summaryrefslogtreecommitdiffhomepage
path: root/mrbgems/mruby-compiler/core/parse.y
diff options
context:
space:
mode:
Diffstat (limited to 'mrbgems/mruby-compiler/core/parse.y')
-rw-r--r--mrbgems/mruby-compiler/core/parse.y276
1 files changed, 221 insertions, 55 deletions
diff --git a/mrbgems/mruby-compiler/core/parse.y b/mrbgems/mruby-compiler/core/parse.y
index d143344c3..b7d47118f 100644
--- a/mrbgems/mruby-compiler/core/parse.y
+++ b/mrbgems/mruby-compiler/core/parse.y
@@ -220,6 +220,26 @@ parser_strdup(parser_state *p, const char *s)
#undef strdup
#define strdup(s) parser_strdup(p, s)
+static void
+dump_int(short i, char *s)
+{
+ char *p = s;
+ char *t = s;
+
+ while (i > 0) {
+ *p++ = (i % 10)+'0';
+ i /= 10;
+ }
+ if (p == s) *p++ = '0';
+ *p = 0;
+ p--; /* point the last char */
+ while (t < p) {
+ char c = *t;
+ *t++ = *p;
+ *p-- = c;
+ }
+}
+
/* xxx ----------------------------- */
static node*
@@ -283,6 +303,20 @@ local_add(parser_state *p, mrb_sym sym)
}
}
+static void
+local_add_blk(parser_state *p, mrb_sym blk)
+{
+ /* allocate register for block */
+ local_add_f(p, blk ? blk : mrb_intern_lit(p->mrb, "&"));
+}
+
+static void
+local_add_kw(parser_state *p, mrb_sym kwd)
+{
+ /* allocate register for keywords hash */
+ local_add_f(p, kwd ? kwd : mrb_intern_lit(p->mrb, "**"));
+}
+
static node*
locals_node(parser_state *p)
{
@@ -726,13 +760,11 @@ new_args_tail(parser_state *p, node *kws, node *kwrest, mrb_sym blk)
{
node *k;
- /* allocate register for keywords hash */
if (kws || kwrest) {
- local_add_f(p, (kwrest && kwrest->cdr)? sym(kwrest->cdr) : mrb_intern_lit(p->mrb, "**"));
+ local_add_kw(p, (kwrest && kwrest->cdr)? sym(kwrest->cdr) : 0);
}
- /* allocate register for block */
- local_add_f(p, blk? blk : mrb_intern_lit(p->mrb, "&"));
+ local_add_blk(p, blk);
// allocate register for keywords arguments
// order is for Proc#parameters
@@ -795,6 +827,13 @@ new_masgn(parser_state *p, node *a, node *b)
return cons((node*)NODE_MASGN, cons(a, b));
}
+/* (:masgn mlhs mrhs) no check */
+static node*
+new_masgn_param(parser_state *p, node *a, node *b)
+{
+ return cons((node*)NODE_MASGN, cons(a, b));
+}
+
/* (:asgn lhs rhs) */
static node*
new_op_asgn(parser_state *p, node *a, mrb_sym op, node *b)
@@ -833,6 +872,86 @@ new_dstr(parser_state *p, node *a)
return cons((node*)NODE_DSTR, a);
}
+static int
+string_node_p(node *n)
+{
+ return (int)((enum node_type)(intptr_t)n->car == NODE_STR);
+}
+
+static node*
+composite_string_node(parser_state *p, node *a, node *b)
+{
+ size_t newlen = (size_t)a->cdr + (size_t)b->cdr;
+ char *str = (char*)mrb_pool_realloc(p->pool, a->car, (size_t)a->cdr + 1, newlen + 1);
+ memcpy(str + (size_t)a->cdr, b->car, (size_t)b->cdr);
+ str[newlen] = '\0';
+ a->car = (node*)str;
+ a->cdr = (node*)newlen;
+ cons_free(b);
+ return a;
+}
+
+static node*
+concat_string(parser_state *p, node *a, node *b)
+{
+ if (string_node_p(a)) {
+ if (string_node_p(b)) {
+ /* a == NODE_STR && b == NODE_STR */
