RE: Extract XML from docx file with xproc 3.0 p:unarchive

Matthieu:

You make an excellent point about running Schematron over the XProc. I am doing the same in my project at https://github.com/usnistgov/oscal-xproc3 - saving me many, many hours debugging.

The Schematrons are also applied to the XProc files under CI/CD, i.e. whenever they are pushed into the repository. Using Morgana under Github actions. Another pipeline under CI/CD runs XSpec test suites in XProc 3.0. Everything is public domain / open source.

So yes! There are lots of good ideas out there … I’ve lifted most of mine from elsewhere. 😊

Although I’m not sure I’ll be using the “./” trick except *in extremis*….

Can I have a Schematron to warn me when I am sending email to you or Geert when I mean to write to the list? (Don’t look now, the AIs are coming.)

Regards, Wendell

From: Matthieu RICAUD-DUSSARGET <m.ricaud-dussarget@lefebvre-dalloz.fr>
Sent: Wednesday, October 9, 2024 3:28 AM
To: Piez, Wendell A. (Fed) <wendell.piez@nist.gov>
Subject: RE: Extract XML from docx file with xproc 3.0 p:unarchive

Hi Wendel,

Thanks for your response. Yes I guess I’ll had a try catch on the whole process, especially the XSLT which might crash depending on the word content.

Using href="./{expr}" looks a bit strange, but I would have seen my mistake before ;)
While coding in XSLT within Oxygen, I get a warning when using a variable name without $, this is a schematron control.
I think a good IDE for developing xproc might help avoiding such typos.
I did develop a schematron to control XSLT quality (https://github.com/mricaud/xslt-quality) maybe doing the same with Xproc might help !

Thanks for your feedback and good ideas !


Cheers
Matthieu Ricaud
De : Piez, Wendell A. (Fed) <wendell.piez@nist.gov<mailto:wendell.piez@nist.gov>>
Envoyé : mardi 8 octobre 2024 23:55
À : Matthieu RICAUD-DUSSARGET <m.ricaud-dussarget@lefebvre-dalloz.fr<mailto:m.ricaud-dussarget@lefebvre-dalloz.fr>>
Objet : RE: Extract XML from docx file with xproc 3.0 p:unarchive

[Mail EXTERNE]: Vérifiez bien l’expéditeur de l’email avant de cliquer sur des liens ou pièces-jointes!

Matthieu -- oops!

I wrote also to suggest try/catch for you, but it appears the email went only to Geert. (Sorry Geert.)

Probably not the last time – and of course it wouldn’t have solved this problem, only helped to mitigate similar problems caused by actual errors in inputs, not errors in the code.

For that matter I have also been bitten by the fallback to read the XProc at path “”.

One thing that occurred to me would be to prepend any href to a relative path:

<p:load href="./{expr}"/>

And this does error out (in Morgana) if ‘expr’ evaluates to the empty string.

But part of me says this is bad form (it feels strange and awkward), and I should just rely on runtime messaging to expose the values for debugging.

Comments?

Regards, Wendell

From: Matthieu RICAUD-DUSSARGET <m.ricaud-dussarget@lefebvre-dalloz.fr<mailto:m.ricaud-dussarget@lefebvre-dalloz.fr>>
Sent: Tuesday, October 8, 2024 5:00 PM
To: list.mu@c-moria.com<mailto:list.mu@c-moria.com>; xproc-dev@w3.org<mailto:xproc-dev@w3.org>
Subject: RE: Extract XML from docx file with xproc 3.0 p:unarchive

Hi all,

Thanks for your responses !

Christophe : yes the only docx file I have for the moment in the directory is valid : I can open it with 7-zip and get the xml document inside
It’s actually a .doc which I have converted with my MS Word (« save as  docx »)

Geert, thanks for all details, yes I might be interested with your script (though I have about 4 millions .docx to convert !)
I’ll also have a look to Aspose.

About my pipeline, thanks to your help I find the problem which was .. so dummy !
I forgot a $  before the docx.uri variable reference in <p:load href="{docx.uri}" … />
=> The href was empty, so I guess the fallback is to take the current xproc file as default and raides err:XC0085 "Cannot process document with media-type 'application/xproc+xml' as a ZIP archive"
Then when I added explicit binding to understand, then I specified (and force) the content-type …
I finally get an the err:XC0085 "Error processing ZIP archive: zip END header not found"  which made me more confused !

Sorry for that guys !

As for reminder for later my really simple xpl that works (display the xml inside the docx) :

<p:declare-step xmlns:p=http://www.w3.org/ns/xproc<https://urldefense.com/v3/__http:/www.w3.org/ns/xproc__;!!KEc074MNZw!bdGqb4wbmFbglhcs25hQR8f7Qnyrvexr4Xx8DlazA8WdiuL_0jOFiHxG4XGuGua1DtncTQO3DfW0dKtW16SAa0_sJTYxUk95g6FGku-UMs0$>
  xmlns:c=http://www.w3.org/ns/xproc-step<https://urldefense.com/v3/__http:/www.w3.org/ns/xproc-step__;!!KEc074MNZw!bdGqb4wbmFbglhcs25hQR8f7Qnyrvexr4Xx8DlazA8WdiuL_0jOFiHxG4XGuGua1DtncTQO3DfW0dKtW16SAa0_sJTYxUk95g6FGz5c7UoE$>
  xmlns:xs=http://www.w3.org/2001/XMLSchema<https://urldefense.com/v3/__http:/www.w3.org/2001/XMLSchema__;!!KEc074MNZw!bdGqb4wbmFbglhcs25hQR8f7Qnyrvexr4Xx8DlazA8WdiuL_0jOFiHxG4XGuGua1DtncTQO3DfW0dKtW16SAa0_sJTYxUk95g6FGru9TTl8$>
  version="3.0">

  <p:input port="source" sequence="true"/>
  <p:output port="result" sequence="true"/>

  <p:option name="input-dir" select="resolve-uri('../../test/input-word', static-base-uri())" as="xs:string"/>

  <p:directory-list path="{$input-dir}" include-filter="\.docx$"/>

  <p:for-each>
    <p:with-input select="//c:file"/>
    <p:variable name="docx.uri" select="c:file/base-uri(.)"/>
    <p:identity message="Processing {$docx.uri}"/>
    <p:load href="{$docx.uri}"/>
    <p:unarchive format="zip" include-filter="word/document\.xml"/>
  </p:for-each>

</p:declare-step>

Cheers,
Matthieu Ricaud