Merge branch 'branch-24.06' into 15668

conda/recipes/libcudf/meta.yaml

@@ -86,6 +86,9 @@ outputs:
         {% else %}
         - {{ compiler('cuda') }}
         {% endif %}
+        # TODO: start taking libarrow's run exports again wwhen they're correct for 16.0
+        # ref: https://github.com/conda-forge/arrow-cpp-feedstock/issues/1418
+        - libarrow
     requirements:
       build:
         - cmake {{ cmake_version }}
@@ -105,6 +108,12 @@ outputs:
         - librmm ={{ minor_version }}
         - libkvikio ={{ minor_version }}
         - dlpack {{ dlpack_version }}
+        # TODO: start taking libarrow's run exports again wwhen they're correct for 16.0
+        # ref: https://github.com/conda-forge/arrow-cpp-feedstock/issues/1418
+        - libarrow>=16.0.0,<16.1.0a0
+        - libarrow-acero>=16.0.0,<16.1.0a0
+        - libarrow-dataset>=16.0.0,<16.1.0a0
+        - libparquet>=16.0.0,<16.1.0a0
     test:
       commands:
         - test -f $PREFIX/lib/libcudf.so

cpp/include/cudf/detail/utilities/stream_pool.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2023, NVIDIA CORPORATION.
+ * Copyright (c) 2023-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -81,6 +81,11 @@ class cuda_stream_pool {
  */
 cuda_stream_pool* create_global_cuda_stream_pool();
 
+/**
+ * @brief Get the global stream pool.
+ */
+cuda_stream_pool& global_cuda_stream_pool();
+
 /**
  * @brief Acquire a set of `cuda_stream_view` objects and synchronize them to an event on another
  * stream.

cpp/include/cudf/io/json.hpp

@@ -101,6 +101,8 @@ class json_reader_options {
   bool _lines = false;
   // Parse mixed types as a string column
   bool _mixed_types_as_string = false;
+  // Delimiter separating records in JSON lines
+  char _delimiter = '\n';
   // Prune columns on read, selected based on the _dtypes option
   bool _prune_columns = false;
 
@@ -229,6 +231,13 @@ class json_reader_options {
     return base_padding + num_columns * column_bytes;
   }
 
+  /**
+   * @brief Returns delimiter separating records in JSON lines
+   *
+   * @return Delimiter separating records in JSON lines
+   */
+  char get_delimiter() const { return _delimiter; }
+
   /**
    * @brief Whether to read the file as a json object per line.
    *
@@ -340,6 +349,30 @@ class json_reader_options {
    */
   void set_byte_range_size(size_type size) { _byte_range_size = size; }
 
+  /**
+   * @brief Set delimiter separating records in JSON lines
+   *
+   * @param delimiter Delimiter separating records in JSON lines
+   */
+  void set_delimiter(char delimiter)
+  {
+    switch (delimiter) {
+      case '{':
+      case '[':
+      case '}':
+      case ']':
+      case ',':
+      case ':':
+      case '"':
+      case '\'':
+      case '\\':
+      case ' ':
+      case '\t':
+      case '\r': CUDF_FAIL("Unsupported delimiter character.", std::invalid_argument); break;
+    }
+    _delimiter = delimiter;
+  }
+
   /**
    * @brief Set whether to read the file as a json object per line.
    *
@@ -507,6 +540,18 @@ class json_reader_options_builder {
     return *this;
   }
 
+  /**
+   * @brief Set delimiter separating records in JSON lines
+   *
+   * @param delimiter Delimiter separating records in JSON lines
+   * @return this for chaining
+   */
+  json_reader_options_builder& delimiter(char delimiter)
+  {
+    options.set_delimiter(delimiter);
+    return *this;
+  }
+
   /**
    * @brief Set whether to read the file as a json object per line.
    *

cpp/include/cudf/io/memory_resource.hpp

@@ -18,6 +18,8 @@
 
 #include <rmm/resource_ref.hpp>
 
+#include <optional>
+
 namespace cudf::io {
 
 /**
@@ -41,4 +43,21 @@ rmm::host_async_resource_ref set_host_memory_resource(rmm::host_async_resource_r
  */
 rmm::host_async_resource_ref get_host_memory_resource();
 
+/**
+ * @brief Options to configure the default host memory resource
+ */
+struct host_mr_options {
+  std::optional<size_t> pool_size;  ///< The size of the pool to use for the default host memory
+                                    ///< resource. If not set, the default pool size is used.
+};
+
+/**
+ * @brief Configure the size of the default host memory resource.
+ *
+ * @throws cudf::logic_error if called after the default host memory resource has been created
+ *
+ * @param opts Options to configure the default host memory resource
+ */
+void config_default_host_memory_resource(host_mr_options const& opts);
+
 }  // namespace cudf::io

cpp/src/io/json/nested_json.hpp

@@ -59,8 +59,8 @@ enum class stack_behavior_t : char {
   PushPopWithoutReset,
 
   /// Opening brackets and braces, [, {, push onto the stack, closing brackets and braces, ], }, pop
-  /// from the stack. Newline characters are considered delimiters and therefore reset to an empty
-  /// stack.
+  /// from the stack. Delimiter characters are passed when the stack context is constructed to
+  /// reset to an empty stack.
   ResetOnDelimiter
 };
 
@@ -198,11 +198,13 @@ namespace detail {
  * within the context of a struct, a '[' represents that it is within the context of an array, and a
  * '_' symbol that it is at the root of the JSON.
  * @param[in] stack_behavior Specifies the stack's behavior
+ * @param[in] delimiter Specifies the delimiter to use as separator for JSON lines input
  * @param[in] stream The cuda stream to dispatch GPU kernels to
  */
 void get_stack_context(device_span<SymbolT const> json_in,
                        SymbolT* d_top_of_stack,
                        stack_behavior_t stack_behavior,
+                       SymbolT delimiter,
                        rmm::cuda_stream_view stream);
 
 /**