+ composite_string_node(p, a->cdr, b->cdr);
+ cons_free(b);
+ return a;
+ }
+ else {
+ /* a == NODE_STR && b == NODE_DSTR */
+
+ if (string_node_p(b->cdr->car)) {
+ /* a == NODE_STR && b->[NODE_STR, ...] */
+ composite_string_node(p, a->cdr, b->cdr->car->cdr);
+ cons_free(b->cdr->car);
+ b->cdr->car = a;
+ return b;
+ }
+ }
+ }
+ else {
+ node *c; /* last node of a */
+ for (c = a; c->cdr != NULL; c = c->cdr) ;
+
+ if (string_node_p(b)) {
+ /* a == NODE_DSTR && b == NODE_STR */
+ if (string_node_p(c->car)) {
+ /* a->[..., NODE_STR] && b == NODE_STR */
+ composite_string_node(p, c->car->cdr, b->cdr);
+ cons_free(b);
+ return a;
+ }
+
+ push(a, b);
+ return a;
+ }
+ else {
+ /* a == NODE_DSTR && b == NODE_DSTR */
+ if (string_node_p(c->car) && string_node_p(b->cdr->car)) {
+ /* a->[..., NODE_STR] && b->[NODE_STR, ...] */
+ node *d = b->cdr;
+ cons_free(b);
+ composite_string_node(p, c->car->cdr, d->car->cdr);
+ cons_free(d->car);
+ c->cdr = d->cdr;
+ cons_free(d);
+ return a;
+ }
+ else {
+ c->cdr = b->cdr;
+ cons_free(b);
+ return a;
+ }
+ }
+ }
+
+ return new_dstr(p, list2(a, b));
+}
+
/* (:str . (s . len)) */
static node*
new_xstr(parser_state *p, const char *s, int len)
@@ -1181,7 +1300,7 @@ heredoc_end(parser_state *p)
%token <nd> tNTH_REF tBACK_REF
%token <num> tREGEXP_END
-%type <nd> singleton string string_rep string_interp xstring regexp
+%type <nd> singleton string string_fragment string_rep string_interp xstring regexp
%type <nd> literal numeric cpath symbol
%type <nd> top_compstmt top_stmts top_stmt
%type <nd> bodystmt compstmt stmts stmt expr arg primary command command_call method_call
@@ -2560,9 +2679,9 @@ block_param : f_arg ',' f_block_optarg ',' f_rest_arg opt_block_args_tail
{
$$ = new_args(p, $1, 0, $3, 0, $4);
}
- | f_arg ','
+ | f_arg ',' opt_block_args_tail
{
- $$ = new_args(p, $1, 0, 0, 0, 0);
+ $$ = new_args(p, $1, 0, 0, 0, $3);
}
| f_arg ',' f_rest_arg ',' f_arg opt_block_args_tail
{
@@ -2603,19 +2722,24 @@ block_param : f_arg ',' f_block_optarg ',' f_rest_arg opt_block_args_tail
;
opt_block_param : none
- | block_param_def
{
+ local_add_blk(p, 0);
+ $$ = 0;
+ }
+ | block_param_def
+ {
p->cmd_start = TRUE;
$$ = $1;
}
;
-block_param_def : '|' opt_bv_decl '|'
+block_param_def : '|' {local_add_blk(p, 0);} opt_bv_decl '|'
{
$$ = 0;
}
| tOROP
{
+ local_add_blk(p, 0);
$$ = 0;
}
| '|' block_param opt_bv_decl '|'
@@ -2828,7 +2952,14 @@ literal : numeric
| symbols
;
-string : tCHAR
+string : string_fragment
+ | string string_fragment
+ {
+ $$ = concat_string(p, $1, $2);
+ }
+ ;
+
+string_fragment : tCHAR
| tSTRING
| tSTRING_BEG tSTRING
{
@@ -3050,7 +3181,7 @@ var_ref : variable
}
| keyword__FILE__
{
- const char *fn = p->filename;
+ const char *fn = mrb_sym2name_len(p->mrb, p->filename_sym, NULL);
if (!fn) {
fn = "(null)";
}
@@ -3060,7 +3191,7 @@ var_ref : variable
{
char buf[16];
- snprintf(buf, sizeof(buf), "%d", p->lineno);
+ dump_int(p->lineno, buf);
$$ = new_int(p, buf, 10);
}
| keyword__ENCODING__
