py/objstr: Optimise mp_obj_new_str_from_vstr for known-safe strings.

jimmo · dpgeorge · commit 3a910b156506 · 2022-08-26T16:44:35.000+10:00
The new `mp_obj_new_str_from_utf8_vstr` can be used when you know you
already have a unicode-safe string.

Signed-off-by: Jim Mussared &lt;jim.mussared@gmail.com&gt;
diff --git a/extmod/modujson.c b/extmod/modujson.c
@@ -67,7 +67,7 @@ STATIC mp_obj_t mod_ujson_dump_helper(size_t n_args, const mp_obj_t *pos_args, m
         vstr_t vstr;
         vstr_init_print(&vstr, 8, &print_ext.base);
         mp_obj_print_helper(&print_ext.base, pos_args[0], PRINT_JSON);
-        return mp_obj_new_str_from_vstr(&vstr);
+        return mp_obj_new_str_from_utf8_vstr(&vstr);
     } else {
         // dump(obj, stream)
         print_ext.base.data = MP_OBJ_TO_PTR(pos_args[1]);
@@ -103,7 +103,7 @@ STATIC mp_obj_t mod_ujson_dumps(mp_obj_t obj) {
     mp_print_t print;
     vstr_init_print(&vstr, 8, &print);
     mp_obj_print_helper(&print, obj, PRINT_JSON);
-    return mp_obj_new_str_from_vstr(&vstr);
+    return mp_obj_new_str_from_utf8_vstr(&vstr);
 }
 STATIC MP_DEFINE_CONST_FUN_OBJ_1(mod_ujson_dumps_obj, mod_ujson_dumps);
 
diff --git a/extmod/modure.c b/extmod/modure.c
@@ -373,7 +373,7 @@ STATIC mp_obj_t re_sub_helper(size_t n_args, const mp_obj_t *args) {
     vstr_add_strn(&vstr_return, subj.begin, subj.end - subj.begin);
 
     if (mp_obj_get_type(where) == &mp_type_str) {
-        return mp_obj_new_str_from_vstr(&vstr_return);
+        return mp_obj_new_str_from_utf8_vstr(&vstr_return);
     } else {
         return mp_obj_new_bytes_from_vstr(&vstr_return);
     }
diff --git a/ports/unix/modusocket.c b/ports/unix/modusocket.c
@@ -559,7 +559,7 @@ STATIC mp_obj_t mod_socket_inet_ntop(mp_obj_t family_in, mp_obj_t binaddr_in) {
         mp_raise_OSError(errno);
     }
     vstr.len = strlen(vstr.buf);
-    return mp_obj_new_str_from_vstr(&vstr);
+    return mp_obj_new_str_from_utf8_vstr(&vstr);
 }
 STATIC MP_DEFINE_CONST_FUN_OBJ_2(mod_socket_inet_ntop_obj, mod_socket_inet_ntop);
 
diff --git a/py/modbuiltins.c b/py/modbuiltins.c
@@ -467,7 +467,7 @@ STATIC mp_obj_t mp_builtin_repr(mp_obj_t o_in) {
     mp_print_t print;
     vstr_init_print(&vstr, 16, &print);
     mp_obj_print_helper(&print, o_in, PRINT_REPR);
-    return mp_obj_new_str_from_vstr(&vstr);
+    return mp_obj_new_str_from_utf8_vstr(&vstr);
 }
 MP_DEFINE_CONST_FUN_OBJ_1(mp_builtin_repr_obj, mp_builtin_repr);
 
