Fix regex non-multiline EOL/$ matching strings ending with a new-line (…

…#9715)

Closes #9620 

Fixes an edge case described in https://docs.python.org/3/library/re.html#re.MULTILINE
where the '$' EOL regex pattern character (without `MULTILINE` set) should match at the very end of a string and also just before the end of the string if the end of that string contains a new-line.

Authors:
  - David Wendt (https://github.com/davidwendt)

Approvers:
  - Christopher Harris (https://github.com/cwharris)
  - Vukasin Milovanovic (https://github.com/vuule)
  - Sheilah Kirui (https://github.com/skirui-source)

URL: #9715

cpp/src/strings/regex/regex.inl

@@ -276,7 +276,10 @@ __device__ inline int32_t reprog_device::regexec(
             }
             break;
           case EOL:
-            if (last_character || (inst->u1.c == '$' && c == '\n')) {
+            if (last_character ||
+                (c == '\n' && (inst->u1.c == '$' ||
+                               // edge case where \n appears at the end of the string
+                               pos + 1 == dstr.length()))) {
               id_activate = inst->u2.next_id;
               expanded    = true;
             }

cpp/tests/strings/contains_tests.cpp

@@ -302,28 +302,29 @@ TEST_F(StringsContainsTests, CountTest)
 
 TEST_F(StringsContainsTests, MultiLine)
 {
-  auto input = cudf::test::strings_column_wrapper({"abc\nfff\nabc", "fff\nabc\nlll", "abc", ""});
-  auto view  = cudf::strings_column_view(input);
+  auto input =
+    cudf::test::strings_column_wrapper({"abc\nfff\nabc", "fff\nabc\nlll", "abc", "", "abc\n"});
+  auto view = cudf::strings_column_view(input);
 
   auto results = cudf::strings::contains_re(view, "^abc$", cudf::strings::regex_flags::MULTILINE);
-  auto expected_contains = cudf::test::fixed_width_column_wrapper<bool>({1, 1, 1, 0});
+  auto expected_contains = cudf::test::fixed_width_column_wrapper<bool>({1, 1, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_contains);
   results           = cudf::strings::contains_re(view, "^abc$");
-  expected_contains = cudf::test::fixed_width_column_wrapper<bool>({0, 0, 1, 0});
+  expected_contains = cudf::test::fixed_width_column_wrapper<bool>({0, 0, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_contains);
 
   results = cudf::strings::matches_re(view, "^abc$", cudf::strings::regex_flags::MULTILINE);
-  auto expected_matches = cudf::test::fixed_width_column_wrapper<bool>({1, 0, 1, 0});
+  auto expected_matches = cudf::test::fixed_width_column_wrapper<bool>({1, 0, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_matches);
   results          = cudf::strings::matches_re(view, "^abc$");
-  expected_matches = cudf::test::fixed_width_column_wrapper<bool>({0, 0, 1, 0});
+  expected_matches = cudf::test::fixed_width_column_wrapper<bool>({0, 0, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_matches);
 
   results = cudf::strings::count_re(view, "^abc$", cudf::strings::regex_flags::MULTILINE);
-  auto expected_count = cudf::test::fixed_width_column_wrapper<int32_t>({2, 1, 1, 0});
+  auto expected_count = cudf::test::fixed_width_column_wrapper<int32_t>({2, 1, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_count);
   results        = cudf::strings::count_re(view, "^abc$");
-  expected_count = cudf::test::fixed_width_column_wrapper<int32_t>({0, 0, 1, 0});
+  expected_count = cudf::test::fixed_width_column_wrapper<int32_t>({0, 0, 1, 0, 1});
   CUDF_TEST_EXPECT_COLUMNS_EQUIVALENT(*results, expected_count);
 }
 

python/cudf/cudf/tests/test_string.py

@@ -1746,12 +1746,13 @@ def test_string_wrap(data, width):
         ["A B", "1.5", "3,000"],
         ["23", "³", "⅕", ""],
         [" ", "\t\r\n ", ""],
-        ["$", "B", "Aab$", "$$ca", "C$B$", "cat"],
+        ["$", "B", "Aab$", "$$ca", "C$B$", "cat", "cat\n"],
         ["line\nto be wrapped", "another\nline\nto be wrapped"],
     ],
 )
 @pytest.mark.parametrize(
-    "pat", ["a", " ", "\t", "another", "0", r"\$", "^line$", "line.*be"]
+    "pat",
+    ["a", " ", "\t", "another", "0", r"\$", "^line$", "line.*be", "cat$"],
 )
 @pytest.mark.parametrize("flags", [0, re.MULTILINE, re.DOTALL])
 def test_string_count(data, pat, flags):