fix: Fix padding for non-ASCII strings

crates/polars-ops/src/chunked_array/strings/pad.rs

@@ -6,7 +6,7 @@ pub(super) fn pad_end<'a>(ca: &'a Utf8Chunked, length: usize, fill_char: char) -
     // amortize allocation
     let mut buf = String::new();
     let f = |s: &'a str| {
-        let padding = length.saturating_sub(s.len());
+        let padding = length.saturating_sub(s.chars().count());
         if padding == 0 {
             s
         } else {
@@ -28,7 +28,7 @@ pub(super) fn pad_start<'a>(ca: &'a Utf8Chunked, length: usize, fill_char: char)
     // amortize allocation
     let mut buf = String::new();
     let f = |s: &'a str| {
-        let padding = length.saturating_sub(s.len());
+        let padding = length.saturating_sub(s.chars().count());
         if padding == 0 {
             s
         } else {

py-polars/polars/expr/string.py

@@ -854,6 +854,11 @@ def zfill(self, length: int) -> Expr:
         --------
         pad_start
 
+        Notes
+        -----
+        This method is intended for padding numeric strings. If your data contains
+        non-ASCII characters, use :func:`pad_start` instead.
+
         Examples
         --------
         >>> df = pl.DataFrame({"a": [-1, 123, 999999, None]})

py-polars/polars/series/string.py

@@ -1355,6 +1355,11 @@ def zfill(self, length: int) -> Series:
         --------
         pad_start
 
+        Notes
+        -----
+        This method is intended for padding numeric strings. If your data contains
+        non-ASCII characters, use :func:`pad_start` instead.
+
         Examples
         --------
         >>> s = pl.Series([-1, 123, 999999, None])

py-polars/tests/unit/namespaces/string/test_pad.py

@@ -86,3 +86,30 @@ def test_str_rjust_deprecated() -> None:
 
     expected = pl.Series(["    a", "   bc", "  def"])
     assert_series_equal(result, expected)
+
+
+def test_pad_end_unicode() -> None:
+    lf = pl.LazyFrame({"a": ["Café", "345", "東京", None]})
+
+    result = lf.select(pl.col("a").str.pad_end(6, "日"))
+
+    expected = pl.LazyFrame({"a": ["Café日日", "345日日日", "東京日日日日", None]})
+    assert_frame_equal(result, expected)
+
+
+def test_pad_start_unicode() -> None:
+    lf = pl.LazyFrame({"a": ["Café", "345", "東京", None]})
+
+    result = lf.select(pl.col("a").str.pad_start(6, "日"))
+
+    expected = pl.LazyFrame({"a": ["日日Café", "日日日345", "日日日日東京", None]})
+    assert_frame_equal(result, expected)
+
+
+def test_str_zfill_unicode_not_respected() -> None:
+    lf = pl.LazyFrame({"a": ["Café", "345", "東京", None]})
+
+    result = lf.select(pl.col("a").str.zfill(6))
+
+    expected = pl.LazyFrame({"a": ["0Café", "000345", "東京", None]})
+    assert_frame_equal(result, expected)