No OneTemporary
Actions

Size

17 KB

Referenced Files

None

Subscribers

None

View Options

	diff --git a/include/rosa/support/csv/CSVReader.hpp b/include/rosa/support/csv/CSVReader.hpp
	index 7e29ad8..95c1fda 100755
	--- a/include/rosa/support/csv/CSVReader.hpp
	+++ b/include/rosa/support/csv/CSVReader.hpp
	@@ -1,419 +1,480 @@
	//===-- rosa/support/csv/CSVReader.hpp --------------------------- C++ --===//
	//
	// The RoSA Framework
	//
	//===----------------------------------------------------------------------===//
	///
	/// \file rosa/support/csv/CSVReader.hpp
	///
	/// \author David Juhasz (david.juhasz@tuwien.ac.at)
	///
	/// \date 2017-2019
	///
	/// \brief Facitilities to read CSV files.
	///
	/// \note The implementation is based on the solution at
	/// https://stackoverflow.com/a/1120224
	///
	//===----------------------------------------------------------------------===//

	#ifndef ROSA_SUPPORT_CSV_CSVREADER_HPP
	#define ROSA_SUPPORT_CSV_CSVREADER_HPP

	#include "rosa/support/debug.hpp"

	#include <istream>
	#include <sstream>
	#include <vector>
	+#include <algorithm>

	namespace rosa {
	namespace csv {

	/// Anonymous namespace providing implementation details for
	/// \c rosa::csv::CSVIterator, consider it private.
	namespace {

	/// Provides facility for parsing values from one row CSV data.
	///
	/// \tparam T type of values to parse from the line
	/// \tparam IsSignedInt if \p T is a signed integral type, always use default
	/// \tparam IsUnsignedInt if \p T is an unsigned integral type, always use
	/// default
	/// \tparam IsFloat if \p T is a floating-point type, always use default
	/// \tparam IsString if \p T is \c std::string, always use default
	///
	/// \note Specializations of this `struct` are provided for arithmentic types
	/// and \c std::string.
	template <typename T, bool IsSignedInt = (std::is_integral<T>::value &&
	std::is_signed<T>::value),
	bool IsUnsignedInt =
	(std::is_integral<T>::value && std::is_unsigned<T>::value),
	bool IsFloat = std::is_floating_point<T>::value,
	bool IsString = std::is_same<T, std::string>::value>
	struct CSVRowParser;

	/// Specialization for signed integral types.
	///
	/// \tparam T type of values to parse from the line
	///
	/// \pre \p T is a signed integral type:\code
	/// std::is_integral<T>::value && std::is_signed<T>::value
	/// \endcode
	template <typename T> struct CSVRowParser<T, true, false, false, false> {
	STATIC_ASSERT((std::is_integral<T>::value && std::is_signed<T>::value),
	"wrong type"); // Sanity check.

	/// Parses a given row of CSV data into a given container.
	///
	/// \p Data is cleared and then filled with values parsed from \p LineStream.
	/// Entries in the line are to be separated by commas, the character `,`. A
	/// trailing comma results in an empty entry at the end of the line. No empty
	/// entry should be present otherwise.
	///
	/// \note Parsed values are silently converted to type \p T.
	///
	/// \param [in,out] LineStream the line to parse
	/// \param [in,out] Data the container to store the parsed values
	static void parse(std::stringstream &LineStream, std::vector<T> &Data, char Delimeter) {
	std::string Cell;
	Data.clear();
	while (std::getline(LineStream, Cell, Delimeter)) {
	Data.push_back(static_cast<T>(std::stoll(Cell)));
	}
	// This checks for a trailing comma with no data after it.
	if (!LineStream && Cell.empty()) {
	// If there was a trailing comma then add an empty element.
	Data.push_back(0);
	}
	}
	};

	/// Specialization for unsigned integral types.
	///
	/// \tparam T type of values to parse from the line
	///
	/// \pre \p T is an unsigned integral type:\code
	/// std::is_integral<T>::value && std::is_unsigned<T>::value
	/// \endcode
	template <typename T> struct CSVRowParser<T, false, true, false, false> {
	STATIC_ASSERT((std::is_integral<T>::value && std::is_unsigned<T>::value),
	"wrong type"); // Sanity check.

