Using iconv to convert the text encoding of a file#

In sqlite-utils issue 439 I was testing against a CSV file that used UTF16 little endian encoding, also known as utf-16-le.

I converted it to UTF-8 using iconv like this:

1
iconv -f UTF-16LE -t UTF-8 file-in-utf16le.csv > file-in-utf8.csv

The -f argument here is the input encoding and -t is the desired output encoding.

I figured out the -f argument should be UTF-16LE (after first trying and failing with utf-16-le) by running:

1
iconv -l

This outputs all of the available encoding options. It’s a pretty long list so I filtered it like this:

1
% iconv -l | grep UTF
2
UTF-8 UTF8
3
UTF-8-MAC UTF8-MAC
4
UTF-16
5
UTF-16BE
6
UTF-16LE
7
UTF-32
8
UTF-32BE
9
UTF-32LE
10
UNICODE-1-1-UTF-7 UTF-7 CSUNICODE11UTF7

Discarding invalid characters#

I picked up this tip from Ben Brandwood: you can also use iconv to fix problems when a file includes invalid UTF-8 characters.

The trick is to use the -c option, which iconv --help tells you will “discard unconvertible characters”.

Here’s Ben’s recipe:

1
iconv -f utf-8 -t utf-8 -c FILE.txt -o NEW_FILE

Note that the input encoding (-f) and the output encoding (-t) are the same here. The -c option does all of the work.