Merge branch 'branch-22.12' into udf-string-class

README.md

@@ -65,32 +65,21 @@ Please see the [Demo Docker Repository](https://hub.docker.com/r/rapidsai/rapids
 
 cuDF can be installed with conda ([miniconda](https://conda.io/miniconda.html), or the full [Anaconda distribution](https://www.anaconda.com/download)) from the `rapidsai` channel:
 
-For `cudf version == 22.06` :
 ```bash
-# for CUDA 11.0
-conda install -c rapidsai -c nvidia -c numba -c conda-forge \
-    cudf=22.06 python=3.9 cudatoolkit=11.0
-
-# or, for CUDA 11.2
-conda install -c rapidsai -c nvidia -c numba -c conda-forge \
-    cudf=22.06 python=3.9 cudatoolkit=11.2
-
+# for CUDA 11.5
+conda install -c rapidsai -c conda-forge -c nvidia \
+    cudf=22.10 python=3.9 cudatoolkit=11.5
+# for CUDA 11.2
+conda install -c rapidsai -c conda-forge -c nvidia \
+    cudf=22.10 python=3.9 cudatoolkit=11.2
 ```
 
-For the nightly version of `cudf` :
-```bash
-# for CUDA 11.0
-conda install -c rapidsai-nightly -c nvidia -c numba -c conda-forge \
-    cudf python=3.9 cudatoolkit=11.0
-
-# or, for CUDA 11.2
-conda install -c rapidsai-nightly -c nvidia -c numba -c conda-forge \
-    cudf python=3.9 cudatoolkit=11.2
-```
+We also provide [nightly Conda packages](https://anaconda.org/rapidsai-nightly) built from the HEAD
+of our latest development branch.
 
 Note: cuDF is supported only on Linux, and with Python versions 3.8 and later.
 
-See the [Get RAPIDS version picker](https://rapids.ai/start.html) for more OS and version info. 
+See the [Get RAPIDS version picker](https://rapids.ai/start.html) for more OS and version info.
 
 ## Build/Install from Source
 See build [instructions](CONTRIBUTING.md#setting-up-your-build-environment).

conda/environments/cudf_dev_cuda11.5.yml

@@ -3,10 +3,10 @@
 name: cudf_dev
 channels:
   - rapidsai
-  - nvidia
   - rapidsai-nightly
   - dask/label/dev
   - conda-forge
+  - nvidia
 dependencies:
   - c-compiler
   - cxx-compiler
@@ -38,7 +38,7 @@ dependencies:
   - ipython
   - pandoc<=2.0.0
   - cudatoolkit=11.5
-  - cuda-python>=11.5,<11.7.1
+  - cuda-python>=11.7.1,<12.0
   - pip
   - doxygen=1.8.20
   - typing_extensions

conda/recipes/cudf/meta.yaml

@@ -62,7 +62,7 @@ requirements:
     - packaging
     - cachetools
     - cubinlinker  # [linux64]  # CUDA enhanced compatibility.
-    - cuda-python >=11.5,<11.7.1
+    - cuda-python >=11.7.1,<12.0
 test:                                   # [linux64]
   requires:                             # [linux64]
     - cudatoolkit {{ cuda_version }}.*  # [linux64]

conda/recipes/strings_udf/meta.yaml

@@ -40,7 +40,7 @@ requirements:
     - numba >=0.54
     - libcudf ={{ version }}
     - cudf ={{ version }}
-    - cudatoolkit ={{ cuda_version }}
+    - cudatoolkit {{ cuda_version }}.*
   run:
     - python
     - typing_extensions

cpp/include/cudf/detail/scatter.cuh

@@ -218,7 +218,8 @@ struct column_scatterer_impl<dictionary32> {
     // first combine keys so both dictionaries have the same set
     auto target_matched    = dictionary::detail::add_keys(target, source.keys(), stream, mr);
     auto const target_view = dictionary_column_view(target_matched->view());
-    auto source_matched    = dictionary::detail::set_keys(source, target_view.keys(), stream);
+    auto source_matched    = dictionary::detail::set_keys(
+      source, target_view.keys(), stream, rmm::mr::get_current_device_resource());
     auto const source_view = dictionary_column_view(source_matched->view());
 
     // now build the new indices by doing a scatter on just the matched indices

cpp/include/cudf/dictionary/detail/concatenate.hpp

@@ -37,10 +37,9 @@ namespace detail {
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New column with concatenated results.
  */
-std::unique_ptr<column> concatenate(
-  host_span<column_view const> columns,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> concatenate(host_span<column_view const> columns,
+                                    rmm::cuda_stream_view stream,
+                                    rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace dictionary

cpp/include/cudf/dictionary/detail/encode.hpp

@@ -51,11 +51,10 @@ namespace detail {
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return Returns a dictionary column.
  */
-std::unique_ptr<column> encode(
-  column_view const& column,
-  data_type indices_type              = data_type{type_id::UINT32},
-  rmm::cuda_stream_view stream        = cudf::get_default_stream(),
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> encode(column_view const& column,
+                               data_type indices_type,
+                               rmm::cuda_stream_view stream,
+                               rmm::mr::device_memory_resource* mr);
 
