diff options
Diffstat (limited to 'awkgram.y')
| -rw-r--r-- | awkgram.y | 625 |
1 files changed, 535 insertions, 90 deletions
@@ -57,6 +57,7 @@ static int include_source(INSTRUCTION *file); static int load_library(INSTRUCTION *file); static void next_sourcefile(void); static char *tokexpand(void); +static NODE *make_profile_number(double d, const char *str, size_t len); #define instruction(t) bcalloc(t, 1, 0) @@ -83,10 +84,18 @@ static void check_funcs(void); static ssize_t read_one_line(int fd, void *buffer, size_t count); static int one_line_close(int fd); +static void split_comment(void); +static void check_comment(void); +static void add_sign_to_num(NODE *n, char sign); static bool at_seen = false; static bool want_source = false; static bool want_regexp = false; /* lexical scanning kludge */ +static enum { + FUNC_HEADER, + FUNC_BODY, + DONT_CHECK +} want_param_names = DONT_CHECK; /* ditto */ static char *in_function; /* parsing kludge */ static int rule = 0; @@ -141,11 +150,23 @@ static INSTRUCTION *ip_atexit = NULL; static INSTRUCTION *ip_end; static INSTRUCTION *ip_endfile; static INSTRUCTION *ip_beginfile; +INSTRUCTION *main_beginfile; + +static INSTRUCTION *comment = NULL; +static INSTRUCTION *prior_comment = NULL; +static INSTRUCTION *comment_to_save = NULL; +static INSTRUCTION *program_comment = NULL; +static INSTRUCTION *function_comment = NULL; +static INSTRUCTION *block_comment = NULL; + +static bool func_first = true; +static bool first_rule = true; static inline INSTRUCTION *list_create(INSTRUCTION *x); static inline INSTRUCTION *list_append(INSTRUCTION *l, INSTRUCTION *x); static inline INSTRUCTION *list_prepend(INSTRUCTION *l, INSTRUCTION *x); static inline INSTRUCTION *list_merge(INSTRUCTION *l1, INSTRUCTION *l2); +static inline INSTRUCTION *add_pending_comment(INSTRUCTION *stmt); extern double fmod(double x, double y); @@ -218,6 +239,7 @@ rule : pattern action { (void) append_rule($1, $2); + first_rule = false; } | pattern statement_term { @@ -234,6 +256,7 @@ rule { in_function = NULL; (void) mk_function($1, $2); + want_param_names = DONT_CHECK; yyerrok; } | '@' LEX_INCLUDE source statement_term @@ -282,9 +305,24 @@ library pattern : /* empty */ - { $$ = NULL; rule = Rule; } + { + rule = Rule; + if (comment != NULL) { + $$ = list_create(comment); + comment = NULL; + } else + $$ = NULL; + } | exp - { $$ = $1; rule = Rule; } + { + rule = Rule; + if (comment != NULL) { + $$ = list_prepend($1, comment); + comment = NULL; + } else + $$ = $1; + } + | exp ',' opt_nls exp { INSTRUCTION *tp; @@ -308,41 +346,55 @@ pattern ($1->nexti + 1)->condpair_left = $1->lasti; ($1->nexti + 1)->condpair_right = $4->lasti; } - $$ = list_append(list_merge($1, $4), tp); + if (comment != NULL) { + $$ = list_append(list_merge(list_prepend($1, comment), $4), tp); + comment = NULL; + } else + $$ = list_append(list_merge($1, $4), tp); rule = Rule; } | LEX_BEGIN { static int begin_seen = 0; + + func_first = false; if (do_lint_old && ++begin_seen == 2) warning_ln($1->source_line, _("old awk does not support multiple `BEGIN' or `END' rules")); $1->in_rule = rule = BEGIN; $1->source_file = source; + check_comment(); $$ = $1; } | LEX_END { static int end_seen = 0; + + func_first = false; if (do_lint_old && ++end_seen == 2) warning_ln($1->source_line, _("old awk does not support multiple `BEGIN' or `END' rules")); $1->in_rule = rule = END; $1->source_file = source; + check_comment(); $$ = $1; } | LEX_BEGINFILE { + func_first = false; $1->in_rule = rule = BEGINFILE; $1->source_file = source; + check_comment(); $$ = $1; } | LEX_ENDFILE { + func_first = false; $1->in_rule = rule = ENDFILE; $1->source_file = source; + check_comment(); $$ = $1; } ; @@ -350,10 +402,12 @@ pattern action : l_brace statements r_brace opt_semi opt_nls { + INSTRUCTION *ip; if ($2 == NULL) - $$ = list_create(instruction(Op_no_op)); + ip = list_create(instruction(Op_no_op)); else - $$ = $2; + ip = $2; + $$ = ip; } ; @@ -381,16 +435,43 @@ lex_builtin ; function_prologue - : LEX_FUNCTION func_name '(' opt_param_list r_paren opt_nls + : LEX_FUNCTION func_name '(' { want_param_names = FUNC_HEADER; } opt_param_list r_paren opt_nls { + /* + * treat any comments between BOF and the first function + * definition (with no intervening BEGIN etc block) as + * program comments. Special kludge: iff there are more + * than one such comments, treat the last as a function + * comment. + */ + if (prior_comment != NULL) { + comment_to_save = prior_comment; + prior_comment = NULL; + } else if (comment != NULL) { + comment_to_save = comment; + comment = NULL; + } else + comment_to_save = NULL; + + if (comment_to_save != NULL && func_first + && strstr(comment_to_save->memory->stptr, "\n\n") != NULL) + split_comment(); + + /* save any other pre-function comment as function comment */ + if (comment_to_save != NULL) { + function_comment = comment_to_save; + comment_to_save = NULL; + } + func_first = false; $1->source_file = source; - if (install_function($2->lextok, $1, $4) < 0) + if (install_function($2->lextok, $1, $5) < 0) YYABORT; in_function = $2->lextok; $2->lextok = NULL; bcfree($2); - /* $4 already free'd in install_function */ + /* $5 already free'd in install_function */ $$ = $1; + want_param_names = FUNC_BODY; } ; @@ -440,19 +521,62 @@ a_slash statements : /* empty */ - { $$ = NULL; } + { + if (prior_comment != NULL) { + $$ = list_create(prior_comment); + prior_comment = NULL; + } else if (comment != NULL) { + $$ = list_create(comment); + comment = NULL; + } else + $$ = NULL; + } | statements statement { - if ($2 == NULL) - $$ = $1; - else { + if ($2 == NULL) { + if (prior_comment != NULL) { + $$ = list_append($1, prior_comment); + prior_comment = NULL; + if (comment != NULL) { + $$ = list_append($$, comment); + comment = NULL; + } + } else if (comment != NULL) { + $$ = list_append($1, comment); + comment = NULL; + } else + $$ = $1; + } else { add_lint($2, LINT_no_effect); - if ($1 == NULL) - $$ = $2; - else + if ($1 == NULL) { + if (prior_comment != NULL) { + $$ = list_append($2, prior_comment); + prior_comment = NULL; + if (comment != NULL) { + $$ = list_append($$, comment); + comment = NULL; + } + } else if (comment != NULL) { + $$ = list_append($2, comment); + comment = NULL; + } else + $$ = $2; + } else { + if (prior_comment != NULL) { + list_append($2, prior_comment); + prior_comment = NULL; + if (comment != NULL) { + list_append($2, comment); + comment = NULL; + } + } else if (comment != NULL) { + list_append($2, comment); + comment = NULL; + } $$ = list_merge($1, $2); + } } - yyerrok; + yyerrok; } | statements error { $$ = NULL; } @@ -496,7 +620,7 @@ statement } /* else curr = NULL; */ - for(; curr != NULL; curr = nextc) { + for (; curr != NULL; curr = nextc) { INSTRUCTION *caseexp = curr->case_exp; INSTRUCTION *casestmt = curr->case_stmt; @@ -782,6 +906,7 @@ regular_loop: $$ = list_prepend($1, instruction(Op_exec_count)); else $$ = $1; + $$ = add_pending_comment($$); } ; @@ -793,6 +918,7 @@ non_compound_stmt _("`break' is not allowed outside a loop or switch")); $1->target_jmp = NULL; $$ = list_create($1); + $$ = add_pending_comment($$); } | LEX_CONTINUE statement_term @@ -802,6 +928,7 @@ non_compound_stmt _("`continue' is not allowed outside a loop")); $1->target_jmp = NULL; $$ = list_create($1); + $$ = add_pending_comment($$); } | LEX_NEXT statement_term @@ -812,6 +939,7 @@ non_compound_stmt _("`next' used in %s action"), ruletab[rule]); $1->target_jmp = ip_rec; $$ = list_create($1); + $$ = add_pending_comment($$); } | LEX_NEXTFILE statement_term { @@ -823,6 +951,7 @@ non_compound_stmt $1->target_newfile = ip_newfile; $1->target_endfile = ip_endfile; $$ = list_create($1); + $$ = add_pending_comment($$); } | LEX_EXIT opt_exp statement_term { @@ -838,6 +967,7 @@ non_compound_stmt $$->nexti->memory = dupnode(Nnull_string); } else $$ = list_append($2, $1); + $$ = add_pending_comment($$); } | LEX_RETURN { @@ -862,6 +992,7 @@ non_compound_stmt $$ = list_append($3, $1); } + $$ = add_pending_comment($$); } | simple_stmt statement_term ; @@ -971,6 +1102,7 @@ regular_print: } } } + $$ = add_pending_comment($$); } | LEX_DELETE NAME { sub_counter = 0; } delete_subscript_list @@ -1005,6 +1137,7 @@ regular_print: $1->expr_count = sub_counter; $$ = list_append(list_append($4, $2), $1); } + $$ = add_pending_comment($$); } | LEX_DELETE '(' NAME ')' /* @@ -1035,9 +1168,13 @@ regular_print: else if ($3->memory == func_table) fatal(_("`delete' is not allowed with FUNCTAB")); } + $$ = add_pending_comment($$); } | exp - { $$ = optimize_assignment($1); } + { + $$ = optimize_assignment($1); + $$ = add_pending_comment($$); + } ; opt_simple_stmt @@ -1100,14 +1237,19 @@ case_value } | '+' YNUMBER %prec UNARY { + NODE *n = $2->lasti->memory; bcfree($1); + add_sign_to_num(n, '+'); $$ = $2; } | YSTRING { $$ = $1; } | regexp { - $1->opcode = Op_push_re; + if ($1->memory->type == Node_regex) + $1->opcode = Op_push_re; + else + $1->opcode = Op_push; $$ = $1; } ; @@ -1184,7 +1326,7 @@ opt_param_list : /* empty */ { $$ = NULL; } | param_list - { $$ = $1 ; } + { $$ = $1; } ; param_list @@ -1255,6 +1397,47 @@ expression_list } ; +opt_fcall_expression_list + : /* empty */ + { $$ = NULL; } + | fcall_expression_list + { $$ = $1; } + ; + +fcall_expression_list + : fcall_exp + { $$ = mk_expression_list(NULL, $1); } + | fcall_expression_list comma fcall_exp + { + $$ = mk_expression_list($1, $3); + yyerrok; + } + | error + { $$ = NULL; } + | fcall_expression_list error + { + /* + * Returning the expression list instead of NULL lets + * snode get a list of arguments that it can count. + */ + $$ = $1; + } + | fcall_expression_list error fcall_exp + { + /* Ditto */ + $$ = mk_expression_list($1, $3); + } + | fcall_expression_list comma error + { + /* Ditto */ + $$ = $1; + } + ; + +fcall_exp + : exp { $$ = $1; } + ; + /* Expressions, not including the comma operator. */ exp : variable assign_operator exp %prec ASSIGNOP @@ -1275,6 +1458,7 @@ exp _("regular expression on left of `~' or `!~' operator")); if ($3->lasti == $3->nexti && $3->nexti->opcode == Op_match_rec) { + /* RHS is /.../ */ $2->memory = $3->nexti->memory; bcfree($3->nexti); /* Op_match_rec */ bcfree($3); /* Op_list */ @@ -1364,10 +1548,15 @@ common_exp NODE *n2 = $2->nexti->memory; size_t nlen; + // 1.5 "" # can't fold this if program mucks with CONVFMT. + // See test #12 in test/posix.awk. + if ((n1->flags & (NUMBER|NUMINT)) != 0 || (n2->flags & (NUMBER|NUMINT)) != 0) + goto plain_concat; + n1 = force_string(n1); n2 = force_string(n2); nlen = n1->stlen + n2->stlen; - erealloc(n1->stptr, char *, nlen + 2, "constant fold"); + erealloc(n1->stptr, char *, nlen + 1, "constant fold"); memcpy(n1->stptr + n1->stlen, n2->stptr, n2->stlen); n1->stlen = nlen; n1->stptr[nlen] = '\0'; @@ -1378,6 +1567,7 @@ common_exp bcfree($2); $$ = $1; } else { + plain_concat: $$ = list_append(list_merge($1, $2), instruction(Op_concat)); $$->lasti->concat_flag = (is_simple_var ? CSVAR : 0); $$->lasti->expr_count = count; @@ -1481,7 +1671,7 @@ non_post_simp_exp if ($2->opcode == Op_match_rec) { $2->opcode = Op_nomatch; $1->opcode = Op_push_i; - $1->memory = make_number(0.0); + $1->memory = make_profile_number(0.0, "0", 1); $$ = list_append(list_append(list_create($1), instruction(Op_field_spec)), $2); } else { @@ -1490,7 +1680,7 @@ non_post_simp_exp && ($2->nexti->memory->flags & (MPFN|MPZN)) == 0 ) { NODE *n = $2->nexti->memory; - if ((n->flags & (STRCUR|STRING)) != 0) { + if ((n->flags & STRING) != 0) { n->numbr = (AWKNUM) (n->stlen == 0); n->flags &= ~(STRCUR|STRING); n->flags |= (NUMCUR|NUMBER); @@ -1510,13 +1700,13 @@ non_post_simp_exp } | '(' exp