Is there a way to split a string in fixed width chunks in XPath?

Question

Using xidel I'm extracting the //Assertion//Signature//KeyInfo//X509Certificate/text() from a SAMLResponse, this is a X509 certificate as a long base64 string.

I want to split this string into 64 chars blocks

I tried with tokenize() and replace() but I could make those work,

It seems that replace() does not allow me to use newlines \n in the replacement string:

echo "$SAMLRESPONSE" | base64 -D | xidel --xpath 'replace(//Assertion//Signature//KeyInfo//X509Certificate/text(),"(.{64})","$1\n")' -
**** Processing: stdin:/// ****
Error:
err:FORX0004: Invalid replacement: $1\n after $1\n
Possible backtrace:
  $000000010203F668: perhaps TXQTermTryCatch + 222920 ? but unlikely
  $0000000102068BBE: perhaps Q{http://www.w3.org/2005/xpath-functions}tokenize + 166350 ? but unlikely
  $000000010203FF78: Q{http://www.w3.org/2005/xpath-functions}replace + 376
  $0000000101FF853F: TXQTermNamedFunction + 767
  $0000000101F71CE7: perhaps ? ? but unlikely

Call xidel with --trace-stack to get an actual backtrace

And tokenize will treat the whole match as separator, and separator are not included in the output

echo "$SAMLRESPONSE" | base64 -D | xidel --xpath 'tokenize(//Assertion//Signature//KeyInfo//X509Certificate/text(),"(?:.{64})")' -
**** Processing: stdin:/// ****















XACcI5tcJbgsvr+ivGPos/WrhywkROwbEBh6OTNXTnaBiiIK

Is there any way to do split a string in fixed width chunks in XPath?

Answer 1

Your first idea wasn't wrong, you just have to use the codepoints-to-string function to generate the newline character:

printf %s "$SAMLRESPONSE" |
base64 -D |
xidel --xpath '
    let
        $cert := //Assertion//Signature//KeyInfo//X509Certificate
    return
        "-----BEGIN CERTIFICATE-----" || codepoints-to-string(10) ||
        replace( $cert, ".{1,64}", "$0" || codepoints-to-string(10) ) ||
        "-----END CERTIFICATE-----" || codepoints-to-string(10)
' -

^{note: I modified the regex to .{1,64} to make sure that the "replaced" string always ends with a linefeed}

---

ASIDE: In the first place, you don't even need to build the full output with XPath.

{
    echo '-----BEGIN CERTIFICATE-----'

    printf %s "$SAMLRESPONSE" |
    base64 -D |
    xidel --xpath '//Assertion//Signature//KeyInfo//X509Certificate' - |
    fold -w 64

    echo '-----END CERTIFICATE-----'
}

Answer 2

It seems that replace() does not allow me to use newlines \n in the replacement string:

That's because regular expressions can't be used in the replacement string. You have to use HTML entities or x:cps():

replace(...,"(.{1,64})","$1
")
replace(...,"(.{1,64})","$1
")
replace(...,"(.{1,64})","$1"||x:cps(10))

And tokenize will treat the whole match as separator

https://www.w3.org/TR/xpath-functions-31/#func-tokenize:

Returns a sequence of strings constructed by splitting the input wherever a separator is found

You want to split the input based on a separator it doesn't have. So tokenize() is unsuitable. Instead, as an alternative to replace(), you could use Xidel's own x:extract(). But above all, together with parse-xml() and x:binary-to-string() this can be done much simpler and all with Xidel:

$ echo "$SAMLRESPONSE" | xidel -se '
  "-----BEGIN CERTIFICATE-----",
  binary-to-string(base64Binary($raw)) ! extract(
    parse-xml(.)//Assertion//Signature//KeyInfo//X509Certificate,
    ".{1,64}",0,"*"
  ),
  "-----END CERTIFICATE-----"
'

And because a newline is the default value for --output-separator, there's no need for codepoints-to-string(10) either.