fix: get element oid from the array itself in binary format

Getting it from the type was causing problems in CRDB because they seem
to have a different interpretation of what Int2 and Int4 are. Refactor
the binary parser to take a transformer, instead of a loader function,
in order to look up the loader from the oid parsed from the array.

Also refactor the text array parser to take the whole loader instead of
the load function only, which will be useful in order to optimize the
use of C loaders.

diff --git a/psycopg/psycopg/types/array.py b/psycopg/psycopg/types/array.py
index 4fd234660faf524a5863b8db90e7be8b5035dd6e..f358cdedf7ffd9bb68317a6c771b262b6ae2cf4f 100644 (file)
--- a/psycopg/psycopg/types/array.py
+++ b/psycopg/psycopg/types/array.py
@@ -11,7 +11,7 @@ from typing import Any, cast, Callable, List, Optional, Pattern, Set, Tuple, Typ
 from .. import pq
 from .. import errors as e
 from .. import postgres
-from ..abc import AdaptContext, Buffer, Dumper, DumperKey, NoneType, LoadFunc
+from ..abc import AdaptContext, Buffer, Dumper, DumperKey, NoneType, Loader, Transformer
 from ..adapt import RecursiveDumper, RecursiveLoader, PyFormat
 from .._compat import cache, prod
 from .._struct import pack_len, unpack_len
@@ -28,6 +28,9 @@ _unpack_dim = cast(Callable[[Buffer, int], Tuple[int, int]], _struct_dim.unpack_
 
 TEXT_ARRAY_OID = postgres.types["text"].array_oid
 
+PY_TEXT = PyFormat.TEXT
+PQ_BINARY = pq.Format.BINARY
+
 
 class BaseListDumper(RecursiveDumper):
     element_oid = 0
@@ -187,7 +190,7 @@ class ListDumper(BaseListDumper):
         if self.sub_dumper:
             return self.sub_dumper.dump(item)
         else:
-            return self._tx.get_dumper(item, PyFormat.TEXT).dump(item)
+            return self._tx.get_dumper(item, PY_TEXT).dump(item)
 
 
 @cache
@@ -299,8 +302,8 @@ class ArrayLoader(BaseArrayLoader):
     delimiter = b","
 
     def load(self, data: Buffer) -> List[Any]:
-        load = self._tx.get_loader(self.base_oid, self.format).load
-        return load_text(data, load, self.delimiter)
+        loader = self._tx.get_loader(self.base_oid, self.format)
+        return load_text(data, loader, self.delimiter)
 
 
 class ArrayBinaryLoader(BaseArrayLoader):
@@ -308,8 +311,7 @@ class ArrayBinaryLoader(BaseArrayLoader):
     format = pq.Format.BINARY
 
     def load(self, data: Buffer) -> List[Any]:
-        load = self._tx.get_loader(self.base_oid, self.format).load
-        return load_binary(data, load)
+        return load_binary(data, self._tx)
 
 
 def register_array(info: TypeInfo, context: Optional[AdaptContext] = None) -> None:
@@ -374,12 +376,13 @@ def register_all_arrays(context: AdaptContext) -> None:
 
 def _load_text(
     data: Buffer,
-    load: LoadFunc,
+    loader: Loader,
     delimiter: bytes = b",",
     __re_unescape: Pattern[bytes] = re.compile(rb"\\(.)"),
 ) -> List[Any]:
     rv = None
     stack: List[Any] = []
+    load = loader.load
 
     # Remove the dimensions information prefix (``[...]=``)
     if data and data[0] == b"["[0]:
@@ -439,8 +442,9 @@ def _get_array_parse_regexp(delimiter: bytes) -> Pattern[bytes]:
     )
 
 
-def _load_binary(data: Buffer, load: LoadFunc) -> List[Any]:
+def _load_binary(data: Buffer, tx: Transformer) -> List[Any]:
     ndims, hasnull, oid = _unpack_head(data)
+    load = tx.get_loader(oid, PQ_BINARY).load
 
     if not ndims:
         return []

diff --git a/psycopg_c/psycopg_c/_psycopg.pyi b/psycopg_c/psycopg_c/_psycopg.pyi
index 86b3977df168cbca553f280cf58da96d68225cba..9bff1ee1ab2be314e056bf2c2e912547d090715a 100644 (file)
--- a/psycopg_c/psycopg_c/_psycopg.pyi
+++ b/psycopg_c/psycopg_c/_psycopg.pyi
@@ -74,8 +74,8 @@ def parse_row_binary(data: abc.Buffer, tx: abc.Transformer) -> Tuple[Any, ...]:
 
 # Arrays optimization
 def array_load_text(
-    data: abc.Buffer, load: abc.LoadFunc, delimiter: bytes = b","
+    data: abc.Buffer, loader: abc.Loader, delimiter: bytes = b","
 ) -> List[Any]: ...
-def array_load_binary(data: abc.Buffer, load: abc.LoadFunc) -> List[Any]: ...
+def array_load_binary(data: abc.Buffer, tx: abc.Transformer) -> List[Any]: ...
 
 # vim: set syntax=python:

diff --git a/psycopg_c/psycopg_c/types/array.pyx b/psycopg_c/psycopg_c/types/array.pyx
index 9495eeb751f38bd6651255239e9cd7ef3864f78b..3b8d828a9caabd8569129388c3e04faed1c1da0d 100644 (file)
--- a/psycopg_c/psycopg_c/types/array.pyx
+++ b/psycopg_c/psycopg_c/types/array.pyx
@@ -17,13 +17,15 @@ from psycopg import errors as e
 
 
 def array_load_text(
-    data: Buffer, load: LoadFunc, delimiter: bytes = b","
+    data: Buffer, loader: Loader, delimiter: bytes = b","
 ) -> List[Any]:
     cdef char cdelim = delimiter[0]
 
     cdef char *buf
     cdef Py_ssize_t length
     _buffer_as_string_and_size(data, &buf, &length)
+    load = loader.load
+
     if length == 0:
         raise e.DataError("malformed array: empty data")
 
@@ -132,7 +134,7 @@ cdef object parse_token(char **bufptr, char *bufend, char cdelim, object load):
             PyMem_Free(unesc)
 
 
-def array_load_binary(data: Buffer, load: LoadFunc) -> List[Any]:
+def array_load_binary(data: Buffer, tx: Transformer) -> List[Any]:
     cdef char *buf
     cdef Py_ssize_t length
     _buffer_as_string_and_size(data, &buf, &length)
@@ -140,6 +142,9 @@ def array_load_binary(data: Buffer, load: LoadFunc) -> List[Any]:
     # head is ndims, hasnull, elem oid
     cdef int32_t *buf32 = <int32_t *>buf
     cdef int ndims = endian.be32toh(buf32[0])
+    cdef int oid = endian.be32toh(buf32[2])
+
+    load = tx.get_loader(oid, PQ_BINARY).load
 
     if not ndims:
         return []