diff --git a/py/obj.h b/py/obj.h
@@ -792,6 +792,11 @@ mp_obj_t mp_obj_new_int_from_ull(unsigned long long val); // this must return a
 mp_obj_t mp_obj_new_str(const char *data, size_t len);
 mp_obj_t mp_obj_new_str_via_qstr(const char *data, size_t len);
 mp_obj_t mp_obj_new_str_from_vstr(vstr_t *vstr);
+#if MICROPY_PY_BUILTINS_STR_UNICODE && MICROPY_PY_BUILTINS_STR_UNICODE_CHECK
+mp_obj_t mp_obj_new_str_from_utf8_vstr(vstr_t *vstr); // only use when vstr is already known to be utf-8 encoded
+#else
+#define mp_obj_new_str_from_utf8_vstr mp_obj_new_str_from_vstr
+#endif
 mp_obj_t mp_obj_new_bytes_from_vstr(vstr_t *vstr);
 mp_obj_t mp_obj_new_bytes(const byte *data, size_t len);
 mp_obj_t mp_obj_new_bytearray(size_t n, const void *items);
diff --git a/py/objstr.c b/py/objstr.c
@@ -2256,6 +2256,13 @@ mp_obj_t mp_obj_new_str_from_vstr(vstr_t *vstr) {
     return mp_obj_new_str_type_from_vstr(&mp_type_str, vstr);
 }
 
+#if MICROPY_PY_BUILTINS_STR_UNICODE && MICROPY_PY_BUILTINS_STR_UNICODE_CHECK
+mp_obj_t mp_obj_new_str_from_utf8_vstr(vstr_t *vstr) {
+    // bypasses utf8_check.
+    return mp_obj_new_str_type_from_vstr(&mp_type_str, vstr);
+}
+#endif // MICROPY_PY_BUILTINS_STR_UNICODE && MICROPY_PY_BUILTINS_STR_UNICODE_CHECK
+
 mp_obj_t mp_obj_new_bytes_from_vstr(vstr_t *vstr) {
     return mp_obj_new_str_type_from_vstr(&mp_type_bytes, vstr);
 }
diff --git a/py/parsenum.c b/py/parsenum.c
@@ -160,7 +160,7 @@ mp_obj_t mp_parse_num_integer(const char *restrict str_, size_t len, int base, m
         mp_printf(&print, "invalid syntax for integer with base %d: ", base);
         mp_str_print_quoted(&print, str_val_start, top - str_val_start, true);
         mp_obj_t exc = mp_obj_new_exception_arg1(&mp_type_ValueError,
-            mp_obj_new_str_from_vstr(&vstr));
+            mp_obj_new_str_from_utf8_vstr(&vstr));
         raise_exc(exc, lex);
         #endif
     }
diff --git a/py/persistentcode.c b/py/persistentcode.c
@@ -203,7 +203,7 @@ STATIC mp_obj_t load_obj(mp_reader_t *reader) {
         if (obj_type == MP_PERSISTENT_OBJ_STR || obj_type == MP_PERSISTENT_OBJ_BYTES) {
             read_byte(reader); // skip null terminator
             if (obj_type == MP_PERSISTENT_OBJ_STR) {
-                return mp_obj_new_str_from_vstr(&vstr);
+                return mp_obj_new_str_from_utf8_vstr(&vstr);
             } else {
                 return mp_obj_new_bytes_from_vstr(&vstr);
             }

