XProc 3.1: Invisible XML

In this first example, no grammar is provided, so the pipeline parses the Invisible XML grammar on the source port and returns its XML representation:

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" version="3.1"> <p:output port="result"/> <p:ixml> <p:with-input port="grammar"><p:empty /></p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain"> date: s?, day, s, month, (s, year)? . -s: -" "+ . day: digit, digit? . -digit: "0"; "1"; "2"; "3"; "4"; "5"; "6"; "7"; "8"; "9". month: "January"; "February"; "March"; "April"; "May"; "June"; "July"; "August"; "September"; "October"; "November"; "December". year: (digit, digit)?, digit, digit . </p:inline> </p:with-input> </p:ixml> </p:declare-step>

This would produce an XML version of the grammar:

<ixml> <rule name="date"> <alt> <option> <nonterminal name="s"/> </option> <nonterminal name="day"/> <nonterminal name="s"/> <nonterminal name="month"/> <option> <alts> <alt> <nonterminal name="s"/> <nonterminal name="year"/> </alt> </alts> </option> </alt> </rule> <!-- … remaining rules elided for brevity … --> </ixml>

If the grammar is provided on the grammar port, it can be used to parse input, the string “31 December 2021” in this case:

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" version="3.1"> <p:output port="result"/> <p:ixml> <p:with-input port="grammar"> <p:inline content-type="text/plain"> date: s?, day, s, month, (s, year)? . -s: -" "+ . day: digit, digit? . -digit: "0"; "1"; "2"; "3"; "4"; "5"; "6"; "7"; "8"; "9". month: "January"; "February"; "March"; "April"; "May"; "June"; "July"; "August"; "September"; "October"; "November"; "December". year: (digit, digit)?, digit, digit . </p:inline> </p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain">31 December 2021</p:inline> </p:with-input> </p:ixml> </p:declare-step>

This would produce an XML version of the date:

<date><day>31</day><month>December</month><year>2021</year></date>

If a parse fails, the implementation must indicate this, but it may also provide information about where the processing failed.

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" version="3.1"> <p:output port="result"/> <p:ixml fail-on-error="false"> <p:with-input port="grammar"> <p:inline content-type="text/plain"> date: s?, day, s, month, (s, year)? . -s: -" "+ . day: digit, digit? . -digit: "0"; "1"; "2"; "3"; "4"; "5"; "6"; "7"; "8"; "9". month: "January"; "February"; "March"; "April"; "May"; "June"; "July"; "August"; "September"; "October"; "November"; "December". year: (digit, digit)?, digit, digit . </p:inline> </p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain">31 Mumble 2021</p:inline> </p:with-input> </p:ixml> </p:declare-step>

Here the output might be something like this:

<error xmlns:ixml="http://invisiblexml.org/NS" xmlns:ex="http://example.com/NS" ixml:state="failed" ex:lastChar="4"> <parse> month -> • M a r c h month -> M • a r c h </parse> <parse> month -> • M a y month -> M • a y </parse> </error>

In the case of failure, Invisible XML requires that the ixml:state attribute appear on the root element containing the token “failed”. It doesn’t constrain the implementation’s choice of the root element or the content of the document.

An ixml grammar may be ambiguous. In the grammar below, there are three different possible ways to parse the input. By default, one of them is returned.

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" version="3.1"> <p:output port="result"/> <p:ixml> <p:with-input port="grammar"> <p:inline content-type="text/plain"> letters: X, (A; B; C) . A: digits . B: digits . C: digits . X: "a" . digits: ["0"-"9"]+ . </p:inline> </p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain">a123</p:inline> </p:with-input> </p:ixml> </p:declare-step>

This might return any one of these parses:

<letters ixml:state="ambiguous" xmlns:ixml="http://invisiblexml.org/NS"><X>a</X><C><digits>123</digits></C></letters>

or

<letters ixml:state="ambiguous" xmlns:ixml="http://invisiblexml.org/NS"><X>a</X><A><digits>123</digits></A></letters>

or

<letters ixml:state="ambiguous" xmlns:ixml="http://invisiblexml.org/NS"><X>a</X><B><digits>123</digits></B></letters>

All are equally correct.

An implementation might provide a parameter to allow the author to select a particular parse:

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" xmlns:ex="http://example.com/" version="3.1"> <p:output port="result"/> <p:ixml parameters="map{'ex:select':2}"> <p:with-input port="grammar"> <p:inline content-type="text/plain"> letters: X, (A; B; C) . A: digits . B: digits . C: digits . X: "a" . digits: ["0"-"9"]+ . </p:inline> </p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain">a123</p:inline> </p:with-input> </p:ixml> </p:declare-step>

This might return:

<letters ixml:state="ambiguous"><X>a</X><A><digits>123</digits></A></letters>

Or a processor might provide a parameter to return all of the parses.

<p:declare-step xmlns:p="http://www.w3.org/ns/xproc" xmlns:ex="http://example.com/" version="3.1"> <p:output port="result"/> <p:ixml parameters="map{'ex:select':'all'}"> <p:with-input port="grammar"> <p:inline content-type="text/plain"> letters: X, (A; B; C) . A: digits . B: digits . C: digits . X: "a" . digits: ["0"-"9"]+ . </p:inline> </p:with-input> <p:with-input port="source"> <p:inline content-type="text/plain">a123</p:inline> </p:with-input> </p:ixml> </p:declare-step>

This might return three documents:

<letters ixml:state="ambiguous"><X>a</X><C><digits>123</digits></C></letters> <letters ixml:state="ambiguous"><X>a</X><B><digits>123</digits></B></letters> <letters ixml:state="ambiguous"><X>a</X><A><digits>123</digits></A></letters>

As before, there is nothing standardized about the results in this case.