diff options
Diffstat (limited to '.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs')
8 files changed, 2818 insertions, 0 deletions
diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/__init__.py b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/__init__.py new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/__init__.py diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/array.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/array.pyx new file mode 100644 index 00000000..f8f9b8dd --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/array.pyx @@ -0,0 +1,875 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +from collections.abc import (Iterable as IterableABC, + Mapping as MappingABC, + Sized as SizedABC) + +from asyncpg import exceptions + + +DEF ARRAY_MAXDIM = 6 # defined in postgresql/src/includes/c.h + +# "NULL" +cdef Py_UCS4 *APG_NULL = [0x004E, 0x0055, 0x004C, 0x004C, 0x0000] + + +ctypedef object (*encode_func_ex)(ConnectionSettings settings, + WriteBuffer buf, + object obj, + const void *arg) + + +ctypedef object (*decode_func_ex)(ConnectionSettings settings, + FRBuffer *buf, + const void *arg) + + +cdef inline bint _is_trivial_container(object obj): + return cpython.PyUnicode_Check(obj) or cpython.PyBytes_Check(obj) or \ + cpythonx.PyByteArray_Check(obj) or cpythonx.PyMemoryView_Check(obj) + + +cdef inline _is_array_iterable(object obj): + return ( + isinstance(obj, IterableABC) and + isinstance(obj, SizedABC) and + not _is_trivial_container(obj) and + not isinstance(obj, MappingABC) + ) + + +cdef inline _is_sub_array_iterable(object obj): + # Sub-arrays have a specialized check, because we treat + # nested tuples as records. + return _is_array_iterable(obj) and not cpython.PyTuple_Check(obj) + + +cdef _get_array_shape(object obj, int32_t *dims, int32_t *ndims): + cdef: + ssize_t mylen = len(obj) + ssize_t elemlen = -2 + object it + + if mylen > _MAXINT32: + raise ValueError('too many elements in array value') + + if ndims[0] > ARRAY_MAXDIM: + raise ValueError( + 'number of array dimensions ({}) exceed the maximum expected ({})'. + format(ndims[0], ARRAY_MAXDIM)) + + dims[ndims[0] - 1] = <int32_t>mylen + + for elem in obj: + if _is_sub_array_iterable(elem): + if elemlen == -2: + elemlen = len(elem) + if elemlen > _MAXINT32: + raise ValueError('too many elements in array value') + ndims[0] += 1 + _get_array_shape(elem, dims, ndims) + else: + if len(elem) != elemlen: + raise ValueError('non-homogeneous array') + else: + if elemlen >= 0: + raise ValueError('non-homogeneous array') + else: + elemlen = -1 + + +cdef _write_array_data(ConnectionSettings settings, object obj, int32_t ndims, + int32_t dim, WriteBuffer elem_data, + encode_func_ex encoder, const void *encoder_arg): + if dim < ndims - 1: + for item in obj: + _write_array_data(settings, item, ndims, dim + 1, elem_data, + encoder, encoder_arg) + else: + for item in obj: + if item is None: + elem_data.write_int32(-1) + else: + try: + encoder(settings, elem_data, item, encoder_arg) + except TypeError as e: + raise ValueError( + 'invalid array element: {}'.format(e.args[0])) from None + + +cdef inline array_encode(ConnectionSettings settings, WriteBuffer buf, + object obj, uint32_t elem_oid, + encode_func_ex encoder, const void *encoder_arg): + cdef: + WriteBuffer elem_data + int32_t dims[ARRAY_MAXDIM] + int32_t ndims = 1 + int32_t i + + if not _is_array_iterable(obj): + raise TypeError( + 'a sized iterable container expected (got type {!r})'.format( + type(obj).__name__)) + + _get_array_shape(obj, dims, &ndims) + + elem_data = WriteBuffer.new() + + if ndims > 1: + _write_array_data(settings, obj, ndims, 0, elem_data, + encoder, encoder_arg) + else: + for i, item in enumerate(obj): + if item is None: + elem_data.write_int32(-1) + else: + try: + encoder(settings, elem_data, item, encoder_arg) + except TypeError as e: + raise ValueError( + 'invalid array element at index {}: {}'.format( + i, e.args[0])) from None + + buf.write_int32(12 + 8 * ndims + elem_data.len()) + # Number of dimensions + buf.write_int32(ndims) + # flags + buf.write_int32(0) + # element type + buf.write_int32(<int32_t>elem_oid) + # upper / lower bounds + for i in range(ndims): + buf.write_int32(dims[i]) + buf.write_int32(1) + # element data + buf.write_buffer(elem_data) + + +cdef _write_textarray_data(ConnectionSettings settings, object obj, + int32_t ndims, int32_t dim, WriteBuffer array_data, + encode_func_ex encoder, const void *encoder_arg, + Py_UCS4 typdelim): + cdef: + ssize_t i = 0 + int8_t delim = <int8_t>typdelim + WriteBuffer elem_data + Py_buffer pybuf + const char *elem_str + char ch + ssize_t elem_len + ssize_t quoted_elem_len + bint need_quoting + + array_data.write_byte(b'{') + + if dim < ndims - 1: + for item in obj: + if i > 0: + array_data.write_byte(delim) + array_data.write_byte(b' ') + _write_textarray_data(settings, item, ndims, dim + 1, array_data, + encoder, encoder_arg, typdelim) + i += 1 + else: + for item in obj: + elem_data = WriteBuffer.new() + + if i > 0: + array_data.write_byte(delim) + array_data.write_byte(b' ') + + if item is None: + array_data.write_bytes(b'NULL') + i += 1 + continue + else: + try: + encoder(settings, elem_data, item, encoder_arg) + except TypeError as e: + raise ValueError( + 'invalid array element: {}'.format( + e.args[0])) from None + + # element string length (first four bytes are the encoded length.) + elem_len = elem_data.len() - 4 + + if elem_len == 0: + # Empty string + array_data.write_bytes(b'""') + else: + cpython.PyObject_GetBuffer( + elem_data, &pybuf, cpython.PyBUF_SIMPLE) + + elem_str = <const char*>(pybuf.buf) + 4 + + try: + if not apg_strcasecmp_char(elem_str, b'NULL'): + array_data.write_byte(b'"') + array_data.write_cstr(elem_str, 4) + array_data.write_byte(b'"') + else: + quoted_elem_len = elem_len + need_quoting = False + + for i in range(elem_len): + ch = elem_str[i] + if ch == b'"' or ch == b'\\': + # Quotes and backslashes need escaping. + quoted_elem_len += 1 + need_quoting = True + elif (ch == b'{' or ch == b'}' or ch == delim or + apg_ascii_isspace(<uint32_t>ch)): + need_quoting = True + + if need_quoting: + array_data.write_byte(b'"') + + if quoted_elem_len == elem_len: + array_data.write_cstr(elem_str, elem_len) + else: + # Escaping required. + for i in range(elem_len): + ch = elem_str[i] + if ch == b'"' or ch == b'\\': + array_data.write_byte(b'\\') + array_data.write_byte(ch) + + array_data.write_byte(b'"') + else: + array_data.write_cstr(elem_str, elem_len) + finally: + cpython.PyBuffer_Release(&pybuf) + + i += 1 + + array_data.write_byte(b'}') + + +cdef inline textarray_encode(ConnectionSettings settings, WriteBuffer buf, + object obj, encode_func_ex encoder, + const void *encoder_arg, Py_UCS4 typdelim): + cdef: + WriteBuffer array_data + int32_t dims[ARRAY_MAXDIM] + int32_t ndims = 1 + int32_t i + + if not _is_array_iterable(obj): + raise TypeError( + 'a sized iterable container expected (got type {!r})'.format( + type(obj).__name__)) + + _get_array_shape(obj, dims, &ndims) + + array_data = WriteBuffer.new() + _write_textarray_data(settings, obj, ndims, 0, array_data, + encoder, encoder_arg, typdelim) + buf.write_int32(array_data.len()) + buf.write_buffer(array_data) + + +cdef inline array_decode(ConnectionSettings settings, FRBuffer *buf, + decode_func_ex decoder, const void *decoder_arg): + cdef: + int32_t ndims = hton.unpack_int32(frb_read(buf, 4)) + int32_t flags = hton.unpack_int32(frb_read(buf, 4)) + uint32_t elem_oid = <uint32_t>hton.unpack_int32(frb_read(buf, 4)) + list result + int i + int32_t elem_len + int32_t elem_count = 1 + FRBuffer elem_buf + int32_t dims[ARRAY_MAXDIM] + Codec elem_codec + + if ndims == 0: + return [] + + if ndims > ARRAY_MAXDIM: + raise exceptions.ProtocolError( + 'number of array dimensions ({}) exceed the maximum expected ({})'. + format(ndims, ARRAY_MAXDIM)) + elif ndims < 0: + raise exceptions.ProtocolError( + 'unexpected array dimensions value: {}'.format(ndims)) + + for i in range(ndims): + dims[i] = hton.unpack_int32(frb_read(buf, 4)) + if dims[i] < 0: + raise exceptions.ProtocolError( + 'unexpected array dimension size: {}'.format(dims[i])) + # Ignore the lower bound information + frb_read(buf, 4) + + if ndims == 1: + # Fast path for flat arrays + elem_count = dims[0] + result = cpython.PyList_New(elem_count) + + for i in range(elem_count): + elem_len = hton.unpack_int32(frb_read(buf, 4)) + if elem_len == -1: + elem = None + else: + frb_slice_from(&elem_buf, buf, elem_len) + elem = decoder(settings, &elem_buf, decoder_arg) + + cpython.Py_INCREF(elem) + cpython.PyList_SET_ITEM(result, i, elem) + + else: + result = _nested_array_decode(settings, buf, + decoder, decoder_arg, ndims, dims, + &elem_buf) + + return result + + +cdef _nested_array_decode(ConnectionSettings settings, + FRBuffer *buf, + decode_func_ex decoder, + const void *decoder_arg, + int32_t ndims, int32_t *dims, + FRBuffer *elem_buf): + + cdef: + int32_t elem_len + int64_t i, j + int64_t array_len = 1 + object elem, stride + # An array of pointers to lists for each current array level. + void *strides[ARRAY_MAXDIM] + # An