Write UTF-16 encoded CSV using Golang

Question

I need to write a UTF-16 conded csv and I can't figure out how to. I found a lot of questions and answers about reading UTF-16 csvs but nothing about writing.

This is what I've tried so far:

package main

import (
    "encoding/csv"
    "fmt"
    "os"

    "golang.org/x/text/encoding/unicode"

)

func main() {
    csvFile, err := os.Create("test.csv")
    if err != nil {
        panic(err)
    }
    defer csvFile.Close()

    message := "weird characters: дгодг"

    message, err = convertUtf8ToUtf16LE(message)
    if err != nil {
        panic(err)
    }
    fmt.Println(message)

    csvWriter := csv.NewWriter(csvFile)
    defer csvWriter.Flush()

    csvWriter.Write([]string{message})
    csvWriter.Write([]string{message})
}

func convertUtf8ToUtf16LE(message string) (string, error) {
    encoder := unicode.UTF16(unicode.LittleEndian, unicode.UseBOM).NewEncoder()
    return encoder.String(message)
}

But I get the following output in the csv:

weird characters: дгодг＊矾攀椀爀搀 挀栀愀爀愀挀琀攀爀猀㨀 㐀㌄㸄㐄㌄਄

What am I doing wrong?

Answer 1

I'm new to transforming, but I think this does the same thing you wrote for yourself, and it uses the same text package.

From your question, it looks like you were just missing the transformer.NewWriter() component.

I took this answer, and just turned it around to make it encode UTF16LE (oh. my. stars. Go makes these things so easy):

package main

import (
    "encoding/csv"
    "os"

    "golang.org/x/text/encoding/unicode"
    "golang.org/x/text/transform"
)

func main() {
    records := [][]string{
        {"first_name", "last_name", "username"},
        {"Rob", "Pike", "rob"},
        {"Ken", "Thompson", "ken"},
        {"Robert", "Griesemer", "gri"},
    }

    f, _ := os.Create("utf16le.csv")
    t := transform.NewWriter(f, unicode.UTF16(unicode.LittleEndian, unicode.UseBOM).NewEncoder())
    w := csv.NewWriter(t)

    for _, record := range records {
        w.Write(record)
    }

    w.Flush()
}

And viewing that CSV file:

hexdump -C utf16le.csv
00000000  ff fe 66 00 69 00 72 00  73 00 74 00 5f 00 6e 00  |��f.i.r.s.t._.n.|
00000010  61 00 6d 00 65 00 2c 00  6c 00 61 00 73 00 74 00  |a.m.e.,.l.a.s.t.|
00000020  5f 00 6e 00 61 00 6d 00  65 00 2c 00 75 00 73 00  |_.n.a.m.e.,.u.s.|
00000030  65 00 72 00 6e 00 61 00  6d 00 65 00 0a 00 52 00  |e.r.n.a.m.e...R.|
00000040  6f 00 62 00 2c 00 50 00  69 00 6b 00 65 00 2c 00  |o.b.,.P.i.k.e.,.|
00000050  72 00 6f 00 62 00 0a 00  4b 00 65 00 6e 00 2c 00  |r.o.b...K.e.n.,.|
00000060  54 00 68 00 6f 00 6d 00  70 00 73 00 6f 00 6e 00  |T.h.o.m.p.s.o.n.|
00000070  2c 00 6b 00 65 00 6e 00  0a 00 52 00 6f 00 62 00  |,.k.e.n...R.o.b.|
00000080  65 00 72 00 74 00 2c 00  47 00 72 00 69 00 65 00  |e.r.t.,.G.r.i.e.|
00000090  73 00 65 00 6d 00 65 00  72 00 2c 00 67 00 72 00  |s.e.m.e.r.,.g.r.|
000000a0  69 00 0a 00                                       |i...|
000000a4

Answer 2

What I ended up doing is I created a struct that implements io.Writer for a file but converts the input to UTF-16LE before writing:

type UTF16LEWriter struct {
    file    *os.File
    encoder *encoding.Encoder
}

func NewUTF16LEWriter(file *os.File) (*UTF16LEWriter, error) {
    _, err := file.Write([]byte{0xFF, 0xFE}) // UTF-16LE BOM
    if err != nil {
        return &UTF16LEWriter{}, err
    }

    return &UTF16LEWriter{
        file:    file,
        encoder: unicode.UTF16(unicode.LittleEndian, unicode.IgnoreBOM).NewEncoder(),
    }, nil
}

func (w *UTF16LEWriter) Write(b []byte) (int, error) {
    b, err := w.encoder.Bytes(b)
    if err != nil {
        return 0, err
    }
    w.file.Write(b)
    return len(b), err
}

Then I only need to replace the io.Writer provided by os.Create with mine and pass that to the csvWriter:

package main

import (
    "encoding/csv"
    "fmt"
    "os"

    "golang.org/x/text/encoding/unicode"

)

func main() {
    csvFile, err := os.Create("test.csv")
    if err != nil {
        panic(err)
    }
    defer csvFile.Close()

    utf16Writer, err := NewUTF16LEWriter(csvFile)
    if err != nil {
        panic(err)
    }

    csvWriter := csv.NewWriter(utf16Writer)
    defer csvWriter.Flush()

    message := "weird characters: дгодг"
    csvWriter.Write([]string{message})
    csvWriter.Write([]string{message})
}