diff options
Diffstat (limited to 'py')
| -rw-r--r-- | py/emitinlinethumb.c | 3 | ||||
| -rw-r--r-- | py/formatfloat.c | 2 | ||||
| -rw-r--r-- | py/objarray.c | 2 | ||||
| -rw-r--r-- | py/stream.c | 88 |
4 files changed, 90 insertions, 5 deletions
diff --git a/py/emitinlinethumb.c b/py/emitinlinethumb.c index 435e1b64d9..3d0efd5645 100644 --- a/py/emitinlinethumb.c +++ b/py/emitinlinethumb.c @@ -315,7 +315,7 @@ STATIC void emit_inline_thumb_op(emit_inline_asm_t *emit, qstr op, int n_args, m uint reg_dest = get_arg_reg(emit, op_str, pn_args[0], 15); uint reg_src = get_arg_reg(emit, op_str, pn_args[1], 15); asm_thumb_mov_reg_reg(emit->as, reg_dest, reg_src); - } else if (strcmp(op_str, "and") == 0) { + } else if (strcmp(op_str, "and_") == 0) { op_code = ASM_THUMB_FORMAT_4_AND; uint reg_dest, reg_src; op_format_4: @@ -323,7 +323,6 @@ STATIC void emit_inline_thumb_op(emit_inline_asm_t *emit, qstr op, int n_args, m reg_src = get_arg_reg(emit, op_str, pn_args[1], 7); asm_thumb_format_4(emit->as, op_code, reg_dest, reg_src); // TODO probably uses less ROM if these ops are in a lookup table - } else if (strcmp(op_str, "and") == 0) { op_code = ASM_THUMB_FORMAT_4_AND; goto op_format_4; } else if (strcmp(op_str, "eor") == 0) { op_code = ASM_THUMB_FORMAT_4_EOR; goto op_format_4; } else if (strcmp(op_str, "lsl") == 0) { op_code = ASM_THUMB_FORMAT_4_LSL; goto op_format_4; } else if (strcmp(op_str, "lsr") == 0) { op_code = ASM_THUMB_FORMAT_4_LSR; goto op_format_4; diff --git a/py/formatfloat.c b/py/formatfloat.c index caecb0f29b..d1b12cdd5b 100644 --- a/py/formatfloat.c +++ b/py/formatfloat.c @@ -75,7 +75,7 @@ int format_float(float f, char *buf, size_t buf_size, char fmt, int prec, char s if (buf_size < 7) { // Smallest exp notion is -9e+99 which is 6 chars plus terminating - // nulll. + // null. if (buf_size >= 2) { *s++ = '?'; diff --git a/py/objarray.c b/py/objarray.c index ca907fd01d..27da662a61 100644 --- a/py/objarray.c +++ b/py/objarray.c @@ -59,7 +59,7 @@ STATIC mp_obj_t array_append(mp_obj_t self_in, mp_obj_t arg); STATIC void array_print(void (*print)(void *env, const char *fmt, ...), void *env, mp_obj_t o_in, mp_print_kind_t kind) { mp_obj_array_t *o = o_in; if (o->typecode == BYTEARRAY_TYPECODE) { - print(env, "bytearray(b", o->typecode); + print(env, "bytearray(b"); mp_str_print_quoted(print, env, o->items, o->len, true); } else { print(env, "array('%c'", o->typecode); diff --git a/py/stream.c b/py/stream.c index 2b4410728f..4c8b8a570a 100644 --- a/py/stream.c +++ b/py/stream.c @@ -67,6 +67,9 @@ STATIC mp_obj_t stream_read(uint n_args, const mp_obj_t *args) { nlr_raise(mp_obj_new_exception_msg(&mp_type_OSError, "Operation not supported")); } + // What to do if sz < -1? Python docs don't specify this case. + // CPython does a readall, but here we silently let negatives through, + // and they will cause a MemoryError. mp_int_t sz; if (n_args == 1 || ((sz = mp_obj_get_int(args[1])) == -1)) { return stream_readall(args[0]); @@ -74,7 +77,90 @@ STATIC mp_obj_t stream_read(uint n_args, const mp_obj_t *args) { #if MICROPY_PY_BUILTINS_STR_UNICODE if (!o->type->stream_p->is_bytes) { - mp_not_implemented("Reading from unicode text streams by character count"); + // We need to read sz number of unicode characters. Because we don't have any + // buffering, and because the stream API can only read bytes, we must read here + // in units of bytes and must never over read. If we want sz chars, then reading + // sz bytes will never over-read, so we follow this approach, in a loop to keep + // reading until we have exactly enough chars. This will be 1 read for text + // with ASCII-only chars, and about 2 reads for text with a couple of non-ASCII + // chars. For text with lots of non-ASCII chars, it'll be pretty inefficient + // in time and memory. + + vstr_t vstr; + vstr_init(&vstr, sz); + mp_uint_t more_bytes = sz; + mp_uint_t last_buf_offset = 0; + while (more_bytes > 0) { + char *p = vstr_add_len(&vstr, more_bytes); + if (p == NULL) { + nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_MemoryError, "out of memory")); + } + int error; + mp_int_t out_sz = o->type->stream_p->read(o, p, more_bytes, &error); + if (out_sz == -1) { + vstr_cut_tail_bytes(&vstr, more_bytes); + if (is_nonblocking_error(error)) { + // With non-blocking streams, we read as much as we can. + // If we read nothing, return None, just like read(). + // Otherwise, return data read so far. + // TODO what if we have read only half a non-ASCII char? + if (vstr.len == 0) { + vstr_clear(&vstr); + return mp_const_none; + } + break; + } + nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_OSError, "[Errno %d]", error)); + } + + if (out_sz == 0) { + // Finish reading. + // TODO what if we have read only half a non-ASCII char? + vstr_cut_tail_bytes(&vstr, more_bytes); + break; + } + + // count chars from bytes just read + for (mp_uint_t off = last_buf_offset;;) { + byte b = vstr.buf[off]; + int n; + if (!UTF8_IS_NONASCII(b)) { + // 1-byte ASCII char + n = 1; + } else if ((b & 0xe0) == 0xc0) { + // 2-byte char + n = 2; + } else if ((b & 0xf0) == 0xe0) { + // 3-byte char + n = 3; + } else if ((b & 0xf8) == 0xf0) { + // 4-byte char + n = 4; + } else { + // TODO + n = 5; + } + if (off + n <= vstr.len) { + // got a whole char in n bytes + off += n; + sz -= 1; + last_buf_offset = off; + if (off >= vstr.len) { + more_bytes = sz; + break; + } + } else { + // didn't get a whole char, so work out how many extra bytes are needed for + // this partial char, plus bytes for additional chars that we want + more_bytes = (off + n - vstr.len) + (sz - 1); + break; + } + } + } + + mp_obj_t ret = mp_obj_new_str_of_type(&mp_type_str, (byte*)vstr.buf, vstr.len); + vstr_clear(&vstr); + return ret; } #endif |