Merge remote-tracking branch 'upstream/branch-22.02' into remove-meth…

…od-parameter-from-merge-and-join

CONTRIBUTING.md

@@ -86,7 +86,7 @@ git submodule update --init --remote --recursive
 ```bash
 # create the conda environment (assuming in base `cudf` directory)
 # note: RAPIDS currently doesn't support `channel_priority: strict`; use `channel_priority: flexible` instead
-conda env create --name cudf_dev --file conda/environments/cudf_dev_cuda11.0.yml
+conda env create --name cudf_dev --file conda/environments/cudf_dev_cuda11.5.yml
 # activate the environment
 conda activate cudf_dev
 ```

cpp/CMakeLists.txt

@@ -343,6 +343,7 @@ add_library(
   src/lists/lists_column_factories.cu
   src/lists/lists_column_view.cu
   src/lists/segmented_sort.cu
+  src/lists/sequences.cu
   src/merge/merge.cu
   src/partitioning/partitioning.cu
   src/partitioning/round_robin.cu
@@ -416,7 +417,8 @@ add_library(
   src/strings/copying/concatenate.cu
   src/strings/copying/copying.cu
   src/strings/copying/shift.cu
-  src/strings/extract.cu
+  src/strings/extract/extract.cu
+  src/strings/extract/extract_all.cu
   src/strings/filling/fill.cu
   src/strings/filter_chars.cu
   src/strings/findall.cu

cpp/cmake/thirdparty/get_cucollections.cmake

@@ -21,7 +21,7 @@ function(find_and_configure_cucollections)
     cuco 0.0
     GLOBAL_TARGETS cuco::cuco
     CPM_ARGS GITHUB_REPOSITORY NVIDIA/cuCollections
-    GIT_TAG 6433e8ad7571f14cc5384051b049029c60dd1ce0
+    GIT_TAG 193de1aa74f5721717f991ca757dc610c852bb17
     OPTIONS "BUILD_TESTS OFF" "BUILD_BENCHMARKS OFF" "BUILD_EXAMPLES OFF"
   )
 

cpp/cmake/thirdparty/get_thrust.cmake

@@ -80,6 +80,6 @@ function(find_and_configure_thrust VERSION)
   endif()
 endfunction()
 
-set(CUDF_MIN_VERSION_Thrust 1.12.0)
+set(CUDF_MIN_VERSION_Thrust 1.15.0)
 
 find_and_configure_thrust(${CUDF_MIN_VERSION_Thrust})

cpp/include/cudf/datetime.hpp

@@ -285,280 +285,66 @@ std::unique_ptr<cudf::column> extract_quarter(
   cudf::column_view const& column,
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
-/** @} */  // end of group
-
-/**
- * @brief Round up to the nearest day
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> ceil_day(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest hour
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> ceil_hour(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest minute
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> ceil_minute(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest second
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> ceil_second(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest millisecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> ceil_millisecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest microsecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> ceil_microsecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round up to the nearest nanosecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> ceil_nanosecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
 /**
- * @brief Round down to the nearest day
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
+ * @brief Fixed frequencies supported by datetime rounding functions ceil, floor, round.
  *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
  */
-std::unique_ptr<cudf::column> floor_day(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+enum class rounding_frequency : int32_t {
+  DAY,
+  HOUR,
+  MINUTE,
+  SECOND,
+  MILLISECOND,
+  MICROSECOND,
+  NANOSECOND
+};
 
 /**
- * @brief Round down to the nearest hour
+ * @brief Round datetimes up to the nearest multiple of the given frequency.
  *
- * @param column cudf::column_view of the input datetime values
+ * @param column cudf::column_view of the input datetime values.
+ * @param freq rounding_frequency indicating the frequency to round up to.
  * @param mr Device memory resource used to allocate device memory of the returned column.
  *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
+ * @throw cudf::logic_error if input column datatype is not TIMESTAMP.
+ * @return cudf::column of the same datetime resolution as the input column.
  */
-std::unique_ptr<cudf::column> floor_hour(
+std::unique_ptr<cudf::column> ceil_datetimes(
   cudf::column_view const& column,
+  rounding_frequency freq,
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
 /**
- * @brief Round down to the nearest minute
+ * @brief Round datetimes down to the nearest multiple of the given frequency.
  *
- * @param column cudf::column_view of the input datetime values
+ * @param column cudf::column_view of the input datetime values.
+ * @param freq rounding_frequency indicating the frequency to round down to.
  * @param mr Device memory resource used to allocate device memory of the returned column.
  *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
+ * @throw cudf::logic_error if input column datatype is not TIMESTAMP.
+ * @return cudf::column of the same datetime resolution as the input column.
  */
-std::unique_ptr<cudf::column> floor_minute(
+std::unique_ptr<cudf::column> floor_datetimes(
   cudf::column_view const& column,
+  rounding_frequency freq,
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
 /**
- * @brief Round down to the nearest second
+ * @brief Round datetimes to the nearest multiple of the given frequency.
  *
- * @param column cudf::column_view of the input datetime values
+ * @param column cudf::column_view of the input datetime values.
+ * @param freq rounding_frequency indicating the frequency to round to.
  * @param mr Device memory resource used to allocate device memory of the returned column.
  *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
+ * @throw cudf::logic_error if input column datatype is not TIMESTAMP.
+ * @return cudf::column of the same datetime resolution as the input column.
  */
-std::unique_ptr<cudf::column> floor_second(
+std::unique_ptr<cudf::column> round_datetimes(
   cudf::column_view const& column,
+  rounding_frequency freq,
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
-/**
- * @brief Round down to the nearest millisecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> floor_millisecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round down to the nearest microsecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> floor_microsecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round down to the nearest nanosecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> floor_nanosecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest day
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> round_day(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest hour
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> round_hour(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest minute
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> round_minute(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest second
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<cudf::column> round_second(
-  cudf::column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest millisecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> round_millisecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest microsecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> round_microsecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
-
-/**
- * @brief Round to the nearest nanosecond
- *
- * @param column cudf::column_view of the input datetime values
- * @param mr Device memory resource used to allocate device memory of the returned column.
- *
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- * @return cudf::column of the same datetime resolution as the input column
- */
-std::unique_ptr<column> round_nanosecond(
-  column_view const& column,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+/** @} */  // end of group
 
 }  // namespace datetime
 }  // namespace cudf

cpp/include/cudf/detail/hashing.hpp

@@ -32,17 +32,15 @@ namespace detail {
  */
 std::unique_ptr<column> hash(
   table_view const& input,
-  hash_id hash_function                        = hash_id::HASH_MURMUR3,
-  cudf::host_span<uint32_t const> initial_hash = {},
-  uint32_t seed                                = 0,
-  rmm::cuda_stream_view stream                 = rmm::cuda_stream_default,
-  rmm::mr::device_memory_resource* mr          = rmm::mr::get_current_device_resource());
+  hash_id hash_function               = hash_id::HASH_MURMUR3,
+  uint32_t seed                       = 0,
+  rmm::cuda_stream_view stream        = rmm::cuda_stream_default,
+  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
 std::unique_ptr<column> murmur_hash3_32(
   table_view const& input,
-  cudf::host_span<uint32_t const> initial_hash = {},
-  rmm::cuda_stream_view stream                 = rmm::cuda_stream_default,
-  rmm::mr::device_memory_resource* mr          = rmm::mr::get_current_device_resource());
+  rmm::cuda_stream_view stream        = rmm::cuda_stream_default,
+  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
 std::unique_ptr<column> md5_hash(
   table_view const& input,