Merge remote-tracking branch 'upstream/branch-24.10' into dask-cudf-a…

…rrow-filesystem

ci/cudf_pandas_scripts/pandas-tests/job-summary.py

@@ -68,8 +68,18 @@ def emoji_failed(x):
 pr_df = pd.DataFrame.from_dict(pr_results, orient="index").sort_index()
 main_df = pd.DataFrame.from_dict(main_results, orient="index").sort_index()
 diff_df = pr_df - main_df
+total_usage = pr_df['_slow_function_call'] + pr_df['_fast_function_call']
+pr_df['CPU Usage'] = ((pr_df['_slow_function_call']/total_usage)*100.0).round(1)
+pr_df['GPU Usage'] = ((pr_df['_fast_function_call']/total_usage)*100.0).round(1)
 
-pr_df = pr_df[["total", "passed", "failed", "skipped"]]
+cpu_usage_mean = pr_df['CPU Usage'].mean().round(2)
+gpu_usage_mean = pr_df['GPU Usage'].mean().round(2)
+
+# Add '%' suffix to 'CPU Usage' and 'GPU Usage' columns
+pr_df['CPU Usage'] = pr_df['CPU Usage'].fillna(0).astype(str) + '%'
+pr_df['GPU Usage'] = pr_df['GPU Usage'].fillna(0).astype(str) + '%'
+
+pr_df = pr_df[["total", "passed", "failed", "skipped", 'CPU Usage', 'GPU Usage']]
 diff_df = diff_df[["total", "passed", "failed", "skipped"]]
 diff_df.columns = diff_df.columns + "_diff"
 diff_df["passed_diff"] = diff_df["passed_diff"].map(emoji_passed)
@@ -95,6 +105,8 @@ def emoji_failed(x):
 
 print(comment)
 print()
+print(f"Average CPU and GPU usage for the tests: {cpu_usage_mean}% and {gpu_usage_mean}%")
+print()
 print("Here are the results of running the Pandas tests against this PR:")
 print()
 print(df.to_markdown())

cpp/include/cudf/detail/aggregation/aggregation.hpp

@@ -1497,8 +1497,7 @@ AGG_KIND_MAPPING(aggregation::VARIANCE, var_aggregation);
  *
  * @tparam F Type of callable
  * @param k The `aggregation::Kind` value to dispatch
- * aram f The callable that accepts an `aggregation::Kind` non-type template
- * argument.
+ * @param f The callable that accepts an `aggregation::Kind` callable function object.
  * @param args Parameter pack forwarded to the `operator()` invocation
  * @return Forwards the return value of the callable.
  */
@@ -1626,6 +1625,7 @@ struct dispatch_source {
  * parameter of the callable `F`
  * @param k The `aggregation::Kind` used to dispatch an `aggregation::Kind`
  * non-type template parameter for the second template parameter of the callable
+ * @param f The callable that accepts `data_type` and `aggregation::Kind` function object.
  * @param args Parameter pack forwarded to the `operator()` invocation
  * `F`.
  */
@@ -1644,8 +1644,8 @@ CUDF_HOST_DEVICE inline constexpr decltype(auto) dispatch_type_and_aggregation(d
  * @brief Returns the target `data_type` for the specified aggregation  k
  * performed on elements of type  source_type.
  *
- * aram source_type The element type to be aggregated
- * aram k The aggregation
+ * @param source_type The element type to be aggregated
+ * @param k The aggregation kind
  * @return data_type The target_type of  k performed on  source_type
  * elements
  */

docs/cudf/source/user_guide/api_docs/pylibcudf/strings/extract.rst

@@ -0,0 +1,6 @@
+=======
+extract
+=======
+
+.. automodule:: pylibcudf.strings.extract
+   :members:

docs/cudf/source/user_guide/api_docs/pylibcudf/strings/index.rst

@@ -7,8 +7,10 @@ strings
     capitalize
     char_types
     contains
+    extract
     find
     regex_flags
     regex_program
+    repeat
     replace
     slice

docs/cudf/source/user_guide/api_docs/pylibcudf/strings/repeat.rst

@@ -0,0 +1,6 @@
+======
+repeat
+======
+
+.. automodule:: pylibcudf.strings.repeat
+   :members:

python/cudf/cudf/_lib/strings/contains.pyx

@@ -1,27 +1,10 @@
 # Copyright (c) 2020-2024, NVIDIA CORPORATION.
 