array of current positions at each array level. + int32_t indexes[ARRAY_MAXDIM] + + for i in range(ndims): + array_len *= dims[i] + indexes[i] = 0 + strides[i] = NULL + + if array_len == 0: + # A multidimensional array with a zero-sized dimension? + return [] + + elif array_len < 0: + # Array length overflow + raise exceptions.ProtocolError('array length overflow') + + for i in range(array_len): + # Decode the element. + elem_len = hton.unpack_int32(frb_read(buf, 4)) + if elem_len == -1: + elem = None + else: + elem = decoder(settings, + frb_slice_from(elem_buf, buf, elem_len), + decoder_arg) + + # Take an explicit reference for PyList_SET_ITEM in the below + # loop expects this. + cpython.Py_INCREF(elem) + + # Iterate over array dimentions and put the element in + # the correctly nested sublist. + for j in reversed(range(ndims)): + if indexes[j] == 0: + # Allocate the list for this array level. + stride = cpython.PyList_New(dims[j]) + + strides[j] = <void*><cpython.PyObject>stride + # Take an explicit reference for PyList_SET_ITEM below + # expects this. + cpython.Py_INCREF(stride) + + stride = <object><cpython.PyObject*>strides[j] + cpython.PyList_SET_ITEM(stride, indexes[j], elem) + indexes[j] += 1 + + if indexes[j] == dims[j] and j != 0: + # This array level is full, continue the + # ascent in the dimensions so that this level + # sublist will be appened to the parent list. + elem = stride + # Reset the index, this will cause the + # new list to be allocated on the next + # iteration on this array axis. + indexes[j] = 0 + else: + break + + stride = <object><cpython.PyObject*>strides[0] + # Since each element in strides has a refcount of 1, + # returning strides[0] will increment it to 2, so + # balance that. + cpython.Py_DECREF(stride) + return stride + + +cdef textarray_decode(ConnectionSettings settings, FRBuffer *buf, + decode_func_ex decoder, const void *decoder_arg, + Py_UCS4 typdelim): + cdef: + Py_UCS4 *array_text + str s + + # Make a copy of array data since we will be mutating it for + # the purposes of element decoding. + s = pgproto.text_decode(settings, buf) + array_text = cpythonx.PyUnicode_AsUCS4Copy(s) + + try: + return _textarray_decode( + settings, array_text, decoder, decoder_arg, typdelim) + except ValueError as e: + raise exceptions.ProtocolError( + 'malformed array literal {!r}: {}'.format(s, e.args[0])) + finally: + cpython.PyMem_Free(array_text) + + +cdef _textarray_decode(ConnectionSettings settings, + Py_UCS4 *array_text, + decode_func_ex decoder, + const void *decoder_arg, + Py_UCS4 typdelim): + + cdef: + bytearray array_bytes + list result + list new_stride + Py_UCS4 *ptr + int32_t ndims = 0 + int32_t ubound = 0 + int32_t lbound = 0 + int32_t dims[ARRAY_MAXDIM] + int32_t inferred_dims[ARRAY_MAXDIM] + int32_t inferred_ndims = 0 + void *strides[ARRAY_MAXDIM] + int32_t indexes[ARRAY_MAXDIM] + int32_t nest_level = 0 + int32_t item_level = 0 + bint end_of_array = False + + bint end_of_item = False + bint has_quoting = False + bint strip_spaces = False + bint in_quotes = False + Py_UCS4 *item_start + Py_UCS4 *item_ptr + Py_UCS4 *item_end + + int i + object item + str item_text + FRBuffer item_buf + char *pg_item_str + ssize_t pg_item_len + + ptr = array_text + + while True: + while apg_ascii_isspace(ptr[0]): + ptr += 1 + + if ptr[0] != '[': + # Finished parsing dimensions spec. + break + + ptr += 1 # '[' + + if ndims > ARRAY_MAXDIM: + raise ValueError( + 'number of array dimensions ({}) exceed the ' + 'maximum expected ({})'.format(ndims, ARRAY_MAXDIM)) + + ptr = apg_parse_int32(ptr, &ubound) + if ptr == NULL: + raise ValueError('missing array dimension value') + + if ptr[0] == ':': + ptr += 1 + lbound = ubound + + # [lower:upper] spec. We disregard the lbound for decoding. + ptr = apg_parse_int32(ptr, &ubound) + if ptr == NULL: + raise ValueError('missing array dimension value') + else: + lbound = 1 + + if ptr[0] != ']': + raise ValueError('missing \']\' after array dimensions') + + ptr += 1 # ']' + + dims[ndims] = ubound - lbound + 1 + ndims += 1 + + if ndims != 0: + # If dimensions were given, the '=' token is expected. + if ptr[0] != '=': + raise ValueError('missing \'=\' after array dimensions') + + ptr += 1 # '=' + + # Skip any whitespace after the '=', whitespace + # before was consumed in the above loop. + while apg_ascii_isspace(ptr[0]): + ptr += 1 + + # Infer the dimensions from the brace structure in the + # array literal body, and check that it matches the explicit + # spec. This also validates that the array literal is sane. + _infer_array_dims(ptr, typdelim, inferred_dims, &inferred_ndims) + + if inferred_ndims != ndims: + raise ValueError( + 'specified array dimensions do not match array content') + + for i in range(ndims): + if inferred_dims[i] != dims[i]: + raise ValueError( + 'specified array dimensions do not match array content') + else: + # Infer the dimensions from the brace structure in the array literal + # body. This also validates that the array literal is sane. + _infer_array_dims(ptr, typdelim, dims, &ndims) + + while not end_of_array: + # We iterate over the literal character by character + # and modify the string in-place removing the array-specific + # quoting and determining the boundaries of each element. + end_of_item = has_quoting = in_quotes = False + strip_spaces = True + + # Pointers to array element start, end, and the current pointer + # tracking the position where characters are written when + # escaping is folded. + item_start = item_end = item_ptr = ptr + item_level = 0 + + while not end_of_item: + if ptr[0] == '"': + in_quotes = not in_quotes + if in_quotes: + strip_spaces = False + else: + item_end = item_ptr + has_quoting = True + + elif ptr[0] == '\\': + # Quoted character, collapse the backslash. + ptr += 1 + has_quoting = True + item_ptr[0] = ptr[0] + item_ptr += 1 + strip_spaces = False + item_end = item_ptr + + elif in_quotes: + # Consume the string until we see the closing quote. + item_ptr[0] = ptr[0] + item_ptr += 1 + + elif ptr[0] == '{': + # Nesting level increase. + nest_level += 1 + + indexes[nest_level - 1] = 0 + new_stride = cpython.PyList_New(dims[nest_level - 1]) + strides[nest_level - 1] = \ + <void*>(<cpython.PyObject>new_stride) + + if nest_level > 1: + cpython.Py_INCREF(new_stride) + cpython.PyList_SET_ITEM( + <object><cpython.PyObject*>strides[nest_level - 2], + indexes[nest_level - 2], + new_stride) + else: + result = new_stride + + elif ptr[0] == '}': + if item_level == 0: + # Make sure we keep track of which nesting + # level the item belongs to, as the loop + # will continue to consume closing braces + # until the delimiter or the end of input. + item_level = nest_level + + nest_level -= 1 + + if nest_level == 0: + end_of_array = end_of_item = True + + elif ptr[0] == typdelim: + # Array element delimiter, + end_of_item = True + if item_level == 0: + item_level = nest_level + + elif apg_ascii_isspace(ptr[0]): + if not strip_spaces: + item_ptr[0] = ptr[0] + item_ptr += 1 + # Ignore the leading literal whitespace. + + else: + item_ptr[0] = ptr[0] + item_ptr += 1 + strip_spaces = False + item_end = item_ptr + + ptr += 1 + + # end while not end_of_item + + if item_end == item_start: + # Empty array + continue + + item_end[0] = '\0' + + if not has_quoting and apg_strcasecmp(item_start, APG_NULL) == 0: + # NULL element. + item = None + else: + # XXX: find a way to avoid the redundant encode/decode + # cycle here. + item_text = cpythonx.PyUnicode_FromKindAndData( + cpythonx.PyUnicode_4BYTE_KIND, + <void *>item_start, + item_end - item_start) + + # Prepare the element buffer and call the text decoder + # for the element type. + pgproto.as_pg_string_and_size( + settings, item_text, &pg_item_str, &pg_item_len) + frb_init(&item_buf, pg_item_str, pg_item_len) + item = decoder(settings, &item_buf, decoder_arg) + + # Place the decoded element in the array. + cpython.Py_INCREF(item) + cpython.PyList_SET_ITEM( + <object><cpython.PyObject*>strides[item_level - 1], + indexes[item_level - 1], + item) + + if nest_level > 0: + indexes[nest_level - 1] += 1 + + return result + + +cdef enum _ArrayParseState: + APS_START = 1 + APS_STRIDE_STARTED = 2 + APS_STRIDE_DONE = 3 + APS_STRIDE_DELIMITED = 4 + APS_ELEM_STARTED = 5 + APS_ELEM_DELIMITED = 6 + + +cdef _UnexpectedCharacter(const Py_UCS4 *array_text, const Py_UCS4 *ptr): + return ValueError('unexpected character {!r} at position {}'.format( + cpython.PyUnicode_FromOrdinal(<int>ptr[0]), ptr - array_text + 1)) + + +cdef _infer_array_dims(const Py_UCS4 *array_text, + Py_UCS4 typdelim, + int32_t *dims, + int32_t *ndims): + cdef: + const Py_UCS4 *ptr = array_text + int i + int nest_level = 0 + bint end_of_array = False + bint end_of_item = False + bint in_quotes = False + bint array_is_empty = True + int stride_len[ARRAY_MAXDIM] + int prev_stride_len[ARRAY_MAXDIM] + _ArrayParseState parse_state = APS_START + + for i in range(ARRAY_MAXDIM): + dims[i] = prev_stride_len[i] = 0 + stride_len[i] = 1 + + while not end_of_array: + end_of_item = False + + while not end_of_item: + if ptr[0] == '\0': + raise ValueError('unexpected end of string') + + elif ptr[0] == '"': + if (parse_state not in (APS_STRIDE_STARTED, + APS_ELEM_DELIMITED) and + not (parse_state == APS_ELEM_STARTED and in_quotes)): + raise _UnexpectedCharacter(array_text, ptr) + + in_quotes = not in_quotes + if in_quotes: + parse_state = APS_ELEM_STARTED + array_is_empty = False + + elif ptr[0] == '\\': + if parse_state not in (APS_STRIDE_STARTED, + APS_ELEM_STARTED, + APS_ELEM_DELIMITED): + raise _UnexpectedCharacter(array_text, ptr) + + parse_state = APS_ELEM_STARTED + array_is_empty = False + + if ptr[1] != '\0': + ptr += 1 + else: + raise ValueError('unexpected end of string') + + elif in_quotes: + # Ignore everything inside the quotes. + pass + + elif ptr[0] == '{': + if parse_state not in (APS_START, + APS_STRIDE_STARTED, + APS_STRIDE_DELIMITED): + raise _UnexpectedCharacter(array_text, ptr) + + parse_state = APS_STRIDE_STARTED + if nest_level >= ARRAY_MAXDIM: + raise ValueError( + 'number of array dimensions ({}) exceed the ' + 'maximum expected ({})'.format( + nest_level, ARRAY_MAXDIM)) + + dims[nest_level] = 0 + nest_level += 1 + if ndims[0] < nest_level: + ndims[0] = nest_level + + elif ptr[0] == '}': + if (parse_state not in (APS_ELEM_STARTED, APS_STRIDE_DONE) and + not (nest_level == 1 and + parse_state == APS_STRIDE_STARTED)): + raise _UnexpectedCharacter(array_text, ptr) + + parse_state = APS_STRIDE_DONE + + if nest_level == 0: + raise _UnexpectedCharacter(array_text, ptr) + + nest_level -= 1 + + if (prev_stride_len[nest_level] != 0 and + stride_len[nest_level] != prev_stride_len[nest_level]): + raise ValueError( + 'inconsistent sub-array dimensions' + ' at position {}'.format( + ptr - array_text + 1)) + + prev_stride_len[nest_level] = stride_len[nest_level] + stride_len[nest_level] = 1 + if nest_level == 0: + end_of_array = end_of_item = True + else: + dims[nest_level - 1] += 1 + + elif ptr[0] == typdelim: + if parse_state not in (APS_ELEM_STARTED, APS_STRIDE_DONE): + raise _UnexpectedCharacter(array_text, ptr) + + if parse_state == APS_STRIDE_DONE: + parse_state = APS_STRIDE_DELIMITED + else: + parse_state = APS_ELEM_DELIMITED + end_of_item = True + stride_len[nest_level - 1] += 1 + + elif not apg_ascii_isspace(ptr[0]): + if parse_state not in (APS_STRIDE_STARTED, + APS_ELEM_STARTED, + APS_ELEM_DELIMITED): + raise _UnexpectedCharacter(array_text, ptr) + + parse_state = APS_ELEM_STARTED + array_is_empty = False + + if not end_of_item: + ptr += 1 + + if not array_is_empty: + dims[ndims[0] - 1] += 1 + + ptr += 1 + + # only whitespace is allowed after the closing brace + while ptr[0] != '\0': + if not apg_ascii_isspace(ptr[0]): + raise _UnexpectedCharacter(array_text, ptr) + + ptr += 1 + + if array_is_empty: + ndims[0] = 0 + + +cdef uint4_encode_ex(ConnectionSettings settings, WriteBuffer buf, object obj, + const void *arg): + return pgproto.uint4_encode(settings, buf, obj) + + +cdef uint4_decode_ex(ConnectionSettings settings, FRBuffer *buf, + const void *arg): + return pgproto.uint4_decode(settings, buf) + + +cdef arrayoid_encode(ConnectionSettings settings, WriteBuffer buf, items): + array_encode(settings, buf, items, OIDOID, + <encode_func_ex>&uint4_encode_ex, NULL) + + +cdef arrayoid_decode(ConnectionSettings settings, FRBuffer *buf): + return array_decode(settings, buf, <decode_func_ex>&uint4_decode_ex, NULL) + + +cdef text_encode_ex(ConnectionSettings settings, WriteBuffer buf, object obj, + const void *arg): + return pgproto.text_encode(settings, buf, obj) + + +cdef text_decode_ex(ConnectionSettings settings, FRBuffer *buf, + const void *arg): + return pgproto.text_decode(settings, buf) + + +cdef arraytext_encode(ConnectionSettings settings, WriteBuffer buf, items): + array_encode(settings, buf, items, TEXTOID, + <encode_func_ex>&text_encode_ex, NULL) + + +cdef arraytext_decode(ConnectionSettings settings, FRBuffer *buf): + return array_decode(settings, buf, <decode_func_ex>&text_decode_ex, NULL) + + +cdef init_array_codecs(): + # oid[] and text[] are registered as core codecs + # to make type introspection query work + # + register_core_codec(_OIDOID, + <encode_func>&arrayoid_encode, + <decode_func>&arrayoid_decode, + PG_FORMAT_BINARY) + + register_core_codec(_TEXTOID, + <encode_func>&arraytext_encode, + <decode_func>&arraytext_decode, + PG_FORMAT_BINARY) + +init_array_codecs() diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pxd b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pxd new file mode 100644 index 00000000..1cfed833 --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pxd @@ -0,0 +1,187 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +ctypedef object (*encode_func)(ConnectionSettings settings, + WriteBuffer buf, + object obj) + +ctypedef object (*decode_func)(ConnectionSettings settings, + FRBuffer *buf) + +ctypedef object (*codec_encode_func)(Codec codec, + ConnectionSettings settings, + WriteBuffer buf, + object obj) + +ctypedef object (*codec_decode_func)(Codec codec, + ConnectionSettings settings, + FRBuffer *buf) + + +cdef enum CodecType: + CODEC_UNDEFINED = 0 + CODEC_C = 1 + CODEC_PY = 2 + CODEC_ARRAY = 3 + CODEC_COMPOSITE = 4 + CODEC_RANGE = 5 + CODEC_MULTIRANGE = 6 + + +cdef enum ServerDataFormat: + PG_FORMAT_ANY = -1 + PG_FORMAT_TEXT = 0 + PG_FORMAT_BINARY = 1 + + +cdef enum ClientExchangeFormat: + PG_XFORMAT_OBJECT = 1 + PG_XFORMAT_TUPLE = 2 + + +cdef class Codec: + cdef: + uint32_t oid + + str name + str schema + str kind + + CodecType type + ServerDataFormat format + ClientExchangeFormat xformat + + encode_func c_encoder + decode_func c_decoder + Codec base_codec + + object py_encoder + object py_decoder + + # arrays + Codec element_codec + Py_UCS4 element_delimiter + + # composite types + tuple element_type_oids + object element_names + object record_desc + list element_codecs + + # Pointers to actual encoder/decoder functions for this codec + codec_encode_func encoder + codec_decode_func decoder + + cdef init(self, str name, str schema, str kind, + CodecType type, ServerDataFormat format, + ClientExchangeFormat xformat, + encode_func c_encoder, decode_func c_decoder, + Codec base_codec, + object py_encoder, object py_decoder, + Codec element_codec, tuple element_type_oids, + object element_names, list element_codecs, + Py_UCS4 element_delimiter) + + cdef encode_scalar(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_array(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_array_text(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_range(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_multirange(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_composite(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef encode_in_python(self, ConnectionSettings settings, WriteBuffer buf, + object obj) + + cdef decode_scalar(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_array(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_array_text(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_range(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_multirange(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_composite(self, ConnectionSettings settings, FRBuffer *buf) + + cdef decode_in_python(self, ConnectionSettings settings, FRBuffer *buf) + + cdef inline encode(self, + ConnectionSettings settings, + WriteBuffer buf, + object obj) + + cdef inline decode(self, ConnectionSettings settings, FRBuffer *buf) + + cdef has_encoder(self) + cdef has_decoder(self) + cdef is_binary(self) + + cdef inline Codec copy(self) + + @staticmethod + cdef Codec new_array_codec(uint32_t oid, + str name, + str schema, + Codec element_codec, + Py_UCS4 element_delimiter) + + @staticmethod + cdef Codec new_range_codec(uint32_t oid, + str name, + str schema, + Codec element_codec) + + @staticmethod + cdef Codec new_multirange_codec(uint32_t oid, + str name, + str schema, + Codec element_codec) + + @staticmethod + cdef Codec new_composite_codec(uint32_t oid, + str name, + str schema, + ServerDataFormat format, + list element_codecs, + tuple element_type_oids, + object element_names) + + @staticmethod + cdef Codec new_python_codec(uint32_t oid, + str name, + str schema, + str kind, + object encoder, + object decoder, + encode_func c_encoder, + decode_func c_decoder, + Codec base_codec, + ServerDataFormat format, + ClientExchangeFormat xformat) + + +cdef class DataCodecConfig: + cdef: + dict _derived_type_codecs + dict _custom_type_codecs + + cdef inline Codec get_codec(self, uint32_t oid, ServerDataFormat format, + bint ignore_custom_codec=*) + cdef inline Codec get_custom_codec(self, uint32_t oid, + ServerDataFormat format) diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pyx new file mode 100644 index 00000000..c269e374 --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/base.pyx @@ -0,0 +1,895 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +from collections.abc import Mapping as MappingABC + +import asyncpg +from asyncpg import exceptions + + +cdef void* binary_codec_map[(MAXSUPPORTEDOID + 1) * 2] +cdef void* text_codec_map[(MAXSUPPORTEDOID + 1) * 2] +cdef dict EXTRA_CODECS = {} + + +@cython.final +cdef class Codec: + + def __cinit__(self, uint32_t oid): + self.oid = oid + self.type = CODEC_UNDEFINED + + cdef init( + self, + str name, + str schema, + str kind, + CodecType type, + ServerDataFormat format, + ClientExchangeFormat xformat, + encode_func c_encoder, + decode_func c_decoder, + Codec base_codec, + object py_encoder, + object py_decoder, + Codec element_codec, + tuple element_type_oids, + object element_names, + list element_codecs, + Py_UCS4 element_delimiter, + ): + + self.name = name + self.schema = schema + self.kind = kind + self.type = type + self.format = format + self.xformat = xformat + self.c_encoder = c_encoder + self.c_decoder = c_decoder + self.base_codec = base_codec + self.py_encoder = py_encoder + self.py_decoder = py_decoder + self.element_codec = element_codec + self.element_type_oids = element_type_oids + self.element_codecs = element_codecs + self.element_delimiter = element_delimiter + self.element_names = element_names + + if base_codec is not None: + if c_encoder != NULL or c_decoder != NULL: + raise exceptions.InternalClientError( + 'base_codec is mutually exclusive with c_encoder/c_decoder' + ) + + if element_names is not None: + self.record_desc = record.ApgRecordDesc_New( + element_names, tuple(element_names)) + else: + self.record_desc = None + + if type == CODEC_C: + self.encoder = <codec_encode_func>&self.encode_scalar + self.decoder = <codec_decode_func>&self.decode_scalar + elif type == CODEC_ARRAY: + if format == PG_FORMAT_BINARY: + self.encoder = <codec_encode_func>&self.encode_array + self.decoder = <codec_decode_func>&self.decode_array + else: + self.encoder = <codec_encode_func>&self.encode_array_text + self.decoder = <codec_decode_func>&self.decode_array_text + elif type == CODEC_RANGE: + if format != PG_FORMAT_BINARY: + raise exceptions.UnsupportedClientFeatureError( + 'cannot decode type "{}"."{}": text encoding of ' + 'range types is not supported'.format(schema, name)) + self.encoder = <codec_encode_func>&self.encode_range + self.decoder = <codec_decode_func>&self.decode_range + elif type == CODEC_MULTIRANGE: + if format != PG_FORMAT_BINARY: + raise exceptions.UnsupportedClientFeatureError( + 'cannot decode type "{}"."{}": text encoding of ' + 'range types is not supported'.format(schema, name)) + self.encoder = <codec_encode_func>&self.encode_multirange + self.decoder = <codec_decode_func>&self.decode_multirange + elif type == CODEC_COMPOSITE: + if format != PG_FORMAT_BINARY: + raise exceptions.UnsupportedClientFeatureError( + 'cannot decode type "{}"."{}": text encoding of ' + 'composite types is not supported'.format(schema, name)) + self.encoder = <codec_encode_func>&self.encode_composite + self.decoder = <codec_decode_func>&self.decode_composite + elif type == CODEC_PY: + self.encoder = <codec_encode_func>&self.encode_in_python + self.decoder = <codec_decode_func>&self.decode_in_python + else: + raise exceptions.InternalClientError( + 'unexpected codec type: {}'.format(type)) + + cdef Codec copy(self): + cdef Codec codec + + codec = Codec(self.oid) + codec.init(self.name, self.schema, self.kind, + self.type, self.format, self.xformat, + self.c_encoder, self.c_decoder, self.base_codec, + self.py_encoder, self.py_decoder, + self.element_codec, + self.element_type_oids, self.element_names, + self.element_codecs, self.element_delimiter) + + return codec + + cdef encode_scalar(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + self.c_encoder(settings, buf, obj) + + cdef encode_array(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + array_encode(settings, buf, obj, self.element_codec.oid, + codec_encode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef encode_array_text(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + return textarray_encode(settings, buf, obj, + codec_encode_func_ex, + <void*>(<cpython.PyObject>self.element_codec), + self.element_delimiter) + + cdef encode_range(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + range_encode(settings, buf, obj, self.element_codec.oid, + codec_encode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef encode_multirange(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + multirange_encode(settings, buf, obj, self.element_codec.oid, + codec_encode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef encode_composite(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + cdef: + WriteBuffer elem_data + int i + list elem_codecs = self.element_codecs + ssize_t count + ssize_t composite_size + tuple rec + + if isinstance(obj, MappingABC): + # Input is dict-like, form a tuple + composite_size = len(self.element_type_oids) + rec = cpython.PyTuple_New(composite_size) + + for i in range(composite_size): + cpython.Py_INCREF(None) + cpython.PyTuple_SET_ITEM(rec, i, None) + + for field in obj: + try: + i = self.element_names[field] + except KeyError: + raise ValueError( + '{!r} is not a valid element of composite ' + 'type {}'.format(field, self.name)) from None + + item = obj[field] + cpython.Py_INCREF(item) + cpython.PyTuple_SET_ITEM(rec, i, item) + + obj = rec + + count = len(obj) + if count > _MAXINT32: + raise ValueError('too many elements in composite type record') + + elem_data = WriteBuffer.new() + i = 0 + for item in obj: + elem_data.write_int32(<int32_t>self.element_type_oids[i]) + if item is None: + elem_data.write_int32(-1) + else: + (<Codec>elem_codecs[i]).encode(settings, elem_data, item) + i += 1 + + record_encode_frame(settings, buf, elem_data, <int32_t>count) + + cdef encode_in_python(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + data = self.py_encoder(obj) + if self.xformat == PG_XFORMAT_OBJECT: + if self.format == PG_FORMAT_BINARY: + pgproto.bytea_encode(settings, buf, data) + elif self.format == PG_FORMAT_TEXT: + pgproto.text_encode(settings, buf, data) + else: + raise exceptions.InternalClientError( + 'unexpected data format: {}'.format(self.format)) + elif self.xformat == PG_XFORMAT_TUPLE: + if self.base_codec is not None: + self.base_codec.encode(settings, buf, data) + else: + self.c_encoder(settings, buf, data) + else: + raise exceptions.InternalClientError( + 'unexpected exchange format: {}'.format(self.xformat)) + + cdef encode(self, ConnectionSettings settings, WriteBuffer buf, + object obj): + return self.encoder(self, settings, buf, obj) + + cdef decode_scalar(self, ConnectionSettings settings, FRBuffer *buf): + return self.c_decoder(settings, buf) + + cdef decode_array(self, ConnectionSettings settings, FRBuffer *buf): + return array_decode(settings, buf, codec_decode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef decode_array_text(self, ConnectionSettings settings, + FRBuffer *buf): + return textarray_decode(settings, buf, codec_decode_func_ex, + <void*>(<cpython.PyObject>self.element_codec), + self.element_delimiter) + + cdef decode_range(self, ConnectionSettings settings, FRBuffer *buf): + return range_decode(settings, buf, codec_decode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef decode_multirange(self, ConnectionSettings settings, FRBuffer *buf): + return multirange_decode(settings, buf, codec_decode_func_ex, + <void*>(<cpython.PyObject>self.element_codec)) + + cdef decode_composite(self, ConnectionSettings settings, + FRBuffer *buf): + cdef: + object result + ssize_t elem_count + ssize_t i + int32_t elem_len + uint32_t elem_typ + uint32_t received_elem_typ + Codec elem_codec + FRBuffer elem_buf + + elem_count = <ssize_t><uint32_t>hton.unpack_int32(frb_read(buf, 4)) + if elem_count != len(self.element_type_oids): + raise exceptions.OutdatedSchemaCacheError( + 'unexpected number of attributes of composite type: ' + '{}, expected {}' + .format( + elem_count, + len(self.element_type_oids), + ), + schema=self.schema, + data_type=self.name, + ) + result = record.ApgRecord_New(asyncpg.Record, self.record_desc, elem_count) + for i in range(elem_count): + elem_typ = self.element_type_oids[i] + received_elem_typ = <uint32_t>hton.unpack_int32(frb_read(buf, 4)) + + if received_elem_typ != elem_typ: + raise exceptions.OutdatedSchemaCacheError( + 'unexpected data type of composite type attribute {}: ' + '{!r}, expected {!r}' + .format( + i, + BUILTIN_TYPE_OID_MAP.get( + received_elem_typ, received_elem_typ), + BUILTIN_TYPE_OID_MAP.get( + elem_typ, elem_typ) + ), + schema=self.schema, + data_type=self.name, + position=i, + ) + + elem_len = hton.unpack_int32(frb_read(buf, 4)) + if elem_len == -1: + elem = None + else: + elem_codec = self.element_codecs[i] + elem = elem_codec.decode( + settings, frb_slice_from(&elem_buf, buf, elem_len)) + + cpython.Py_INCREF(elem) + record.ApgRecord_SET_ITEM(result, i, elem) + + return result + + cdef decode_in_python(self, ConnectionSettings settings, + FRBuffer *buf): + if self.xformat == PG_XFORMAT_OBJECT: + if self.format == PG_FORMAT_BINARY: + data = pgproto.bytea_decode(settings, buf) + elif self.format == PG_FORMAT_TEXT: + data = pgproto.text_decode(settings, buf) + else: + raise exceptions.InternalClientError( + 'unexpected data format: {}'.format(self.format)) + elif self.xformat == PG_XFORMAT_TUPLE: + if self.base_codec is not None: + data = self.base_codec.decode(settings, buf) + else: + data = self.c_decoder(settings, buf) + else: + raise