diff options
Diffstat (limited to 'awkgram.y')
| -rw-r--r-- | awkgram.y | 244 |
1 files changed, 217 insertions, 27 deletions
@@ -87,6 +87,7 @@ 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 bool want_source = false; static bool want_regexp = false; /* lexical scanning kludge */ @@ -146,6 +147,12 @@ static INSTRUCTION *ip_end; static INSTRUCTION *ip_endfile; static INSTRUCTION *ip_beginfile; +static INSTRUCTION *comment = NULL; +static INSTRUCTION *program_comment = NULL; +static INSTRUCTION *function_comment = NULL; + +static bool func_first = 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); @@ -286,9 +293,24 @@ 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; @@ -318,6 +340,8 @@ pattern | 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")); @@ -329,6 +353,8 @@ pattern | 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")); @@ -339,12 +365,14 @@ pattern } | LEX_BEGINFILE { + func_first = false; $1->in_rule = rule = BEGINFILE; $1->source_file = source; $$ = $1; } | LEX_ENDFILE { + func_first = false; $1->in_rule = rule = ENDFILE; $1->source_file = source; $$ = $1; @@ -354,10 +382,12 @@ 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; + $$ = ip; } ; @@ -384,6 +414,22 @@ lex_builtin function_prologue : LEX_FUNCTION func_name '(' opt_param_list r_paren opt_nls { + /* + * treat any comments between BOF and the first function + * definition (with no intervening BEGIN etc block) as + * program comments. Special kludge: iff there are more + * than one such comments, treat the last as a function + * comment. + */ + if (comment != NULL && func_first + && strstr(comment->memory->stptr, "\n\n") != NULL) + split_comment(); + /* save any other pre-function comment as function comment */ + if (comment != NULL) { + function_comment = comment; + comment = NULL; + } + func_first = false; $1->source_file = source; if (install_function($2->lextok, $1, $4) < 0) YYABORT; @@ -441,19 +487,39 @@ a_slash statements : /* empty */ - { $$ = NULL; } + { + if (comment != NULL) { + $$ = list_create(comment); + comment = NULL; + } else $$ = NULL; + } | statements statement { - if ($2 == NULL) - $$ = $1; - else { + if ($2 == NULL) { + if (comment == NULL) + $$ = $1; + else { + $$ = list_append($1, comment); + comment = NULL; + } + } else { add_lint($2, LINT_no_effect); - if ($1 == NULL) - $$ = $2; - else + if ($1 == NULL) { + if (comment == NULL) + $$ = $2; + else { + $$ = 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; } @@ -497,7 +563,7 @@ statement } /* else curr = NULL; */ - for(; curr != NULL; curr = nextc) { + for (; curr != NULL; curr = nextc) { INSTRUCTION *caseexp = curr->case_exp; INSTRUCTION *casestmt = curr->case_stmt; @@ -1185,7 +1251,7 @@ opt_param_list : /* empty */ { $$ = NULL; } | param_list - { $$ = $1 ; } + { $$ = $1; } ; param_list @@ -1859,6 +1925,7 @@ static const struct token tokentab[] = { {"dcngettext", Op_builtin, LEX_BUILTIN, GAWKX|A(1)|A(2)|A(3)|A(4)|A(5), do_dcngettext, 0}, {"default", Op_K_default, LEX_DEFAULT, GAWKX, 0, 0}, {"delete", Op_K_delete, LEX_DELETE, NOT_OLD, 0, 0}, +{"div", Op_builtin, LEX_BUILTIN, GAWKX|A(3), do_div, MPF(div)}, {"do", Op_K_do, LEX_DO, NOT_OLD|BREAK|CONTINUE, 0, 0}, {"else", Op_K_else, LEX_ELSE, 0, 0, 0}, {"eval", Op_symbol, LEX_EVAL, 0, 0, 0}, @@ -2220,6 +2287,13 @@ mk_program() cp = end_block; else cp = list_merge(begin_block, end_block); + /* + * We don't need to clear the comment variables + * since they're not used anymore after this + * function is called. + */ + if (comment != NULL) + (void) list_append(cp, comment); (void) list_append(cp, ip_atexit); (void) list_append(cp, instruction(Op_stop)); @@ -2252,6 +2326,12 @@ 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)); @@ -2753,7 +2833,7 @@ 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); @@ -2846,7 +2926,7 @@ 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); @@ -2925,6 +3005,79 @@ pushback(void) } +/* get_comment --- collect comment text */ + +int +get_comment(void) +{ + int c; + int sl; + tok = tokstart; + tokadd('#'); + sl = sourceline; + + while (true) { + while ((c = nextc(false)) != '\n' && c != END_FILE) { + tokadd(c); + } + 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(); + break; + } else + tokadd(c); + } else + break; + } + comment = bcalloc(Op_comment, 1, sl); + comment->source_file = source; + comment->memory = make_str_node(tokstart, tok - tokstart, 0); + + 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->memory->stptr; + l = comment->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; + 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->source_file; + p[l + 2] = 0; + program_comment->memory = make_str_node(p, l + 2, 0); + comment = NULL; + freenode(n); + break; + } + else + l--; + } +} + /* allow_newline --- allow newline after &&, ||, ? and : */ static void @@ -2939,8 +3092,13 @@ 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(); + } else { + while ((c = nextc(false)) != '\n' && c != END_FILE) + continue; + } if (c == END_FILE) { pushback(); break; @@ -2963,7 +3121,8 @@ 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) { @@ -3144,9 +3303,20 @@ 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. + */ + c = get_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; @@ -3157,7 +3327,7 @@ 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. @@ -3174,9 +3344,13 @@ 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(); + else { + while ((c = nextc(false)) != '\n') + if (c == END_FILE) + break; + } } pushback(); } @@ -3388,7 +3562,7 @@ retry: lastline = sourceline; return lasttok = c; } - did_newline++; + did_newline = true; --lexptr; /* pick up } next time */ return lasttok = NEWLINE; @@ -3908,7 +4082,7 @@ 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 @@ -3937,6 +4111,15 @@ snode(INSTRUCTION *subn, INSTRUCTION *r) 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_div +#ifdef HAVE_MPFR + || r->builtin == MPF(div) +#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; @@ -4232,6 +4415,14 @@ 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 */ @@ -5051,7 +5242,6 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action) action), tp); } - } list_append(rule_list, rp + 1); |