Commit c7de20db authored by Paul Asmuth's avatar Paul Asmuth
Browse files

csvfile -> csvinputstream

parent e94b9c29

CMakeLists.txt

+1 −1
Original line number Diff line number Diff line
@@ -39,7 +39,7 @@ set(FNORDMETRIC_QUERY_SOURCES 
    src/query/tokenize.cc)



set(FNORDMETRIC_CSV_BACKEND_SOURCES

    src/backends/csv/csvfile.cc)

    src/backends/csv/csvinputstream.cc)



include_directories(./src)

set(CMAKE_CXX_FLAGS "-std=c++0x -stdlib=libc++")

src/backends/csv/csvbackend_test.cc

+12 −8
Original line number Diff line number Diff line
@@ -10,32 +10,36 @@ 
#include <assert.h>

#include <sys/fcntl.h>

#include <unistd.h>

#include "csvfile.h"

#include "csvinputstream.h"

//#include "csv_backend.h"

#include "../../util/runtimeexception.h"

#include "../../util/unittest.h"



using namespace fnordmetric::csv_backend;



UNIT_TEST(CSVFileTest);

UNIT_TEST(CSVInputStreamTest);



TEST_CASE(CSVInputStreamTest, TestOpenFile, [] () {

  auto csv_file = CSVInputStream::openFile(

      "test/fixtures/gbp_per_country.csv");



TEST_CASE(CSVFileTest, TestOpenFile, [] () {

  auto csv_file = CSVFile::openFile("test/fixtures/gbp_per_country.csv");

  EXPECT(csv_file.get() != nullptr);

});



TEST_CASE(CSVFileTest, TestInvalidFileName, [] () {

TEST_CASE(CSVInputStreamTest, TestInvalidFileName, [] () {

  auto errmsg = "error opening file 'test/fixtures/invalid.csv': "

      "No such file or directory";



  EXPECT_EXCEPTION(errmsg, [] () {

    auto csv_file = CSVFile::openFile("test/fixtures/invalid.csv");

    auto csv_file = CSVInputStream::openFile("test/fixtures/invalid.csv");

    EXPECT(csv_file.get() != nullptr);

  });

});



TEST_CASE(CSVFileTest, TestReadHeaders, [] () {

  auto csv_file = CSVFile::openFile("test/fixtures/gbp_per_country_simple.csv");

TEST_CASE(CSVInputStreamTest, TestReadHeaders, [] () {

  auto csv_file = CSVInputStream::openFile(

      "test/fixtures/gbp_per_country_simple.csv");



  EXPECT(csv_file.get() != nullptr);

  std::vector<std::string> headers;

  csv_file->readNextRow(&headers);

src/backends/csv/csvinputstream.cc

0 → 100644
+47 −0
Original line number Diff line number Diff line 
/**

 * This file is part of the "FnordMetric" project

 *   Copyright (c) 2014 Paul Asmuth, Google Inc.

 *

 * Licensed under the MIT license (see LICENSE).

 */

#include <assert.h>

#include <unistd.h>

#include <fcntl.h>

#include "csvinputstream.h"

#include "../../util/inputstream.h"

#include "../../util/runtimeexception.h"



namespace fnordmetric {

namespace csv_backend {



std::unique_ptr<CSVInputStream> CSVInputStream::openFile(

    const std::string& file_path,

    char column_seperator /* = ',' */,

    char row_seperator /* = '\n' */,

    char quote_char /* = '"' */) {

  auto csv_file = new CSVInputStream(

      fnordmetric::util::FileInputStream::openFile(file_path));



  return std::unique_ptr<CSVInputStream>(csv_file);

}



CSVInputStream::CSVInputStream(

    std::unique_ptr<fnordmetric::util::InputStream>&& input_stream,

    char column_seperator /* = ',' */,

    char row_seperator /* = '\n' */,

    char quote_char /* = '"' */) :

    input_(std::move(input_stream)) {}



void CSVInputStream::readNextRow(std::vector<std::string>* target) {

  target->emplace_back(readNextColumn());

}



// FIXPAUL optimize?

std::string CSVInputStream::readNextColumn() {

  std::string column;



  return column;

}



}

}

src/backends/csv/csvfile.h→src/backends/csv/csvinputstream.h

+59 −0
Original line number Diff line number Diff line
@@ -4,48 +4,37 @@ 
 *

 * Licensed under the MIT license (see LICENSE).

