0, 0, 5, 6, 0, 13, 0, 0, + 14, 0, 10, 0, 0, 0, 0, 0, 19, 0, + 17, 0, 12, 0, 18, +}; +static YYCONST short yydgoto[] = { 1, + 12, 30, 25, 2, 13, 8, 10, +}; +static YYCONST short yysindex[] = { 0, + 0, -4, -2, 0, -250, 0, 0, 0, 0, -9, + 0, 1, -10, -49, 0, 0, 3, 0, -44, -248, + 0, -244, 0, -22, -242, -244, -245, -37, 0, 10, + 0, -244, 0, -20, 0, +}; +static YYCONST short yyrindex[] = { 0, + 0, 22, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 23, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, -39, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, +}; +static YYCONST short yygindex[] = { 0, + 11, 0, -23, 0, 0, 0, 0, +}; +#define YYTABLESIZE 255 +static YYCONST short yytable[] = { 18, + 15, 16, 28, 31, 16, 7, 32, 9, 34, 11, + 16, 20, 21, 22, 23, 24, 29, 26, 27, 33, + 35, 2, 1, 19, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 16, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 17, 0, 0, 0, 11, + 14, 3, 4, 5, 6, +}; +static YYCONST short yycheck[] = { 10, + 10, 41, 26, 41, 44, 10, 44, 10, 32, 260, + 10, 61, 10, 58, 263, 260, 262, 40, 261, 10, + 41, 0, 0, 13, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, 261, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, 256, -1, -1, -1, 260, + 260, 256, 257, 258, 259, +}; +#define YYFINAL 1 +#ifndef YYDEBUG +#define YYDEBUG 0 +#endif +#define YYMAXTOKEN 263 +#if YYDEBUG +static YYCONST char *YYCONST yyname[] = { +"end-of-file",0,0,0,0,0,0,0,0,0,"'\\n'",0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,"'('","')'",0,0,"','",0,0,0,0,0,0,0,0,0,0,0,0,0,"':'",0,0, +"'='",0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +"TERMINAL","START","PPERCENT","ID","TEMPLATE","CODE","INT", +}; +static YYCONST char *YYCONST yyrule[] = { +"$accept : spec", +"spec : decls PPERCENT rules", +"spec : decls", +"decls :", +"decls : decls decl", +"decl : TERMINAL blist '\\n'", +"decl : START nonterm '\\n'", +"decl : '\\n'", +"decl : error '\\n'", +"blist :", +"blist : blist ID '=' INT", +"rules :", +"rules : rules nonterm ':' tree TEMPLATE cost '\\n'", +"rules : rules '\\n'", +"rules : rules error '\\n'", +"nonterm : ID", +"tree : ID", +"tree : ID '(' tree ')'", +"tree : ID '(' tree ',' tree ')'", +"cost : CODE", +}; +#endif +#define YYLEX yylex() +#define YYEMPTY -1 +#define yyclearin (yychar=(YYEMPTY)) +#define yyerrok (yyerrflag=0) +#ifndef YYINITDEPTH +#define YYINITDEPTH 200 +#endif +#ifdef YYSTACKSIZE +#ifndef YYMAXDEPTH +#define YYMAXDEPTH YYSTACKSIZE +#endif +#else +#ifdef YYMAXDEPTH +#define YYSTACKSIZE YYMAXDEPTH +#else +#define YYSTACKSIZE 500 +#define YYMAXDEPTH 500 +#endif +#endif +#ifndef YYMAXSTACKSIZE +#define YYMAXSTACKSIZE 10000 +#endif +int yydebug; +int yynerrs; +int yyerrflag; +int yychar; +YYSTYPE yyval; +YYSTYPE yylval; +static short *yyss; +static YYSTYPE *yyvs; +static int yystacksize; +#define yyfree(x) free(x) +extern int yylex(); + +static YYPTR +YYDEFUN (yymalloc, (bytes), unsigned bytes) +{ + YYPTR ptr = (YYPTR) malloc (bytes); + if (ptr != 0) return (ptr); + yyerror ("yyparse: memory exhausted"); + return (0); +} + +static YYPTR +YYDEFUN (yyrealloc, (old, bytes), YYPTR old YYAND unsigned bytes) +{ + YYPTR ptr = (YYPTR) realloc (old, bytes); + if (ptr != 0) return (ptr); + yyerror ("yyparse: memory exhausted"); + return (0); +} + +static int +#ifdef __GNUC__ +inline +#endif +yygrow () +{ +#if YYDEBUG + int old_stacksize = yystacksize; +#endif + short *new_yyss; + YYSTYPE *new_yyvs; + + if (yystacksize == YYMAXSTACKSIZE) + return (1); + yystacksize += (yystacksize + 1 ) / 2; + if (yystacksize > YYMAXSTACKSIZE) + yystacksize = YYMAXSTACKSIZE; +#if YYDEBUG + if (yydebug) + printf("yydebug: growing stack size from %d to %d\n", + old_stacksize, yystacksize); +#endif + new_yyss = (short *) yyrealloc ((char *)yyss, yystacksize * sizeof (short)); + if (new_yyss == 0) + return (1); + new_yyvs = (YYSTYPE *) yyrealloc ((char *)yyvs, yystacksize * sizeof (YYSTYPE)); + if (new_yyvs == 0) + { + yyfree (new_yyss); + return (1); + } + yyss = new_yyss; + yyvs = new_yyvs; + return (0); +} +#line 60 "lburg/gram.y" +#include +#include +#include +#include +#include + +int errcnt = 0; +FILE *infp = NULL; +FILE *outfp = NULL; +static char buf[BUFSIZ], *bp = buf; +static int ppercent = 0; +static int code = 0; + +static int get(void) { + if (*bp == 0) { + bp = buf; + *bp = 0; + if (fgets(buf, sizeof buf, infp) == NULL) + return EOF; + yylineno++; + while (buf[0] == '%' && buf[1] == '{' && buf[2] == '\n') { + for (;;) { + if (fgets(buf, sizeof buf, infp) == NULL) { + yywarn("unterminated %{...%}\n"); + return EOF; + } + yylineno++; + if (strcmp(buf, "%}\n") == 0) + break; + fputs(buf, outfp); + } + if (fgets(buf, sizeof buf, infp) == NULL) + return EOF; + yylineno++; + } + } + return *bp++; +} + +void yyerror(char *fmt, ...) { + va_list ap; + + va_start(ap, fmt); + if (yylineno > 0) + fprintf(stderr, "line %d: ", yylineno); + vfprintf(stderr, fmt, ap); + if (fmt[strlen(fmt)-1] != '\n') + fprintf(stderr, "\n"); + errcnt++; + va_end(ap); +} + +int yylex(void) { + int c; + + if (code) { + char *p; + bp += strspn(bp, " \t\f"); + p = strchr(bp, '\n'); + if (p == NULL) + p = strchr(bp, '\n'); + while (p > bp && isspace(p[-1])) + p--; + yylval.string = alloc(p - bp + 1); + strncpy(yylval.string, bp, p - bp); + yylval.string[p - bp] = 0; + bp = p; + code--; + return CODE; + } + while ((c = get()) != EOF) { + switch (c) { + case ' ': case '\f': case '\t': + continue; + case '\n': + case '(': case ')': case ',': + case ':': case '=': + return c; + } + if (c == '%' && *bp == '%') { + bp++; + return ppercent++ ? 