warning message about column mismatch

dp_wizard/app/dataset_panel.py

@@ -1,8 +1,10 @@
 from pathlib import Path
+import csv
 
 from shiny import ui, reactive, render, Inputs, Outputs, Session
 
 from dp_wizard.utils.argparse_helpers import get_cli_info
+from dp_wizard.utils.csv_helper import csv_names_mismatch
 from dp_wizard.app.components.outputs import output_code_sample, demo_tooltip
 from dp_wizard.utils.code_generators import make_privacy_unit_block
 
@@ -93,12 +95,30 @@ def _on_public_csv_path_change():
     def _on_private_csv_path_change():
         private_csv_path.set(input.private_csv_path()[0]["datapath"])
 
-    @render.ui
-    def csv_column_match_ui():
+    @reactive.calc
+    def csv_column_match_calc() -> tuple[set, set] | None:
         public = public_csv_path()
         private = private_csv_path()
         if public and private:
-            return f"TODO: read files and check columns: public: {public_csv_path()}, private: {private_csv_path()}"
+            just_public, just_private = csv_names_mismatch(Path(public), Path(private))
+            if just_public or just_private:
+                return just_public, just_private
+
+    @render.ui
+    def csv_column_match_ui():
+        mismatch = csv_column_match_calc()
+        if mismatch:
+            just_public, just_private = mismatch
+            messages = []
+            if just_public:
+                messages.append(
+                    f"- Only the public CSV contains: {', '.join(just_public)}."
+                )
+            if just_private:
+                messages.append(
+                    f"- Only the private CSV contains: {', '.join(just_private)}."
+                )
+            return ui.markdown("\n".join(messages))
 
     @reactive.effect
     @reactive.event(input.contributions)
@@ -109,8 +129,13 @@ def _on_contributions_change():
     def button_enabled():
         contributions_is_set = input.contributions() is not None
         csv_path_is_set = (
-            input.csv_path() is not None and len(input.csv_path()) > 0
-        ) or is_demo
+            (input.public_csv_path() is not None and len(input.public_csv_path()) > 0)
+            or (
+                input.private_csv_path() is not None
+                and len(input.private_csv_path()) > 0
+            )
+            or is_demo
+        )
         return contributions_is_set and csv_path_is_set
 
     @render.ui

dp_wizard/utils/csv_helper.py

@@ -8,9 +8,11 @@
 
 import re
 import polars as pl
+from tempfile import tempdir
+from pathlib import Path
 
 
-def read_csv_names(csv_path: str):
+def read_csv_names(csv_path: Path):
     # Polars is overkill, but it is more robust against
     # variations in encoding than Python stdlib csv.
     # However, it could be slow:
@@ -21,14 +23,22 @@ def read_csv_names(csv_path: str):
     return lf.collect_schema().names()
 
 
-def read_csv_ids_labels(csv_path: str):
+def csv_names_mismatch(public_csv_path: Path, private_csv_path: Path):
+    public_names = set(read_csv_names(public_csv_path))
+    private_names = set(read_csv_names(private_csv_path))
+    extra_public = public_names - private_names
+    extra_private = private_names - public_names
+    return (extra_public, extra_private)
+
+
+def read_csv_ids_labels(csv_path: Path):
     return {
         name_to_id(name): f"{i+1}: {name or '[blank]'}"
         for i, name in enumerate(read_csv_names(csv_path))
     }
 
 
-def read_csv_ids_names(csv_path: str):
+def read_csv_ids_names(csv_path: Path):
     return {name_to_id(name): name for name in read_csv_names(csv_path)}
 
 

tests/utils/test_csv_helper.py

@@ -4,7 +4,24 @@
 import tempfile
 import pytest
 
-from dp_wizard.utils.csv_helper import read_csv_ids_labels, read_csv_ids_names
+from pathlib import Path
+
+from dp_wizard.utils.csv_helper import (
+    read_csv_ids_labels,
+    read_csv_ids_names,
+    csv_names_mismatch,
+)
+
+
+def test_csv_names_mismatch():
+    with tempfile.TemporaryDirectory() as tmp:
+        a_path = Path(tmp) / "a.csv"
+        a_path.write_text("a,b,c")
+        b_path = Path(tmp) / "b.csv"
+        b_path.write_text("b,c,d")
+        just_a, just_b = csv_names_mismatch(a_path, b_path)
+        assert just_a == {"a"}
+        assert just_b == {"d"}
 
 
 # We will not reference the encoding when reading: