From 1c91d6590f3ec2e358f6d77d6bcf9b9f05d889f9 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <417981+cpcloud@users.noreply.github.com>
Date: Thu, 19 Sep 2024 10:47:20 -0400
Subject: [PATCH] fix(datatype-parsing): ensure that geospatial types are round
 trippable through the data type parser (#10171)

Fixes #10170.
---
 ibis/expr/datatypes/parse.py            | 27 ++++++++++++++++++-------
 ibis/expr/datatypes/tests/test_parse.py |  2 +-
 2 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/ibis/expr/datatypes/parse.py b/ibis/expr/datatypes/parse.py
index 22ec355333bb..dc709d64e31d 100644
--- a/ibis/expr/datatypes/parse.py
+++ b/ibis/expr/datatypes/parse.py
@@ -1,8 +1,8 @@
 from __future__ import annotations
 
 import ast
-import functools
 import re
+from functools import lru_cache, partial
 from operator import methodcaller
 
 import parsy
@@ -50,7 +50,7 @@ def spaceless_string(*strings: str):
 
 
 @public
-@functools.lru_cache(maxsize=100)
+@lru_cache(maxsize=100)
 def parse(
     text: str, default_decimal_parameters: tuple[int | None, int | None] = (None, None)
 ) -> dt.DataType:
@@ -88,12 +88,15 @@ def parse(
     geotype = spaceless_string("geography", "geometry")
 
     srid_geotype = SEMICOLON.then(parsy.seq(srid=NUMBER.skip(COLON), geotype=geotype))
+    geotype_srid = COLON.then(parsy.seq(geotype=geotype, srid=SEMICOLON.then(NUMBER)))
     geotype_part = COLON.then(parsy.seq(geotype=geotype))
     srid_part = SEMICOLON.then(parsy.seq(srid=NUMBER))
 
     def geotype_parser(typ: type[dt.DataType]) -> dt.DataType:
         return spaceless_string(typ.__name__.lower()).then(
-            (srid_geotype | geotype_part | srid_part).optional(dict()).combine_dict(typ)
+            (srid_geotype | geotype_srid | geotype_part | srid_part)
+            .optional(dict())
+            .combine_dict(typ)
         )
 
     primitive = (
@@ -116,15 +119,25 @@ def geotype_parser(typ: type[dt.DataType]) -> dt.DataType:
             "time",
             "date",
             "null",
-        ).map(functools.partial(getattr, dt))
-        | spaceless_string("bytes").result(dt.binary)
-        | geotype.map(dt.GeoSpatial)
+        ).map(partial(getattr, dt))
         | geotype_parser(dt.LineString)
         | geotype_parser(dt.Polygon)
         | geotype_parser(dt.Point)
         | geotype_parser(dt.MultiLineString)
         | geotype_parser(dt.MultiPolygon)
         | geotype_parser(dt.MultiPoint)
+        | spaceless_string("bytes").result(dt.binary)
+        | spaceless_string("geospatial:geography").then(
+            srid_part.optional(dict()).combine_dict(
+                partial(dt.GeoSpatial, geotype="geography")
+            )
+        )
+        | spaceless_string("geospatial:geometry").then(
+            srid_part.optional(dict()).combine_dict(
+                partial(dt.GeoSpatial, geotype="geometry")
+            )
+        )
+        | geotype.map(dt.GeoSpatial)
     )
 
     varchar_or_char = (
@@ -201,7 +214,7 @@ def geotype_parser(typ: type[dt.DataType]) -> dt.DataType:
         | map
         | struct
         | spaceless_string("jsonb", "json", "uuid", "macaddr", "inet").map(
-            functools.partial(getattr, dt)
+            partial(getattr, dt)
         )
         | spaceless_string("int").result(dt.int64)
         | spaceless_string("str").result(dt.string)
diff --git a/ibis/expr/datatypes/tests/test_parse.py b/ibis/expr/datatypes/tests/test_parse.py
index 7666ae95cfb1..b020f96d4ca3 100644
--- a/ibis/expr/datatypes/tests/test_parse.py
+++ b/ibis/expr/datatypes/tests/test_parse.py
@@ -266,7 +266,6 @@ def test_parse_null():
 
 
 # corresponds to its.all_dtypes() but without:
-# - geospacial types, the string representation is different from what the parser expects
 # - struct types, the generated struct field names contain special characters
 
 field_names = st.text(
@@ -286,6 +285,7 @@ def test_parse_null():
         | its.struct_dtypes(names=field_names)
         | its.array_dtypes(roundtrippable_dtypes)
         | its.map_dtypes(roundtrippable_dtypes, roundtrippable_dtypes)
+        | its.geospatial_dtypes()
     )
 )