0 : PPERCENT; + } else if (c == '%' && strncmp(bp, "term", 4) == 0 + && isspace(bp[4])) { + bp += 4; + return TERMINAL; + } else if (c == '%' && strncmp(bp, "start", 5) == 0 + && isspace(bp[5])) { + bp += 5; + return START; + } else if (c == '"') { + char *p = strchr(bp, '"'); + if (p == NULL) { + yyerror("missing \" in assembler template\n"); + p = strchr(bp, '\n'); + if (p == NULL) + p = strchr(bp, '\0'); + } + assert(p); + yylval.string = alloc(p - bp + 1); + strncpy(yylval.string, bp, p - bp); + yylval.string[p - bp] = 0; + bp = *p == '"' ? p + 1 : p; + code++; + return TEMPLATE; + } else if (isdigit(c)) { + int n = 0; + do { + int d = c - '0'; + if (n > (INT_MAX - d)/10) + yyerror("integer greater than %d\n", INT_MAX); + else + n = 10*n + d; + c = get(); + } while (c != EOF && isdigit(c)); + bp--; + yylval.n = n; + return INT; + } else if (isalpha(c)) { + char *p = bp - 1; + while (isalpha(*bp) || isdigit(*bp) || *bp == '_') + bp++; + yylval.string = alloc(bp - p + 1); + strncpy(yylval.string, p, bp - p); + yylval.string[bp - p] = 0; + return ID; + } else if (isprint(c)) + yyerror("invalid character `%c'\n", c); + else + yyerror("invalid character `\\%03o'\n", (unsigned char)c); + } + return 0; +} + +void yywarn(char *fmt, ...) { + va_list ap; + + va_start(ap, fmt); + if (yylineno > 0) + fprintf(stderr, "line %d: ", yylineno); + fprintf(stderr, "warning: "); + vfprintf(stderr, fmt, ap); +} +#line 403 "y.tab.c" +#define YYABORT goto yyabort +#define YYACCEPT goto yyaccept +#define YYERROR goto yyerrlab + +#if YYDEBUG +#ifdef __cplusplus +extern "C" char *getenv(); +#else +extern char *getenv(); +#endif +#endif + +int +yyparse() +{ + register int yym, yyn, yystate; + register YYSTYPE *yyvsp; + register short *yyssp; + short *yysse; +#if YYDEBUG + register YYCONST char *yys; + + if (yys = getenv("YYDEBUG")) + { + yyn = *yys; + if (yyn >= '0' && yyn <= '9') + yydebug = yyn - '0'; + } +#endif + + yynerrs = 0; + yyerrflag = 0; + yychar = (-1); + + if (yyss == 0) + { + yyss = (short *) yymalloc (YYSTACKSIZE * sizeof (short)); + if (yyss == 0) + goto yyabort; + yyvs = (YYSTYPE *) yymalloc (YYSTACKSIZE * sizeof (YYSTYPE)); + if (yyvs == 0) + { + yyfree (yyss); + goto yyabort; + } + yystacksize = YYSTACKSIZE; + } + yysse = yyss + yystacksize - 1; + yyssp = yyss; + yyvsp = yyvs; + *yyssp = yystate = 0; + goto yyloop; + +yypush_lex: + yyval = yylval; + yystate = yytable[yyn]; +yypush: + if (yyssp >= yysse) + { + int depth = yyssp - yyss; + if (yygrow() != 0) + goto yyoverflow; + yysse = yyss + yystacksize -1; + yyssp = depth + yyss; + yyvsp = depth + yyvs; + } + *++yyssp = yystate; + *++yyvsp = yyval; + +yyloop: + if ((yyn = yydefred[yystate])) goto yyreduce; + yyn = yysindex[yystate]; + if (yychar < 0) + { + if ((yychar = yylex()) < 0) yychar = 0; +#if YYDEBUG + if (yydebug) + { + yys = 0; + if (yychar <= YYMAXTOKEN) yys = yyname[yychar]; + if (!yys) yys = "illegal-symbol"; + printf("yydebug: state %d, reading %d (%s)\n", yystate, + yychar, yys); + } +#endif + } + if (yyn != 0 + && ((yyn += yychar), ((unsigned)yyn <= (unsigned)YYTABLESIZE)) + && yycheck[yyn] == yychar) + { +#if YYDEBUG + if (yydebug) + printf("yydebug: state %d, shifting to state %d\n", + yystate, yytable[yyn]); +#endif + if (yyerrflag > 0) --yyerrflag; + yychar = (-1); + goto yypush_lex; + } + yyn = yyrindex[yystate]; + if (yyn != 0 + && ((yyn += yychar), ((unsigned)yyn <= (unsigned)YYTABLESIZE)) + && yycheck[yyn] == yychar) + { + yyn = yytable[yyn]; + goto yyreduce; + } + if (yyerrflag) goto yyinrecovery; +#ifdef lint + goto yynewerror; +yynewerror: +#endif + yyerror("syntax error"); +#ifdef lint + goto yyerrlab; +yyerrlab: +#endif + ++yynerrs; +yyinrecovery: + if (yyerrflag < 3) + { + yyerrflag = 3; + for (;;) + { + yyn = yysindex[*yyssp]; + if (yyn != 0 + && ((yyn += YYERRCODE), ((unsigned)yyn <= (unsigned)YYTABLESIZE)) + && yycheck[yyn] == YYERRCODE) + { +#if YYDEBUG + if (yydebug) + printf("yydebug: state %d, error recovery shifting\ + to state %d\n", *yyssp, yytable[yyn]); +#endif + goto yypush_lex; + } + else + { +#if YYDEBUG + if (yydebug) + printf("yydebug: error recovery discarding state %d\n", + *yyssp); +#endif + if (yyssp <= yyss) goto yyabort; + --yyssp; + --yyvsp; + } + } + } + else + { + if (yychar == 0) goto yyabort; +#if YYDEBUG + if (yydebug) + { + yys = 0; + if (yychar <= YYMAXTOKEN) yys = yyname[yychar]; + if (!yys) yys = "illegal-symbol"; + printf("yydebug: state %d, error recovery discards token %d (%s)\n", + yystate, yychar, yys); + } +#endif + yychar = (-1); + goto yyloop; + } +yyreduce: +#if YYDEBUG + if (yydebug) + printf("yydebug: state %d, reducing by rule %d (%s)\n", + yystate, yyn, yyrule[yyn]); +#endif + yym = yylen[yyn]; + yyval = yyvsp[1-yym]; + switch (yyn) + { +case 1: +#line 22 "lburg/gram.y" +{ yylineno = 0; } +break; +case 2: +#line 23 "lburg/gram.y" +{ yylineno = 0; } +break; +case 6: +#line 31 "lburg/gram.y" +{ + if (nonterm(yyvsp[-1].string)->number != 1) + yyerror("redeclaration of the start