@@ -3302,7 +3433,7 @@ f_arg_item : f_norm_arg
}
f_margs rparen
{
- $$ = new_masgn(p, $3, p->locals->car);
+ $$ = new_masgn_param(p, $3, p->locals->car);
local_resume(p, $<nd>2);
local_add_f(p, 0);
}
@@ -3454,7 +3585,7 @@ assoc : arg tASSOC arg
void_expr_error(p, $3);
$$ = cons(new_sym(p, $1), $3);
}
- | string tLABEL_TAG arg
+ | string_fragment tLABEL_TAG arg
{
void_expr_error(p, $3);
if ($1->car == (node*)NODE_DSTR) {
@@ -3559,8 +3690,9 @@ yyerror(parser_state *p, const char *s)
if (! p->capture_errors) {
#ifndef MRB_DISABLE_STDIO
- if (p->filename) {
- fprintf(stderr, "%s:%d:%d: %s\n", p->filename, p->lineno, p->column, s);
+ if (p->filename_sym) {
+ const char *filename = mrb_sym2name_len(p->mrb, p->filename_sym, NULL);
+ fprintf(stderr, "%s:%d:%d: %s\n", filename, p->lineno, p->column, s);
}
else {
fprintf(stderr, "line %d:%d: %s\n", p->lineno, p->column, s);
@@ -3579,11 +3711,13 @@ yyerror(parser_state *p, const char *s)
}
static void
-yyerror_i(parser_state *p, const char *fmt, int i)
+yyerror_c(parser_state *p, const char *msg, char c)
{
char buf[256];
- snprintf(buf, sizeof(buf), fmt, i);
+ strncpy(buf, msg, sizeof(buf) - 2);
+ buf[sizeof(buf) - 2] = '\0';
+ strncat(buf, &c, 1);
yyerror(p, buf);
}
@@ -3595,11 +3729,12 @@ yywarn(parser_state *p, const char *s)
if (! p->capture_errors) {
#ifndef MRB_DISABLE_STDIO
- if (p->filename) {
- fprintf(stderr, "%s:%d:%d: %s\n", p->filename, p->lineno, p->column, s);
+ if (p->filename_sym) {
+ const char *filename = mrb_sym2name_len(p->mrb, p->filename_sym, NULL);
+ fprintf(stderr, "%s:%d:%d: warning: %s\n", filename, p->lineno, p->column, s);
}
else {
- fprintf(stderr, "line %d:%d: %s\n", p->lineno, p->column, s);
+ fprintf(stderr, "line %d:%d: warning: %s\n", p->lineno, p->column, s);
}
#endif
}
@@ -3621,11 +3756,14 @@ yywarning(parser_state *p, const char *s)
}
static void
-yywarning_s(parser_state *p, const char *fmt, const char *s)
+yywarning_s(parser_state *p, const char *msg, const char *s)
{
char buf[256];
- snprintf(buf, sizeof(buf), fmt, s);
+ strncpy(buf, msg, sizeof(buf) - 1);
+ buf[sizeof(buf) - 1] = '\0';
+ strncat(buf, ": ", sizeof(buf) - strlen(buf) - 1);
+ strncat(buf, s, sizeof(buf) - strlen(buf) - 1);
yywarning(p, buf);
}
@@ -3637,10 +3775,10 @@ backref_error(parser_state *p, node *n)
c = intn(n->car);
if (c == NODE_NTH_REF) {
- yyerror_i(p, "can't set variable $%" MRB_PRId, intn(n->cdr));
+ yyerror_c(p, "can't set variable $", (char)intn(n->cdr)+'0');
}
else if (c == NODE_BACK_REF) {
- yyerror_i(p, "can't set variable $%c", intn(n->cdr));
+ yyerror_c(p, "can't set variable $", (char)intn(n->cdr));
}
else {
mrb_bug(p->mrb, "Internal error in backref_error() : n=>car == %S", mrb_fixnum_value(c));
@@ -3664,8 +3802,10 @@ void_expr_error(parser_state *p, node *n)
break;
case NODE_AND:
case NODE_OR:
- void_expr_error(p, n->cdr->car);
- void_expr_error(p, n->cdr->cdr);
+ if (n->cdr) {
+ void_expr_error(p, n->cdr->car);
+ void_expr_error(p, n->cdr->cdr);
+ }
break;
case NODE_BEGIN:
if (n->cdr) {