-from cython.operator cimport dereference
 from libc.stdint cimport uint32_t
 
 from cudf.core.buffer import acquire_spill_lock
 
-from libcpp.memory cimport unique_ptr
-from libcpp.string cimport string
-from libcpp.utility cimport move
-
-from pylibcudf.libcudf.column.column cimport column
-from pylibcudf.libcudf.column.column_view cimport column_view
-from pylibcudf.libcudf.scalar.scalar cimport string_scalar
-from pylibcudf.libcudf.strings.contains cimport (
-    count_re as cpp_count_re,
-    like as cpp_like,
-    matches_re as cpp_matches_re,
-)
-from pylibcudf.libcudf.strings.regex_flags cimport regex_flags
-from pylibcudf.libcudf.strings.regex_program cimport regex_program
-
 from cudf._lib.column cimport Column
-from cudf._lib.scalar cimport DeviceScalar
 
 from pylibcudf.strings import contains
 from pylibcudf.strings.regex_program import RegexProgram
@@ -45,21 +28,10 @@ def count_re(Column source_strings, object reg_ex, uint32_t flags):
     Returns a Column with count of occurrences of `reg_ex` in
     each string of `source_strings`
     """
-    cdef unique_ptr[column] c_result
-    cdef column_view source_view = source_strings.view()
-
-    cdef string reg_ex_string = <string>str(reg_ex).encode()
-    cdef regex_flags c_flags = <regex_flags>flags
-    cdef unique_ptr[regex_program] c_prog
-
-    with nogil:
-        c_prog = move(regex_program.create(reg_ex_string, c_flags))
-        c_result = move(cpp_count_re(
-            source_view,
-            dereference(c_prog)
-        ))
-
-    return Column.from_unique_ptr(move(c_result))
+    prog = RegexProgram.create(str(reg_ex), flags)
+    return Column.from_pylibcudf(
+        contains.count_re(source_strings.to_pylibcudf(mode="read"), prog)
+    )
 
 
 @acquire_spill_lock()
@@ -68,21 +40,10 @@ def match_re(Column source_strings, object reg_ex, uint32_t flags):
     Returns a Column with each value True if the string matches `reg_ex`
     regular expression with each record of `source_strings`
     """
-    cdef unique_ptr[column] c_result
-    cdef column_view source_view = source_strings.view()
-
-    cdef string reg_ex_string = <string>str(reg_ex).encode()
-    cdef regex_flags c_flags = <regex_flags>flags
-    cdef unique_ptr[regex_program] c_prog
-
-    with nogil:
-        c_prog = move(regex_program.create(reg_ex_string, c_flags))
-        c_result = move(cpp_matches_re(
-            source_view,
-            dereference(c_prog)
-        ))
-
-    return Column.from_unique_ptr(move(c_result))
+    prog = RegexProgram.create(str(reg_ex), flags)
+    return Column.from_pylibcudf(
+        contains.matches_re(source_strings.to_pylibcudf(mode="read"), prog)
+    )
 
 
 @acquire_spill_lock()
@@ -91,24 +52,9 @@ def like(Column source_strings, object py_pattern, object py_escape):
     Returns a Column with each value True if the string matches the
     `py_pattern` like expression with each record of `source_strings`
     """
-    cdef unique_ptr[column] c_result
-    cdef column_view source_view = source_strings.view()
-
-    cdef DeviceScalar pattern = py_pattern.device_value
-    cdef DeviceScalar escape = py_escape.device_value
-
-    cdef const string_scalar* scalar_ptn = <const string_scalar*>(
-        pattern.get_raw_ptr()
-    )
-    cdef const string_scalar* scalar_esc = <const string_scalar*>(
-        escape.get_raw_ptr()
+    plc_column = contains.like(
+        source_strings.to_pylibcudf(mode="read"),
+        py_pattern.device_value.c_value,
+        py_escape.device_value.c_value,
     )
-
-    with nogil:
-        c_result = move(cpp_like(
-            source_view,
-            scalar_ptn[0],
-            scalar_esc[0]
-        ))
-
-    return Column.from_unique_ptr(move(c_result))
+    return Column.from_pylibcudf(plc_column)

python/cudf/cudf/_lib/strings/extract.pyx

@@ -1,21 +1,12 @@
 # Copyright (c) 2020-2024, NVIDIA CORPORATION.
 
