aboutsummaryrefslogtreecommitdiffstats
path: root/awkgram.c
diff options
context:
space:
mode:
Diffstat (limited to 'awkgram.c')
-rw-r--r--awkgram.c130
1 files changed, 88 insertions, 42 deletions
diff --git a/awkgram.c b/awkgram.c
index b36816c9..28776494 100644
--- a/awkgram.c
+++ b/awkgram.c
@@ -4199,6 +4199,7 @@ 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},
@@ -5130,12 +5131,40 @@ tokexpand()
return tok;
}
+/* check_bad_char --- fatal if c isn't allowed in gawk source code */
+
+/*
+ * The error message was inspired by someone who decided to put
+ * a physical \0 byte into the source code to see what would
+ * happen and then filed a bug report about it. Sigh.
+ */
+
+static void
+check_bad_char(int c)
+{
+ /* allow escapes. needed for autoconf. bleah. */
+ switch (c) {
+ case '\a':
+ case '\b':
+ case '\f':
+ case '\n':
+ case '\r':
+ case '\t':
+ return;
+ default:
+ break;
+ }
+
+ if (iscntrl(c) && ! isspace(c))
+ fatal(_("PEBKAC error: invalid character '\\%03o' in source code"), c);
+}
+
/* nextc --- get the next input character */
#if MBS_SUPPORT
static int
-nextc(void)
+nextc(bool check_for_bad)
{
if (gawk_mb_cur_max > 1) {
again:
@@ -5186,14 +5215,19 @@ again:
0 : work_ring_idx + 1;
cur_char_ring[work_ring_idx] = 0;
}
+ if (check_for_bad)
+ check_bad_char(*lexptr);
return (int) (unsigned char) *lexptr++;
} else {
do {
if (lexeof)
return END_FILE;
- if (lexptr && lexptr < lexend)
- return ((int) (unsigned char) *lexptr++);
+ if (lexptr && lexptr < lexend) {
+ if (check_for_bad)
+ check_bad_char(*lexptr);
+ return ((int) (unsigned char) *lexptr++);
+ }
} while (get_src_buf());
return END_SRC;
}
@@ -5202,13 +5236,16 @@ again:
#else /* MBS_SUPPORT */
int
-nextc()
+nextc(bool check_for_bad)
{
do {
if (lexeof)
return END_FILE;
- if (lexptr && lexptr < lexend)
+ if (lexptr && lexptr < lexend) {
+ if (check_for_bad)
+ check_bad_char(*lexptr);
return ((int) (unsigned char) *lexptr++);
+ }
} while (get_src_buf());
return END_SRC;
}
@@ -5237,13 +5274,13 @@ allow_newline(void)
int c;
for (;;) {
- c = nextc();
+ c = nextc(true);
if (c == END_FILE) {
pushback();
break;
}
if (c == '#') {
- while ((c = nextc()) != '\n' && c != END_FILE)
+ while ((c = nextc(false)) != '\n' && c != END_FILE)
continue;
if (c == END_FILE) {
pushback();
@@ -5315,7 +5352,7 @@ yylex(void)
if (lasttok == LEX_EOF) /* error earlier in current source, must give up !! */
return 0;
- c = nextc();
+ c = nextc(true);
if (c == END_SRC)
return 0;
if (c == END_FILE)
@@ -5357,12 +5394,12 @@ yylex(void)
want_regexp = false;
tok = tokstart;
for (;;) {
- c = nextc();
+ c = nextc(true);
if (gawk_mb_cur_max == 1 || nextc_is_1stbyte) switch (c) {
case '[':
/* one day check for `.' and `=' too */
- if (nextc() == ':' || in_brack == 0)
+ if (nextc(true) == ':' || in_brack == 0)
in_brack++;
pushback();
break;
@@ -5376,7 +5413,7 @@ yylex(void)
in_brack--;
break;
case '\\':
- if ((c = nextc()) == END_FILE) {
+ if ((c = nextc(true)) == END_FILE) {
