Lesson 35 · Decode a chunked request body

Decode a chunked request body

When a client does not know a body's size up front it sends it in chunks, each prefixed by its length. Today you decode that framing into the whole body.

The goal

Decode a chunked transfer-encoded body into the concatenated bytes it represents.

Start here - the target

TO DO

Scenario: Reassembling a chunked body

Givena body encoded as "5\r\nhello\r\n6\r\n world\r\n0\r\n\r\n"

Whenit is decoded

Thenthe reassembled body is "hello world"

Andthe terminating "0" chunk ends the body

Background

Sometimes a client streams a body whose total size it does not know in advance — so it cannot send a Content-Length. Chunked transfer encoding solves this: the body is a series of chunks, each written as its size in hexadecimal, a CRLF, the chunk bytes, and another CRLF. A zero-size chunk marks the end.

Decoding is a loop: read the size line, parse it as hex, read exactly that many bytes (reusing your read-exactly primitive), consume the trailing CRLF, and append. Stop at the 0 chunk. The result is the same body a Content-Length request would have carried — the framing just arrived differently.

Make it work

for {
    line, _ := readLine(r)
    n, _ := strconv.ParseInt(line, 16, 64) // chunk size is HEX
    if n == 0 { break }                    // last chunk
    chunk := readBody(r, int(n))
    readLine(r)                            // consume the trailing CRLF
    out = append(out, chunk...)
}

CheckpointDONE

A chunked request body reassembles into the bytes it represents. Commit.