 /**
  * @brief Create a column by gathering the keys from the provided
@@ -72,10 +71,9 @@ std::unique_ptr<column> encode(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New column with type matching the dictionary_column's keys.
  */
-std::unique_ptr<column> decode(
-  dictionary_column_view const& dictionary_column,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> decode(dictionary_column_view const& dictionary_column,
+                               rmm::cuda_stream_view stream,
+                               rmm::mr::device_memory_resource* mr);
 
 /**
  * @brief Return minimal integer type for the given number of elements.

cpp/include/cudf/dictionary/detail/replace.hpp

@@ -39,11 +39,10 @@ namespace detail {
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New dictionary column with null rows replaced.
  */
-std::unique_ptr<column> replace_nulls(
-  dictionary_column_view const& input,
-  dictionary_column_view const& replacement,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> replace_nulls(dictionary_column_view const& input,
+                                      dictionary_column_view const& replacement,
+                                      rmm::cuda_stream_view stream,
+                                      rmm::mr::device_memory_resource* mr);
 
 /**
  * @brief Create a new dictionary column by replacing nulls with a
@@ -57,11 +56,10 @@ std::unique_ptr<column> replace_nulls(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New dictionary column with null rows replaced.
  */
-std::unique_ptr<column> replace_nulls(
-  dictionary_column_view const& input,
-  scalar const& replacement,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> replace_nulls(dictionary_column_view const& input,
+                                      scalar const& replacement,
+                                      rmm::cuda_stream_view stream,
+                                      rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace dictionary

cpp/include/cudf/dictionary/detail/search.hpp

@@ -31,11 +31,10 @@ namespace detail {
  *
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
-std::unique_ptr<scalar> get_index(
-  dictionary_column_view const& dictionary,
-  scalar const& key,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<scalar> get_index(dictionary_column_view const& dictionary,
+                                  scalar const& key,
+                                  rmm::cuda_stream_view stream,
+                                  rmm::mr::device_memory_resource* mr);
 
 /**
  * @brief Get the index for a key if it were added to the given dictionary.
@@ -56,11 +55,10 @@ std::unique_ptr<scalar> get_index(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return Numeric scalar index value of the key within the dictionary
  */
-std::unique_ptr<scalar> get_insert_index(
-  dictionary_column_view const& dictionary,
-  scalar const& key,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<scalar> get_insert_index(dictionary_column_view const& dictionary,
+                                         scalar const& key,
+                                         rmm::cuda_stream_view stream,
+                                         rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace dictionary

cpp/include/cudf/dictionary/detail/update_keys.hpp

@@ -32,46 +32,42 @@ namespace detail {
  *
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
-std::unique_ptr<column> add_keys(
-  dictionary_column_view const& dictionary_column,
-  column_view const& new_keys,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> add_keys(dictionary_column_view const& dictionary_column,
+                                 column_view const& new_keys,
+                                 rmm::cuda_stream_view stream,
+                                 rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc cudf::dictionary::remove_keys(dictionary_column_view const&,column_view
  * const&,mm::mr::device_memory_resource*)
  *
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
-std::unique_ptr<column> remove_keys(
-  dictionary_column_view const& dictionary_column,
-  column_view const& keys_to_remove,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> remove_keys(dictionary_column_view const& dictionary_column,
+                                    column_view const& keys_to_remove,
+                                    rmm::cuda_stream_view stream,
+                                    rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc cudf::dictionary::remove_unused_keys(dictionary_column_view
  * const&,mm::mr::device_memory_resource*)
  *
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
-std::unique_ptr<column> remove_unused_keys(
-  dictionary_column_view const& dictionary_column,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> remove_unused_keys(dictionary_column_view const& dictionary_column,
+                                           rmm::cuda_stream_view stream,
+                                           rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc cudf::dictionary::set_keys(dictionary_column_view
  * const&,mm::mr::device_memory_resource*)
  *
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
-std::unique_ptr<column> set_keys(
-  dictionary_column_view const& dictionary_column,
-  column_view const& keys,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> set_keys(dictionary_column_view const& dictionary_column,
+                                 column_view const& keys,
+                                 rmm::cuda_stream_view stream,
+                                 rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc
@@ -82,7 +78,7 @@ std::unique_ptr<column> set_keys(
 std::vector<std::unique_ptr<column>> match_dictionaries(
   cudf::host_span<dictionary_column_view const> input,
   rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+  rmm::mr::device_memory_resource* mr);
 