	/// Parses a given row of CSV data into a given container.
	///
	/// \p Data is cleared and then filled with values parsed from \p LineStream.
	/// Entries in the line are to be separated by commas, the character `,`. A
	/// trailing comma results in an empty entry at the end of the line. No empty
	/// entry should be present otherwise.
	///
	/// \note Parsed values are silently converted to type \p T.
	///
	/// \param [in,out] LineStream the line to parse
	/// \param [in,out] Data the container to store the parsed values
	static void parse(std::stringstream &LineStream, std::vector<T> &Data, char Delimeter = ',') {
	std::string Cell;
	Data.clear();
	while (std::getline(LineStream, Cell, Delimeter)) {
	Data.push_back(static_cast<T>(std::stoull(Cell)));
	}
	// This checks for a trailing comma with no data after it.
	if (!LineStream && Cell.empty()) {
	// If there was a trailing comma then add an empty element.
	Data.push_back(0);
	}
	}
	};

	/// Specialization for floating-point types.
	///
	/// \tparam T type of values to parse from the line
	///
	/// \pre \p T is a floating-point type:\code
	/// std::is_floating_point<T>::value
	/// \endcode
	template <typename T> struct CSVRowParser<T, false, false, true, false> {
	STATIC_ASSERT((std::is_floating_point<T>::value),
	"wrong type"); // Sanity check.

	/// Parses a given row of CSV data into a given container.
	///
	/// \p Data is cleared and then filled with values parsed from \p LineStream.
	/// Entries in the line are to be separated by commas, the character `,`. A
	/// trailing comma results in an empty entry at the end of the line. No empty
	/// entry should be present otherwise.
	///
	/// \note Parsed values are silently converted to type \p T.
	///
	/// \param [in,out] LineStream the line to parse
	/// \param [in,out] Data the container to store the parsed values
	static void parse(std::stringstream &LineStream, std::vector<T> &Data, char Delimeter = ',') {
	std::string Cell;
	Data.clear();
	while (std::getline(LineStream, Cell, Delimeter)) {
	Data.push_back(static_cast<T>(std::stold(Cell)));
	}
	// This checks for a trailing comma with no data after it.
	if (!LineStream && Cell.empty()) {
	// If there was a trailing comma then add an empty element.
	Data.push_back(0);
	}
	}
	};

	/// Specialization for \c std::string.
	///
	/// \tparam T type of values to parse from the line
	///
	/// \pre \p T is \c std::string:\code
	/// std::is_same<T, std::string>::value
	/// \endcode
	template <typename T> struct CSVRowParser<T, false, false, false, true> {
	STATIC_ASSERT((std::is_same<T, std::string>::value),
	"wrong type"); // Sanity check.

	/// Parses a given row of CSV data into a given container.
	///
	/// \p Data is cleared and then filled with values parsed from \p LineStream.
	/// Entries in the line are to be separated by commas, the character `,`. A
	/// trailing comma results in an empty entry at the end of the line. No empty
	/// entry should be present otherwise.
	///
	/// \param [in,out] LineStream the line to parse
	/// \param [in,out] Data the container to store the parsed values
	static void parse(std::stringstream &LineStream, std::vector<T> &Data, char Delimeter = ',') {
	std::string Cell;
	Data.clear();
	while (std::getline(LineStream, Cell, Delimeter)) {
	Data.push_back(Cell);
	}
	// This checks for a trailing comma with no data after it.
	if (!LineStream && Cell.empty()) {
	// If there was a trailing comma then add an empty element.
	Data.push_back("");
	}
	}
	};

	/// Parses and stores entries from a row of CSV data.
	///
	/// \tparam T type of values to parse and store, i.e. entries in the row
	///
	/// \note The implementation relies on \c rosa::csv::CSVRowParser, which is
	/// implemented only for `arithmetic` types -- signed and unsigned integral and
	/// floating-point types -- and for \c std::string. Those are the valid values
	/// for \p T.
	template <typename T>
	class CSVRow {
	public:
	/// Gives a constant reference for an entry at a given position of the row.
	///
	/// \note No bounds checking is performed.
	///
	/// \param Index the position of the entry
	///
	/// \return constant reference for the stored entry at position \p Index
	const T &operator[](const size_t Index) const noexcept { return Data[Index]; }

	/// Tells the number of entries stored in the row.
	///
	/// \return number of stored entries.
	size_t size(void) const noexcept { return Data.size(); }

	/// Parses and stores one row of CSV data.
	///
	/// The function reads one line from \p Str and parses it into
	/// \c rosa::csv::CSVRow::Data using \c rosa::csv::CSVRowParser.
	///
	/// \param [in,out] Str input stream of a CSV file
	void readNextRow(std::istream &Str) {
	std::string Line;
	std::getline(Str, Line);
	std::stringstream LineStream(Line);
	+
	CSVRowParser<T>::parse(LineStream, Data, Delimeter);
	}

