diff options
Diffstat (limited to 'py')
| -rw-r--r-- | py/builtin.c | 12 | ||||
| -rw-r--r-- | py/builtin.h | 1 | ||||
| -rw-r--r-- | py/obj.c | 12 | ||||
| -rw-r--r-- | py/obj.h | 5 | ||||
| -rw-r--r-- | py/objlist.c | 9 | ||||
| -rw-r--r-- | py/objstr.c | 87 | ||||
| -rw-r--r-- | py/py.mk | 1 | ||||
| -rw-r--r-- | py/qstrdefs.h | 1 | ||||
| -rw-r--r-- | py/runtime.c | 1 | ||||
| -rw-r--r-- | py/sequence.c | 25 | ||||
| -rw-r--r-- | py/stream.c | 14 | ||||
| -rw-r--r-- | py/stream.h | 3 |
12 files changed, 140 insertions, 31 deletions
diff --git a/py/builtin.c b/py/builtin.c index 8340ad3045..9cbc037677 100644 --- a/py/builtin.c +++ b/py/builtin.c @@ -350,3 +350,15 @@ static mp_obj_t mp_builtin_str(mp_obj_t o_in) { } MP_DEFINE_CONST_FUN_OBJ_1(mp_builtin_str_obj, mp_builtin_str); + +// TODO: This should be type, this is just quick CPython compat hack +static mp_obj_t mp_builtin_bytes(uint n_args, const mp_obj_t *args) { + if (!MP_OBJ_IS_QSTR(args[0]) && !MP_OBJ_IS_TYPE(args[0], &str_type)) { + assert(0); + } + // Currently, MicroPython strings are mix between CPython byte and unicode + // strings. So, conversion is null so far. + return args[0]; +} + +MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(mp_builtin_bytes_obj, 1, 3, mp_builtin_bytes); diff --git a/py/builtin.h b/py/builtin.h index 050a2161c1..4257de5bdb 100644 --- a/py/builtin.h +++ b/py/builtin.h @@ -5,6 +5,7 @@ MP_DECLARE_CONST_FUN_OBJ(mp_builtin___repl_print___obj); MP_DECLARE_CONST_FUN_OBJ(mp_builtin_abs_obj); MP_DECLARE_CONST_FUN_OBJ(mp_builtin_all_obj); MP_DECLARE_CONST_FUN_OBJ(mp_builtin_any_obj); +MP_DECLARE_CONST_FUN_OBJ(mp_builtin_bytes_obj); // Temporary hack MP_DECLARE_CONST_FUN_OBJ(mp_builtin_callable_obj); MP_DECLARE_CONST_FUN_OBJ(mp_builtin_chr_obj); MP_DECLARE_CONST_FUN_OBJ(mp_builtin_divmod_obj); @@ -28,6 +28,8 @@ mp_obj_t mp_obj_get_type(mp_obj_t o_in) { const char *mp_obj_get_type_str(mp_obj_t o_in) { if (MP_OBJ_IS_SMALL_INT(o_in)) { return "int"; + } else if (MP_OBJ_IS_QSTR(o_in)) { + return "str"; } else { mp_obj_base_t *o = o_in; return o->type->name; @@ -222,7 +224,9 @@ void mp_obj_get_complex(mp_obj_t arg, mp_float_t *real, mp_float_t *imag) { #endif qstr mp_obj_get_qstr(mp_obj_t arg) { - if (MP_OBJ_IS_TYPE(arg, &str_type)) { + if (MP_OBJ_IS_QSTR(arg)) { + return MP_OBJ_QSTR_VALUE(arg); + } else if (MP_OBJ_IS_TYPE(arg, &str_type)) { return mp_obj_str_get(arg); } else { assert(0); @@ -286,3 +290,9 @@ mp_obj_t mp_obj_len_maybe(mp_obj_t o_in) { } return MP_OBJ_NEW_SMALL_INT(len); } + +// Return input argument. Useful as .getiter for objects which are +// their own iterators, etc. +mp_obj_t mp_identity(mp_obj_t self) { + return self; +} @@ -351,6 +351,8 @@ extern const mp_obj_type_t fun_native_type; extern const mp_obj_type_t fun_bc_type; void mp_obj_fun_bc_get(mp_obj_t self_in, int *n_args, uint *n_state, const byte **code); +mp_obj_t mp_identity(mp_obj_t self); + // generator extern const mp_obj_type_t gen_instance_type; @@ -374,3 +376,6 @@ typedef struct _mp_obj_classmethod_t { mp_obj_base_t base; mp_obj_t fun; } mp_obj_classmethod_t; + +// sequence helpers +void mp_seq_multiply(const void *items, uint item_sz, uint len, uint times, void *dest); diff --git a/py/objlist.c b/py/objlist.c index bc363d38fd..2e9a8705ff 100644 --- a/py/objlist.c +++ b/py/objlist.c @@ -153,13 +153,8 @@ static mp_obj_t list_binary_op(int op, mp_obj_t lhs, mp_obj_t rhs) { return NULL; } int n = MP_OBJ_SMALL_INT_VALUE(rhs); - int len = o->len; - mp_obj_list_t *s = list_new(len * n); - mp_obj_t *dest = s->items; - for (int i = 0; i < n; i++) { - memcpy(dest, o->items, sizeof(mp_obj_t) * len); - dest += len; - } + mp_obj_list_t *s = list_new(o->len * n); + mp_seq_multiply(o->items, sizeof(*o->items), o->len, n, s->items); return s; } case RT_COMPARE_OP_EQUAL: diff --git a/py/objstr.c b/py/objstr.c index 3552058430..f4dc857398 100644 --- a/py/objstr.c +++ b/py/objstr.c @@ -65,9 +65,8 @@ mp_obj_t str_binary_op(int op, mp_obj_t lhs_in, mp_obj_t rhs_in) { // TODO: need predicate to check for int-like type (bools are such for example) // ["no", "yes"][1 == 2] is common idiom if (MP_OBJ_IS_SMALL_INT(rhs_in)) { - // TODO: This implements byte string access for single index so far - // TODO: Handle negative indexes. - return mp_obj_new_int(lhs_data[mp_obj_get_int(rhs_in)]); + uint index = mp_get_index(lhs->base.type, lhs_len, rhs_in); + return mp_obj_new_str(qstr_from_strn((const char*)lhs_data + index, 1)); #if MICROPY_ENABLE_SLICE } else if (MP_OBJ_IS_TYPE(rhs_in, &slice_type)) { machine_int_t start, stop, step; @@ -122,6 +121,16 @@ mp_obj_t str_binary_op(int op, mp_obj_t lhs_in, mp_obj_t rhs_in) { return mp_const_false; } break; + case RT_BINARY_OP_MULTIPLY: + { + if (!MP_OBJ_IS_SMALL_INT(rhs_in)) { + return NULL; + } + int n = MP_OBJ_SMALL_INT_VALUE(rhs_in); + char *s = m_new(char, lhs_len * n); + mp_seq_multiply(lhs_data, sizeof(*lhs_data), lhs_len, n, s); + return MP_OBJ_NEW_QSTR(qstr_from_strn_take(s, lhs_len * n, lhs_len * n)); + } } return MP_OBJ_NULL; // op not supported @@ -184,6 +193,45 @@ bad_arg: nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "?str.join expecting a list of str's")); } +#define is_ws(c) ((c) == ' ' || (c) == '\t') + +static mp_obj_t str_split(uint n_args, const mp_obj_t *args) { + int splits = -1; + mp_obj_t sep = mp_const_none; + if (n_args > 1) { + sep = args[1]; + if (n_args > 2) { + splits = MP_OBJ_SMALL_INT_VALUE(args[2]); + } + } + assert(sep == mp_const_none); + (void)sep; // unused; to hush compiler warning + mp_obj_t res = mp_obj_new_list(0, NULL); + const char *s = qstr_str(mp_obj_str_get(args[0])); + const char *start; + + // Initial whitespace is not counted as split, so we pre-do it + while (is_ws(*s)) s++; + while (*s && splits != 0) { + start = s; + while (*s != 0 && !is_ws(*s)) s++; + rt_list_append(res, MP_OBJ_NEW_QSTR(qstr_from_strn(start, s - start))); + if (*s == 0) { + break; + } + while (is_ws(*s)) s++; + if (splits > 0) { + splits--; + } + } + + if (*s != 0) { + rt_list_append(res, MP_OBJ_NEW_QSTR(qstr_from_str(s))); + } + + return res; +} + static bool chr_in_str(const char* const str, const size_t str_len, const char c) { for (size_t i = 0; i < str_len; i++) { if (str[i] == c) { @@ -195,16 +243,8 @@ static bool chr_in_str(const char* const str, const size_t str_len, const char c static mp_obj_t str_find(uint n_args, const mp_obj_t *args) { assert(2 <= n_args && n_args <= 4); - assert(MP_OBJ_IS_TYPE(args[0], &str_type)); - if (!MP_OBJ_IS_TYPE(args[1], &str_type)) { - nlr_jump(mp_obj_new_exception_msg_1_arg( - MP_QSTR_TypeError, - "Can't convert '%s' object to str implicitly", - mp_obj_get_type_str(args[1]))); - } - - const char* haystack = qstr_str(((mp_obj_str_t*)args[0])->qstr); - const char* needle = qstr_str(((mp_obj_str_t*)args[1])->qstr); + const char* haystack = qstr_str(mp_obj_str_get(args[0])); + const char* needle = qstr_str(mp_obj_str_get(args[1])); size_t haystack_len = strlen(haystack); size_t needle_len = strlen(needle); @@ -242,14 +282,11 @@ mp_obj_t str_strip(uint n_args, const mp_obj_t *args) { if (n_args == 1) { chars_to_del = whitespace; } else { - assert(MP_OBJ_IS_TYPE(args[1], &str_type)); - mp_obj_str_t *chars_to_del_obj = args[1]; - chars_to_del = qstr_str(chars_to_del_obj->qstr); + chars_to_del = qstr_str(mp_obj_str_get(args[1])); } const size_t chars_to_del_len = strlen(chars_to_del); - mp_obj_str_t *self = args[0]; - const char *orig_str = qstr_str(self->qstr); + const char *orig_str = qstr_str(mp_obj_str_get(args[0])); const size_t orig_str_len = strlen(orig_str); size_t first_good_char_pos = 0; @@ -307,12 +344,14 @@ mp_obj_t str_format(uint n_args, const mp_obj_t *args) { static MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_find_obj, 2, 4, str_find); static MP_DEFINE_CONST_FUN_OBJ_2(str_join_obj, str_join); +static MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_split_obj, 1, 3, str_split); static MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(str_strip_obj, 1, 2, str_strip); static MP_DEFINE_CONST_FUN_OBJ_VAR(str_format_obj, 1, str_format); static const mp_method_t str_type_methods[] = { { "find", &str_find_obj }, { "join", &str_join_obj }, + { "split", &str_split_obj }, { "strip", &str_strip_obj }, { "format", &str_format_obj }, { NULL, NULL }, // end-of-list sentinel @@ -335,9 +374,15 @@ mp_obj_t mp_obj_new_str(qstr qstr) { } qstr mp_obj_str_get(mp_obj_t self_in) { - assert(MP_OBJ_IS_TYPE(self_in, &str_type)); - mp_obj_str_t *self = self_in; - return self->qstr; + if (MP_OBJ_IS_QSTR(self_in)) { + return MP_OBJ_QSTR_VALUE(self_in); + } + if (MP_OBJ_IS_TYPE(self_in, &str_type)) { + mp_obj_str_t *self = self_in; + return self->qstr; + } + nlr_jump(mp_obj_new_exception_msg_varg(MP_QSTR_TypeError, "Can't convert '%s' object to str implicitly", + mp_obj_get_type_str(self_in))); } /******************************************************************************/ @@ -97,6 +97,7 @@ PY_O_BASENAME = \ objtuple.o \ objtype.o \ objzip.o \ + sequence.o \ stream.o \ builtin.o \ builtinimport.o \ diff --git a/py/qstrdefs.h b/py/qstrdefs.h index 10b1fc0d39..9bc01c5851 100644 --- a/py/qstrdefs.h +++ b/py/qstrdefs.h @@ -40,6 +40,7 @@ Q(any) Q(array) Q(bool) Q(bytearray) +Q(bytes) Q(callable) Q(chr) Q(complex) diff --git a/py/runtime.c b/py/runtime.c index 210047ac0a..ccd3d7d0af 100644 --- a/py/runtime.c +++ b/py/runtime.c @@ -130,6 +130,7 @@ void rt_init(void) { mp_map_add_qstr(&map_builtins, MP_QSTR_abs, (mp_obj_t)&mp_builtin_abs_obj); mp_map_add_qstr(&map_builtins, MP_QSTR_all, (mp_obj_t)&mp_builtin_all_obj); mp_map_add_qstr(&map_builtins, MP_QSTR_any, (mp_obj_t)&mp_builtin_any_obj); + mp_map_add_qstr(&map_builtins, MP_QSTR_bytes, (mp_obj_t)&mp_builtin_bytes_obj); mp_map_add_qstr(&map_builtins, MP_QSTR_callable, (mp_obj_t)&mp_builtin_callable_obj); mp_map_add_qstr(&map_builtins, MP_QSTR_chr, (mp_obj_t)&mp_builtin_chr_obj); mp_map_add_qstr(&map_builtins, MP_QSTR_divmod, (mp_obj_t)&mp_builtin_divmod_obj); diff --git a/py/sequence.c b/py/sequence.c new file mode 100644 index 0000000000..56718c6f85 --- /dev/null +++ b/py/sequence.c @@ -0,0 +1,25 @@ +#include <stdlib.h> +#include <stdint.h> +#include <string.h> +#include <assert.h> + +#include "nlr.h" +#include "misc.h" +#include "mpconfig.h" +#include "qstr.h" +#include "obj.h" +#include "map.h" +#include "runtime0.h" +#include "runtime.h" + +// Helpers for sequence types + +// Implements backend of sequence * integer operation. Assumes elements are +// memory-adjacent in sequence. +void mp_seq_multiply(const void *items, uint item_sz, uint len, uint times, void *dest) { + for (int i = 0; i < times; i++) { + uint copy_sz = item_sz * len; + memcpy(dest, items, copy_sz); + dest = (char*)dest + copy_sz; + } +} diff --git a/py/stream.c b/py/stream.c index 88ddc5e6c9..be560d3c22 100644 --- a/py/stream.c +++ b/py/stream.c @@ -30,7 +30,8 @@ static mp_obj_t stream_read(uint n_args, const mp_obj_t *args) { nlr_jump(mp_obj_new_exception_msg_varg(MP_QSTR_OSError, "[Errno %d]", error)); } else { // TODO don't intern this string - return mp_obj_new_str(qstr_from_strn_take(buf, sz, out_sz)); + buf = m_realloc(buf, sz, out_sz); + return mp_obj_new_str(qstr_from_strn_take(buf, out_sz, out_sz)); } } @@ -134,9 +135,18 @@ static mp_obj_t stream_unbuffered_readline(uint n_args, const mp_obj_t *args) { } // TODO don't intern this string vstr_shrink(vstr); - return mp_obj_new_str(qstr_from_strn_take(vstr_str(vstr), vstr->alloc, vstr_len(vstr))); + return MP_OBJ_NEW_QSTR(qstr_from_strn_take(vstr_str(vstr), vstr->alloc, vstr_len(vstr))); } +mp_obj_t mp_stream_unbuffered_iter(mp_obj_t self) { + mp_obj_t l_in = stream_unbuffered_readline(1, &self); + const char *l = qstr_str(MP_OBJ_QSTR_VALUE(l_in)); + // TODO: \0 + if (*l != 0) { + return l_in; + } + return mp_const_stop_iteration; +} MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(mp_stream_read_obj, 1, 2, stream_read); MP_DEFINE_CONST_FUN_OBJ_1(mp_stream_readall_obj, stream_readall); diff --git a/py/stream.h b/py/stream.h index 58e8072549..a0cc34797b 100644 --- a/py/stream.h +++ b/py/stream.h @@ -2,3 +2,6 @@ extern const mp_obj_fun_native_t mp_stream_read_obj; extern const mp_obj_fun_native_t mp_stream_readall_obj; extern const mp_obj_fun_native_t mp_stream_unbuffered_readline_obj; extern const mp_obj_fun_native_t mp_stream_write_obj; + +// Iterator which uses mp_stream_unbuffered_readline_obj +mp_obj_t mp_stream_unbuffered_iter(mp_obj_t self); |