pushback();
yyerror(_("unterminated regexp ends with `\\' at end of file"));
goto end_regexp; /* kludge */
@@ -5396,7 +5433,7 @@ end_regexp:
yylval = GET_INSTRUCTION(Op_token);
yylval->lextok = estrdup(tokstart, tok - tokstart);
if (do_lint) {
- int peek = nextc();
+ int peek = nextc(true);
pushback();
if (peek == 'i' || peek == 's') {
@@ -5426,7 +5463,7 @@ end_regexp:
retry:
/* skipping \r is a hack, but windows is just too pervasive. sigh. */
- while ((c = nextc()) == ' ' || c == '\t' || c == '\r')
+ while ((c = nextc(true)) == ' ' || c == '\t' || c == '\r')
continue;
lexeme = lexptr ? lexptr - 1 : lexptr;
@@ -5448,7 +5485,7 @@ retry:
return lasttok = NEWLINE;
case '#': /* it's a comment */
- while ((c = nextc()) != '\n') {
+ while ((c = nextc(false)) != '\n') {
if (c == END_FILE)
return lasttok = NEWLINE_EOF;
}
@@ -5468,7 +5505,7 @@ retry:
*/
if (! do_traditional) {
/* strip trailing white-space and/or comment */
- while ((c = nextc()) == ' ' || c == '\t' || c == '\r')
+ while ((c = nextc(true)) == ' ' || c == '\t' || c == '\r')
continue;
if (c == '#') {
static bool warned = false;
@@ -5478,16 +5515,16 @@ retry:
lintwarn(
_("use of `\\ #...' line continuation is not portable"));
}
- while ((c = nextc()) != '\n')
+ while ((c = nextc(false)) != '\n')
if (c == END_FILE)
break;
}
pushback();
}
#endif /* RELAXED_CONTINUATION */
- c = nextc();
+ c = nextc(true);
if (c == '\r') /* allow MS-DOS files. bleah */
- c = nextc();
+ c = nextc(true);
if (c == '\n') {
sourceline++;
goto retry;
@@ -5526,7 +5563,7 @@ retry:
case '[':
return lasttok = c;
case ']':
- c = nextc();
+ c = nextc(true);
pushback();
if (c == '[') {
yylval = GET_INSTRUCTION(Op_sub_array);
@@ -5538,7 +5575,7 @@ retry:
return ']';
case '*':
- if ((c = nextc()) == '=') {
+ if ((c = nextc(true)) == '=') {
yylval = GET_INSTRUCTION(Op_assign_times);
return lasttok = ASSIGNOP;
} else if (do_posix) {
@@ -5549,7 +5586,7 @@ retry:
/* make ** and **= aliases for ^ and ^= */
static bool did_warn_op = false, did_warn_assgn = false;
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
if (! did_warn_assgn) {
did_warn_assgn = true;
if (do_lint)
@@ -5577,7 +5614,7 @@ retry:
return lasttok = '*';
case '/':
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
pushback();
return lasttok = SLASH_BEFORE_EQUAL;
}
@@ -5586,7 +5623,7 @@ retry:
return lasttok = '/';
case '%':
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
yylval = GET_INSTRUCTION(Op_assign_mod);
return lasttok = ASSIGNOP;
}
@@ -5598,7 +5635,7 @@ retry:
{
static bool did_warn_op = false, did_warn_assgn = false;
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
if (do_lint_old && ! did_warn_assgn) {
did_warn_assgn = true;
warning(_("operator `^=' is not supported in old awk"));
@@ -5616,7 +5653,7 @@ retry:
}
case '+':
- if ((c = nextc()) == '=') {
+ if ((c = nextc(true)) == '=') {
yylval = GET_INSTRUCTION(Op_assign_plus);
return lasttok = ASSIGNOP;
}
@@ -5629,7 +5666,7 @@ retry:
