fix(checkpoint): support tinker paths with run-id suffix

Fixes Issue #24: 'tinker checkpoint delete' now accepts paths with run-id
suffix (e.g., tinker://run-id🚋0/sampler_weights/final).

The previous implementation assumed a strict 3-part path format, which
failed when the run-id contained a suffix like '🚋0'.

This change updates ParsedCheckpointTinkerPath.from_tinker_path() to:
- Handle run-ids with arbitrary suffixes
- Provide better error messages for invalid formats

Good day,

Thank you for your work on this excellent library!

Warmly,
RoomWithOutRoof

src/tinker/types/checkpoint.py

@@ -46,18 +46,45 @@ class ParsedCheckpointTinkerPath(BaseModel):
 
     @classmethod
     def from_tinker_path(cls, tinker_path: str) -> "ParsedCheckpointTinkerPath":
-        """Parse a tinker path to an instance of ParsedCheckpointTinkerPath"""
+        """Parse a tinker path to an instance of ParsedCheckpointTinkerPath.
+
+        Supports two formats:
+        - Standard: tinker://run-id/weights/0001
+        - With suffix: tinker://run-id:suffix/weights/0001
+          (e.g., tinker://run-id:train:0/weights/0001)
+        """
         if not tinker_path.startswith("tinker://"):
             raise ValueError(f"Invalid tinker path: {tinker_path}")
-        parts = tinker_path[9:].split("/")
-        if len(parts) != 3:
-            raise ValueError(f"Invalid tinker path: {tinker_path}")
-        if parts[1] not in ["weights", "sampler_weights"]:
-            raise ValueError(f"Invalid tinker path: {tinker_path}")
-        checkpoint_type = "training" if parts[1] == "weights" else "sampler"
+
+        # Remove the tinker:// prefix
+        path_parts = tinker_path[9:]
+
+        # Split into segments
+        # Format: run_id_with_type/checkpoint_type/checkpoint_id
+        segments = path_parts.split("/")
+        if len(segments) != 3:
+            raise ValueError(
+                f"Invalid tinker path: {tinker_path}. "
+                f"Expected: tinker://run-id/weights/0001 or tinker://run-id:train:0/weights/0001"
+            )
+
+        run_id_with_type = segments[0]
+        checkpoint_type_segment = segments[1]
+        checkpoint_id = segments[2]
+
+        # Validate checkpoint type
+        if checkpoint_type_segment not in ["weights", "sampler_weights"]:
+            raise ValueError(
+                f"Invalid checkpoint type: {checkpoint_type_segment}. "
+                f"Expected: weights or sampler_weights"
+            )
+
+
+        checkpoint_type = "training" if checkpoint_type_segment == "weights" else "sampler"
+
         return cls(
             tinker_path=tinker_path,
-            training_run_id=parts[0],
+            training_run_id=run_id_with_type,
             checkpoint_type=checkpoint_type,
-            checkpoint_id="/".join(parts[1:]),
+            checkpoint_id="/".join([checkpoint_type_segment, checkpoint_id]),
         )

tests/test_checkpoint_delete.py

@@ -194,3 +194,65 @@ class TestDeleteCLIValidation:
         )
         assert result.exit_code != 0
         assert "--run-id" in self._get_error_message(result)
+
+
+class TestParsedCheckpointTinkerPath:
+    """Tests for ParsedCheckpointTinkerPath.from_tinker_path()."""
+
+    def test_standard_format(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        result = ParsedCheckpointTinkerPath.from_tinker_path(
+            "tinker://run-id/weights/0001"
+        )
+        assert result.training_run_id == "run-id"
+        assert result.checkpoint_type == "training"
+        assert result.checkpoint_id == "weights/0001"
+
+    def test_sampler_format(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        result = ParsedCheckpointTinkerPath.from_tinker_path(
+            "tinker://run-id/sampler_weights/0001"
+        )
+        assert result.training_run_id == "run-id"
+        assert result.checkpoint_type == "sampler"
+        assert result.checkpoint_id == "sampler_weights/0001"
+
+    def test_with_train_suffix(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        result = ParsedCheckpointTinkerPath.from_tinker_path(
+            "tinker://5f2d7413-3980-502a-b012-9b7e122b3305:train:0/sampler_weights/final"
+        )
+        assert result.training_run_id == "5f2d7413-3980-502a-b012-9b7e122b3305:train:0"
+        assert result.checkpoint_type == "sampler"
+        assert result.checkpoint_id == "sampler_weights/final"
+
+    def test_with_sampler_suffix(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        result = ParsedCheckpointTinkerPath.from_tinker_path(
+            "tinker://run-id:sampler/weights/0001"
+        )
+        assert result.training_run_id == "run-id:sampler"
+        assert result.checkpoint_type == "training"
+        assert result.checkpoint_id == "weights/0001"
+
+    def test_invalid_missing_prefix(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        with pytest.raises(ValueError, match="Invalid tinker path"):
+            ParsedCheckpointTinkerPath.from_tinker_path("run-id/weights/0001")
+
+    def test_invalid_wrong_checkpoint_type(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        with pytest.raises(ValueError, match="Invalid checkpoint type"):
+            ParsedCheckpointTinkerPath.from_tinker_path("tinker://run-id/invalid/0001")
+
+    def test_invalid_not_enough_parts(self) -> None:
+        from tinker.types.checkpoint import ParsedCheckpointTinkerPath
+
+        with pytest.raises(ValueError, match="Invalid tinker path"):
+            ParsedCheckpointTinkerPath.from_tinker_path("tinker://run-id/weights")