Merge pull request #1604 from fsprojects/repo-assist/fix-csv-schema-paren-946-6367eea86dd341c8

[Repo Assist] Fix CSV schema parsing: column names containing parentheses corrupt type annotation

Author: dsyme

src/FSharp.Data.Csv.Core/CsvInference.fs

@@ -36,7 +36,10 @@ let private nameToTypeForCsv =
     |> dict
 
 let private nameAndTypeRegex =
-    lazy Regex(@"^(?<name>.+)\((?<type>.+)\)$", RegexOptions.Compiled ||| RegexOptions.RightToLeft)
+    // Note: do NOT use RightToLeft here — it causes incorrect splits when the column name
+    // itself contains parentheses (e.g. "Na(  )me (type)" would give name="Na", type="  )me (type").
+    // Without RightToLeft, greedy backtracking correctly matches the *last* "(type)" group.
+    lazy Regex(@"^(?<name>.+)\((?<type>.+)\)$", RegexOptions.Compiled)
 
 let private overrideByNameRegex =
     lazy

tests/FSharp.Data.DesignTime.Tests/InferenceTests.fs

@@ -359,6 +359,20 @@ let ``Inference schema override by parameter``() =
 
   actual |> should equal expected
 
+[<Test>]
+let ``Column name with parentheses is parsed correctly in schema`` () =
+  // Regression test for https://github.com/fsprojects/FSharp.Data/issues/946
+  // A column name like "Na(  )me" must not be split at the first '(' when the
+  // schema type annotation "(int)" appears at the end.
+  let source = CsvFile.Parse("Na(  )me,other\n1,2")
+  let actual =
+    inferType source Int32.MaxValue [||] culture "Na(  )me (int),other (string)" false false
+    ||> CsvInference.getFields false
+    |> List.map (fun field -> field.Name, field.Value.RuntimeType)
+
+  let expected = [ "Na(  )me", typeof<int>; "other", typeof<string> ]
+  actual |> should equal expected
+
 [<Test>]
 let ``Doesn't infer 12-002 as a date``() =
   // a previous version inferred a IntOrStringOrDateTime