/Dataframes-xlsx

Primary LanguageC++

Introduction

This is a C++ library providing support for Luna Datasets framework. It provides function both for dataset file I/O operations and basic array splicing.

Third party dependencies

  • xlnt library (C++/CMake) — needed for .xlsx file format support. NOTE: On MacOS mwu-tow's fork is needed to fix the compilation issue. On other platforms, official library repo can be used.

Build & Install

  • have installed xlnt library either:
    • system-wide install (cmake /PATH/TO/XLNT && make install)
    • local install and set CMAKE_PREFIX_PATH to the location
    • separately provide library's include and library directories with CMAKE_LIBRARY_PATH
  • clone the repo
  • cmake repo/DataframeHelper
  • make
  • Binary will be placed in the repo/native_libs/os/ directory where os is either windows, macos or linux.

Structure

  • Core/
    • Common — defines an EXPORT macro
    • Error — helpers for translating between C++ error reporting (exceptions) and Luna-consumable error messages
    • Matrix2d — class that manages a Luna-consumable 2D array with strings
  • IO/
    • xlsx — read/write dataframe from/to files in Microsoft Excel Open XML Format;
    • csv — read/write dataframe from/to file in CSV format.

C API

Types

using MatrixDataPtr = const char * const *;

MatrixDataPtr is a an array of string pointers. It represents a two-dimensional array of C-style strings stored in a single row-major linear storage. The storage (both for the array itself and the strings) is managed by the Matrix2d class.

Pointer of that type is meant for Luna — to allow easy access to array elements.

Luna is not allowed to directly modify neither array elements nor the strings itself, the storage should be treated as read-only.

Functions

Core

void mat_delete(MatrixDataPtr mat)

Deletes a Matrix2d object associated with given matrix data pointer.

IO

MatrixDataPtr read_csv(const char *filename, size_t *rows, size_t cols, int error)

Reads file from the given path. File is required to be in CSV format. Upon successfull read a pointer to array with Dataframe contents is returned, and array sizes are written to rows and cols arguments. Upon failure, NULL pointer is returned and error is set to a non-zero value.

void write_csv(const char *filename, char *mat, size_t rows, size_t cols, int error)

Writes data from given array to CSV file. Result code will be stored under error pointer: 0 if success, non-zero in case of error.