	+ bool isNumeric(const std::string& input){
	+ return std::all_of(input.begin(), input.end(), ::isdigit);
	+ }
	+
	+ void checkIfHeader(std::istream &Str){
	+ std::string Line;
	+ std::getline(Str, Line);
	+ std::vector<std::string> FirstRowValues;
	+ std::stringstream LineStream(Line);
	+ std::string Value;
	+ bool HasHeaderLocal = true;
	+
	+ CSVRowParser<std::string>::parse(LineStream, FirstRowValues, Delimeter);
	+ for(std::vector<std::string>::iterator it = FirstRowValues.begin();
	+ it != FirstRowValues.end(); ++it){
	+ Value = *it;
	+ if(isNumeric(Value)){
	+ HasHeaderLocal = false;
	+ }
	+ }
	+ if(HasHeaderLocal){
	+ Header.swap(FirstRowValues);
	+ /* // only for debugging purpose.
	+ for(std::vector<std::string>::iterator it = Header.begin();
	+ it != Header.end(); ++it){
	+ std::cout << *it << ", ";
	+ }
	+ std::cout << std::endl; */
	+ }else {
	+ std::stringstream LineStream2(Line);
	+ CSVRowParser<T>::parse(LineStream2, Data, Delimeter);
	+ }
	+ HasHeader = HasHeaderLocal;
	+ isFirstRow = false;
	+ }
	+
	+ bool isHavingHeader(){
	+ return HasHeader;
	+ }
	+
	inline void setDelimeter(char Delimeter){
	this->Delimeter = Delimeter;
	}

	inline char getDelimeter(){
	return this->Delimeter;
	}

	inline void setEndOfLine(char EndOfLine){
	this->EndOfLine = EndOfLine;
	}

	inline char getEndOfLine(){
	return this->EndOfLine;
	}

	+ inline bool isThisFirstRow(){
	+ return this->isFirstRow;
	+ }
	+
	private:
	- std::vector<T> Data; ///< Stores parsed entries
	- uint64_t RowNumber = 0; ///< Current row number
	- bool HasHeader = false; ///< Has the current csv file a header
	- std::vector<std::string> Header; /// < stores the header entries if available
	- char Delimeter = ','; /// < stores the delimeter between data entries
	- char EndOfLine = '\n'; ///< stores the end of line character
	+ std::vector<T> Data; ///< Stores parsed entries
	+ uint64_t RowNumber = 0; ///< Current row number
	+ bool isFirstRow = true; ///< is this the first row
	+ bool isFirstRowRead = false; ///< is the first row read already
	+ bool HasHeader = false; ///< Has the current csv file a header
	+ std::vector<std::string> Header; /// < stores the header entries if available
	+ char Delimeter = ','; /// < stores the delimeter between data entries
	+ char EndOfLine = '\n'; ///< stores the end of line character
	};

	/// Reads a row of CSV data into \c rosa::csv::CSVRow.
	///
	/// The next line is read from \p Str by calling
	/// \c rosa::csv::CSVRow::readNextRow on \p Data.
	///
	/// \note A CSV file should contain no empty lines.
	///
	/// \param [in,out] Str input stream of a CSV file
	/// \param [in,out] Data object to read the next line into
	///
	/// \return \p Str after reading one line from it
	template <typename T>
	std::istream &operator>>(std::istream &Str, CSVRow<T> &Data) {
	- Data.readNextRow(Str);
	+ if (Data.isThisFirstRow()){
	+ Data.checkIfHeader(Str);
	+ if(Data.isHavingHeader()){
	+ Data.readNextRow(Str);
	+ }
	+ }else {
	+ Data.readNextRow(Str);
	+ }
	+ /* // just for debugging purpose
	+ char c;
	+ while(Str.get(c)){
	+ std::cout << c;
	+ }
	+ std::cout << std::endl;
	+ */
	return Str;
	}