@@ -4182,8 +4322,17 @@ parse_string(parser_state *p)
}
if (c < 0) {
char buf[256];
- snprintf(buf, sizeof(buf), "can't find heredoc delimiter \"%s\" anywhere before EOF", hinf->term);
- yyerror(p, buf);
+ const char s1[] = "can't find heredoc delimiter \"";
+ const char s2[] = "\" anywhere before EOF";
+
+ if (sizeof(s1)+sizeof(s2)+strlen(hinf->term)+1 >= sizeof(buf)) {
+ yyerror(p, "can't find heredoc delimiter anywhere before EOF");
+ } else {
+ strcpy(buf, s1);
+ strcat(buf, hinf->term);
+ strcat(buf, s2);
+ yyerror(p, buf);
+ }
return 0;
}
pylval.nd = new_str(p, tok(p), toklen(p));
@@ -4333,9 +4482,14 @@ parse_string(parser_state *p)
pushback(p, re_opt);
if (toklen(p)) {
char msg[128];
+
+ strcpy(msg, "unknown regexp option");
tokfix(p);
- snprintf(msg, sizeof(msg), "unknown regexp option%s - %s",
- toklen(p) > 1 ? "s" : "", tok(p));
+ if (toklen(p) > 1) {
+ strcat(msg, "s");
+ }
+ strcat(msg, " - ");
+ strncat(msg, tok(p), sizeof(msg) - strlen(msg) - 1);
yyerror(p, msg);
}
if (f != 0) {
@@ -4542,7 +4696,7 @@ parser_yylex(parser_state *p)
}
pushback(p, c);
if (IS_SPCARG(c)) {
- yywarning(p, "`**' interpreted as argument prefix");
+ yywarning(p, "'**' interpreted as argument prefix");
c = tDSTAR;
}
else if (IS_BEG()) {
@@ -4763,7 +4917,10 @@ parser_yylex(parser_state *p)
}
if (c2) {
char buf[256];
- snprintf(buf, sizeof(buf), "invalid character syntax; use ?\\%c", c2);
+ char cc = (char)c2;
+
+ strcpy(buf, "invalid character syntax; use ?\\");
+ strncat(buf, &cc, 1);
yyerror(p, buf);
}
}
@@ -4774,10 +4931,10 @@ parser_yylex(parser_state *p)
}
newtok(p);
/* need support UTF-8 if configured */
- if ((isalnum(c) || c == '_')) {
+ if ((ISALNUM(c) || c == '_')) {
int c2 = nextc(p);
pushback(p, c2);
- if ((isalnum(c2) || c2 == '_')) {
+ if ((ISALNUM(c2) || c2 == '_')) {
goto ternary;
}
}
@@ -5136,12 +5293,12 @@ parser_yylex(parser_state *p)
pushback(p, c);
if (nondigit) {
trailing_uc:
- yyerror_i(p, "trailing '%c' in number", nondigit);
+ yyerror_c(p, "trailing non digit in number: ", (char)nondigit);
}
tokfix(p);
if (is_float) {
#ifdef MRB_WITHOUT_FLOAT
- yywarning_s(p, "floating point numbers are not supported", tok(p));
+ yywarning(p, "floating point numbers are not supported");
pylval.nd = new_int(p, "0", 10);
return tINTEGER;
#else
@@ -5151,10 +5308,10 @@ parser_yylex(parser_state *p)
errno = 0;
d = mrb_float_read(tok(p), &endp);
if (d == 0 && endp == tok(p)) {
- yywarning_s(p, "corrupted float value %s", tok(p));
+ yywarning_s(p, "corrupted float value", tok(p));
}
else if (errno == ERANGE) {
- yywarning_s(p, "float %s out of range", tok(p));
+ yywarning_s(p, "float out of range", tok(p));
errno = 0;
}
pylval.nd = new_float(p, tok(p));
@@ -5345,7 +5502,7 @@ parser_yylex(parser_state *p)
}
else {
term = nextc(p);
- if (isalnum(term)) {
+ if (ISALNUM(term)) {
yyerror(p, "unknown type of %string");
return 0;
}
@@ -5489,14 +5646,14 @@ parser_yylex(parser_state *p)
do {
tokadd(p, c);
c = nextc(p);
- } while (c >= 0 && isdigit(c));
