From 3342c1ae6e955007f89deea2cc9797f6d92dc28a Mon Sep 17 00:00:00 2001
From: z3z1ma <butler.alex2010@gmail.com>
Date: Sat, 8 Jun 2024 14:57:18 -0700
Subject: [PATCH] feat: add a duckdb handle with the project filesystem
 registered under cdf protocol

---
 src/cdf/core/filesystem.py |  3 ++-
 src/cdf/core/project.py    | 20 ++++++++++++++++++++
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/src/cdf/core/filesystem.py b/src/cdf/core/filesystem.py
index e6e8085..046584d 100644
--- a/src/cdf/core/filesystem.py
+++ b/src/cdf/core/filesystem.py
@@ -52,7 +52,8 @@ def wrapped(self) -> fsspec.AbstractFileSystem:
 
         options = self.options.copy()
         options.setdefault("auto_mkdir", True)
-        return DirFileSystem(
+        CdfFs = type("CdfFs", (DirFileSystem,), {"protocol": "cdf"})
+        return CdfFs(
             path=posixpath.join(strip_protocol(self.uri), "x")[:-1],
             fs=fsspec.filesystem(self.protocol, **options),
             auto_mkdir=True,
diff --git a/src/cdf/core/project.py b/src/cdf/core/project.py
index a97682d..a3610c6 100644
--- a/src/cdf/core/project.py
+++ b/src/cdf/core/project.py
@@ -48,6 +48,7 @@
 from functools import cached_property
 from pathlib import Path
 
+import duckdb
 import dynaconf
 import pydantic
 from dynaconf.utils.boxing import DynaBox
@@ -753,6 +754,25 @@ def feature_flags(self) -> FeatureFlagAdapter:
             self.ff_settings, filesystem=self.filesystem
         ).unwrap()
 
+    @cached_property
+    def duckdb(self) -> duckdb.DuckDBPyConnection:
+        """Get a handle to the project's DuckDB connection"""
+        conn = duckdb.connect(":memory:")
+        conn.install_extension("httpfs")
+        conn.install_extension("json")
+        conn.register_filesystem(self.filesystem.wrapped)
+        conn.execute("CREATE TABLE workspaces (name TEXT PRIMARY KEY, path TEXT)")
+        for workspace in self.workspaces:
+            conn.execute(
+                "INSERT INTO workspaces (name, path) VALUES (?, ?)",
+                (workspace.name, workspace.path.as_posix()),
+            )
+        return conn
+
+    def get_workspace_path(self, name: str) -> M.Result[Path, Exception]:
+        """Get the path to a workspace by name"""
+        return self.get_workspace(name).map(lambda ws: ws.path)
+
     @classmethod
     def from_path(cls, root: PathLike):
         """Load configuration data from a project root path using dynaconf.