ETag Regular Expressions

Question

I noticed that Feedzirra uses this regex to get the ETag from response header:

/.*ETag:\s(.*)\r/

Personally I would have written this one:

/ETag:\s(.*)\n/

Here the questions:

1. Why does it put .* at the beginning even if it is unnecessary (\A is not specified)?
2. Why does it use \r instead of \n? What is the difference?

Answer 1

The HTTP RFC mandates CRLF as a linebreak for HTTP message. So \n would match extra \r with properly formatted message:

    generic-message = start-line
                      *(message-header CRLF)
                      CRLF
                      [ message-body ]
    start-line      = Request-Line | Status-Line

That said, I would make it [\r\n] for the sake of robustness.

Answer 2

I would agree .* at the beginning is not needed

/r and /n are different characters. 

/r = line ending for old macs
/n = line ending for *nix
/r/n = line ending for windows

probably [\r\n] would be best

Answer 3

1. Completeness I dare say. It's not part of a capture. There may be an implicit start of line anchor though, depending on language and implementation, and in this case it may be necessary.
2. The HTTP spec says that HTTP is to use "\r\n" as a line ending. In most programming languages, only "\n" is treated as a line ending. The \r makes sure that the \r is not swallowed inside the .* which would give erroneous whitespace at the end of the capture.

Answer 4

Let’s take a look into the HTTP specification:

The ETag header field is defined as:

ETag = "ETag" ":" entity-tag

The entitiy-tag is defined as:

entity-tag = [ weak ] opaque-tag
weak       = "W/"
opaque-tag = quoted-string

And quoted-string is defined as:

quoted-string  = ( <"> *(qdtext | quoted-pair ) <"> )

So the ETag header field value may contain a line break. And the correct regular expression would be:

/ETag:\s+(?:W\/)?"(?:[ !#-\x7E\x80-\xFF]*|\r\n[\t ]|\\.)*"/