r_paren { $$ = $2; } - | LEX_BUILTIN '(' opt_expression_list r_paren + | LEX_BUILTIN '(' opt_fcall_expression_list r_paren { $$ = snode($3, $1); if ($$ == NULL) YYABORT; } - | LEX_LENGTH '(' opt_expression_list r_paren + | LEX_LENGTH '(' opt_fcall_expression_list r_paren { $$ = snode($3, $1); if ($$ == NULL) @@ -1558,7 +1748,7 @@ non_post_simp_exp | '-' simp_exp %prec UNARY { if ($2->lasti->opcode == Op_push_i - && ($2->lasti->memory->flags & (STRCUR|STRING)) == 0 + && ($2->lasti->memory->flags & STRING) == 0 ) { NODE *n = $2->lasti->memory; (void) force_number(n); @@ -1572,13 +1762,21 @@ non_post_simp_exp } | '+' simp_exp %prec UNARY { - /* - * was: $$ = $2 - * POSIX semantics: force a conversion to numeric type - */ - $1->opcode = Op_plus_i; - $1->memory = make_number(0.0); - $$ = list_append($2, $1); + if ($2->lasti->opcode == Op_push_i + && ($2->lasti->memory->flags & STRING) == 0 + && ($2->lasti->memory->flags & NUMCONSTSTR) != 0) { + NODE *n = $2->lasti->memory; + add_sign_to_num(n, '+'); + $$ = $2; + bcfree($1); + } else { + /* + * was: $$ = $2 + * POSIX semantics: force a conversion to numeric type + */ + $1->opcode = Op_unary_plus; + $$ = list_append($2, $1); + } } ; @@ -1625,14 +1823,14 @@ func_call ; direct_func_call - : FUNC_CALL '(' opt_expression_list r_paren + : FUNC_CALL '(' opt_fcall_expression_list r_paren { NODE *n; if (! at_seen) { n = lookup($1->func_name); if (n != NULL && n->type != Node_func - && n->type != Node_ext_func && n->type != Node_old_ext_func) { + && n->type != Node_ext_func) { error_ln($1->source_line, _("attempt to use non-function `%s' in function call"), $1->func_name); @@ -1878,9 +2076,6 @@ static const struct token tokentab[] = { {"eval", Op_symbol, LEX_EVAL, 0, 0, 0}, {"exit", Op_K_exit, LEX_EXIT, 0, 0, 0}, {"exp", Op_builtin, LEX_BUILTIN, A(1), do_exp, MPF(exp)}, -#ifdef DYNAMIC -{"extension", Op_builtin, LEX_BUILTIN, GAWKX|A(1)|A(2)|A(3), do_ext, 0}, -#endif {"fflush", Op_builtin, LEX_BUILTIN, A(0)|A(1), do_fflush, 0}, {"for", Op_K_for, LEX_FOR, BREAK|CONTINUE, 0, 0}, {"func", Op_func, LEX_FUNCTION, NOT_POSIX|NOT_OLD, 0, 0}, @@ -1893,6 +2088,7 @@ static const struct token tokentab[] = { {"include", Op_symbol, LEX_INCLUDE, GAWKX, 0, 0}, {"index", Op_builtin, LEX_BUILTIN, A(2), do_index, 0}, {"int", Op_builtin, LEX_BUILTIN, A(1), do_int, MPF(int)}, +{"intdiv", Op_builtin, LEX_BUILTIN, GAWKX|A(3), do_intdiv, MPF(intdiv)}, {"isarray", Op_builtin, LEX_BUILTIN, GAWKX|A(1), do_isarray, 0}, {"length", Op_builtin, LEX_LENGTH, A(0)|A(1), do_length, 0}, {"load", Op_symbol, LEX_LOAD, GAWKX, 0, 0}, @@ -1926,6 +2122,7 @@ static const struct token tokentab[] = { {"systime", Op_builtin, LEX_BUILTIN, GAWKX|A(0), do_systime, 0}, {"tolower", Op_builtin, LEX_BUILTIN, NOT_OLD|A(1), do_tolower, 0}, {"toupper", Op_builtin, LEX_BUILTIN, NOT_OLD|A(1), do_toupper, 0}, +{"typeof", Op_builtin, LEX_BUILTIN, GAWKX|A(1), do_typeof, 0}, {"while", Op_K_while, LEX_WHILE, BREAK|CONTINUE, 0, 0}, {"xor", Op_builtin, LEX_BUILTIN, GAWKX, do_xor, MPF(xor)}, }; @@ -1967,6 +2164,8 @@ negate_num(NODE *n) int tval = 0; #endif + add_sign_to_num(n, '-'); + if (! is_mpg_number(n)) { n->numbr = -n->numbr; return; @@ -1999,6 +2198,21 @@ negate_num(NODE *n) #endif } +/* add_sign_to_num --- make a constant unary plus or minus for profiling */ + +static void +add_sign_to_num(NODE *n, char sign) +{ + if ((n->flags & NUMCONSTSTR) != 0) { + char *s; + + s = n->stptr; + memmove(& s[1], & s[0], n->stlen + 1); + s[0] = sign; + n->stlen++; + } +} + /* print_included_from --- print `Included from ..' file names and locations */ static void @@ -2101,7 +2315,6 @@ yyerror(const char *m, ...) char *buf; int count; static char end_of_file_line[] = "(END OF FILE)"; - char save; print_included_from(); @@ -2132,25 +2345,17 @@ yyerror(const char *m, ...) bp = thisline + strlen(thisline); } - /* - * Saving and restoring *bp keeps valgrind happy, - * since the guts of glibc uses strlen, even though - * we're passing an explict precision. Sigh. - * - * 8/2003: We may not need this anymore. - */ - save = *bp; - *bp = '\0'; - msg("%.