symbol\n"); + } +break; +case 8: +#line 36 "lburg/gram.y" +{ yyerrok; } +break; +case 10: +#line 40 "lburg/gram.y" +{ term(yyvsp[-2].string, yyvsp[0].n); } +break; +case 12: +#line 44 "lburg/gram.y" +{ rule(yyvsp[-5].string, yyvsp[-3].tree, yyvsp[-2].string, yyvsp[-1].string); } +break; +case 14: +#line 46 "lburg/gram.y" +{ yyerrok; } +break; +case 15: +#line 49 "lburg/gram.y" +{ nonterm(yyval.string = yyvsp[0].string); } +break; +case 16: +#line 52 "lburg/gram.y" +{ yyval.tree = tree(yyvsp[0].string, 0, 0); } +break; +case 17: +#line 53 "lburg/gram.y" +{ yyval.tree = tree(yyvsp[-3].string, yyvsp[-1].tree, 0); } +break; +case 18: +#line 54 "lburg/gram.y" +{ yyval.tree = tree(yyvsp[-5].string, yyvsp[-3].tree, yyvsp[-1].tree); } +break; +case 19: +#line 57 "lburg/gram.y" +{ if (*yyvsp[0].string == 0) yyval.string = "0"; } +break; +#line 630 "y.tab.c" + } + yyssp -= yym; + yyvsp -= yym; + yym = yylhs[yyn]; + yystate = *yyssp; + if (yystate == 0 && yym == 0) + { +#if YYDEBUG + if (yydebug) + printf("yydebug: after reduction, shifting from state 0 to\ + state %d\n", YYFINAL); +#endif + yystate = YYFINAL; + *++yyssp = YYFINAL; + *++yyvsp = yyval; + if (yychar < 0) + { + if ((yychar = yylex()) < 0) yychar = 0; +#if YYDEBUG + if (yydebug) + { + yys = 0; + if (yychar <= YYMAXTOKEN) yys = yyname[yychar]; + if (!yys) yys = "illegal-symbol"; + printf("yydebug: state %d, reading %d (%s)\n", + YYFINAL, yychar, yys); + } +#endif + } + if (yychar == 0) goto yyaccept; + goto yyloop; + } + yyn = yygindex[yym]; + if (yyn != 0 + && ((yyn += yystate), ((unsigned)yyn <= (unsigned)YYTABLESIZE)) + && yycheck[yyn] == yystate) + yystate = yytable[yyn]; + else + yystate = yydgoto[yym]; +#if YYDEBUG + if (yydebug) + printf("yydebug: after reduction, shifting from state %d \ +to state %d\n", *yyssp, yystate); +#endif + goto yypush; +yyoverflow: + yyerror("yacc stack overflow"); +yyabort: + return (1); +yyaccept: + return (0); +} diff --git a/src/tools/lcc/lburg/gram.y b/src/tools/lcc/lburg/gram.y new file mode 100644 index 0000000..1ecd8a9 --- /dev/null +++ b/src/tools/lcc/lburg/gram.y @@ -0,0 +1,202 @@ +%{ +#include +#include "lburg.h" +static char rcsid[] = "$Id: gram.y 145 2001-10-17 21:53:10Z timo $"; +/*lint -e616 -e527 -e652 -esym(552,yynerrs) -esym(563,yynewstate,yyerrlab) */ +static int yylineno = 0; +%} +%union { + int n; + char *string; + Tree tree; +} +%term TERMINAL +%term START +%term PPERCENT + +%token ID TEMPLATE CODE +%token INT +%type nonterm cost +%type tree +%% +spec : decls PPERCENT rules { yylineno = 0; } + | decls { yylineno = 0; } + ; + +decls : /* lambda */ + | decls decl + ; + +decl : TERMINAL blist '\n' + | START nonterm '\n' { + if (nonterm($2)->number != 1) + yyerror("redeclaration of the start symbol\n"); + } + | '\n' + | error '\n' { yyerrok; } + ; + +blist : /* lambda */ + | blist ID '=' INT { term($2, $4); } + ; + +rules : /* lambda */ + | rules nonterm ':' tree TEMPLATE cost '\n' { rule($2, $4, $5, $6); } + | rules '\n' + | rules error '\n' { yyerrok; } + ; + +nonterm : ID { nonterm($$ = $1); } + ; + +tree : ID { $$ = tree($1, 0, 0); } + | ID '(' tree ')' { $$ = tree($1, $3, 0); } + | ID '(' tree ',' tree ')' { $$ = tree($1, $3, $5); } + ; + +cost : CODE { if (*$1 == 0) $$ = "0"; } + ; +%% +#include +#include +#include +#include +#include + +int errcnt = 0; +FILE *infp = NULL; +FILE *outfp = NULL; +static char buf[BUFSIZ], *bp = buf; +static int ppercent = 0; +static int code = 0; + +static int get(void) { + if (*bp == 0) { + bp = buf; + *bp = 0; + if (fgets(buf, sizeof buf, infp) == NULL) + return EOF; + yylineno++; + while (buf[0] == '%' && buf[1] == '{' && buf[2] == '\n') { + for (;;) { + if (fgets(buf, sizeof buf, infp) == NULL) { + yywarn("unterminated %{...%}\n"); + return EOF; + } + yylineno++; + if (strcmp(buf, "%}\n") == 0) + break; + fputs(buf, outfp); + } + if (fgets(buf, sizeof buf, infp) == NULL) + return EOF; + yylineno++; + } + } + return *bp++; +} + +void yyerror(char *fmt, ...) { + va_list ap; + + va_start(ap, fmt); + if (yylineno > 0) + fprintf(stderr, "line %d: ", yylineno); + vfprintf(stderr, fmt, ap); + if (fmt[strlen(fmt)-1] != '\n') + fprintf(stderr, "\n"); + errcnt++; + va_end(ap); +} + +int yylex(void) { + int c; + + if (code) { + char *p; + bp += strspn(bp, " \t\f"); + p = strchr(bp, '\n'); + if (p == NULL) + p = strchr(bp, '\n'); + while (p > bp && isspace(p[-1])) + p--; + yylval.string = alloc(p - bp + 1); + strncpy(yylval.string, bp, p - bp); + yylval.string[p - bp] = 0; + bp = p; + code--; + return CODE; + } + while ((c = get()) != EOF) { + switch (c) { + case ' ': case '\f': case '\t': + continue; + case '\n': + case '(': case ')': case ',': + case ':': case '=': + return c; + } + if (c == '%' && *bp == '%') { + bp++; + return ppercent++ ? 