CSV

A Go CSV implementation inspired by Python's CSV module. It supports various CSV dialects (see below) and is fully backward compatible with the encoding/csv package in the Go standard library.

Examples

Here's a basic writing example:

f, err := os.Create("output.csv")
checkError(err)
defer func() {
  err := f.Close()
  checkError(err)
}
w := NewWriter(f)
w.Write([]string{
  "a",
  "b",
  "c",
})
w.Flush()
// output.csv will now contains the line "a b c" with a trailing newline.

Here's a basic reading example:

f, err := os.Open('myfile.csv')
checkError(err)
defer func() {
  err := f.Close()
  checkError(err)
}

r := NewReader(f)
for {
  fields, err := r.Read()
  if err == io.EOF {
    break
  }
  checkOtherErrors(err)
  handleFields(fields)
}

CSV dialects

To modify CSV dialect, have a look at csv.Dialect, csv.NewDialectWriter(...) and csv.NewDialectReader(...). It supports changing:

• separator/delimiter.
• quoting modes:
  • Always quote.
  • Never quote.
  • Quote when needed (minimal quoting).
  • Quote all non-numerical fields.
  • Quote all non-empty, non-numerical fields.
• line terminator.
• how quote character escaping should be done - using double escape, or using a custom escape character.

Have a look at the documentation in csv_test.go for example on how to use these. All values above have sane defaults (that makes the module behave the same as the csv module in the Go standard library).

Documentation

Package documentation can be found here.

Why was this developed?

I needed it for mysqlcsvdump to support variations of CSV output. The csv module in the Go (1.2) standard library was inadequate as it it does not support any CSV dialect modifications except changing separator and partially line termination.

Who developed this?

I'm Jens Rantil. Have a look at my blog for more info on what I'm working on.