*s", (int) (bp - thisline), thisline); - *bp = save; va_start(args, m); if (mesg == NULL) mesg = m; - count = (bp - thisline) + strlen(mesg) + 2 + 1; - emalloc(buf, char *, count, "yyerror"); + count = strlen(mesg) + 1; + if (lexptr != NULL) + count += (lexeme - thisline) + 2; + emalloc(buf, char *, count+1, "yyerror"); + memset(buf, 0, count+1); bp = buf; @@ -2232,6 +2437,11 @@ mk_program() cp = end_block; else cp = list_merge(begin_block, end_block); + if (program_comment != NULL) { + (void) list_prepend(cp, program_comment); + } + if (comment != NULL) + (void) list_append(cp, comment); (void) list_append(cp, ip_atexit); (void) list_append(cp, instruction(Op_stop)); @@ -2264,6 +2474,12 @@ mk_program() if (begin_block != NULL) cp = list_merge(begin_block, cp); + if (program_comment != NULL) { + (void) list_prepend(cp, program_comment); + } + if (comment != NULL) { + (void) list_append(cp, comment); + } (void) list_append(cp, ip_atexit); (void) list_append(cp, instruction(Op_stop)); @@ -2271,6 +2487,10 @@ out: /* delete the Op_list, not needed */ tmp = cp->nexti; bcfree(cp); + /* these variables are not used again but zap them anyway. */ + comment = NULL; + function_comment = NULL; + program_comment = NULL; return tmp; #undef begin_block @@ -2297,7 +2517,7 @@ parse_program(INSTRUCTION **pcode) ip_newfile = ip_rec = ip_atexit = ip_beginfile = ip_endfile = NULL; else { ip_endfile = instruction(Op_no_op); - ip_beginfile = instruction(Op_no_op); + main_beginfile = ip_beginfile = instruction(Op_no_op); ip_rec = instruction(Op_get_record); /* target for `next', also ip_newfile */ ip_newfile = bcalloc(Op_newfile, 2, 0); /* target for `nextfile' */ ip_newfile->target_jmp = ip_end; @@ -2768,7 +2988,7 @@ get_src_buf() lexend = lexptr + n; if (n == 0) { static bool warned = false; - if (do_lint && newfile && ! warned){ + if (do_lint && newfile && ! warned) { warned = true; sourceline = 0; lintwarn(_("source file `%s' is empty"), source); @@ -2859,7 +3079,7 @@ again: mbstate_t tmp_state; size_t mbclen; - for (idx = 0 ; lexptr + idx < lexend ; idx++) { + for (idx = 0; lexptr + idx < lexend; idx++) { tmp_state = cur_mbstate; mbclen = mbrlen(lexptr, idx + 1, &tmp_state); @@ -2916,6 +3136,122 @@ pushback(void) (! lexeof && lexptr && lexptr > lexptr_begin ? lexptr-- : lexptr); } +/* check_comment --- check for block comment */ + +void +check_comment(void) +{ + if (comment != NULL) { + if (first_rule) { + program_comment = comment; + } else + block_comment = comment; + comment = NULL; + } + first_rule = false; +} + +/* + * get_comment --- collect comment text. + * Flag = EOL_COMMENT for end-of-line comments. + * Flag = FULL_COMMENT for self-contained comments. + */ + +int +get_comment(int flag) +{ + int c; + int sl; + tok = tokstart; + tokadd('#'); + sl = sourceline; + char *p1; + char *p2; + + while (true) { + while ((c = nextc(false)) != '\n' && c != END_FILE) { + /* ignore \r characters */ + if (c != '\r') + tokadd(c); + } + if (flag == EOL_COMMENT) { + /* comment at end of line. */ + if (c == '\n') + tokadd(c); + break; + } + if (c == '\n') { + tokadd(c); + sourceline++; + do { + c = nextc(false); + if (c == '\n') { + sourceline++; + tokadd(c); + } + } while (isspace(c) && c != END_FILE); + if (c == END_FILE) + break; + else if (c != '#') { + pushback(); + sourceline--; + break; + } else + tokadd(c); + } else + break; + } + + if (comment != NULL) + prior_comment = comment; + + /* remove any trailing blank lines (consecutive \n) from comment */ + p1 = tok - 1; + p2 = tok - 2; + while (*p1 == '\n' && *p2 == '\n') { + p1--; + p2--; + tok--; + } + + comment = bcalloc(Op_comment, 1, sl); + comment->source_file = source; + comment->memory = make_str_node(tokstart, tok - tokstart, 0); + comment->memory->comment_type = flag; + + return c; +} + +/* split_comment --- split initial comment text into program and function parts */ + +static void +split_comment(void) +{ + char *p; + int l; + NODE *n; + + p = comment_to_save->memory->stptr; + l = comment_to_save->memory->stlen - 3; + /* have at least two comments so split at last blank line (\n\n) */ + while (l >= 0) { + if (p[l] == '\n' && p[l+1] == '\n') { + function_comment = comment_to_save; + n = function_comment->memory; + function_comment->memory = make_str_node(p + l + 2, n->stlen - l - 2, 0); + /* create program comment */ + program_comment = bcalloc(Op_comment, 1, sourceline); + program_comment->source_file = comment_to_save->source_file; + p[l + 2] = 0; + program_comment->memory = make_str_node(p, l + 2, 0); + comment_to_save = NULL; + freenode(n); + break; + } + else + l--; + } +} /* allow_newline --- allow newline after &&, ||, ? and : */ @@ -2931,8 +3267,13 @@ allow_newline(void) break; } if (c == '#') { - while ((c = nextc(false)) != '\n' && c != END_FILE) - continue; + if (do_pretty_print && ! do_profile) { + /* collect comment byte code iff doing pretty print but not profiling. */ + c = get_comment(EOL_COMMENT); + } else { + while ((c = nextc(false)) != '\n' && c != END_FILE) + continue; + } if (c == END_FILE) { pushback(); break; @@ -2955,7 +3296,8 @@ allow_newline(void) * removes the warnings. */ -static int newline_eof() +static int +newline_eof() { /* NB: a newline at end does not start a source line. */ if (lasttok != NEWLINE) { @@ -3027,6 +3369,7 @@ yylex(void) lexeme = lexptr; thisline = NULL; + if (want_regexp) { int in_brack = 0; /* count brackets, [[:alnum:]] allowed */ int b_index = -1; @@ -3113,7 +3456,9 @@ end_regexp: peek); } } - return lasttok = REGEXP; + lasttok = REGEXP; + + return lasttok; case '\n': pushback(); yyerror(_("unterminated regexp")); @@ -3149,9 +3494,23 @@ retry: return lasttok = NEWLINE; case '#': /* it's a comment */ - while ((c = nextc(false)) != '\n') { + if (do_pretty_print && ! do_profile) { + /* + * Collect comment byte code iff doing pretty print + * but not profiling. + */ + if (lasttok == NEWLINE || lasttok == 0) + c = get_comment(FULL_COMMENT); + else + c = get_comment(EOL_COMMENT); + if (c == END_FILE) return lasttok = NEWLINE_EOF; + } else { + while ((c = nextc(false)) != '\n') { + if (c == END_FILE) + return lasttok = NEWLINE_EOF; + } } sourceline++; return lasttok = NEWLINE; @@ -3163,7 +3522,7 @@ retry: case '\\': #ifdef RELAXED_CONTINUATION /* - * This code puports to allow comments and/or whitespace + * This code purports to allow comments and/or whitespace * after the `\' at the end of a line used for continuation. * Use it at your own risk. We think it's a bad idea, which * is why it's not on by default. @@ -3180,9 +3539,13 @@ retry: lintwarn( _("use of `\\ #...' line continuation is not portable")); } - while ((c = nextc(false)) != '\n') - if (c == END_FILE) - break; + if (do_pretty_print && ! do_profile) + c = get_comment(EOL_COMMENT); + else { + while ((c = nextc(false)) != '\n') + if (c == END_FILE) + break; + } } pushback(); } @@ -3398,7 +3761,7 @@ retry: lastline = sourceline; return lasttok = c; } - did_newline++; + did_newline = true; --lexptr; /* pick up } next time */ return lasttok = NEWLINE; @@ -3606,14 +3969,19 @@ retry: IEEE_FMT(r->mpg_numbr, tval); } yylval->memory = r; + if (do_pretty_print) { + yylval->memory->stptr = estrdup(tokstart, strlen(tokstart)-1); + yylval->memory->stlen = strlen(tokstart)-1; + yylval->memory->flags |= NUMCONSTSTR; + } return lasttok = YNUMBER; } #endif if (base != 10) - d = nondec2awknum(tokstart, strlen(tokstart)); + d = nondec2awknum(tokstart, strlen(tokstart), NULL); else d = atof(tokstart); - yylval->memory = make_number(d); + yylval->memory = make_profile_number(d, tokstart, strlen(tokstart) - 1); if (d <= INT32_MAX && d >= INT32_MIN && d == (int32_t) d) yylval->memory->flags |= NUMINT; return lasttok = YNUMBER; @@ -3696,6 +4064,33 @@ retry: && lasttok != '@') goto out; + /* allow parameter names to shadow the names of gawk extension built-ins */ + if ((tokentab[mid].flags & GAWKX) != 0) { + NODE *f; + + switch (want_param_names) { + case FUNC_HEADER: + /* in header, defining parameter names */ + goto out; + case FUNC_BODY: + /* in body, name must be in symbol table for it to be a parameter */ + if ((f = lookup(tokstart)) != NULL) { + if (f->type == Node_builtin_func) + break; + else + goto out; + } + /* else + fall through */ + case DONT_CHECK: + /* regular code */ + break; + default: + cant_happen(); + break; + } + } + if (do_lint) { if ((tokentab[mid].flags & GAWKX) != 0 && (warntab[mid] & GAWKX) == 0) { lintwarn(_("`%s' is a gawk extension"), @@ -3896,7 +4291,7 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) INSTRUCTION *expr; expr = list_create(instruction(Op_push_i)); - expr->nexti->memory = make_number(0.0); + expr->nexti->memory = make_profile_number(0.0, "0", 1); (void) mk_expression_list(subn, list_append(expr, instruction(Op_field_spec))); } @@ -3944,7 +4339,7 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) r->sub_flags |= GENSUB; if (nexp == 3) { ip = instruction(Op_push_i); - ip->memory = make_number(0.0); + ip->memory = make_profile_number(0.0, "0", 1); (void) mk_expression_list(subn, list_append(list_create(ip), instruction(Op_field_spec))); } @@ -3955,7 +4350,7 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) } #ifdef HAVE_MPFR - /* N.B.: There isn't any special processing for an alternate function below */ + /* N.B.: If necessary, add special processing for alternate builtin, below */ if (do_mpfr && tokentab[idx].ptr2) r->builtin = tokentab[idx].ptr2; else @@ -3973,17 +4368,26 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) list = list_create(r); (void) list_prepend(list, instruction(Op_field_spec)); (void) list_prepend(list, instruction(Op_push_i)); - list->nexti->memory = make_number(0.0); + list->nexti->memory = make_profile_number(0.0, "0", 1); return list; } else { arg = subn->nexti; if (arg->nexti == arg->lasti && arg->nexti->opcode == Op_push) arg->nexti->opcode = Op_push_arg; /* argument may be array */ } - } else if (r->builtin == do_isarray) { + } else if (r->builtin == do_isarray || r->builtin == do_typeof) { arg = subn->nexti; if (arg->nexti == arg->lasti && arg->nexti->opcode == Op_push) - arg->nexti->opcode = Op_push_arg; /* argument may be array */ + arg->nexti->opcode = Op_push_arg_untyped; /* argument may be untyped */ + } else if (r->builtin == do_intdiv +#ifdef HAVE_MPFR + || r->builtin == MPF(intdiv) +#endif + ) { + arg = subn->nexti->lasti->nexti->lasti->nexti; /* 3rd arg list */ + ip = arg->lasti; + if (ip->opcode == Op_push) + ip->opcode = Op_push_array; } else if (r->builtin == do_match) { static bool warned = false; @@ -4060,7 +4464,7 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) } else if (do_intl /* --gen-po */ && r->builtin == do_dcgettext /* dcgettext(...) */ && subn->nexti->lasti->opcode == Op_push_i /* 1st arg is constant */ - && (subn->nexti->lasti->memory->flags & STRCUR) != 0) { /* it's a string constant */ + && (subn->nexti->lasti->memory->flags & STRING) != 0) { /* it's a string constant */ /* ala xgettext, dcgettext("some string" ...) dumps the string */ NODE *str = subn->nexti->lasti->memory; @@ -4072,9 +4476,9 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) } else if (do_intl /* --gen-po */ && r->builtin == do_dcngettext /* dcngettext(...) */ && subn->nexti->lasti->opcode == Op_push_i /* 1st arg is constant */ - && (subn->nexti->lasti->memory->flags & STRCUR) != 0 /* it's a string constant */ + && (subn->nexti->lasti->memory->flags & STRING) != 0 /* it's a string