0 : PPERCENT; + } else if (c == '%' && strncmp(bp, "term", 4) == 0 + && isspace(bp[4])) { + bp += 4; + return TERMINAL; + } else if (c == '%' && strncmp(bp, "start", 5) == 0 + && isspace(bp[5])) { + bp += 5; + return START; + } else if (c == '"') { + char *p = strchr(bp, '"'); + if (p == NULL) { + yyerror("missing \" in assembler template\n"); + p = strchr(bp, '\n'); + if (p == NULL) + p = strchr(bp, '\0'); + } + assert(p); + yylval.string = alloc(p - bp + 1); + strncpy(yylval.string, bp, p - bp); + yylval.string[p - bp] = 0; + bp = *p == '"' ? p + 1 : p; + code++; + return TEMPLATE; + } else if (isdigit(c)) { + int n = 0; + do { + int d = c - '0'; + if (n > (INT_MAX - d)/10) + yyerror("integer greater than %d\n", INT_MAX); + else + n = 10*n + d; + c = get(); + } while (c != EOF && isdigit(c)); + bp--; + yylval.n = n; + return INT; + } else if (isalpha(c)) { + char *p = bp - 1; + while (isalpha(*bp) || isdigit(*bp) || *bp == '_') + bp++; + yylval.string = alloc(bp - p + 1); + strncpy(yylval.string, p, bp - p); + yylval.string[bp - p] = 0; + return ID; + } else if (isprint(c)) + yyerror("invalid character `%c'\n", c); + else + yyerror("invalid character `\\%03o'\n", (unsigned char)c); + } + return 0; +} + +void yywarn(char *fmt, ...) { + va_list ap; + + va_start(ap, fmt); + if (yylineno > 0) + fprintf(stderr, "line %d: ", yylineno); + fprintf(stderr, "warning: "); + vfprintf(stderr, fmt, ap); +} diff --git a/src/tools/lcc/lburg/lburg.1 b/src/tools/lcc/lburg/lburg.1 new file mode 100644 index 0000000..8cf7250 --- /dev/null +++ b/src/tools/lcc/lburg/lburg.1 @@ -0,0 +1,179 @@ +.TH LBURG 1 "local \- 11/30/94" +.\" $Id: lburg.1 145 2001-10-17 21:53:10Z timo $ +.SH NAME +lburg \- lcc's code-generator generator +.SH SYNOPSIS +.B lburg +[ +.I option +]... +[ [ +.I input +] +.I output +] +.br +.SH DESCRIPTION +.PP +.I lburg +reads an lcc-style BURG specification from +.I input +and writes a pattern-matching code generator to +.IR output . +If +.I input +is `\-' or is omitted, +.I lburg +reads the standard input; +If +.I output +is `\-' or is omitted, +.I lburg +writes to the standard output. +.PP +.I lburg +accepts specifications that conform to the following EBNF grammar. +Terminals are enclosed in single quotes or are +given in uppercase, all other symbols are nonterminals or English phrases, +{X} denotes zero or more instances of X, and [X] denotes an optional X. +.PP +.nf +.RS +.ft CW +spec: `%{' configuration `%}' { dcl } `%%' { rule } + [ `%%' C code ] + +dcl: `%start' nonterm + `%term' { ID `=' INT } + +rule: nonterm `:' tree template [ C expression ] + +tree: term `(' tree `,' tree `)' + term `(' tree `)' + term + nonterm + +nonterm: ID + +template: `"' { any character except double quote } `"' +.RE +.fi +.PP +Specifications are structurally similar to +.IR yacc 's. +Text between +`\f(CW%{\fP' +and +`\f(CW%}\fP' +is called the configuration section; there may be several such segments. +All are concatenated and copied verbatim into the head of the output. +Text after the second +`\f(CW%%\fP', +if any, is also copied verbatim into the output, at the end. +.PP +Specifications consist of declarations, a +`\f(CW%%\fP' +separator, and rules. +Input is line-oriented; each declaration and rule must appear on a separate line, +and declarations must begin in column 1. +Declarations declare terminals \(em the operators in subject +trees \(em and associate a unique, positive external symbol +number with each one. +Nonterminals are declared by their presence +on the left side of rules. The +\f(CW%start\fP +declaration optionally declares a nonterminal as the start symbol. +In the grammar above, +\f(CWterm\fP +and +\f(CWnonterm\fP +denote identifiers that are terminals and nonterminals. +.PP +Rules define tree patterns in a fully parenthesized prefix +form. Every nonterminal denotes a tree. +Each operator has a fixed +arity, which is inferred from the rules in which it is used. +A chain rule is a rule whose pattern is another nonterminal. +If no start symbol is declared, the nonterminal defined by the first rule is used. +.PP +Each rule ends with an expression that computes the cost of matching +that rule; omitted costs +default to zero. Costs of chain rules must be constants. +.PP +The configuration section configures the output +for the trees being parsed and the client's environment. +As shown, this section must define +\f(CWNODEPTR_TYPE\fP +to be a visible typedef symbol for a pointer to a +node in the subject tree. +The labeller invokes +\f(CWOP_LABEL(p)\fP, +\f(CWLEFT\_CHILD(p)\fP, and +\f(CWRIGHT\_CHILD(p)\fP +to read the operator and children from the node pointed to by \f(CWp\fP. +If the configuration section defines these operations as macros, they are implemented in-line; +otherwise, they must be implemented as functions. +.PP +The matcher +computes and stores a single integral state in each node of the subject tree. +The configuration section must define a macro +\f(CWSTATE_LABEL(p)\fP +to access the state field of the node pointed to +by \f(CWp\fP. It must be large enough to hold a pointer, and +a macro is required because it is used as an lvalue. +.PP +.SH OPTIONS +.TP +.BI \-p \ prefix +.br +.ns +.TP +.BI \-p prefix +Use +.I prefix +as the disambiquating prefix for visible names and fields. +The default is `\f(CW_\fP'. +.TP +.B \-T +Arrange for +.sp +.nf +.ft CW + void _trace(NODEPTR_TYPE p, int eruleno, + int cost, int bestcost); +.sp +.fi +.ft R +to be called at each successful match. +\f(CWp\fP +identifies the node and +\f(CWeruleno\fP +identifies the matching rule; the rules are numbered +beginning at 1 in the order they appear in the input. +\f(CWcost\fP +is the cost of the match and +\f(CWbestcost\fP +is the cost of the best previous match. The current match +wins only if +\f(CWcost\fP +is less than \f(CWbestcost\fP. +32767 represents the infinite cost of no previous match. +\f(CW_trace\fP must be declared in the configuration section. +.SH "SEE ALSO" +.IR lcc (1) +.PP +C. W. Fraser and D. R. Hanson, +.IR A Retargetable C Compiler: Design and Implementation , +Benjamin/Cummings, Redwood City, CA, 1995, +ISBN 0-8053-1670-1. Chapter 14. +.PP +C. W. Fraser, D. R. Hanson and T. A. Proebsting, +`Engineering a simple, efficient code generator generator,' +.I +ACM Letters on Programming Languages and Systems +.BR 1 , +3 (Sep. 1992), 213-226. +.br +.SH BUGS +Mail bug reports along with the shortest input +that exposes them to drh@cs.princeton.edu. diff --git a/src/tools/lcc/lburg/lburg.c b/src/tools/lcc/lburg/lburg.c new file mode 100644 index 0000000..c43c96a --- /dev/null +++ b/src/tools/lcc/lburg/lburg.c @@ -0,0 +1,671 @@ +#include +#include +#include +#include +#include +#include +#include +#include "lburg.h" + +static char rcsid[] = "lburg.c - faked rcsid"; + +static char *prefix = ""; +static int Tflag = 0; +static int ntnumber = 0; +static Nonterm start = 0; +static Term terms; +static Nonterm nts; +static Rule rules; +static int nrules; +static struct block { + struct block *link; +} *memlist; /* list of allocated blocks */ + +static char *stringf(char *fmt, ...); +static void print(char *fmt, ...); +static void ckreach(Nonterm p); +static void emitclosure(Nonterm nts); +static void emitcost(Tree t, char *v); +static void emitdefs(Nonterm nts, int ntnumber); +static void emitheader(void); +static void emitkids(Rule rules, int nrules); +static void emitnts(Rule rules, int nrules); +static void emitrecalc(char *pre, Term root, Term kid); +static void emitrecord(char *pre, Rule r, char *c, int cost); +static void emitrule(Nonterm nts); +static void emitlabel(Term terms, Nonterm start, int ntnumber); +static void emitstring(Rule rules); +static void emitstruct(Nonterm nts, int ntnumber); +static void emittest(Tree t, char *v, char *suffix); + +int main(int argc, char *argv[]) { + int c, i; + Nonterm p; + + for (i = 1; i < argc; i++) + if (strcmp(argv[i], "-T") == 0) + Tflag = 1; + else if (strncmp(argv[i], "-p", 2) == 0 && argv[i][2]) + prefix = &argv[i][2]; + else if (strncmp(argv[i], "-p", 2) == 0 && i + 1 < argc) + prefix = argv[++i]; + else if (*argv[i] == '-' && argv[i][1]) { + yyerror("usage: %s [-T | -p prefix]... [ [ input ] output ] \n", + argv[0]); + exit(1); + } else if (infp == NULL) { + if (strcmp(argv[i], "-") == 0) + infp = stdin; + else if ((infp = fopen(argv[i], "r")) == NULL) { + yyerror("%s: can't read `%s'\n", argv[0], argv[i]); + exit(1); + } + } else if (outfp == NULL) { + if (strcmp(argv[i], "-") == 0) + outfp = stdout; + if ((outfp = fopen(argv[i], "w")) == NULL) { + yyerror("%s: can't write `%s'\n", argv[0], argv[i]); + exit(1); + } + } + if (infp == NULL) + infp = stdin; + if (outfp == NULL) + outfp = stdout; + yyparse(); + if (start) + ckreach(start); + for (p = nts; p; p = p->link) { + if (p->rules == NULL) + yyerror("undefined nonterminal `%s'\n", p->name); + if (!p->reached) + yyerror("can't reach nonterminal `%s'\n", p->name); + } + emitheader(); + emitdefs(nts, ntnumber); + emitstruct(nts, ntnumber); + emitnts(rules, nrules); + emitstring(rules); + emitrule(nts); + emitclosure(nts); + if (start) + emitlabel(terms, start, ntnumber); + emitkids(rules, nrules); + if (!feof(infp)) + while ((c = getc(infp)) != EOF) + putc(c, outfp); + while (memlist) { /* for purify */ + struct block *q = memlist->link; + free(memlist); + memlist = q; + } + return errcnt > 0; +} + +/* alloc - allocate nbytes or issue fatal error */ +void *alloc(int nbytes) { + struct block *p = calloc(1, sizeof *p + nbytes); + + if (p == NULL) { + yyerror("out of memory\n"); + exit(1); + } + p->link = memlist; + memlist = p; + return p + 1; +} + +/* stringf - format and save a string */ +static char *stringf(char *fmt, ...) { + va_list ap; + char buf[512]; + + va_start(ap, fmt); + vsprintf(buf, fmt, ap); + va_end(ap); + return strcpy(alloc(strlen(buf) + 1), buf); +} + +struct entry { + union { + char *name; + struct term t; + struct nonterm nt; + } sym; + struct entry *link; +} *table[211]; +#define HASHSIZE (sizeof table/sizeof table[0]) + +/* hash - return hash number for str */ +static unsigned hash(char *str) { + unsigned h = 0; + + while (*str) + h = (h<<1) + *str++; + return h; +} + +/* lookup - lookup symbol name */ +static void *lookup(char *name) { + struct entry *p = table[hash(name)%HASHSIZE]; + + for ( ; p; p = p->link) + if (strcmp(name, p->sym.name) == 0) + return &p->sym; + return 0; +} + +/* install - install symbol name */ +static void *install(char *name) { + struct entry *p = alloc(sizeof *p); + int i = hash(name)%HASHSIZE; + + p->sym.name = name; + p->link = table[i]; + table[i] = p; + return &p->sym; +} + +/* nonterm - create a new terminal id, if necessary */ +Nonterm nonterm(char *id) { + Nonterm p = lookup(id), *q = &nts; + + if (p && p->kind == NONTERM) + return p; + if (p && p->kind == TERM) + yyerror("`%s' is a terminal\n", id); + p = install(id); + p->kind = NONTERM; + p->number = ++ntnumber; + if (p->number == 1) + start = p; + while (*q && (*q)->number < p->number) + q = &(*q)->link; + assert(*q == 0 || (*q)->number != p->number); + p->link = *q; + *q = p; + return p; +} + +/* term - create a new terminal id with external symbol number esn */ +Term term(char *id, int esn) { + Term p = lookup(id), *q = &terms; + + if (p) + yyerror("redefinition of terminal `%s'\n", id); + else + p = install(id); + p->kind = TERM; + p->esn = esn; + p->arity = -1; + while (*q && (*q)->esn < p->esn) + q = &(*q)->link; + if (*q && (*q)->esn == p->esn) + yyerror("duplicate external symbol number `%s=%d'\n", + p->name, p->esn); + p->link = *q; + *q = p; + return p; +} + +/* tree - create & initialize a tree node with the given fields */ +Tree tree(char *id, Tree left, Tree right) { + Tree t = alloc(sizeof *t); + Term p = lookup(id); + int arity = 0; + + if (left && right) + arity = 2; + else if (left) + arity = 1; + if (p == NULL && arity > 0) { + yyerror("undefined terminal `%s'\n", id); + p = term(id, -1); + } else if (p == NULL && arity == 0) + p = (Term)nonterm(id); + else if (p && p->kind == NONTERM && arity > 0) { + yyerror("`%s' is a nonterminal\n", id); + p = term(id, -1); + } + if (p->kind == TERM && p->arity == -1) + p->arity = arity; + if (p->kind == TERM && arity != p->arity) + yyerror("inconsistent arity for terminal `%s'\n", id); + t->op = p; + t->nterms = p->kind == TERM; + if ((t->left = left) != NULL) + t->nterms += left->nterms; + if ((t->right = right) != NULL) + t->nterms += right->nterms; + return t; +} + +/* rule - create & initialize a rule with the given fields */ +Rule rule(char *id, Tree pattern, char *template, char *code) { + Rule r = alloc(sizeof *r), *q; + Term p = pattern->op; + char *end; + + r->lhs = nonterm(id); + r->packed = ++r->lhs->lhscount; + for (q = &r->lhs->rules; *q; q = &(*q)->decode) + ; + *q = r; + r->pattern = pattern; + r->ern = ++nrules; + r->template = template; + r->code = code; + r->cost = strtol(code, &end, 10); + if (*end) { + r->cost = -1; + r->code = stringf("(%s)", code); + } + if (p->kind == TERM) { + for (q = &p->rules; *q; q = &(*q)->next) + ; + *q = r; + } else if (pattern->left == NULL && pattern->right == NULL) { + Nonterm p = pattern->op; + r->chain = p->chain; + p->chain = r; + if (r->cost == -1) + yyerror("illegal nonconstant cost `%s'\n", code); + } + for (q = &rules; *q; q = &(*q)->link) + ; + r->link = *q; + *q = r; + return r; +} + +/* print - formatted output */ +static void print(char *fmt, ...) { + va_list ap; + + va_start(ap, fmt); + for ( ; *fmt; fmt++) + if (*fmt == '%') + switch (*++fmt) { + case 'd': fprintf(outfp, "%d", va_arg(ap, int)); break; + case 's': fputs(va_arg(ap, char *), outfp); break; + case 'P': fprintf(outfp, "%s_", prefix); break; + case 'T': { + Tree t = va_arg(ap, Tree); + print("%S", t->op); + if (t->left && t->right) + print("(%T,%T)", t->left, t->right); + else if (t->left) + print("(%T)", t->left); + break; + } + case 'R': { + Rule r = va_arg(ap, Rule); + print("%S: %T", r->lhs, r->pattern); + break; + } + case 'S': fputs(va_arg(ap, Term)->name, outfp); break; + case '1': case '2': case '3': case '4': case '5': { + int n = *fmt - '0'; + while (n-- > 0) + putc('\t', outfp); + break; + } + default: putc(*fmt, outfp); break; + } + else + putc(*fmt, outfp); + va_end(ap); +} + +/* reach - mark all nonterminals in tree t as reachable */ +static void reach(Tree t) { + Nonterm p = t->op; + + if (p->kind == NONTERM) + if (!p->reached) + ckreach(p); + if (t->left) + reach(t->left); + if (t->right) + reach(t->right); +} + +/* ckreach - mark all nonterminals reachable from p */ +static void ckreach(Nonterm p) { + Rule r; + + p->reached = 1; + for (r = p->rules; r; r = r->decode) + reach(r->pattern); +} + +/* emitcase - emit one case in function state */ +static void emitcase(Term p, int ntnumber) { + Rule r; + + print("%1case %d: /* %S */\n", p->esn, p); + switch (p->arity) { + case 0: case -1: + break; + case 1: + print("%2%Plabel(LEFT_CHILD(a));\n"); + break; + case 2: + print("%2%Plabel(LEFT_CHILD(a));\n"); + print("%2%Plabel(RIGHT_CHILD(a));\n"); + break; + default: assert(0); + } + for (r = p->rules; r; r = r->next) { + char *indent = "\t\t\0"; + switch (p->arity) { + case 0: case -1: + print("%2/* %R */\n", r); + if (r->cost == -1) { + print("%2c = %s;\n", r->code); + emitrecord("\t\t", r, "c", 0); + } else + emitrecord("\t\t", r, r->code, 0); + break; + case 1: + if (r->pattern->nterms > 1) { + print("%2if (%1/* %R */\n", r); + emittest(r->pattern->left, "LEFT_CHILD(a)", " "); + print("%2) {\n"); + indent = "\t\t\t"; + } else + print("%2/* %R */\n", r); + if (r->pattern->nterms == 2 && r->pattern->left + && r->pattern->right == NULL) + emitrecalc(indent, r->pattern->op, r->pattern->left->op); + print("%sc = ", indent); + emitcost(r->pattern->left, "LEFT_CHILD(a)"); + print("%s;\n", r->code); + emitrecord(indent, r, "c", 0); + if (indent[2]) + print("%2}\n"); + break; + case 2: + if (r->pattern->nterms > 1) { + print("%2if (%1/* %R */\n", r); + emittest(r->pattern->left, "LEFT_CHILD(a)", + r->pattern->right->nterms ? " && " : " "); + emittest(r->pattern->right, "RIGHT_CHILD(a)", " "); + print("%2) {\n"); + indent = "\t\t\t"; + } else + print("%2/* %R */\n", r); + print("%sc = ", indent); + emitcost(r->pattern->left, "LEFT_CHILD(a)"); + emitcost(r->pattern->right, "RIGHT_CHILD(a)"); + print("%s;\n", r->code); + emitrecord(indent, r, "c", 