main: fix bug deleting all data for output path with trailing slash

carte_cli/publisher/remove_deleted_publisher.py

@@ -17,8 +17,12 @@ def init(self, conf: ConfigTree) -> None:
     def _get_datasets_to_delete(
         self, datasets: Set[str], file_paths: List[str]
     ) -> Iterator[str]:
+        prefix_length = len(self.tables_path)
+        if not self.tables_path.endswith("/"):
+            prefix_length += 1
+
         file_ids = [
-            path[(len(self.tables_path) + 1) : -(len(".md"))] for path in file_paths
+            path[prefix_length : -(len(".md"))] for path in file_paths
         ]
 
         for file_path, file_id in zip(file_paths, file_ids):

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "carte-cli"
-version = "0.3.6"
+version = "0.3.7"
 description = "A static site generator for data catalogs"
 authors = ["Balint Haller <balint@hey.com>"]
 license = "GPL-3.0-or-later"

tests/publisher/test_remove_deleted_publisher.py

@@ -9,11 +9,11 @@ def test_get_datasets_to_delete():
 
     test_datasets = set(
         [
-            f"db1/dataset1",
-            f"db1/dataset3",
-            f"db1/dataset4",
-            f"db1/dataset5",
-            f"db2/dataset1",
+            "db1/dataset1",
+            "db1/dataset3",
+            "db1/dataset4",
+            "db1/dataset5",
+            "db2/dataset1",
         ]
     )
 
@@ -34,3 +34,26 @@ def test_get_datasets_to_delete():
     assert len(to_delete) == 2
     assert "test-tables-path/db3/dataset1.md" in to_delete
     assert "test-tables-path/db2/dataset2.md" in to_delete
+
+
+def test_output_path_with_trailing_slash():
+    publisher = RemoveDeletedPublisher()
+    tables_path = "test-tables-path/"
+
+    publisher.tables_path = tables_path
+
+    test_datasets = set(["db1/dataset1", "db2/dataset1"])
+
+    test_file_paths = [
+        f"{tables_path}db1/dataset1.md",
+        f"{tables_path}db2/dataset1.md",
+        f"{tables_path}db2/dataset2.md",
+    ]
+
+    to_delete = [
+        dataset
+        for dataset in publisher._get_datasets_to_delete(test_datasets, test_file_paths)
+    ]
+
+    assert len(to_delete) == 1
+    assert to_delete[0] == "test-tables-path/db2/dataset2.md"