exceptions.InternalClientError( + 'unexpected exchange format: {}'.format(self.xformat)) + + return self.py_decoder(data) + + cdef inline decode(self, ConnectionSettings settings, FRBuffer *buf): + return self.decoder(self, settings, buf) + + cdef inline has_encoder(self): + cdef Codec elem_codec + + if self.c_encoder is not NULL or self.py_encoder is not None: + return True + + elif ( + self.type == CODEC_ARRAY + or self.type == CODEC_RANGE + or self.type == CODEC_MULTIRANGE + ): + return self.element_codec.has_encoder() + + elif self.type == CODEC_COMPOSITE: + for elem_codec in self.element_codecs: + if not elem_codec.has_encoder(): + return False + return True + + else: + return False + + cdef has_decoder(self): + cdef Codec elem_codec + + if self.c_decoder is not NULL or self.py_decoder is not None: + return True + + elif ( + self.type == CODEC_ARRAY + or self.type == CODEC_RANGE + or self.type == CODEC_MULTIRANGE + ): + return self.element_codec.has_decoder() + + elif self.type == CODEC_COMPOSITE: + for elem_codec in self.element_codecs: + if not elem_codec.has_decoder(): + return False + return True + + else: + return False + + cdef is_binary(self): + return self.format == PG_FORMAT_BINARY + + def __repr__(self): + return '<Codec oid={} elem_oid={} core={}>'.format( + self.oid, + 'NA' if self.element_codec is None else self.element_codec.oid, + has_core_codec(self.oid)) + + @staticmethod + cdef Codec new_array_codec(uint32_t oid, + str name, + str schema, + Codec element_codec, + Py_UCS4 element_delimiter): + cdef Codec codec + codec = Codec(oid) + codec.init(name, schema, 'array', CODEC_ARRAY, element_codec.format, + PG_XFORMAT_OBJECT, NULL, NULL, None, None, None, + element_codec, None, None, None, element_delimiter) + return codec + + @staticmethod + cdef Codec new_range_codec(uint32_t oid, + str name, + str schema, + Codec element_codec): + cdef Codec codec + codec = Codec(oid) + codec.init(name, schema, 'range', CODEC_RANGE, element_codec.format, + PG_XFORMAT_OBJECT, NULL, NULL, None, None, None, + element_codec, None, None, None, 0) + return codec + + @staticmethod + cdef Codec new_multirange_codec(uint32_t oid, + str name, + str schema, + Codec element_codec): + cdef Codec codec + codec = Codec(oid) + codec.init(name, schema, 'multirange', CODEC_MULTIRANGE, + element_codec.format, PG_XFORMAT_OBJECT, NULL, NULL, None, + None, None, element_codec, None, None, None, 0) + return codec + + @staticmethod + cdef Codec new_composite_codec(uint32_t oid, + str name, + str schema, + ServerDataFormat format, + list element_codecs, + tuple element_type_oids, + object element_names): + cdef Codec codec + codec = Codec(oid) + codec.init(name, schema, 'composite', CODEC_COMPOSITE, + format, PG_XFORMAT_OBJECT, NULL, NULL, None, None, None, + None, element_type_oids, element_names, element_codecs, 0) + return codec + + @staticmethod + cdef Codec new_python_codec(uint32_t oid, + str name, + str schema, + str kind, + object encoder, + object decoder, + encode_func c_encoder, + decode_func c_decoder, + Codec base_codec, + ServerDataFormat format, + ClientExchangeFormat xformat): + cdef Codec codec + codec = Codec(oid) + codec.init(name, schema, kind, CODEC_PY, format, xformat, + c_encoder, c_decoder, base_codec, encoder, decoder, + None, None, None, None, 0) + return codec + + +# Encode callback for arrays +cdef codec_encode_func_ex(ConnectionSettings settings, WriteBuffer buf, + object obj, const void *arg): + return (<Codec>arg).encode(settings, buf, obj) + + +# Decode callback for arrays +cdef codec_decode_func_ex(ConnectionSettings settings, FRBuffer *buf, + const void *arg): + return (<Codec>arg).decode(settings, buf) + + +cdef uint32_t pylong_as_oid(val) except? 0xFFFFFFFFl: + cdef: + int64_t oid = 0 + bint overflow = False + + try: + oid = cpython.PyLong_AsLongLong(val) + except OverflowError: + overflow = True + + if overflow or (oid < 0 or oid > UINT32_MAX): + raise OverflowError('OID value too large: {!r}'.format(val)) + + return <uint32_t>val + + +cdef class DataCodecConfig: + def __init__(self, cache_key): + # Codec instance cache for derived types: + # composites, arrays, ranges, domains and their combinations. + self._derived_type_codecs = {} + # Codec instances set up by the user for the connection. + self._custom_type_codecs = {} + + def add_types(self, types): + cdef: + Codec elem_codec + list comp_elem_codecs + ServerDataFormat format + ServerDataFormat elem_format + bint has_text_elements + Py_UCS4 elem_delim + + for ti in types: + oid = ti['oid'] + + if self.get_codec(oid, PG_FORMAT_ANY) is not None: + continue + + name = ti['name'] + schema = ti['ns'] + array_element_oid = ti['elemtype'] + range_subtype_oid = ti['range_subtype'] + if ti['attrtypoids']: + comp_type_attrs = tuple(ti['attrtypoids']) + else: + comp_type_attrs = None + base_type = ti['basetype'] + + if array_element_oid: + # Array type (note, there is no separate 'kind' for arrays) + + # Canonicalize type name to "elemtype[]" + if name.startswith('_'): + name = name[1:] + name = '{}[]'.format(name) + + elem_codec = self.get_codec(array_element_oid, PG_FORMAT_ANY) + if elem_codec is None: + elem_codec = self.declare_fallback_codec( + array_element_oid, ti['elemtype_name'], schema) + + elem_delim = <Py_UCS4>ti['elemdelim'][0] + + self._derived_type_codecs[oid, elem_codec.format] = \ + Codec.new_array_codec( + oid, name, schema, elem_codec, elem_delim) + + elif ti['kind'] == b'c': + # Composite type + + if not comp_type_attrs: + raise exceptions.InternalClientError( + f'type record missing field types for composite {oid}') + + comp_elem_codecs = [] + has_text_elements = False + + for typoid in comp_type_attrs: + elem_codec = self.get_codec(typoid, PG_FORMAT_ANY) + if elem_codec is None: + raise exceptions.InternalClientError( + f'no codec for composite attribute type {typoid}') + if elem_codec.format is PG_FORMAT_TEXT: + has_text_elements = True + comp_elem_codecs.append(elem_codec) + + element_names = collections.OrderedDict() + for i, attrname in enumerate(ti['attrnames']): + element_names[attrname] = i + + # If at least one element is text-encoded, we must + # encode the whole composite as text. + if has_text_elements: + elem_format = PG_FORMAT_TEXT + else: + elem_format = PG_FORMAT_BINARY + + self._derived_type_codecs[oid, elem_format] = \ + Codec.new_composite_codec( + oid, name, schema, elem_format, comp_elem_codecs, + comp_type_attrs, element_names) + + elif ti['kind'] == b'd': + # Domain type + + if not base_type: + raise exceptions.InternalClientError( + f'type record missing base type for domain {oid}') + + elem_codec = self.get_codec(base_type, PG_FORMAT_ANY) + if elem_codec is None: + elem_codec = self.declare_fallback_codec( + base_type, ti['basetype_name'], schema) + + self._derived_type_codecs[oid, elem_codec.format] = elem_codec + + elif ti['kind'] == b'r': + # Range type + + if not range_subtype_oid: + raise exceptions.InternalClientError( + f'type record missing base type for range {oid}') + + elem_codec = self.get_codec(range_subtype_oid, PG_FORMAT_ANY) + if elem_codec is None: + elem_codec = self.declare_fallback_codec( + range_subtype_oid, ti['range_subtype_name'], schema) + + self._derived_type_codecs[oid, elem_codec.format] = \ + Codec.new_range_codec(oid, name, schema, elem_codec) + + elif ti['kind'] == b'm': + # Multirange type + + if not range_subtype_oid: + raise exceptions.InternalClientError( + f'type record missing base type for multirange {oid}') + + elem_codec = self.get_codec(range_subtype_oid, PG_FORMAT_ANY) + if elem_codec is None: + elem_codec = self.declare_fallback_codec( + range_subtype_oid, ti['range_subtype_name'], schema) + + self._derived_type_codecs[oid, elem_codec.format] = \ + Codec.new_multirange_codec(oid, name, schema, elem_codec) + + elif ti['kind'] == b'e': + # Enum types are essentially text + self._set_builtin_type_codec(oid, name, schema, 'scalar', + TEXTOID, PG_FORMAT_ANY) + else: + self.declare_fallback_codec(oid, name, schema) + + def add_python_codec(self, typeoid, typename, typeschema, typekind, + typeinfos, encoder, decoder, format, xformat): + cdef: + Codec core_codec = None + encode_func c_encoder = NULL + decode_func c_decoder = NULL + Codec base_codec = None + uint32_t oid = pylong_as_oid(typeoid) + bint codec_set = False + + # Clear all previous overrides (this also clears type cache). + self.remove_python_codec(typeoid, typename, typeschema) + + if typeinfos: + self.add_types(typeinfos) + + if format == PG_FORMAT_ANY: + formats = (PG_FORMAT_TEXT, PG_FORMAT_BINARY) + else: + formats = (format,) + + for fmt in formats: + if xformat == PG_XFORMAT_TUPLE: + if typekind == "scalar": + core_codec = get_core_codec(oid, fmt, xformat) + if core_codec is None: + continue + c_encoder = core_codec.c_encoder + c_decoder = core_codec.c_decoder + elif typekind == "composite": + base_codec = self.get_codec(oid, fmt) + if base_codec is None: + continue + + self._custom_type_codecs[typeoid, fmt] = \ + Codec.new_python_codec(oid, typename, typeschema, typekind, + encoder, decoder, c_encoder, c_decoder, + base_codec, fmt, xformat) + codec_set = True + + if not