	} // End namespace

	/// Provides `InputIterator` features for iterating over a CSV file in a
	/// flat way.
	///
	/// The iterator hides rows of the CSV file, and iterates over the entries
	/// row-by-row.
	///
	/// \note A CSV file should contain no empty lines.
	///
	/// \tparam T type of values to iterate over, i.e. entries in the CSV file.
	///
	/// \note The implementation relies on \c rosa::csv::CSVRow, which in turn
	/// relies on \c rosa::csv::CSVRowParser, which is implemented only for
	/// `arithmetic` types -- signed and unsigned integral types and floating-point
	/// types -- and for \c std::string. Those are the valid values for \p T.
	template <typename T>
	class CSVFlatIterator {
	public:
	/// \defgroup CSVFlatIteratorTypedefs Typedefs of rosa::csv::CSVFlatIterator
	///
	/// Standard `typedef`s for iterators.
	///
	///@{
	typedef std::input_iterator_tag
	iterator_category; ///< Category of the iterator.
	typedef T value_type; ///< Type of values iterated over.
	typedef std::size_t difference_type; ///< Type to identify distance.
	typedef T *pointer; ///< Pointer to the type iterated over.
	typedef T &reference; ///< Reference to the type iterated over.
	///@}

	/// Creates a new instance.
	///
	/// \param [in,out] S input stream to iterate over
	CSVFlatIterator(std::istream &S)
	: Str(S.good() ? &S : nullptr),
	Pos((size_t)(-1)),
	Delimeter(','), EndOfLine('\n') {
	Row.setDelimeter(Delimeter);
	- Row.setDelimeter('a');
	Row.setEndOfLine(EndOfLine);
	// \c rosa::csv::CSVFlatIterator::Pos is initialized to `-1` so the first
	// incrementation here will set it properly.
	++(*this);
	}

	/// Creates an empty new instance.
	CSVFlatIterator(void) noexcept : Str(nullptr) {}

	/// Pre-increment operator.
	///
	/// The implementation moves over the entries in the current row and advances
	/// to the next row when the end of the current row is reached. If the end of
	/// the input stream is reached, the operator becomes empty and has no
	/// further effect.
	///
	/// \return \p this object after incrementing it.
	CSVFlatIterator &operator++() {
	if (Str) {
	++Pos;
	if (Pos == Row.size()) {
	if (!((*Str) >> Row)) {
	Str = nullptr;
	--Pos; // Stay on the last entry forever.
	} else {
	Pos = 0;
	}
	}
	}
	return *this;
	}

	/// Post-increment operator.
	///
	/// The implementation uses the pre-increment operator and returns a copy of
	/// the original state of \p this object.
	///
	/// \return \p this object before incrementing it.
	CSVFlatIterator operator++(int) {
	CSVFlatIterator Tmp(*this);
	++(*this);
	return Tmp;
	}

	/// Returns a constant reference to the current entry.
	///
	/// \note Should not dereference the iterator when it is empty.
	///
	/// \return constant reference to the current entry.
	const T &operator*(void)const noexcept { return Row[Pos]; }

	/// Returns a constant pointer to the current entry.
	///
	/// \note Should not dereference the iterator when it is empty.
	///
	/// \return constant pointer to the current entry.
	const T *operator->(void)const noexcept { return &Row[Pos]; }

	/// Tells if \p this object is equal to another one.
	///
	/// Two \c rosa::csv::CSVReader instances are equal if and only if they are
	/// the same or both are empty.
	///
	/// \param RHS other object to compare to
	///
	/// \return whether \p this object is equal with \p RHS
	bool operator==(const CSVFlatIterator &RHS) const noexcept {
	return ((this == &RHS) \|\| ((this->Str == nullptr) && (RHS.Str == nullptr)));
	}

	/// Tells if \p this object is not equal to another one.
	///
	/// \see rosa::csv::CSVReader::operator==
	///
	/// \param RHS other object to compare to
	///
	/// \return whether \p this object is not equal with \p RHS.
	bool operator!=(const CSVFlatIterator &RHS) const noexcept {
	return !((*this) == RHS);
	}

	inline void setDelimeter(char Delimter){
	this->Delimeter = Delimter;
	}

	inline char getDelimeter(){
	return this->Delimeter;
	}

	private:
	std::istream *Str; ///< Input stream of a CSV file to iterate over.
	CSVRow<T> Row; ///< Content of the current row iterating over.
	size_t Pos; ///< Current position within the current row.
	char Delimeter; ///< Delimeter between the entries
	char EndOfLine; ///< stores the end of line character

	};

	} // End namespace csv
	} // End namespace rosa

	#endif // ROSA_SUPPORT_CSV_CSVREADER_HPP

File Metadata

Mime Type: text/x-diff
Expires: Mon, Oct 20, 6:06 PM (21 h, 2 m)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 199743
Default Alt Text: (17 KB)

No OneTemporaryActions

View Options

File Metadata

Event Timeline

No OneTemporary
Actions