awdeorio / csvstream

An easy-to-use CSV file parser for C++

awdeorio's dot files

This repository contains all of the configuration files in my home directory.

Files of particular interest

.always_forget.txt is an extensive cheat sheet of UNIX commands.

.bashrc contains my shell customizations.

Install

git clone [email protected]:awdeorio/dotfiles.git
rsync -av dotfiles/ ~/
rm -rf dotfiles/

csvstream's Issues

Line 289

Line 289 supposed to be if (row.size() != header.size()), please check?

Crash in the case if header columns count less then general columns count

I have CSV where 2nd row has more values than header columns. Library doesn't work in this case.
Expected behaviour: no crash, skipping of redundant values at the end of row.

Quoted data

Add support for quoted data. For example:

name,age
"DeOrio, Drew",31

Code coverage

Measure code coverage of unit tests. Report coverage using a badge on the README.

csvstream_exception should derive from std::exception and override what()

[Feature request] Retreive data keeping the CSV column order

Hi,

the std::map does not grant the order of inserted items. This is bad in my case because I want to analyze spatial time series that are contained in a CSV files, where each column represent a sensor in a fixed spatial position. I need to retrieve first the first column, then the second column, and so on. I'm not using a header to know what column come first (I could resolve modifying the CSV), so the column name can not help.

while(csvin >> row)
{
    // from column 0 to N
    //     do something
}

Thanks,
Riccardo

DEBUG output

There's a line that prints DEBUG output in csvstream.h

Support escaped quotes

Add support for escaped quotes. For example:

stringstream iss("\"a\",b,c\n\"\\\"1\\\"\",2,3\n");

csvstream_exception::what() definition should use override keyword

Minor, and doesn't affect functionality, but best practice would be to add the override keyword since the std::exception::what() function is virtual.

const char * what () const noexcept override {
  return msg.c_str();
}

include guards

For the 280 project, at least, I think we need to have these.

include guards

For the 280 project, at least, I think we need to have these.

GCC-7 support

Compiling with GCC-7 yields the following error:

csvstream.h:113:13: error: this statement may fall through [-Werror=implicit-fallthrough=]
       state = UNQUOTED;
       ~~~~~~^~~~~~~~~~

Be sure to update .travis.yml to include gcc-7 in the build matrix.

.h -> .hpp

Suggest considering either of the following:

1. Change the library to a .hpp extension
2. Make a parallel version of the library available as .hpp

This suggestion follows a change in EECS 280 at the University of Michigan (where csvstream is used in course projects) from .h to .hpp as the preferred extension for C++ development.

Add example to documentation for eecs280 p5

Adding an example of using the square bracket operator to the main example will be helpful for our students so that they know that can do e.g. row["tag"] or row["content"] to just pick out the piece they want.

ignore sorry my bad

this would mean the stream never gets closed

Problematic exception specification for what()

csvstream/csvstream.h

Line 26 in 13f65a3

const char * what () const throw () {

Report from student:

./csvstream.h:25:7: error: exception specification of overriding
      function is more lax than base version
class csvstream_exception : public std::exception {

Not sure what system or compiler version. Will update if I found out.

Possible fix:

  const char * what () const noexcept { 

Description of operator bool() is incorrect

It states:

  // Return true if an error flag on underlying stream is set
  explicit operator bool() const;

However, it actually returns false when an error flag is set.

Bug

terminate called after throwing an instance of 'csvstream_exception'

When trying to build example with cmake it gives this error

terminate called after throwing an instance of 'csvstream_exception'
  what():  Error opening file: dataset.csv
Aborted (core dumped)

Can any please help !! Thank you.

Better traversal by index example

csvstream can return a row of data either a map or as a vector of pair. Do a better job of showing how to use the vector in example 3.

Examples 2 and 3 in the README should iterate by reference instead of value

For example:

    for (auto& col:row) {
      string column_name = col.first;
      string datum = col.second;
      cout << "  " << column_name << ": " << datum << "\n";
    }

Multiple definition linker error generated when csvstream.h is included in multiple *.cpp files

Including csvstream.h in multiple *.cpp files will generate the following linker error:
multiple definition of `split(std::string const&, char)'

Support Windows line endings

Add support for Windows line endings (\r\n), old OSX line endings (\r) in addition to UNIX line endings (\n).

separation of definition and declaration but both inside header file generates multiple definition error

this breaks ODR (also see https://stackoverflow.com/questions/4955159/is-is-a-good-practice-to-put-the-definition-of-c-classes-into-the-header-file)

You should either revert this separation or do the definition in a .cpp file

Otherwise it won't work if you include this header in multiple .cpp files.