codec_set: + raise exceptions.InterfaceError( + "{} type does not support the 'tuple' exchange format".format( + typename)) + + def remove_python_codec(self, typeoid, typename, typeschema): + for fmt in (PG_FORMAT_BINARY, PG_FORMAT_TEXT): + self._custom_type_codecs.pop((typeoid, fmt), None) + self.clear_type_cache() + + def _set_builtin_type_codec(self, typeoid, typename, typeschema, typekind, + alias_to, format=PG_FORMAT_ANY): + cdef: + Codec codec + Codec target_codec + uint32_t oid = pylong_as_oid(typeoid) + uint32_t alias_oid = 0 + bint codec_set = False + + if format == PG_FORMAT_ANY: + formats = (PG_FORMAT_BINARY, PG_FORMAT_TEXT) + else: + formats = (format,) + + if isinstance(alias_to, int): + alias_oid = pylong_as_oid(alias_to) + else: + alias_oid = BUILTIN_TYPE_NAME_MAP.get(alias_to, 0) + + for format in formats: + if alias_oid != 0: + target_codec = self.get_codec(alias_oid, format) + else: + target_codec = get_extra_codec(alias_to, format) + + if target_codec is None: + continue + + codec = target_codec.copy() + codec.oid = typeoid + codec.name = typename + codec.schema = typeschema + codec.kind = typekind + + self._custom_type_codecs[typeoid, format] = codec + codec_set = True + + if not codec_set: + if format == PG_FORMAT_BINARY: + codec_str = 'binary' + elif format == PG_FORMAT_TEXT: + codec_str = 'text' + else: + codec_str = 'text or binary' + + raise exceptions.InterfaceError( + f'cannot alias {typename} to {alias_to}: ' + f'there is no {codec_str} codec for {alias_to}') + + def set_builtin_type_codec(self, typeoid, typename, typeschema, typekind, + alias_to, format=PG_FORMAT_ANY): + self._set_builtin_type_codec(typeoid, typename, typeschema, typekind, + alias_to, format) + self.clear_type_cache() + + def clear_type_cache(self): + self._derived_type_codecs.clear() + + def declare_fallback_codec(self, uint32_t oid, str name, str schema): + cdef Codec codec + + if oid <= MAXBUILTINOID: + # This is a BKI type, for which asyncpg has no + # defined codec. This should only happen for newly + # added builtin types, for which this version of + # asyncpg is lacking support. + # + raise exceptions.UnsupportedClientFeatureError( + f'unhandled standard data type {name!r} (OID {oid})') + else: + # This is a non-BKI type, and as such, has no + # stable OID, so no possibility of a builtin codec. + # In this case, fallback to text format. Applications + # can avoid this by specifying a codec for this type + # using Connection.set_type_codec(). + # + self._set_builtin_type_codec(oid, name, schema, 'scalar', + TEXTOID, PG_FORMAT_TEXT) + + codec = self.get_codec(oid, PG_FORMAT_TEXT) + + return codec + + cdef inline Codec get_codec(self, uint32_t oid, ServerDataFormat format, + bint ignore_custom_codec=False): + cdef Codec codec + + if format == PG_FORMAT_ANY: + codec = self.get_codec( + oid, PG_FORMAT_BINARY, ignore_custom_codec) + if codec is None: + codec = self.get_codec( + oid, PG_FORMAT_TEXT, ignore_custom_codec) + return codec + else: + if not ignore_custom_codec: + codec = self.get_custom_codec(oid, PG_FORMAT_ANY) + if codec is not None: + if codec.format != format: + # The codec for this OID has been overridden by + # set_{builtin}_type_codec with a different format. + # We must respect that and not return a core codec. + return None + else: + return codec + + codec = get_core_codec(oid, format) + if codec is not None: + return codec + else: + try: + return self._derived_type_codecs[oid, format] + except KeyError: + return None + + cdef inline Codec get_custom_codec( + self, + uint32_t oid, + ServerDataFormat format + ): + cdef Codec codec + + if format == PG_FORMAT_ANY: + codec = self.get_custom_codec(oid, PG_FORMAT_BINARY) + if codec is None: + codec = self.get_custom_codec(oid, PG_FORMAT_TEXT) + else: + codec = self._custom_type_codecs.get((oid, format)) + + return codec + + +cdef inline Codec get_core_codec( + uint32_t oid, ServerDataFormat format, + ClientExchangeFormat xformat=PG_XFORMAT_OBJECT): + cdef: + void *ptr = NULL + + if oid > MAXSUPPORTEDOID: + return None + if format == PG_FORMAT_BINARY: + ptr = binary_codec_map[oid * xformat] + elif format == PG_FORMAT_TEXT: + ptr = text_codec_map[oid * xformat] + + if ptr is NULL: + return None + else: + return <Codec>ptr + + +cdef inline Codec get_any_core_codec( + uint32_t oid, ServerDataFormat format, + ClientExchangeFormat xformat=PG_XFORMAT_OBJECT): + """A version of get_core_codec that accepts PG_FORMAT_ANY.""" + cdef: + Codec codec + + if format == PG_FORMAT_ANY: + codec = get_core_codec(oid, PG_FORMAT_BINARY, xformat) + if codec is None: + codec = get_core_codec(oid, PG_FORMAT_TEXT, xformat) + else: + codec = get_core_codec(oid, format, xformat) + + return codec + + +cdef inline int has_core_codec(uint32_t oid): + return binary_codec_map[oid] != NULL or text_codec_map[oid] != NULL + + +cdef register_core_codec(uint32_t oid, + encode_func encode, + decode_func decode, + ServerDataFormat format, + ClientExchangeFormat xformat=PG_XFORMAT_OBJECT): + + if oid > MAXSUPPORTEDOID: + raise exceptions.InternalClientError( + 'cannot register core codec for OID {}: it is greater ' + 'than MAXSUPPORTEDOID ({})'.format(oid, MAXSUPPORTEDOID)) + + cdef: + Codec codec + str name + str kind + + name = BUILTIN_TYPE_OID_MAP[oid] + kind = 'array' if oid in ARRAY_TYPES else 'scalar' + + codec = Codec(oid) + codec.init(name, 'pg_catalog', kind, CODEC_C, format, xformat, + encode, decode, None, None, None, None, None, None, None, 0) + cpython.Py_INCREF(codec) # immortalize + + if format == PG_FORMAT_BINARY: + binary_codec_map[oid * xformat] = <void*>codec + elif format == PG_FORMAT_TEXT: + text_codec_map[oid * xformat] = <void*>codec + else: + raise exceptions.InternalClientError( + 'invalid data format: {}'.format(format)) + + +cdef register_extra_codec(str name, + encode_func encode, + decode_func decode, + ServerDataFormat format): + cdef: + Codec codec + str kind + + kind = 'scalar' + + codec = Codec(INVALIDOID) + codec.init(name, None, kind, CODEC_C, format, PG_XFORMAT_OBJECT, + encode, decode, None, None, None, None, None, None, None, 0) + EXTRA_CODECS[name, format] = codec + + +cdef inline Codec get_extra_codec(str name, ServerDataFormat format): + return EXTRA_CODECS.get((name, format)) diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/pgproto.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/pgproto.pyx new file mode 100644 index 00000000..51d650d0 --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/pgproto.pyx @@ -0,0 +1,484 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +cdef init_bits_codecs(): + register_core_codec(BITOID, + <encode_func>pgproto.bits_encode, + <decode_func>pgproto.bits_decode, + PG_FORMAT_BINARY) + + register_core_codec(VARBITOID, + <encode_func>pgproto.bits_encode, + <decode_func>pgproto.bits_decode, + PG_FORMAT_BINARY) + + +cdef init_bytea_codecs(): + register_core_codec(BYTEAOID, + <encode_func>pgproto.bytea_encode, + <decode_func>pgproto.bytea_decode, + PG_FORMAT_BINARY) + + register_core_codec(CHAROID, + <encode_func>pgproto.bytea_encode, + <decode_func>pgproto.bytea_decode, + PG_FORMAT_BINARY) + + +cdef init_datetime_codecs(): + register_core_codec(DATEOID, + <encode_func>pgproto.date_encode, + <decode_func>pgproto.date_decode, + PG_FORMAT_BINARY) + + register_core_codec(DATEOID, + <encode_func>pgproto.date_encode_tuple, + <decode_func>pgproto.date_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + register_core_codec(TIMEOID, + <encode_func>pgproto.time_encode, + <decode_func>pgproto.time_decode, + PG_FORMAT_BINARY) + + register_core_codec(TIMEOID, + <encode_func>pgproto.time_encode_tuple, + <decode_func>pgproto.time_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + register_core_codec(TIMETZOID, + <encode_func>pgproto.timetz_encode, + <decode_func>pgproto.timetz_decode, + PG_FORMAT_BINARY) + + register_core_codec(TIMETZOID, + <encode_func>pgproto.timetz_encode_tuple, + <decode_func>pgproto.timetz_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + register_core_codec(TIMESTAMPOID, + <encode_func>pgproto.timestamp_encode, + <decode_func>pgproto.timestamp_decode, + PG_FORMAT_BINARY) + + register_core_codec(TIMESTAMPOID, + <encode_func>pgproto.timestamp_encode_tuple, + <decode_func>pgproto.timestamp_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + register_core_codec(TIMESTAMPTZOID, + <encode_func>pgproto.timestamptz_encode, + <decode_func>pgproto.timestamptz_decode, + PG_FORMAT_BINARY) + + register_core_codec(TIMESTAMPTZOID, + <encode_func>pgproto.timestamp_encode_tuple, + <decode_func>pgproto.timestamp_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + register_core_codec(INTERVALOID, + <encode_func>pgproto.interval_encode, + <decode_func>pgproto.interval_decode, + PG_FORMAT_BINARY) + + register_core_codec(INTERVALOID, + <encode_func>pgproto.interval_encode_tuple, + <decode_func>pgproto.interval_decode_tuple, + PG_FORMAT_BINARY, + PG_XFORMAT_TUPLE) + + # For obsolete abstime/reltime/tinterval, we do not bother to + # interpret the value, and simply return and pass it as text. + # + register_core_codec(ABSTIMEOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + register_core_codec(RELTIMEOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + register_core_codec(TINTERVALOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + +cdef init_float_codecs(): + register_core_codec(FLOAT4OID, + <encode_func>pgproto.float4_encode, + <decode_func>pgproto.float4_decode, + PG_FORMAT_BINARY) + + register_core_codec(FLOAT8OID, + <encode_func>pgproto.float8_encode, + <decode_func>pgproto.float8_decode, + PG_FORMAT_BINARY) + + +cdef init_geometry_codecs(): + register_core_codec(BOXOID, + <encode_func>pgproto.box_encode, + <decode_func>pgproto.box_decode, + PG_FORMAT_BINARY) + + register_core_codec(LINEOID, + <encode_func>pgproto.line_encode, + <decode_func>pgproto.line_decode, + PG_FORMAT_BINARY) + + register_core_codec(LSEGOID, + <encode_func>pgproto.lseg_encode, + <decode_func>pgproto.lseg_decode, + PG_FORMAT_BINARY) + + register_core_codec(POINTOID, + <encode_func>pgproto.point_encode, + <decode_func>pgproto.point_decode, + PG_FORMAT_BINARY) + + register_core_codec(PATHOID, + <encode_func>pgproto.path_encode, + <decode_func>pgproto.path_decode, + PG_FORMAT_BINARY) + + register_core_codec(POLYGONOID, + <encode_func>pgproto.poly_encode, + <decode_func>pgproto.poly_decode, + PG_FORMAT_BINARY) + + register_core_codec(CIRCLEOID, + <encode_func>pgproto.circle_encode, + <decode_func>pgproto.circle_decode, + PG_FORMAT_BINARY) + + +cdef init_hstore_codecs(): + register_extra_codec('pg_contrib.hstore', + <encode_func>pgproto.hstore_encode, + <decode_func>pgproto.hstore_decode, + PG_FORMAT_BINARY) + + +cdef init_json_codecs(): + register_core_codec(JSONOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_BINARY) + register_core_codec(JSONBOID, + <encode_func>pgproto.jsonb_encode, + <decode_func>pgproto.jsonb_decode, + PG_FORMAT_BINARY) + register_core_codec(JSONPATHOID, + <encode_func>pgproto.jsonpath_encode, + <decode_func>pgproto.jsonpath_decode, + PG_FORMAT_BINARY) + + +cdef init_int_codecs(): + + register_core_codec(BOOLOID, + <encode_func>pgproto.bool_encode, + <decode_func>pgproto.bool_decode, + PG_FORMAT_BINARY) + + register_core_codec(INT2OID, + <encode_func>pgproto.int2_encode, + <decode_func>pgproto.int2_decode, + PG_FORMAT_BINARY) + + register_core_codec(INT4OID, + <encode_func>pgproto.int4_encode, + <decode_func>pgproto.int4_decode, + PG_FORMAT_BINARY) + + register_core_codec(INT8OID, + <encode_func>pgproto.int8_encode, + <decode_func>pgproto.int8_decode, + PG_FORMAT_BINARY) + + +cdef init_pseudo_codecs(): + # Void type is returned by SELECT void_returning_function() + register_core_codec(VOIDOID, + <encode_func>pgproto.void_encode, + <decode_func>pgproto.void_decode, + PG_FORMAT_BINARY) + + # Unknown type, always decoded as text + register_core_codec(UNKNOWNOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # OID and friends + oid_types = [ + OIDOID, XIDOID, CIDOID + ] + + for oid_type in oid_types: + register_core_codec(oid_type, + <encode_func>pgproto.uint4_encode, + <decode_func>pgproto.uint4_decode, + PG_FORMAT_BINARY) + + # 64-bit OID types + oid8_types = [ + XID8OID, + ] + + for oid_type in oid8_types: + register_core_codec(oid_type, + <encode_func>pgproto.uint8_encode, + <decode_func>pgproto.uint8_decode, + PG_FORMAT_BINARY) + + # reg* types -- these are really system catalog OIDs, but + # allow the catalog object name as an input. We could just + # decode these as OIDs, but handling them as text seems more + # useful. + # + reg_types = [ + REGPROCOID, REGPROCEDUREOID, REGOPEROID, REGOPERATOROID, + REGCLASSOID, REGTYPEOID, REGCONFIGOID, REGDICTIONARYOID, + REGNAMESPACEOID, REGROLEOID, REFCURSOROID, REGCOLLATIONOID, + ] + + for reg_type in reg_types: + register_core_codec(reg_type, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # cstring type is used by Postgres' I/O functions + register_core_codec(CSTRINGOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_BINARY) + + # various system pseudotypes with no I/O + no_io_types = [ + ANYOID, TRIGGEROID, EVENT_TRIGGEROID, LANGUAGE_HANDLEROID, + FDW_HANDLEROID, TSM_HANDLEROID, INTERNALOID, OPAQUEOID, + ANYELEMENTOID, ANYNONARRAYOID, ANYCOMPATIBLEOID, + ANYCOMPATIBLEARRAYOID, ANYCOMPATIBLENONARRAYOID, + ANYCOMPATIBLERANGEOID, ANYCOMPATIBLEMULTIRANGEOID, + ANYRANGEOID, ANYMULTIRANGEOID, ANYARRAYOID, + PG_DDL_COMMANDOID, INDEX_AM_HANDLEROID, TABLE_AM_HANDLEROID, + ] + + register_core_codec(ANYENUMOID, + NULL, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + for no_io_type in no_io_types: + register_core_codec(no_io_type, + NULL, + NULL, + PG_FORMAT_BINARY) + + # ACL specification string + register_core_codec(ACLITEMOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # Postgres' serialized expression tree type + register_core_codec(PG_NODE_TREEOID, + NULL, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # pg_lsn type -- a pointer to a location in the XLOG. + register_core_codec(PG_LSNOID, + <encode_func>pgproto.int8_encode, + <decode_func>pgproto.int8_decode, + PG_FORMAT_BINARY) + + register_core_codec(SMGROID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # pg_dependencies and pg_ndistinct are special types + # used in pg_statistic_ext columns. + register_core_codec(PG_DEPENDENCIESOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + register_core_codec(PG_NDISTINCTOID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + # pg_mcv_list is a special type used in pg_statistic_ext_data + # system catalog + register_core_codec(PG_MCV_LISTOID, + <encode_func>pgproto.bytea_encode, + <decode_func>pgproto.bytea_decode, + PG_FORMAT_BINARY) + + # These two are internal to BRIN index support and are unlikely + # to be sent, but since I/O functions for these exist, add decoders + # nonetheless. + register_core_codec(PG_BRIN_BLOOM_SUMMARYOID, + NULL, + <decode_func>pgproto.bytea_decode, + PG_FORMAT_BINARY) + + register_core_codec(PG_BRIN_MINMAX_MULTI_SUMMARYOID, + NULL, + <decode_func>pgproto.bytea_decode, + PG_FORMAT_BINARY) + + +cdef init_text_codecs(): + textoids = [ + NAMEOID, + BPCHAROID, + VARCHAROID, + TEXTOID, + XMLOID + ] + + for oid in textoids: + register_core_codec(oid, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_BINARY) + + register_core_codec(oid, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + +cdef init_tid_codecs(): + register_core_codec(TIDOID, + <encode_func>pgproto.tid_encode, + <decode_func>pgproto.tid_decode, + PG_FORMAT_BINARY) + + +cdef init_txid_codecs(): + register_core_codec(TXID_SNAPSHOTOID, + <encode_func>pgproto.pg_snapshot_encode, + <decode_func>pgproto.pg_snapshot_decode, + PG_FORMAT_BINARY) + + register_core_codec(PG_SNAPSHOTOID, + <encode_func>pgproto.pg_snapshot_encode, + <decode_func>pgproto.pg_snapshot_decode, + PG_FORMAT_BINARY) + + +cdef init_tsearch_codecs(): + ts_oids = [ + TSQUERYOID, + TSVECTOROID, + ] + + for oid in ts_oids: + register_core_codec(oid, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + register_core_codec(GTSVECTOROID, + NULL, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + +cdef init_uuid_codecs(): + register_core_codec(UUIDOID, + <encode_func>pgproto.uuid_encode, + <decode_func>pgproto.uuid_decode, + PG_FORMAT_BINARY) + + +cdef init_numeric_codecs(): + register_core_codec(NUMERICOID, + <encode_func>pgproto.numeric_encode_text, + <decode_func>pgproto.numeric_decode_text, + PG_FORMAT_TEXT) + + register_core_codec(NUMERICOID, + <encode_func>pgproto.numeric_encode_binary, + <decode_func>pgproto.numeric_decode_binary, + PG_FORMAT_BINARY) + + +cdef init_network_codecs(): + register_core_codec(CIDROID, + <encode_func>pgproto.cidr_encode, + <decode_func>pgproto.cidr_decode, + PG_FORMAT_BINARY) + + register_core_codec(INETOID, + <encode_func>pgproto.inet_encode, + <decode_func>pgproto.inet_decode, + PG_FORMAT_BINARY) + + register_core_codec(MACADDROID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + register_core_codec(MACADDR8OID, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + +cdef init_monetary_codecs(): + moneyoids = [ + MONEYOID, + ] + + for oid in moneyoids: + register_core_codec(oid, + <encode_func>pgproto.text_encode, + <decode_func>pgproto.text_decode, + PG_FORMAT_TEXT) + + +cdef init_all_pgproto_codecs(): + # Builtin types, in lexicographical order. + init_bits_codecs() + init_bytea_codecs() + init_datetime_codecs() + init_float_codecs() + init_geometry_codecs() + init_int_codecs() + init_json_codecs() + init_monetary_codecs() + init_network_codecs() + init_numeric_codecs() + init_text_codecs() + init_tid_codecs() + init_tsearch_codecs() + init_txid_codecs() + init_uuid_codecs() + + # Various pseudotypes and system types + init_pseudo_codecs() + + # contrib + init_hstore_codecs() + + +init_all_pgproto_codecs() diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/range.