+ } while (c >= 0 && ISDIGIT(c));
pushback(p, c);
if (last_state == EXPR_FNAME) goto gvar;
tokfix(p);
{
unsigned long n = strtoul(tok(p), NULL, 10);
if (n > INT_MAX) {
- yyerror_i(p, "capture group index must be <= %d", INT_MAX);
+ yyerror(p, "capture group index must be <= " MRB_STRINGIZE(INT_MAX));
return 0;
}
pylval.nd = new_nth_ref(p, (int)n);
@@ -5531,12 +5688,12 @@ parser_yylex(parser_state *p)
}
return 0;
}
- else if (isdigit(c)) {
+ else if (ISDIGIT(c)) {
if (p->tidx == 1) {
- yyerror_i(p, "'@%c' is not allowed as an instance variable name", c);
+ yyerror_c(p, "wrong instance variable name: @", c);
}
else {
- yyerror_i(p, "'@@%c' is not allowed as a class variable name", c);
+ yyerror_c(p, "wrong class variable name: @@", c);
}
return 0;
}
@@ -5552,7 +5709,15 @@ parser_yylex(parser_state *p)
default:
if (!identchar(c)) {
- yyerror_i(p, "Invalid char '\\x%02X' in expression", c);
+ char buf[36];
+ const char s[] = "Invalid char in expression: 0x";
+ const char hexdigits[] = "0123456789ABCDEF";
+
+ strcpy(buf, s);
+ buf[sizeof(s)-1] = hexdigits[(c & 0xf0) >> 4];
+ buf[sizeof(s)] = hexdigits[(c & 0x0f)];
+ buf[sizeof(s)+1] = 0;
+ yyerror(p, buf);
goto retry;
}
@@ -5688,7 +5853,7 @@ parser_yylex(parser_state *p)
mrb_sym ident = intern_cstr(tok(p));
pylval.id = ident;
- if (last_state != EXPR_DOT && islower(tok(p)[0]) && local_var_p(p, ident)) {
+ if (last_state != EXPR_DOT && ISLOWER(tok(p)[0]) && local_var_p(p, ident)) {
p->lstate = EXPR_END;
}
}
@@ -5898,7 +6063,7 @@ mrb_parser_set_filename(struct mrb_parser_state *p, const char *f)
mrb_sym* new_table;
sym = mrb_intern_cstr(p->mrb, f);
- p->filename = mrb_sym2name_len(p->mrb, sym, NULL);
+ p->filename_sym = sym;
p->lineno = (p->filename_table_length > 0)? 0 : 1;
for (i = 0; i < p->filename_table_length; ++i) {
@@ -5922,11 +6087,11 @@ mrb_parser_set_filename(struct mrb_parser_state *p, const char *f)
p->filename_table[p->filename_table_length - 1] = sym;
}
-MRB_API char const*
+MRB_API mrb_sym
mrb_parser_get_filename(struct mrb_parser_state* p, uint16_t idx) {
- if (idx >= p->filename_table_length) return NULL;
+ if (idx >= p->filename_table_length) return 0;
else {
- return mrb_sym2name_len(p->mrb, p->filename_table[idx], NULL);
+ return p->filename_table[idx];
}
}
@@ -5981,11 +6146,12 @@ mrb_load_exec(mrb_state *mrb, struct mrb_parser_state *p, mrbc_context *c)
if (c) c->parser_nerr = p->nerr;
if (p->capture_errors) {
char buf[256];
- int n;
- n = snprintf(buf, sizeof(buf), "line %d: %s\n",
- p->error_buffer[0].lineno, p->error_buffer[0].message);
- mrb->exc = mrb_obj_ptr(mrb_exc_new(mrb, E_SYNTAX_ERROR, buf, n));
+ strcpy(buf, "line ");
+ dump_int(p->error_buffer[0].lineno, buf+5);
+ strcat(buf, ": ");
+ strncat(buf, p->error_buffer[0].message, sizeof(buf) - strlen(buf) - 1);
+ mrb->exc = mrb_obj_ptr(mrb_exc_new(mrb, E_SYNTAX_ERROR, buf, strlen(buf)));
mrb_parser_free(p);
return mrb_undef_value();
}
generated by cgit v1.2.3 (git 2.39.1) at 2026-05-04 23:52:22 +0000