-from cython.operator cimport dereference
 from libc.stdint cimport uint32_t
-from libcpp.memory cimport unique_ptr
-from libcpp.string cimport string
-from libcpp.utility cimport move
 
 from cudf.core.buffer import acquire_spill_lock
 
-from pylibcudf.libcudf.column.column_view cimport column_view
-from pylibcudf.libcudf.strings.extract cimport extract as cpp_extract
-from pylibcudf.libcudf.strings.regex_flags cimport regex_flags
-from pylibcudf.libcudf.strings.regex_program cimport regex_program
-from pylibcudf.libcudf.table.table cimport table
-
 from cudf._lib.column cimport Column
-from cudf._lib.utils cimport data_from_unique_ptr
+
+import pylibcudf as plc
 
 
 @acquire_spill_lock()
@@ -26,21 +17,8 @@ def extract(Column source_strings, object pattern, uint32_t flags):
     The returning data contains one row for each subject string,
     and one column for each group.
     """
-    cdef unique_ptr[table] c_result
-    cdef column_view source_view = source_strings.view()
-
-    cdef string pattern_string = <string>str(pattern).encode()
-    cdef regex_flags c_flags = <regex_flags>flags
-    cdef unique_ptr[regex_program] c_prog
-
-    with nogil:
-        c_prog = move(regex_program.create(pattern_string, c_flags))
-        c_result = move(cpp_extract(
-            source_view,
-            dereference(c_prog)
-        ))
-
-    return data_from_unique_ptr(
-        move(c_result),
-        column_names=range(0, c_result.get()[0].num_columns())
+    prog = plc.strings.regex_program.RegexProgram.create(str(pattern), flags)
+    plc_result = plc.strings.extract.extract(
+        source_strings.to_pylibcudf(mode="read"), prog
     )
+    return dict(enumerate(Column.from_pylibcudf(col) for col in plc_result.columns()))

python/cudf/cudf/_lib/strings/repeat.pyx

@@ -1,17 +1,12 @@
 # Copyright (c) 2021-2024, NVIDIA CORPORATION.
-
-from libcpp.memory cimport unique_ptr
-from libcpp.utility cimport move
-
 from cudf.core.buffer import acquire_spill_lock
 
-from pylibcudf.libcudf.column.column cimport column
-from pylibcudf.libcudf.column.column_view cimport column_view
-from pylibcudf.libcudf.strings cimport repeat as cpp_repeat
 from pylibcudf.libcudf.types cimport size_type
 
 from cudf._lib.column cimport Column
 
+import pylibcudf as plc
+
 
 @acquire_spill_lock()
 def repeat_scalar(Column source_strings,
@@ -21,16 +16,11 @@ def repeat_scalar(Column source_strings,
     each string in `source_strings`
     `repeats` number of times.
     """
-    cdef unique_ptr[column] c_result
-    cdef column_view source_view = source_strings.view()
-
-    with nogil:
-        c_result = move(cpp_repeat.repeat_strings(
-            source_view,
-            repeats
-        ))
-
-    return Column.from_unique_ptr(move(c_result))
+    plc_result = plc.strings.repeat.repeat_strings(
+        source_strings.to_pylibcudf(mode="read"),
+        repeats
+    )
+    return Column.from_pylibcudf(plc_result)
 
 
 @acquire_spill_lock()
@@ -41,14 +31,8 @@ def repeat_sequence(Column source_strings,
     each string in `source_strings`
     `repeats` number of times.
     """
-    cdef unique_ptr[column] c_result
-    cdef column_view source_view = source_strings.view()
-    cdef column_view repeats_view = repeats.view()
-
-    with nogil:
-        c_result = move(cpp_repeat.repeat_strings(
-            source_view,
-            repeats_view
-        ))
-
-    return Column.from_unique_ptr(move(c_result))
+    plc_result = plc.strings.repeat.repeat_strings(
+        source_strings.to_pylibcudf(mode="read"),
+        repeats.to_pylibcudf(mode="read")
+    )
+    return Column.from_pylibcudf(plc_result)

python/cudf/cudf/core/column/string.py

@@ -623,11 +623,9 @@ def extract(
                 "unsupported value for `flags` parameter"
             )
 
-        data, _ = libstrings.extract(self._column, pat, flags)
+        data = libstrings.extract(self._column, pat, flags)
         if len(data) == 1 and expand is False:
-            data = next(iter(data.values()))
-        else:
-            data = data
+            _, data = data.popitem()
         return self._return_or_inplace(data, expand=expand)
 
     def contains(