 */

#ifndef _FNORDMETRIC_CSVFILE_H

#define _FNORDMETRIC_CSVFILE_H

#ifndef _FNORDMETRIC_CSVINPUTSTREAM_H

#define _FNORDMETRIC_CSVINPUTSTREAM_H

#include <string>

#include <vector>

#include <memory>

#include "../../util/inputstream.h"

#include "../../util/unittest.h"



namespace fnordmetric {

namespace csv_backend {



class CSVFile {

  friend class fnordmetric::util::UnitTest::TestCase;

class CSVInputStream {

public:



  enum ErrorCodes {

    ERR_CSV_CANNOT_OPEN_FILE = 4000,

    ERR_CSV_READ_ERROR = 4001

  };



  /**

   * Open a new csv file from the provided file path. Throws an exception if

   * the file cannot be opened.

   * Open a new csv file input stream from the provided file path. Throws an

   * exception if the file cannot be opened.

   *

   * @param file_path the path to the csv file

   */

  static std::unique_ptr<CSVFile> openFile(

  static std::unique_ptr<CSVInputStream> openFile(

      const std::string& file_path,

      char column_seperator = ',',

      char row_seperator = '\n',

      char quote_char = '"');



  CSVFile(const CSVFile& other) = delete;

  CSVFile& operator=(const CSVFile& other) = delete;



  /**

   * Create a new CSVFile instance with the provided filedescriptor. The csv

   * file takes ownership of the fd and will close() it when destructed.

   *

   * @param fd a valid an opened fd, transfers ownership and closes on destruct

   * Create a new CSVInputStream from the provided InputStream.

   */

  explicit CSVFile(

      int fd,

  explicit CSVInputStream(

      std::unique_ptr<fnordmetric::util::InputStream>&& input_stream,

      char column_seperator = ',',

      char row_seperator = '\n',

      char quote_char = '"');
@@ -62,18 +51,7 @@ protected: 
   */

  std::string readNextColumn();





  /**

   * Read the next chunk from disk. Returns true on success and false on EOF.

   * Raises an exception on error.

   */

  bool readNextChunk();





  int fd_;

  char buf_[4];

  size_t buf_len_;

  size_t buf_pos_;

  std::unique_ptr<fnordmetric::util::InputStream> input_;

};



}

src/backends/csv/csvfile.cc→src/util/inputstream.cc

+47 −0
Original line number Diff line number Diff line
@@ -4,16 +4,11 @@ 
 *

 * Licensed under the MIT license (see LICENSE).

 */

#include <assert.h>

#include <unistd.h>

#include <fcntl.h>

#include "csvfile.h"

#include "../../util/runtimeexception.h"



#include "inputstream.h"

namespace fnordmetric {

namespace csv_backend {

namespace util {



std::unique_ptr<CSVFile> CSVFile::openFile(

std::unique_ptr<InputStream> InputStream::openFile(

    const std::string& file_path,

    char column_seperator /* = ',' */,

    char row_seperator /* = '\n' */,
@@ -23,49 +18,30 @@ std::unique_ptr<CSVFile> CSVFile::openFile( 


  if (fd < 1) {

    throw RUNTIME_EXCEPTION_ERRNO(

        &typeid(CSVFile),

        &typeid(InputStream),

        ERR_CSV_CANNOT_OPEN_FILE,

        "error opening file '%s'",

        fp);

  }



  auto csv_file = new CSVFile(fd);

  return std::unique_ptr<CSVFile>(csv_file);

}



CSVFile::CSVFile(

    int fd,

    char column_seperator /* = ',' */,

    char row_seperator /* = '\n' */,

    char quote_char /* = '"' */) :

    fd_(fd),

    buf_len_(0),

    buf_pos_(0) {

  assert(fd > 0);

  auto csv_file = new InputStream(fd);

  return std::unique_ptr<InputStream>(csv_file);

}



void CSVFile::readNextRow(std::vector<std::string>* target) {

  target->emplace_back(readNextColumn());

}



std::string CSVFile::readNextColumn() {

  if (buf_pos_ < buf_len_) {

    readNextChunk();

  }

}



bool CSVFile::readNextChunk() {

/*

bool CSVInputStream::readNextChunk() {

  int bytes_read = read(fd_, buf_, sizeof(buf_));



  if (bytes_read < 0) {

    throw RUNTIME_EXCEPTION_ERRNO(

        &typeid(CSVFile),

        &typeid(CSVInputStream),

        ERR_CSV_READ_ERROR,

        "read() failed");

  }



  return bytes_read == 0;

}

*/



}

}

CRA Git | Maintained and supported by SUSTech CRA and CCSE