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/range.pyx new file mode 100644 index 00000000..1038c18d --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/range.pyx @@ -0,0 +1,207 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +from asyncpg import types as apg_types + +from collections.abc import Sequence as SequenceABC + +# defined in postgresql/src/include/utils/rangetypes.h +DEF RANGE_EMPTY = 0x01 # range is empty +DEF RANGE_LB_INC = 0x02 # lower bound is inclusive +DEF RANGE_UB_INC = 0x04 # upper bound is inclusive +DEF RANGE_LB_INF = 0x08 # lower bound is -infinity +DEF RANGE_UB_INF = 0x10 # upper bound is +infinity + + +cdef enum _RangeArgumentType: + _RANGE_ARGUMENT_INVALID = 0 + _RANGE_ARGUMENT_TUPLE = 1 + _RANGE_ARGUMENT_RANGE = 2 + + +cdef inline bint _range_has_lbound(uint8_t flags): + return not (flags & (RANGE_EMPTY | RANGE_LB_INF)) + + +cdef inline bint _range_has_ubound(uint8_t flags): + return not (flags & (RANGE_EMPTY | RANGE_UB_INF)) + + +cdef inline _RangeArgumentType _range_type(object obj): + if cpython.PyTuple_Check(obj) or cpython.PyList_Check(obj): + return _RANGE_ARGUMENT_TUPLE + elif isinstance(obj, apg_types.Range): + return _RANGE_ARGUMENT_RANGE + else: + return _RANGE_ARGUMENT_INVALID + + +cdef range_encode(ConnectionSettings settings, WriteBuffer buf, + object obj, uint32_t elem_oid, + encode_func_ex encoder, const void *encoder_arg): + cdef: + ssize_t obj_len + uint8_t flags = 0 + object lower = None + object upper = None + WriteBuffer bounds_data = WriteBuffer.new() + _RangeArgumentType arg_type = _range_type(obj) + + if arg_type == _RANGE_ARGUMENT_INVALID: + raise TypeError( + 'list, tuple or Range object expected (got type {})'.format( + type(obj))) + + elif arg_type == _RANGE_ARGUMENT_TUPLE: + obj_len = len(obj) + if obj_len == 2: + lower = obj[0] + upper = obj[1] + + if lower is None: + flags |= RANGE_LB_INF + + if upper is None: + flags |= RANGE_UB_INF + + flags |= RANGE_LB_INC | RANGE_UB_INC + + elif obj_len == 1: + lower = obj[0] + flags |= RANGE_LB_INC | RANGE_UB_INF + + elif obj_len == 0: + flags |= RANGE_EMPTY + + else: + raise ValueError( + 'expected 0, 1 or 2 elements in range (got {})'.format( + obj_len)) + + else: + if obj.isempty: + flags |= RANGE_EMPTY + else: + lower = obj.lower + upper = obj.upper + + if obj.lower_inc: + flags |= RANGE_LB_INC + elif lower is None: + flags |= RANGE_LB_INF + + if obj.upper_inc: + flags |= RANGE_UB_INC + elif upper is None: + flags |= RANGE_UB_INF + + if _range_has_lbound(flags): + encoder(settings, bounds_data, lower, encoder_arg) + + if _range_has_ubound(flags): + encoder(settings, bounds_data, upper, encoder_arg) + + buf.write_int32(1 + bounds_data.len()) + buf.write_byte(<int8_t>flags) + buf.write_buffer(bounds_data) + + +cdef range_decode(ConnectionSettings settings, FRBuffer *buf, + decode_func_ex decoder, const void *decoder_arg): + cdef: + uint8_t flags = <uint8_t>frb_read(buf, 1)[0] + int32_t bound_len + object lower = None + object upper = None + FRBuffer bound_buf + + if _range_has_lbound(flags): + bound_len = hton.unpack_int32(frb_read(buf, 4)) + if bound_len == -1: + lower = None + else: + frb_slice_from(&bound_buf, buf, bound_len) + lower = decoder(settings, &bound_buf, decoder_arg) + + if _range_has_ubound(flags): + bound_len = hton.unpack_int32(frb_read(buf, 4)) + if bound_len == -1: + upper = None + else: + frb_slice_from(&bound_buf, buf, bound_len) + upper = decoder(settings, &bound_buf, decoder_arg) + + return apg_types.Range(lower=lower, upper=upper, + lower_inc=(flags & RANGE_LB_INC) != 0, + upper_inc=(flags & RANGE_UB_INC) != 0, + empty=(flags & RANGE_EMPTY) != 0) + + +cdef multirange_encode(ConnectionSettings settings, WriteBuffer buf, + object obj, uint32_t elem_oid, + encode_func_ex encoder, const void *encoder_arg): + cdef: + WriteBuffer elem_data + ssize_t elem_data_len + ssize_t elem_count + + if not isinstance(obj, SequenceABC): + raise TypeError( + 'expected a sequence (got type {!r})'.format(type(obj).__name__) + ) + + elem_data = WriteBuffer.new() + + for elem in obj: + range_encode(settings, elem_data, elem, elem_oid, encoder, encoder_arg) + + elem_count = len(obj) + if elem_count > INT32_MAX: + raise OverflowError(f'too many elements in multirange value') + + elem_data_len = elem_data.len() + if elem_data_len > INT32_MAX - 4: + raise OverflowError( + f'size of encoded multirange datum exceeds the maximum allowed' + f' {INT32_MAX - 4} bytes') + + # Datum length + buf.write_int32(4 + <int32_t>elem_data_len) + # Number of elements in multirange + buf.write_int32(<int32_t>elem_count) + buf.write_buffer(elem_data) + + +cdef multirange_decode(ConnectionSettings settings, FRBuffer *buf, + decode_func_ex decoder, const void *decoder_arg): + cdef: + int32_t nelems = hton.unpack_int32(frb_read(buf, 4)) + FRBuffer elem_buf + int32_t elem_len + int i + list result + + if nelems == 0: + return [] + + if nelems < 0: + raise exceptions.ProtocolError( + 'unexpected multirange size value: {}'.format(nelems)) + + result = cpython.PyList_New(nelems) + for i in range(nelems): + elem_len = hton.unpack_int32(frb_read(buf, 4)) + if elem_len == -1: + raise exceptions.ProtocolError( + 'unexpected NULL element in multirange value') + else: + frb_slice_from(&elem_buf, buf, elem_len) + elem = range_decode(settings, &elem_buf, decoder, decoder_arg) + cpython.Py_INCREF(elem) + cpython.PyList_SET_ITEM(result, i, elem) + + return result diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/record.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/record.pyx new file mode 100644 index 00000000..6446f2da --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/record.pyx @@ -0,0 +1,71 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +from asyncpg import exceptions + + +cdef inline record_encode_frame(ConnectionSettings settings, WriteBuffer buf, + WriteBuffer elem_data, int32_t elem_count): + buf.write_int32(4 + elem_data.len()) + # attribute count + buf.write_int32(elem_count) + # encoded attribute data + buf.write_buffer(elem_data) + + +cdef anonymous_record_decode(ConnectionSettings settings, FRBuffer *buf): + cdef: + tuple result + ssize_t elem_count + ssize_t i + int32_t elem_len + uint32_t elem_typ + Codec elem_codec + FRBuffer elem_buf + + elem_count = <ssize_t><uint32_t>hton.unpack_int32(frb_read(buf, 4)) + result = cpython.PyTuple_New(elem_count) + + for i in range(elem_count): + elem_typ = <uint32_t>hton.unpack_int32(frb_read(buf, 4)) + elem_len = hton.unpack_int32(frb_read(buf, 4)) + + if elem_len == -1: + elem = None + else: + elem_codec = settings.get_data_codec(elem_typ) + if elem_codec is None or not elem_codec.has_decoder(): + raise exceptions.InternalClientError( + 'no decoder for composite type element in ' + 'position {} of type OID {}'.format(i, elem_typ)) + elem = elem_codec.decode(settings, + frb_slice_from(&elem_buf, buf, elem_len)) + + cpython.Py_INCREF(elem) + cpython.PyTuple_SET_ITEM(result, i, elem) + + return result + + +cdef anonymous_record_encode(ConnectionSettings settings, WriteBuffer buf, obj): + raise exceptions.UnsupportedClientFeatureError( + 'input of anonymous composite types is not supported', + hint=( + 'Consider declaring an explicit composite type and ' + 'using it to cast the argument.' + ), + detail='PostgreSQL does not implement anonymous composite type input.' + ) + + +cdef init_record_codecs(): + register_core_codec(RECORDOID, + <encode_func>anonymous_record_encode, + <decode_func>anonymous_record_decode, + PG_FORMAT_BINARY) + +init_record_codecs() diff --git a/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/textutils.pyx b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/textutils.pyx new file mode 100644 index 00000000..dfaf29e0 --- /dev/null +++ b/.venv/lib/python3.12/site-packages/asyncpg/protocol/codecs/textutils.pyx @@ -0,0 +1,99 @@ +# Copyright (C) 2016-present the asyncpg authors and contributors +# <see AUTHORS file> +# +# This module is part of asyncpg and is released under +# the Apache 2.0 License: http://www.apache.org/licenses/LICENSE-2.0 + + +cdef inline uint32_t _apg_tolower(uint32_t c): + if c >= <uint32_t><Py_UCS4>'A' and c <= <uint32_t><Py_UCS4>'Z': + return c + <uint32_t><Py_UCS4>'a' - <uint32_t><Py_UCS4>'A' + else: + return c + + +cdef int apg_strcasecmp(const Py_UCS4 *s1, const Py_UCS4 *s2): + cdef: + uint32_t c1 + uint32_t c2 + int i = 0 + + while True: + c1 = s1[i] + c2 = s2[i] + + if c1 != c2: + c1 = _apg_tolower(c1) + c2 = _apg_tolower(c2) + if c1 != c2: + return <int32_t>c1 - <int32_t>c2 + + if c1 == 0 or c2 == 0: + break + + i += 1 + + return 0 + + +cdef int apg_strcasecmp_char(const char *s1, const char *s2): + cdef: + uint8_t c1 + uint8_t c2 + int i = 0 + + while True: + c1 = <uint8_t>s1[i] + c2 = <uint8_t>s2[i] + + if c1 != c2: + c1 = <uint8_t>_apg_tolower(c1) + c2 = <uint8_t>_apg_tolower(c2) + if c1 != c2: + return <int8_t>c1 - <int8_t>c2 + + if c1 == 0 or c2 == 0: + break + + i += 1 + + return 0 + + +cdef inline bint apg_ascii_isspace(Py_UCS4 ch): + return ( + ch == ' ' or + ch == '\n' or + ch == '\r' or + ch == '\t' or + ch == '\v' or + ch == '\f' + ) + + +cdef Py_UCS4 *apg_parse_int32(Py_UCS4 *buf, int32_t *num): + cdef: + Py_UCS4 *p + int32_t n = 0 + int32_t neg = 0 + + if buf[0] == '-': + neg = 1 + buf += 1 + elif buf[0] == '+': + buf += 1 + + p = buf + while <int>p[0] >= <int><Py_UCS4>'0' and <int>p[0] <= <int><Py_UCS4>'9': + n = 10 * n - (<int>p[0] - <int32_t><Py_UCS4>'0') + p += 1 + + if p == buf: + return NULL + + if not neg: + n = -n + + num[0] = n + + return p |