Original file line number	Diff line number	Diff line change
`@@ -373,7 +373,7 @@ STATIC mp_obj_t re_sub_helper(size_t n_args, const mp_obj_t *args) {`
`373`	`373`	`vstr_add_strn(&vstr_return, subj.begin, subj.end - subj.begin);`
`374`	`374`
`375`	`375`	`if (mp_obj_get_type(where) == &mp_type_str) {`
`376`		`- return mp_obj_new_str_from_vstr(&vstr_return);`
	`376`	`+ return mp_obj_new_str_from_utf8_vstr(&vstr_return);`
`377`	`377`	`} else {`
`378`	`378`	`return mp_obj_new_bytes_from_vstr(&vstr_return);`
`379`	`379`	`}`
Original file line number	Diff line number	Diff line change
`@@ -559,7 +559,7 @@ STATIC mp_obj_t mod_socket_inet_ntop(mp_obj_t family_in, mp_obj_t binaddr_in) {`
`559`	`559`	`mp_raise_OSError(errno);`
`560`	`560`	`}`
`561`	`561`	`vstr.len = strlen(vstr.buf);`
`562`		`- return mp_obj_new_str_from_vstr(&vstr);`
	`562`	`+ return mp_obj_new_str_from_utf8_vstr(&vstr);`
`563`	`563`	`}`
`564`	`564`	`STATIC MP_DEFINE_CONST_FUN_OBJ_2(mod_socket_inet_ntop_obj, mod_socket_inet_ntop);`
`565`	`565`
Original file line number	Diff line number	Diff line change
`@@ -467,7 +467,7 @@ STATIC mp_obj_t mp_builtin_repr(mp_obj_t o_in) {`
`467`	`467`	`mp_print_t print;`
`468`	`468`	`vstr_init_print(&vstr, 16, &print);`
`469`	`469`	`mp_obj_print_helper(&print, o_in, PRINT_REPR);`
`470`		`- return mp_obj_new_str_from_vstr(&vstr);`
	`470`	`+ return mp_obj_new_str_from_utf8_vstr(&vstr);`
`471`	`471`	`}`
`472`	`472`	`MP_DEFINE_CONST_FUN_OBJ_1(mp_builtin_repr_obj, mp_builtin_repr);`
`473`	`473`
Original file line number	Diff line number	Diff line change
`@@ -2256,6 +2256,13 @@ mp_obj_t mp_obj_new_str_from_vstr(vstr_t *vstr) {`
`2256`	`2256`	`return mp_obj_new_str_type_from_vstr(&mp_type_str, vstr);`
`2257`	`2257`	`}`
`2258`	`2258`
	`2259`	`+#if MICROPY_PY_BUILTINS_STR_UNICODE && MICROPY_PY_BUILTINS_STR_UNICODE_CHECK`
	`2260`	`+mp_obj_t mp_obj_new_str_from_utf8_vstr(vstr_t *vstr) {`
	`2261`	`+ // bypasses utf8_check.`
	`2262`	`+ return mp_obj_new_str_type_from_vstr(&mp_type_str, vstr);`
	`2263`	`+}`
	`2264`	`+#endif // MICROPY_PY_BUILTINS_STR_UNICODE && MICROPY_PY_BUILTINS_STR_UNICODE_CHECK`
	`2265`	`+`
`2259`	`2266`	`mp_obj_t mp_obj_new_bytes_from_vstr(vstr_t *vstr) {`
`2260`	`2267`	`return mp_obj_new_str_type_from_vstr(&mp_type_bytes, vstr);`
`2261`	`2268`	`}`
Original file line number	Diff line number	Diff line change
`@@ -160,7 +160,7 @@ mp_obj_t mp_parse_num_integer(const char *restrict str_, size_t len, int base, m`
`160`	`160`	`mp_printf(&print, "invalid syntax for integer with base %d: ", base);`
`161`	`161`	`mp_str_print_quoted(&print, str_val_start, top - str_val_start, true);`
`162`	`162`	`mp_obj_t exc = mp_obj_new_exception_arg1(&mp_type_ValueError,`
`163`		`- mp_obj_new_str_from_vstr(&vstr));`
	`163`	`+ mp_obj_new_str_from_utf8_vstr(&vstr));`
`164`	`164`	`raise_exc(exc, lex);`
`165`	`165`	`#endif`
`166`	`166`	`}`
Original file line number	Diff line number	Diff line change
`@@ -203,7 +203,7 @@ STATIC mp_obj_t load_obj(mp_reader_t *reader) {`
`203`	`203`	`if (obj_type == MP_PERSISTENT_OBJ_STR \|\| obj_type == MP_PERSISTENT_OBJ_BYTES) {`
`204`	`204`	`read_byte(reader); // skip null terminator`
`205`	`205`	`if (obj_type == MP_PERSISTENT_OBJ_STR) {`
`206`		`- return mp_obj_new_str_from_vstr(&vstr);`
	`206`	`+ return mp_obj_new_str_from_utf8_vstr(&vstr);`
`207`	`207`	`} else {`
`208`	`208`	`return mp_obj_new_bytes_from_vstr(&vstr);`
`209`	`209`	`}`