return lasttok = '+';
case '!':
- if ((c = nextc()) == '=') {
+ if ((c = nextc(true)) == '=') {
yylval = GET_INSTRUCTION(Op_notequal);
return lasttok = RELOP;
}
@@ -5642,7 +5679,7 @@ retry:
return lasttok = '!';
case '<':
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
yylval = GET_INSTRUCTION(Op_leq);
return lasttok = RELOP;
}
@@ -5651,7 +5688,7 @@ retry:
return lasttok = '<';
case '=':
- if (nextc() == '=') {
+ if (nextc(true) == '=') {
yylval = GET_INSTRUCTION(Op_equal);
return lasttok = RELOP;
}
@@ -5660,7 +5697,7 @@ retry:
return lasttok = ASSIGN;
case '>':
- if ((c = nextc()) == '=') {
+ if ((c = nextc(true)) == '=') {
yylval = GET_INSTRUCTION(Op_geq);
return lasttok = RELOP;
} else if (c == '>') {
@@ -5699,7 +5736,7 @@ retry:
case '"':
string:
esc_seen = false;
- while ((c = nextc()) != '"') {
+ while ((c = nextc(true)) != '"') {
if (c == '\n') {
pushback();
yyerror(_("unterminated string"));
@@ -5707,7 +5744,7 @@ retry:
}
if ((gawk_mb_cur_max == 1 || nextc_is_1stbyte) &&
c == '\\') {
- c = nextc();
+ c = nextc(true);
if (c == '\n') {
sourceline++;
continue;
@@ -5741,7 +5778,7 @@ retry:
return lasttok = YSTRING;
case '-':
- if ((c = nextc()) == '=') {
+ if ((c = nextc(true)) == '=') {
yylval = GET_INSTRUCTION(Op_assign_minus);
return lasttok = ASSIGNOP;
}
@@ -5754,7 +5791,7 @@ retry:
return lasttok = '-';
case '.':
- c = nextc();
+ c = nextc(true);
pushback();
if (! isdigit(c))
return lasttok = '.';
@@ -5782,7 +5819,7 @@ retry:
if (do_traditional)
goto done;
if (tok == tokstart + 2) {
- int peek = nextc();
+ int peek = nextc(true);
if (isxdigit(peek)) {
inhex = true;
@@ -5810,8 +5847,8 @@ retry:
break;
}
seen_e = true;
- if ((c = nextc()) == '-' || c == '+') {
- int c2 = nextc();
+ if ((c = nextc(true)) == '-' || c == '+') {
+ int c2 = nextc(true);
if (isdigit(c2)) {
tokadd(c);
@@ -5858,7 +5895,7 @@ retry:
}
if (gotnumber)
break;
- c = nextc();
+ c = nextc(true);
}
pushback();
@@ -5907,7 +5944,7 @@ retry:
return lasttok = YNUMBER;
case '&':
- if ((c = nextc()) == '&') {
+ if ((c = nextc(true)) == '&') {
yylval = GET_INSTRUCTION(Op_and);
allow_newline();
return lasttok = LEX_AND;
@@ -5917,7 +5954,7 @@ retry:
return lasttok = '&';
case '|':
- if ((c = nextc()) == '|') {
+ if ((c = nextc(true)) == '|') {
yylval = GET_INSTRUCTION(Op_or);
allow_newline();
return lasttok = LEX_OR;
@@ -5958,7 +5995,7 @@ retry:
* occasions where the interactions are funny.
*/
if (! do_traditional && c == '_' && lasttok != '$') {
- if ((c = nextc()) == '"') {
+ if ((c = nextc(true)) == '"') {
intlstr = true;
goto string;
}
@@ -5970,7 +6007,7 @@ retry:
tok = tokstart;
while (c != END_FILE && is_identchar(c)) {
tokadd(c);
- c = nextc();
+ c = nextc(true);
}
tokadd('\0');
pushback();
@@ -6208,7 +6245,7 @@ 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
@@ -6237,6 +6274,15 @@ 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;
generated by cgit v1.2.3 (git 2.25.1) at 2025-05-08 16:40:23 +0000