Add infrastructure for multiformat data codecs

Current asyncpg codec architecture allows for only one codec per data
type.  This arrangement works most of the time, but fails utterly to
handle situations when the text format I/O is forced by the lack of
binary I/O support for one of the types in the composite type.

This commit is the preparation for the introduction of text format
parsing and allows to associate multiple format-specific codecs for each
type.

Author: elprans

asyncpg/introspection.py

@@ -7,8 +7,8 @@
 
 INTRO_LOOKUP_TYPES = '''\
 WITH RECURSIVE typeinfo_tree(
-    oid, ns, name, kind, basetype, elemtype, range_subtype,
-    elem_has_bin_input, elem_has_bin_output, attrtypoids, attrnames, depth)
+    oid, ns, name, kind, basetype, has_bin_io, elemtype, elemdelim,
+    range_subtype, elem_has_bin_io, attrtypoids, attrnames, depth)
 AS (
     WITH composite_attrs
     AS (
@@ -58,10 +58,23 @@
 
                ELSE NULL
             END)                            AS basetype,
+            t.typreceive::oid != 0 AND t.typsend::oid != 0
+                                            AS has_bin_io,
             t.typelem                       AS elemtype,
+            elem_t.typdelim                 AS elemdelim,
             range_t.rngsubtype              AS range_subtype,
-            elem_t.typreceive::oid != 0     AS elem_has_bin_input,
-            elem_t.typsend::oid != 0        AS elem_has_bin_output,
+            (CASE WHEN t.typtype = 'r' THEN
+                (SELECT
+                    range_elem_t.typreceive::oid != 0 AND
+                        range_elem_t.typsend::oid != 0
+                FROM
+                    pg_catalog.pg_type AS range_elem_t
+                WHERE
+                    range_elem_t.oid = range_t.rngsubtype)
+            ELSE
+                elem_t.typreceive::oid != 0 AND
+                    elem_t.typsend::oid != 0
+            END)                            AS elem_has_bin_io,
             (CASE WHEN t.typtype = 'c' THEN
                 (SELECT ca.typoids
                 FROM composite_attrs AS ca
@@ -91,8 +104,8 @@
     )
 
     SELECT
-        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.elemtype,
-        ti.range_subtype, ti.elem_has_bin_input, ti.elem_has_bin_output,
+        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.has_bin_io,
+        ti.elemtype, ti.elemdelim, ti.range_subtype, ti.elem_has_bin_io,
         ti.attrtypoids, ti.attrnames, 0
     FROM
         typeinfo AS ti
@@ -102,8 +115,8 @@
     UNION ALL
 
     SELECT
-        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.elemtype,
-        ti.range_subtype, ti.elem_has_bin_input, ti.elem_has_bin_output,
+        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.has_bin_io,
+        ti.elemtype, ti.elemdelim, ti.range_subtype, ti.elem_has_bin_io,
         ti.attrtypoids, ti.attrnames, tt.depth + 1
     FROM
         typeinfo ti,
@@ -126,8 +139,8 @@
 # Prior to 9.2 PostgreSQL did not have range types.
 INTRO_LOOKUP_TYPES_91 = '''\
 WITH RECURSIVE typeinfo_tree(
-    oid, ns, name, kind, basetype, elemtype, range_subtype,
-    elem_has_bin_input, elem_has_bin_output, attrtypoids, attrnames, depth)
+    oid, ns, name, kind, basetype, has_bin_io, elemtype, elemdelim,
+    range_subtype, elem_has_bin_io, attrtypoids, attrnames, depth)
 AS (
     WITH composite_attrs
     AS (
@@ -177,10 +190,14 @@
 
                ELSE NULL
             END)                            AS basetype,
+            t.typreceive::oid != 0 AND t.typsend::oid != 0
+                                            AS has_bin_io,
             t.typelem                       AS elemtype,
+            elem_t.typdelim                 AS elemdelim,
             NULL::oid                       AS range_subtype,
-            elem_t.typreceive::oid != 0     AS elem_has_bin_input,
-            elem_t.typsend::oid != 0        AS elem_has_bin_output,
+            elem_t.typreceive::oid != 0 AND
+                elem_t.typsend::oid != 0
+                                            AS elem_has_bin_io,
             (CASE WHEN t.typtype = 'c' THEN
                 (SELECT ca.typoids
                 FROM composite_attrs AS ca
@@ -207,8 +224,8 @@
     )
 
     SELECT
-        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.elemtype,
-        ti.range_subtype, ti.elem_has_bin_input, ti.elem_has_bin_output,
+        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.has_bin_io,
+        ti.elemtype, ti.elemdelim, ti.range_subtype, ti.elem_has_bin_io,
         ti.attrtypoids, ti.attrnames, 0
     FROM
         typeinfo AS ti
@@ -218,8 +235,8 @@
     UNION ALL
 
     SELECT
-        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.elemtype,
-        ti.range_subtype, ti.elem_has_bin_input, ti.elem_has_bin_output,
+        ti.oid, ti.ns, ti.name, ti.kind, ti.basetype, ti.has_bin_io,
+        ti.elemtype, ti.elemdelim, ti.range_subtype, ti.elem_has_bin_io,
         ti.attrtypoids, ti.attrnames, tt.depth + 1
     FROM
         typeinfo ti,

asyncpg/protocol/codecs/base.pxd

@@ -32,6 +32,7 @@ cdef enum CodecType:
 
 
 cdef enum CodecFormat:
+    PG_FORMAT_ANY = -1
     PG_FORMAT_TEXT = 0
     PG_FORMAT_BINARY = 1
 
@@ -55,6 +56,7 @@ cdef class Codec:
 
         # arrays
         Codec           element_codec
+        Py_UCS4         element_delimiter
 
         # composite types
         tuple           element_type_oids
@@ -70,7 +72,8 @@ cdef class Codec:
               encode_func c_encoder, decode_func c_decoder,
               object py_encoder, object py_decoder,
               Codec element_codec, tuple element_type_oids,
-              object element_names, list element_codecs)
+              object element_names, list element_codecs,
+              Py_UCS4 element_delimiter)
 
     cdef encode_scalar(self, ConnectionSettings settings, WriteBuffer buf,
                        object obj)
@@ -118,7 +121,8 @@ cdef class Codec:
     cdef Codec new_array_codec(uint32_t oid,
                                str name,
                                str schema,
-                               Codec element_codec)
+                               Codec element_codec,
+                               Py_UCS4 element_delimiter)
 
     @staticmethod
     cdef Codec new_range_codec(uint32_t oid,
@@ -149,4 +153,4 @@ cdef class DataCodecConfig:
         dict _type_codecs_cache
         dict _local_type_codecs
 
-    cdef inline Codec get_codec(self, uint32_t oid)
+    cdef inline Codec get_codec(self, uint32_t oid, CodecFormat format)