Staging

requirements.in

@@ -6,6 +6,3 @@ pydantic-settings
 s3fs
 aiofiles
 duckdb
-polars
-pyarrow
-xlsx2csv

src/hellocomputer/analytics.py

@@ -35,7 +35,7 @@ class DDB:
         self.sheets = tuple(
             self.db.query("select Field2 from metadata where Field1 = 'Sheets'")
             .fetchall()[0][0]
-            .split(",")
+            .split(";")
         )
         self.path = path
 
@@ -83,5 +83,18 @@ class DDB:
 
         return self
 
-    def query(self, sql):
+    def load_folder_local(self, path: str):
-        return self.db.query(sql)
+        self.sheets = tuple(
+            self.query(
+                f"select Field2 from read_csv_auto('{path}/metadata.csv') where Field1 = 'Sheets'"
+            )
+            .fetchall()[0][0]
+            .split(";")
+        )
+        return self
+
+    def load_folder_gcs(self, path: str):
+        return self
+
+    def query(self, sql, *args, **kwargs):
+        return self.db.query(sql, *args, **kwargs)

test/test_data.py

@@ -6,7 +6,7 @@ TEST_DATA_FOLDER = Path(hellocomputer.__file__).parents[2] / "test" / "data"
 TEST_OUTPUT_FOLDER = Path(hellocomputer.__file__).parents[2] / "test" / "output"
 
 
-def test_load_data():
+def test_dump():
     db = (
         DDB()
         .load_metadata(TEST_DATA_FOLDER / "TestExcelHelloComputer.xlsx")
@@ -15,3 +15,8 @@ def test_load_data():
 
     assert db.sheets == ("answers",)
     assert (TEST_OUTPUT_FOLDER / "answers.csv").exists()
+
+
+def test_load():
+    db = DDB().load_folder_local(TEST_OUTPUT_FOLDER)
+    assert db.sheets == ("answers",)