constant */ && subn->nexti->lasti->nexti->lasti->opcode == Op_push_i /* 2nd arg is constant too */ - && (subn->nexti->lasti->nexti->lasti->memory->flags & STRCUR) != 0) { /* it's a string constant */ + && (subn->nexti->lasti->nexti->lasti->memory->flags & STRING) != 0) { /* it's a string constant */ /* ala xgettext, dcngettext("some string", "some plural" ...) dumps the string */ NODE *str1 = subn->nexti->lasti->memory; NODE *str2 = subn->nexti->lasti->nexti->lasti->memory; @@ -4181,18 +4585,6 @@ valinfo(NODE *n, Func_print print_func, FILE *fp) else #endif print_func(fp, "%.17g\n", n->numbr); - } else if ((n->flags & STRCUR) != 0) { - pp_string_fp(print_func, fp, n->stptr, n->stlen, '"', false); - print_func(fp, "\n"); - } else if ((n->flags & NUMCUR) != 0) { -#ifdef HAVE_MPFR - if (is_mpg_float(n)) - print_func(fp, "%s\n", mpg_fmt("%.17R*g", ROUND_MODE, n->mpg_numbr)); - else if (is_mpg_integer(n)) - print_func(fp, "%s\n", mpg_fmt("%Zd", n->mpg_i)); - else -#endif - print_func(fp, "%.17g\n", n->numbr); } else print_func(fp, "?? flags %s\n", flags2str(n->flags)); } @@ -4281,6 +4673,14 @@ mk_function(INSTRUCTION *fi, INSTRUCTION *def) (t + 1)->tail_call = true; } + /* add any pre-function comment to start of action for profile.c */ + + if (function_comment != NULL) { + function_comment->source_line = 0; + (void) list_prepend(def, function_comment); + function_comment = NULL; + } + /* add an implicit return at end; * also used by 'return' command in debugger */ @@ -4555,6 +4955,7 @@ make_regnode(int type, NODE *exp) } n->re_exp = exp; n->re_flags = CONSTANT; + n->valref = 1; } return n; } @@ -4570,6 +4971,8 @@ mk_rexp(INSTRUCTION *list) ip = list->nexti; if (ip == list->lasti && ip->opcode == Op_match_rec) ip->opcode = Op_push_re; + else if (ip == list->lasti && ip->opcode == Op_push_re) + ; /* do nothing --- @/.../ */ else { ip = instruction(Op_push_re); ip->memory = make_regnode(Node_dynregex, NULL); @@ -4764,6 +5167,8 @@ mk_binary(INSTRUCTION *s1, INSTRUCTION *s2, INSTRUCTION *op) } op->opcode = Op_push_i; + // We don't need to call make_profile_number() here since + // optimizing is disabled when doing pretty printing. op->memory = make_number(res); unref(n1); unref(n2); @@ -4985,7 +5390,11 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action) (rp + 1)->lasti = action->lasti; (rp + 2)->first_line = pattern->source_line; (rp + 2)->last_line = lastline; - ip = list_prepend(action, rp); + if (block_comment != NULL) { + ip = list_prepend(list_prepend(action, block_comment), rp); + block_comment = NULL; + } else + ip = list_prepend(action, rp); } else { rp = bcalloc(Op_rule, 3, 0); @@ -5027,7 +5436,6 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action) action), tp); } - } list_append(rule_list, rp + 1); @@ -5597,6 +6005,26 @@ list_merge(INSTRUCTION *l1, INSTRUCTION *l2) return l1; } +/* add_pending_comment --- add a pending comment to a statement */ + +static inline INSTRUCTION * +add_pending_comment(INSTRUCTION *stmt) +{ + INSTRUCTION *ret = stmt; + + if (prior_comment != NULL) { + if (function_comment != prior_comment) + ret = list_append(stmt, prior_comment); + prior_comment = NULL; + } else if (comment != NULL && comment->memory->comment_type == EOL_COMMENT) { + if (function_comment != comment) + ret = list_append(stmt, comment); + comment = NULL; + } + + return ret; +} + /* See if name is a special token. */ int @@ -5802,3 +6230,20 @@ is_identchar(int c) { return (is_alnum(c) || c == '_'); } + +/* make_profile_number --- make a number that can be printed when profiling */ + +static NODE * +make_profile_number(double d, const char *str, size_t len) +{ + NODE *n = make_number(d); + if (do_pretty_print) { + // extra byte in case need to add minus sign in negate_num + n->stptr = estrdup(str, len + 1); + n->stptr[len] = '\0'; + n->stlen = len; + n->flags |= NUMCONSTSTR; + } + + return n; +} |