Convert from/to encodings with iconv-lite only about iconv-lite

[ashtuchkin]

You're inadvertently converting the chunk to string by doing buffer += chunk. You should keep it as a buffer, something like this:

        var buffer = new Buffer(0);
        stream.on("data", function(chunk) {
            buffer = Buffer.concat([buffer, chunk]);
        });

        stream.once("end", function(){
            str = iconv.decode(buffer, "ISO-8859-1");
        });

[ashtuchkin]

Yes, sure. iconv.decode(buf, encoding) takes binary data and decodes it into a JS string. So, what you're need here is:

originalData = new Buffer("base64 string with ISO-8859-1 encoding goes here", "base64");  // Notice, no .toString().
jsStr = iconv.decode(originalData, "ISO-8859-1");

// Here you can use jsStr as usual javascript String:
jsStr.replace("hello", "world");

// If you need a buffer with this string UTF-8 encoded, then you can use Buffer like that:
utf8EncodedStringBuf = new Buffer(jsStr);  // utf8 is the default encoding.

// Or, equivalent
utf8EncodedStringBuf = iconv.encode(jsStr, "utf8");

[alexandernst]

I tried concat-ing the buffer instead of converting it to a string and now it works as expected, thank you! :)

[alexandernst]

Ok, but what if I get a string in ISO-8859-1 instead of base64?

[ashtuchkin]

It depends on how that string was converted from the actual bytes.

If you have a JS string which was UTF-8-decoded from a buffer of ISO-8859-1-encoded data, then the recovery is possible only if the original string was all ASCII. UTF-8 is (obviously) not compatible with ISO-8859-1 and will produce all sorts of weird chars and decoding errors otherwise.

In general, you should attempt to get original bytes (in Buffer-s), that way it's easier and more robust.
If its not possible, then the best you can do is to try do encode it back as UTF8, then decode as ISO-8859-1, but it'll surely lose some information.

str = iconv.decode(new Buffer(str), "ISO-8859-1");

[alexandernst]

I'm getting the string from this library: https://github.com/mscdex/node-imap

I fetch a mail part with:

f.on("message", function(msg, seqno){

    msg.on("body", function(stream, info){
        var buffer = "";
        stream.on("data", function(chunk){
            buffer += chunk;
        });

        stream.once("end", function(){
            //convert buffer from whatever encoding has to utf8
        });

..........

I'll ask that there too :)

[alexandernst]

That seems reasonable. I'll try it monday morning as soon as I get to the office :)