0); + if (indent[2]) + print("%2}\n"); + break; + default: assert(0); + } + } + print("%2break;\n"); +} + +/* emitclosure - emit the closure functions */ +static void emitclosure(Nonterm nts) { + Nonterm p; + + for (p = nts; p; p = p->link) + if (p->chain) + print("static void %Pclosure_%S(NODEPTR_TYPE, int);\n", p); + print("\n"); + for (p = nts; p; p = p->link) + if (p->chain) { + Rule r; + print("static void %Pclosure_%S(NODEPTR_TYPE a, int c) {\n" +"%1struct %Pstate *p = STATE_LABEL(a);\n", p); + for (r = p->chain; r; r = r->chain) + emitrecord("\t", r, "c", r->cost); + print("}\n\n"); + } +} + +/* emitcost - emit cost computation for tree t */ +static void emitcost(Tree t, char *v) { + Nonterm p = t->op; + + if (p->kind == TERM) { + if (t->left) + emitcost(t->left, stringf("LEFT_CHILD(%s)", v)); + if (t->right) + emitcost(t->right, stringf("RIGHT_CHILD(%s)", v)); + } else + print("((struct %Pstate *)(%s->x.state))->cost[%P%S_NT] + ", v, p); +} + +/* emitdefs - emit nonterminal defines and data structures */ +static void emitdefs(Nonterm nts, int ntnumber) { + Nonterm p; + + for (p = nts; p; p = p->link) + print("#define %P%S_NT %d\n", p, p->number); + print("\n"); + print("static char *%Pntname[] = {\n%10,\n"); + for (p = nts; p; p = p->link) + print("%1\"%S\",\n", p); + print("%10\n};\n\n"); +} + +/* emitheader - emit initial definitions */ +static void emitheader(void) { + time_t timer = time(NULL); + + print("/*\ngenerated at %sby %s\n*/\n", ctime(&timer), rcsid); + print("static void %Pkids(NODEPTR_TYPE, int, NODEPTR_TYPE[]);\n"); + print("static void %Plabel(NODEPTR_TYPE);\n"); + print("static int %Prule(void*, int);\n\n"); +} + +/* computekids - compute paths to kids in tree t */ +static char *computekids(Tree t, char *v, char *bp, int *ip) { + Term p = t->op; + + if (p->kind == NONTERM) { + sprintf(bp, "\t\tkids[%d] = %s;\n", (*ip)++, v); + bp += strlen(bp); + } else if (p->arity > 0) { + bp = computekids(t->left, stringf("LEFT_CHILD(%s)", v), bp, ip); + if (p->arity == 2) + bp = computekids(t->right, stringf("RIGHT_CHILD(%s)", v), bp, ip); + } + return bp; +} + +/* emitkids - emit _kids */ +static void emitkids(Rule rules, int nrules) { + int i; + Rule r, *rc = alloc((nrules + 1 + 1)*sizeof *rc); + char **str = alloc((nrules + 1 + 1)*sizeof *str); + + for (i = 0, r = rules; r; r = r->link) { + int j = 0; + char buf[1024], *bp = buf; + *computekids(r->pattern, "p", bp, &j) = 0; + for (j = 0; str[j] && strcmp(str[j], buf); j++) + ; + if (str[j] == NULL) + str[j] = strcpy(alloc(strlen(buf) + 1), buf); + r->kids = rc[j]; + rc[j] = r; + } + print("static void %Pkids(NODEPTR_TYPE p, int eruleno, NODEPTR_TYPE kids[]) {\n" +"%1if (!p)\n%2fatal(\"%Pkids\", \"Null tree\\n\", 0);\n" +"%1if (!kids)\n%2fatal(\"%Pkids\", \"Null kids\\n\", 0);\n" +"%1switch (eruleno) {\n"); + for (i = 0; (r = rc[i]) != NULL; i++) { + for ( ; r; r = r->kids) + print("%1case %d: /* %R */\n", r->ern, r); + print("%s%2break;\n", str[i]); + } + print("%1default:\n%2fatal(\"%Pkids\", \"Bad rule number %%d\\n\", eruleno);\n%1}\n}\n\n"); +} + +/* emitlabel - emit label function */ +static void emitlabel(Term terms, Nonterm start, int ntnumber) { + int i; + Term p; + + print("static void %Plabel(NODEPTR_TYPE a) {\n%1int c;\n" +"%1struct %Pstate *p;\n\n" +"%1if (!a)\n%2fatal(\"%Plabel\", \"Null tree\\n\", 0);\n"); + print("%1STATE_LABEL(a) = p = allocate(sizeof *p, FUNC);\n" +"%1p->rule._stmt = 0;\n"); + for (i = 1; i <= ntnumber; i++) + print("%1p->cost[%d] =\n", i); + print("%20x7fff;\n%1switch (OP_LABEL(a)) {\n"); + for (p = terms; p; p = p->link) + emitcase(p, ntnumber); + print("%1default:\n" +"%2fatal(\"%Plabel\", \"Bad terminal %%d\\n\", OP_LABEL(a));\n%1}\n}\n\n"); +} + +/* computents - fill in bp with _nts vector for tree t */ +static char *computents(Tree t, char *bp) { + if (t) { + Nonterm p = t->op; + if (p->kind == NONTERM) { + sprintf(bp, "%s_%s_NT, ", prefix, p->name); + bp += strlen(bp); + } else + bp = computents(t->right, computents(t->left, bp)); + } + return bp; +} + +/* emitnts - emit _nts ragged array */ +static void emitnts(Rule rules, int nrules) { + Rule r; + int i, j, *nts = alloc((nrules + 1)*sizeof *nts); + char **str = alloc((nrules + 1)*sizeof *str); + + for (i = 0, r = rules; r; r = r->link) { + char buf[1024]; + *computents(r->pattern, buf) = 0; + for (j = 0; str[j] && strcmp(str[j], buf); j++) + ; + if (str[j] == NULL) { + print("static short %Pnts_%d[] = { %s0 };\n", j, buf); + str[j] = strcpy(alloc(strlen(buf) + 1), buf); + } + nts[i++] = j; + } + print("\nstatic short *%Pnts[] = {\n"); + for (i = j = 0, r = rules; r; r = r->link) { + for ( ; j < r->ern; j++) + print("%10,%1/* %d */\n", j); + print("%1%Pnts_%d,%1/* %d */\n", nts[i++], j++); + } + print("};\n\n"); +} + +/* emitrecalc - emit code that tests for recalculation of INDIR?(VREGP) */ +static void emitrecalc(char *pre, Term root, Term kid) { + if (root->kind == TERM && strncmp(root->name, "INDIR", 5) == 0 + && kid->kind == TERM && strcmp(kid->name, "VREGP" ) == 0) { + Nonterm p; + print("%sif (mayrecalc(a)) {\n", pre); + print("%s%1struct %Pstate *q = a->syms[RX]->u.t.cse->x.state;\n", pre); + for (p = nts; p; p = p->link) { + print("%s%1if (q->cost[%P%S_NT] == 0) {\n", pre, p); + print("%s%2p->cost[%P%S_NT] = 0;\n", pre, p); + print("%s%2p->rule.