diff options
Diffstat (limited to 'awkgram.y')
| -rw-r--r-- | awkgram.y | 340 |
1 files changed, 307 insertions, 33 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); @@ -154,8 +161,6 @@ static inline INSTRUCTION *list_merge(INSTRUCTION *l1, INSTRUCTION *l2); extern double fmod(double x, double y); #define YYSTYPE INSTRUCTION * - -#define is_identchar(c) (isalnum(c) || (c) == '_') %} %token FUNC_CALL NAME REGEXP FILENAME @@ -288,9 +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; @@ -320,6 +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")); @@ -331,6 +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")); @@ -341,12 +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; @@ -356,10 +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; + $$ = ip; } ; @@ -386,6 +414,22 @@ lex_builtin function_prologue : LEX_FUNCTION func_name '(' opt_param_list r_paren opt_nls { + /* + * treat any comments between BOF and the first function + * definition (with no intervening BEGIN etc block) as + * program comments. Special kludge: iff there are more + * than one such comments, treat the last as a function + * comment. + */ + if (comment != NULL && func_first + && strstr(comment->memory->stptr, "\n\n") != NULL) + split_comment(); + /* save any other pre-function comment as function comment */ + if (comment != NULL) { + function_comment = comment; + comment = NULL; + } + func_first = false; $1->source_file = source; if (install_function($2->lextok, $1, $4) < 0) YYABORT; @@ -443,19 +487,39 @@ a_slash statements : /* empty */ - { $$ = 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; } @@ -499,7 +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; @@ -1187,7 +1251,7 @@ opt_param_list : /* empty */ { $$ = NULL; } | param_list - { $$ = $1 ; } + { $$ = $1; } ; param_list @@ -1805,6 +1869,7 @@ struct token { # define GAWKX 0x0400 /* gawk extension */ # define BREAK 0x0800 /* break allowed inside */ # define CONTINUE 0x1000 /* continue allowed inside */ +# define DEBUG_USE 0x2000 /* for use by developers */ NODE *(*ptr)(int); /* function that implements this keyword */ NODE *(*ptr2)(int); /* alternate arbitrary-precision function */ @@ -1843,7 +1908,7 @@ static const struct token tokentab[] = { {"END", Op_rule, LEX_END, 0, 0, 0}, {"ENDFILE", Op_rule, LEX_ENDFILE, GAWKX, 0, 0}, #ifdef ARRAYDEBUG -{"adump", Op_builtin, LEX_BUILTIN, GAWKX|A(1)|A(2), do_adump, 0}, +{"adump", Op_builtin, LEX_BUILTIN, GAWKX|A(1)|A(2)|DEBUG_USE, do_adump, 0}, #endif {"and", Op_builtin, LEX_BUILTIN, GAWKX, do_and, MPF(and)}, {"asort", Op_builtin, LEX_BUILTIN, GAWKX|A(1)|A(2)|A(3), do_asort, 0}, @@ -1903,7 +1968,7 @@ static const struct token tokentab[] = { {"sqrt", Op_builtin, LEX_BUILTIN, A(1), do_sqrt, MPF(sqrt)}, {"srand", Op_builtin, LEX_BUILTIN, NOT_OLD|A(0)|A(1), do_srand, MPF(srand)}, #if defined(GAWKDEBUG) || defined(ARRAYDEBUG) /* || ... */ -{"stopme", Op_builtin, LEX_BUILTIN, GAWKX|A(0), stopme, 0}, +{"stopme", Op_builtin, LEX_BUILTIN, GAWKX|A(0)|DEBUG_USE, stopme, 0}, #endif {"strftime", Op_builtin, LEX_BUILTIN, GAWKX|A(0)|A(1)|A(2)|A(3), do_strftime, 0}, {"strtonum", Op_builtin, LEX_BUILTIN, GAWKX|A(1), do_strtonum, MPF(strtonum)}, @@ -2222,6 +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)); @@ -2254,6 +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)); @@ -2755,7 +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); @@ -2817,7 +2895,7 @@ check_bad_char(int c) } if (iscntrl(c) && ! isspace(c)) - fatal(_("PEBKAC error: invalid character '\\%03o' in source code"), c); + fatal(_("PEBKAC error: invalid character '\\%03o' in source code"), c & 0xFF); } /* nextc --- get the next input character */ @@ -2848,7 +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); @@ -2927,6 +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 @@ -2941,8 +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; @@ -2965,7 +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) { @@ -3146,9 +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; @@ -3159,7 +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. @@ -3176,9 +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(); } @@ -3390,14 +3562,18 @@ retry: lastline = sourceline; return lasttok = c; } - did_newline++; + did_newline = true; --lexptr; /* pick up } next time */ return lasttok = NEWLINE; case '"': string: esc_seen = false; - while ((c = nextc(true)) != '"') { + /* + * Allow any kind of junk in quoted string, + * so pass false to nextc(). + */ + while ((c = nextc(false)) != '"') { if (c == '\n') { pushback(); yyerror(_("unterminated string")); @@ -3636,7 +3812,7 @@ retry: } } - if (c != '_' && ! isalpha(c)) { + if (c != '_' && ! is_alpha(c)) { yyerror(_("invalid char '%c' in expression"), c); return lasttok = LEX_EOF; } @@ -4239,6 +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 */ @@ -5058,7 +5242,6 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action) action), tp); } - } list_append(rule_list, rp + 1); @@ -5715,3 +5898,94 @@ one_line_close(int fd) } +/* lookup_builtin --- find a builtin function or return NULL */ + +builtin_func_t +lookup_builtin(const char *name) +{ + int mid = check_special(name); + + if (mid == -1 || tokentab[mid].class != LEX_BUILTIN) + return NULL; +#ifdef HAVE_MPFR + if (do_mpfr) + return tokentab[mid].ptr2; +#endif + + return tokentab[mid].ptr; +} + +/* install_builtins --- add built-in functions to FUNCTAB */ + +void +install_builtins(void) +{ + int i, j; + + j = sizeof(tokentab) / sizeof(tokentab[0]); + for (i = 0; i < j; i++) { + if ( tokentab[i].class == LEX_BUILTIN + && (tokentab[i].flags & DEBUG_USE) == 0) { + (void) install_symbol(tokentab[i].operator, Node_builtin_func); + } + } +} + +/* + * 9/2014: Gawk cannot use <ctype.h> isalpha or isalnum when + * parsing the program since that can let through non-English + * letters. So, we supply our own. !@#$%^&*()-ing locales! + */ + +/* is_alpha --- return true if c is an English letter */ + +/* + * The scene of the murder was grisly to look upon. When the inspector + * arrived, the sergeant turned to him and said, "Another programmer stabbed + * in the back. He never knew what happened." + * + * The inspector replied, "Looks like the MO of isalpha, and his even meaner + * big brother, isalnum. The Locale brothers." The sergeant merely + * shuddered in horror. + */ + +bool +is_alpha(int c) +{ +#ifdef I_DONT_KNOW_WHAT_IM_DOING + return isalpha(c); +#else /* ! I_DONT_KNOW_WHAT_IM_DOING */ + switch (c) { + case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': + case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': + case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': + case 's': case 't': case 'u': case 'v': case 'w': case 'x': + case 'y': case 'z': + case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': + case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': + case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': + case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': + case 'Y': case 'Z': + return true; + } + return false; +#endif /* ! I_DONT_KNOW_WHAT_IM_DOING */ +} + +/* is_alnum --- return true for alphanumeric, English only letters */ + +bool +is_alnum(int c) +{ + /* digit test is good for EBCDIC too. so there. */ + return (is_alpha(c) || ('0' <= c && c <= '9')); +} + + +/* is_identchar --- return true if c can be in an identifier */ + +bool +is_identchar(int c) +{ + return (is_alnum(c) || c == '_'); +} |