Program Listing for File file_source.cpp

Return to documentation for file (morpheus/_lib/src/stages/file_source.cpp)

Copy
Copied!
            

#include "morpheus/stages/file_source.hpp" #include "morpheus/io/deserializers.hpp" #include <cudf/column/column.hpp>// for column #include <cudf/io/csv.hpp> #include <cudf/io/json.hpp> #include <cudf/scalar/scalar.hpp>// for string_scalar #include <cudf/strings/replace.hpp> #include <cudf/strings/strings_column_view.hpp>// for strings_column_view #include <cudf/table/table.hpp>// for table #include <cudf/types.hpp> #include <glog/logging.h> #include <mrc/segment/builder.hpp> #include <pybind11/cast.h>// for object_api::operator() #include <pybind11/gil.h> #include <pybind11/pybind11.h>// for str_attr_accessor #include <pybind11/pytypes.h>// for pybind11::int_ #include <algorithm>// for find #include <cstddef>// for size_t #include <filesystem> #include <memory> #include <regex> #include <sstream> #include <stdexcept>// for runtime_error #include <utility> // IWYU thinks we need __alloc_traits<>::value_type for vector assignments // IWYU pragma: no_include <ext/alloc_traits.h> namespace morpheus { // Component public implementations // ************ FileSourceStage ************* // FileSourceStage::FileSourceStage(std::string filename, int repeat) : PythonSource(build()), m_filename(std::move(filename)), m_repeat(repeat) {} FileSourceStage::subscriber_fn_t FileSourceStage::build() { return [this](rxcpp::subscriber<source_type_t> output) { auto data_table = load_table_from_file(m_filename); int index_col_count = get_index_col_count(data_table); // Next, create the message metadata. This gets reused for repeats // When index_col_count is 0 this will cause a new range index to be created auto meta = MessageMeta::create_from_cpp(std::move(data_table), index_col_count); // next_meta stores a copy of the upcoming meta std::shared_ptr<MessageMeta> next_meta = nullptr; for (cudf::size_type repeat_idx = 0; repeat_idx < m_repeat; ++repeat_idx) { if (!output.is_subscribed()) { // Grab the GIL before disposing, just in case pybind11::gil_scoped_acquire gil; // Reset meta to allow the DCHECK after the loop to pass meta.reset(); break; } // Clone the meta object before pushing while we still have access to it if (repeat_idx + 1 < m_repeat) { // GIL must come after get_info pybind11::gil_scoped_acquire gil; // Use the copy function auto df = meta->get_py_table().attr("copy")(); pybind11::int_ df_len = pybind11::len(df); pybind11::object index = df.attr("index"); df.attr("index") = index + df_len; next_meta = MessageMeta::create_from_python(std::move(df)); } DCHECK(meta) << "Cannot push null meta"; output.on_next(std::move(meta)); // Move next_meta into meta std::swap(meta, next_meta); } DCHECK(!meta) << "meta was not properly pushed"; DCHECK(!next_meta) << "next_meta was not properly pushed"; output.on_completed(); }; } // ************ FileSourceStageInterfaceProxy ************ // std::shared_ptr<mrc::segment::Object<FileSourceStage>> FileSourceStageInterfaceProxy::init( mrc::segment::Builder& builder, const std::string& name, std::string filename, int repeat) { auto stage = builder.construct_object<FileSourceStage>(name, filename, repeat); return stage; } } // namespace morpheus

© Copyright 2023, NVIDIA. Last updated on Feb 3, 2023.