 /**
  * @brief Create new dictionaries that have keys merged from dictionary columns
@@ -106,7 +102,7 @@ std::vector<std::unique_ptr<column>> match_dictionaries(
 std::pair<std::vector<std::unique_ptr<column>>, std::vector<table_view>> match_dictionaries(
   std::vector<table_view> tables,
   rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+  rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace dictionary

cpp/include/nvtext/detail/tokenize.hpp

@@ -35,12 +35,10 @@ namespace detail {
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New strings columns of tokens.
  */
-std::unique_ptr<cudf::column> tokenize(
-  cudf::strings_column_view const& strings,
-  cudf::string_scalar const& delimiter = cudf::string_scalar{""},
-  // Move before delimiter?
-  rmm::cuda_stream_view stream        = cudf::get_default_stream(),
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<cudf::column> tokenize(cudf::strings_column_view const& strings,
+                                       cudf::string_scalar const& delimiter,
+                                       rmm::cuda_stream_view stream,
+                                       rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc nvtext::tokenize(strings_column_view const&,strings_column_view
@@ -52,11 +50,10 @@ std::unique_ptr<cudf::column> tokenize(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New strings columns of tokens.
  */
-std::unique_ptr<cudf::column> tokenize(
-  cudf::strings_column_view const& strings,
-  cudf::strings_column_view const& delimiters,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<cudf::column> tokenize(cudf::strings_column_view const& strings,
+                                       cudf::strings_column_view const& delimiters,
+                                       rmm::cuda_stream_view stream,
+                                       rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc nvtext::count_tokens(strings_column_view const&, string_scalar
@@ -69,12 +66,10 @@ std::unique_ptr<cudf::column> tokenize(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New INT32 column of token counts.
  */
-std::unique_ptr<cudf::column> count_tokens(
-  cudf::strings_column_view const& strings,
-  cudf::string_scalar const& delimiter = cudf::string_scalar{""},
-  // Move before delimiter?
-  rmm::cuda_stream_view stream        = cudf::get_default_stream(),
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<cudf::column> count_tokens(cudf::strings_column_view const& strings,
+                                           cudf::string_scalar const& delimiter,
+                                           rmm::cuda_stream_view stream,
+                                           rmm::mr::device_memory_resource* mr);
 
 /**
  * @copydoc nvtext::count_tokens(strings_column_view const&,strings_column_view
@@ -86,11 +81,10 @@ std::unique_ptr<cudf::column> count_tokens(
  * @param mr Device memory resource used to allocate the returned column's device memory.
  * @return New INT32 column of token counts.
  */
-std::unique_ptr<cudf::column> count_tokens(
-  cudf::strings_column_view const& strings,
-  cudf::strings_column_view const& delimiters,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<cudf::column> count_tokens(cudf::strings_column_view const& strings,
+                                           cudf::strings_column_view const& delimiters,
+                                           rmm::cuda_stream_view stream,
+                                           rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace nvtext

cpp/src/copying/copy_range.cu

@@ -172,7 +172,8 @@ std::unique_ptr<cudf::column> out_of_place_copy_range_dispatch::operator()<cudf:
   auto target_matched =
     cudf::dictionary::detail::add_keys(dict_target, dict_source.keys(), stream, mr);
   auto const target_view = cudf::dictionary_column_view(target_matched->view());
-  auto source_matched = cudf::dictionary::detail::set_keys(dict_source, target_view.keys(), stream);
+  auto source_matched    = cudf::dictionary::detail::set_keys(
+    dict_source, target_view.keys(), stream, rmm::mr::get_current_device_resource());
   auto const source_view = cudf::dictionary_column_view(source_matched->view());
 
   // build the new indices by calling in_place_copy_range on just the indices

cpp/src/copying/scatter.cu

@@ -184,8 +184,9 @@ struct column_scalar_scatterer_impl<dictionary32, MapIterator> {
                                    stream,
                                    mr);
     auto dict_view    = dictionary_column_view(dict_target->view());
-    auto scalar_index = dictionary::detail::get_index(dict_view, source.get(), stream);
-    auto scalar_iter  = thrust::make_permutation_iterator(
+    auto scalar_index = dictionary::detail::get_index(
+      dict_view, source.get(), stream, rmm::mr::get_current_device_resource());
+    auto scalar_iter = thrust::make_permutation_iterator(
       indexalator_factory::make_input_iterator(*scalar_index), thrust::make_constant_iterator(0));
     auto new_indices = std::make_unique<column>(dict_view.get_indices_annotated(), stream, mr);
     auto target_iter = indexalator_factory::make_output_iterator(new_indices->mutable_view());

cpp/src/dictionary/add_keys.cu

@@ -44,11 +44,10 @@ namespace detail {
  * d2 is now {[a, b, c, d, e, f], [5, 0, 3, 1, 2, 2, 2, 5, 0]}
  * ```
  */
-std::unique_ptr<column> add_keys(
-  dictionary_column_view const& dictionary_column,
-  column_view const& new_keys,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+std::unique_ptr<column> add_keys(dictionary_column_view const& dictionary_column,
+                                 column_view const& new_keys,
+                                 rmm::cuda_stream_view stream,
+                                 rmm::mr::device_memory_resource* mr)
 {
   CUDF_EXPECTS(!new_keys.has_nulls(), "Keys must not have nulls");
   auto old_keys = dictionary_column.keys();  // [a,b,c,d,f]