%P%S = q->rule.%P%S;\n", pre, p, p); + print("%s%1}\n", pre); + } + print("%s}\n", pre); + } +} + +/* emitrecord - emit code that tests for a winning match of rule r */ +static void emitrecord(char *pre, Rule r, char *c, int cost) { + if (Tflag) + print("%s%Ptrace(a, %d, %s + %d, p->cost[%P%S_NT]);\n", + pre, r->ern, c, cost, r->lhs); + print("%sif (", pre); + print("%s + %d < p->cost[%P%S_NT]) {\n" +"%s%1p->cost[%P%S_NT] = %s + %d;\n%s%1p->rule.%P%S = %d;\n", + c, cost, r->lhs, pre, r->lhs, c, cost, pre, r->lhs, + r->packed); + if (r->lhs->chain) + print("%s%1%Pclosure_%S(a, %s + %d);\n", pre, r->lhs, c, cost); + print("%s}\n", pre); +} + +/* emitrule - emit decoding vectors and _rule */ +static void emitrule(Nonterm nts) { + Nonterm p; + + for (p = nts; p; p = p->link) { + Rule r; + print("static short %Pdecode_%S[] = {\n%10,\n", p); + for (r = p->rules; r; r = r->decode) + print("%1%d,\n", r->ern); + print("};\n\n"); + } + print("static int %Prule(void *state, int goalnt) {\n" +"%1if (goalnt < 1 || goalnt > %d)\n%2fatal(\"%Prule\", \"Bad goal nonterminal %%d\\n\", goalnt);\n" +"%1if (!state)\n%2return 0;\n%1switch (goalnt) {\n", ntnumber); + for (p = nts; p; p = p->link) + print("%1case %P%S_NT:" +"%1return %Pdecode_%S[((struct %Pstate *)state)->rule.%P%S];\n", p, p, p); + print("%1default:\n%2fatal(\"%Prule\", \"Bad goal nonterminal %%d\\n\", goalnt);\n%2return 0;\n%1}\n}\n\n"); +} + +/* emitstring - emit arrays of templates, instruction flags, and rules */ +static void emitstring(Rule rules) { + Rule r; + + print("static char *%Ptemplates[] = {\n"); + print("/* 0 */%10,\n"); + for (r = rules; r; r = r->link) + print("/* %d */%1\"%s\",%1/* %R */\n", r->ern, r->template, r); + print("};\n"); + print("\nstatic char %Pisinstruction[] = {\n"); + print("/* 0 */%10,\n"); + for (r = rules; r; r = r->link) { + int len = strlen(r->template); + print("/* %d */%1%d,%1/* %s */\n", r->ern, + len >= 2 && r->template[len-2] == '\\' && r->template[len-1] == 'n', + r->template); + } + print("};\n"); + print("\nstatic char *%Pstring[] = {\n"); + print("/* 0 */%10,\n"); + for (r = rules; r; r = r->link) + print("/* %d */%1\"%R\",\n", r->ern, r); + print("};\n\n"); +} + +/* emitstruct - emit the definition of the state structure */ +static void emitstruct(Nonterm nts, int ntnumber) { + print("struct %Pstate {\n%1short cost[%d];\n%1struct {\n", ntnumber + 1); + for ( ; nts; nts = nts->link) { + int n = 1, m = nts->lhscount; + while ((m >>= 1) != 0) + n++; + print("%2unsigned int %P%S:%d;\n", nts, n); + } + print("%1} rule;\n};\n\n"); +} + +/* emittest - emit clause for testing a match */ +static void emittest(Tree t, char *v, char *suffix) { + Term p = t->op; + + if (p->kind == TERM) { + print("%3%s->op == %d%s/* %S */\n", v, p->esn, + t->nterms > 1 ? " && " : suffix, p); + if (t->left) + emittest(t->left, stringf("LEFT_CHILD(%s)", v), + t->right && t->right->nterms ? " && " : suffix); + if (t->right) + emittest(t->right, stringf("RIGHT_CHILD(%s)", v), suffix); + } +} diff --git a/src/tools/lcc/lburg/lburg.h b/src/tools/lcc/lburg/lburg.h new file mode 100644 index 0000000..b67e802 --- /dev/null +++ b/src/tools/lcc/lburg/lburg.h @@ -0,0 +1,65 @@ +#ifndef BURG_INCLUDED +#define BURG_INCLUDED + +/* iburg.c: */ +extern void *alloc(int nbytes); + +typedef enum { TERM=1, NONTERM } Kind; +typedef struct rule *Rule; +typedef struct term *Term; +struct term { /* terminals: */ + char *name; /* terminal name */ + Kind kind; /* TERM */ + int esn; /* external symbol number */ + int arity; /* operator arity */ + Term link; /* next terminal in esn order */ + Rule rules; /* rules whose pattern starts with term */ +}; + +typedef struct nonterm *Nonterm; +struct nonterm { /* nonterminals: */ + char *name; /* nonterminal name */ + Kind kind; /* NONTERM */ + int number; /* identifying number */ + int lhscount; /* # times nt appears in a rule lhs */ + int reached; /* 1 iff reached from start nonterminal */ + Rule rules; /* rules w/nonterminal on lhs */ + Rule chain; /* chain rules w/nonterminal on rhs */ + Nonterm link; /* next terminal in number order */ +}; +extern Nonterm nonterm(char *id); +extern Term term(char *id, int esn); + +typedef struct tree *Tree; +struct tree { /* tree patterns: */ + void *op; /* a terminal or nonterminal */ + Tree left, right; /* operands */ + int nterms; /* number of terminal nodes in this tree */ +}; +extern Tree tree(char *op, Tree left, Tree right); + +struct rule { /* rules: */ + Nonterm lhs; /* lefthand side nonterminal */ + Tree pattern; /* rule pattern */ + int ern; /* external rule number */ + int packed; /* packed external rule number */ + int cost; /* cost, if a constant */ + char *code; /* cost, if an expression */ + char *template; /* assembler template */ + Rule link; /* next rule in ern order */ + Rule next; /* next rule with same pattern root */ + Rule chain; /* next chain rule with same rhs */ + Rule decode; /* next rule with same lhs */ + Rule kids; /* next rule with same _kids pattern */ +}; +extern Rule rule(char *id, Tree pattern, char *template, char *code); + +/* gram.y: */ +void yyerror(char *fmt, ...); +int yyparse(void); +void yywarn(char *fmt, ...); +extern int errcnt; +extern FILE *infp; +extern FILE *outfp; + +#endif -- cgit