emo/busybox
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

awk: tighten parsing - disallow extra semicolons

Browse Source 
'; BEGIN {...}' and 'BEGIN {...} ;; {...}' are not accepted by gawk

function                                             old     new   delta
parse_program                                        332     353     +21

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
This commit is contained in:
Denys Vlasenko
2021-07-14 14:25:07 +02:00
parent e6f4145f29
commit d62627487a
1 changed files with 24 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

40
editors/awk.c
View File

@@ -1634,7 +1634,7 @@ static void chain_group(void)
debug_printf_parse("%s: ST_FOR\n", __func__); debug_printf_parse("%s: ST_FOR\n", __func__);
next_token(TC_LPAREN); next_token(TC_LPAREN);
n2 = parse_expr(TC_SEMICOL | TC_RPAREN); n2 = parse_expr(TC_SEMICOL | TC_RPAREN);
if (t_tclass & TC_RPAREN) { /* for-in */ if (t_tclass & TC_RPAREN) { /* for (I in ARRAY) */
if (!n2 || n2->info != TI_IN) if (!n2 || n2->info != TI_IN)
syntax_error(EMSG_UNEXP_TOKEN); syntax_error(EMSG_UNEXP_TOKEN);
n = chain_node(OC_WALKINIT | VV); n = chain_node(OC_WALKINIT | VV);
@@ -1700,20 +1700,15 @@ static void parse_program(char *p)
for (;;) { for (;;) {
uint32_t tclass; uint32_t tclass;
tclass = next_token(TC_EOF | TS_OPSEQ | TC_LBRACE | tclass = next_token(TS_OPSEQ | TC_LBRACE | TC_BEGIN | TC_END | TC_FUNCDECL
TC_SEMICOL | TC_NEWLINE | TC_BEGIN | TC_END | TC_FUNCDECL); | TC_EOF | TC_NEWLINE /* but not TC_SEMICOL */);
got_tok:
if (tclass == TC_EOF) { if (tclass == TC_EOF) {
debug_printf_parse("%s: TC_EOF\n", __func__); debug_printf_parse("%s: TC_EOF\n", __func__);
break; break;
} }
if (tclass & (TC_SEMICOL | TC_NEWLINE)) { if (tclass == TC_NEWLINE) {
debug_printf_parse("%s: TC_SEMICOL | TC_NEWLINE\n", __func__); debug_printf_parse("%s: TC_NEWLINE\n", __func__);
//NB: gawk allows many newlines, but does not allow more than one semicolon:
// BEGIN {...}<newline>;<newline>;
//would complain "each rule must have a pattern or an action part".
//Same message for
// ; BEGIN {...}
continue; continue;
} }
if (tclass == TC_BEGIN) { if (tclass == TC_BEGIN) {
@@ -1722,7 +1717,7 @@ static void parse_program(char *p)
/* ensure there is no newline between BEGIN and { */ /* ensure there is no newline between BEGIN and { */
next_token(TC_LBRACE); next_token(TC_LBRACE);
chain_until_rbrace(); chain_until_rbrace();
continue; goto next_tok;
} }
if (tclass == TC_END) { if (tclass == TC_END) {
debug_printf_parse("%s: TC_END\n", __func__); debug_printf_parse("%s: TC_END\n", __func__);
@@ -1730,7 +1725,7 @@ static void parse_program(char *p)
/* ensure there is no newline between END and { */ /* ensure there is no newline between END and { */
next_token(TC_LBRACE); next_token(TC_LBRACE);
chain_until_rbrace(); chain_until_rbrace();
continue; goto next_tok;
} }
if (tclass == TC_FUNCDECL) { if (tclass == TC_FUNCDECL) {
func *f; func *f;
@@ -1765,7 +1760,7 @@ static void parse_program(char *p)
continue; continue;
chain_until_rbrace(); chain_until_rbrace();
hash_clear(ahash); hash_clear(ahash);
continue; goto next_tok;
} }
seq = &mainseq; seq = &mainseq;
if (tclass & TS_OPSEQ) { if (tclass & TS_OPSEQ) {
@@ -1784,12 +1779,25 @@ static void parse_program(char *p)
chain_node(OC_PRINT); chain_node(OC_PRINT);
} }
cn->r.n = mainseq.last; cn->r.n = mainseq.last;
continue; goto next_tok;
} }
/* tclass == TC_LBRACE */ /* tclass == TC_LBRACE */
debug_printf_parse("%s: TC_LBRACE(?)\n", __func__); debug_printf_parse("%s: TC_LBRACE(?)\n", __func__);
chain_until_rbrace(); chain_until_rbrace();
} next_tok:
/* Same as next_token() at the top of the loop, + TC_SEMICOL */
tclass = next_token(TS_OPSEQ | TC_LBRACE | TC_BEGIN | TC_END | TC_FUNCDECL
| TC_EOF | TC_NEWLINE | TC_SEMICOL);
/* gawk allows many newlines, but does not allow more than one semicolon:
* BEGIN {...}<newline>;<newline>;
* would complain "each rule must have a pattern or an action part".
* Same message for
* ; BEGIN {...}
*/
if (tclass != TC_SEMICOL)
goto got_tok; /* use this token */
/* else: loop back - ate the semicolon, get and use _next_ token */
} /* for (;;) */
} }