fix: respect client's preferred encoding when possible

Previously, `pygls` would always use `UTF-16` except when the client
tried to hide the fact that it supports `UTF-16` (which the LSP spec
requires it to do in all cases). Now, `pygls` will choose the editor's
preferred encoding.

When it is `UTF-32`, `pygls` saves a bit of computation in most
position codec related operations (`X_to_client_units` +
`client_num_units` are faster, `X_from_client_units` is about the
same), which is great. When it is `UTF-16` or `UTF-8`, the
computational load is about the same.

Closes: openlawlibrary#445

pygls/capabilities.py

@@ -15,7 +15,7 @@
 # limitations under the License.                                           #
 ############################################################################
 from functools import reduce
-from typing import Any, Dict, List, Optional, Set, Union, TypeVar
+from typing import Any, Dict, List, Optional, Set, Union, TypeVar, FrozenSet
 import logging
 
 from lsprotocol import types
@@ -24,6 +24,14 @@
 logger = logging.getLogger(__name__)
 T = TypeVar("T")
 
+_SUPPORTED_ENCODINGS = frozenset(
+    [
+        types.PositionEncodingKind.Utf8,
+        types.PositionEncodingKind.Utf16,
+        types.PositionEncodingKind.Utf32,
+    ]
+)
+
 
 def get_capability(
     client_capabilities: types.ClientCapabilities, field: str, default: Any = None
@@ -401,16 +409,11 @@ def _with_position_encodings(self):
         if encodings is None:
             return self
 
-        if types.PositionEncodingKind.Utf16 in encodings:
-            return self
-
-        if types.PositionEncodingKind.Utf32 in encodings:
-            self.server_cap.position_encoding = types.PositionEncodingKind.Utf32
-            return self
-
-        if types.PositionEncodingKind.Utf8 in encodings:
-            self.server_cap.position_encoding = types.PositionEncodingKind.Utf8
-            return self
+        # We match client preference where this an overlap between its and our supported encodings.
+        for encoding in encodings:
+            if encoding in _SUPPORTED_ENCODINGS:
+                self.server_cap.position_encoding = encoding
+                return self
 
         logger.warning(f"Unknown `PositionEncoding`s: {encodings}")
 

tests/test_feature_manager.py

@@ -240,6 +240,19 @@ def server_capabilities(**kwargs):
                     ]
                 )
             ),
+            server_capabilities(position_encoding=lsp.PositionEncodingKind.Utf8),
+        ),
+        (
+            lsp.INITIALIZE,
+            None,
+            lsp.ClientCapabilities(
+                general=lsp.GeneralClientCapabilities(
+                    position_encodings=[
+                        lsp.PositionEncodingKind.Utf32,
+                        lsp.PositionEncodingKind.Utf8,
+                    ]
+                )
+            ),
             server_capabilities(position_encoding=lsp.PositionEncodingKind.Utf32),
         ),
         (