[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[bug#73439] [PATCH v5 1/5] import: Add %libreoffice-updater.
|
From: |
Ludovic Courtès |
|
Subject: |
[bug#73439] [PATCH v5 1/5] import: Add %libreoffice-updater. |
|
Date: |
Fri, 18 Oct 2024 14:31:47 +0200 |
|
User-agent: |
Gnus/5.13 (Gnus v5.13) |
Hi,
Nicolas Graves <ngraves@ngraves.fr> skribis:
> Change-Id: I481b1175db531c4fea4a57838fe190f679cd1a85
> ---
> Makefile.am | 1 +
> guix/import/libreoffice.scm | 98 +++++++++++++++++++++++++++++++++++++
Neat. Could you add a test, an @item in doc/guix.texi next to the other
updaters, and a commit log?
> +(define (libreoffice-latest-version)
Please add a docstring to all top-level procedures.
> + (let* ((response port (http-get libreoffice-latest-url
> + #:streaming? #t))
> + (content (get-string-all port))
PORT is not closed. If you’re going to load it all in memory, just
leave #:streaming? to #f and you’ll get a string instead of a port.
> + ;; xml->sxml is not flexible enough for html.
> + ;; For instance, <img> tags don't have closing </img>.
> + ;; This trick preprocesses html to extract all <a> tags in
> + ;; a <body> wrapper, which sxml-match can handle well.
> + (xml (xml->sxml
> + (string-append
> + "<body><"
> + (string-join
> + (filter (cute string-prefix? "a " <>)
> + (string-split content #\<))
> + "</a><")
> + "></a></body>")
> + #:trim-whitespace? #t)))
This is terrible. :-)
(guix import go) and (guix gnu-maintenance) use (htmlprag) for that.
Would that work for you?
Thinking about it, the strategy looks very similar to that of the
‘generic-html’ updater.
Would it be enough to add a ‘release-monitoring-url’ property to the
relevant LibreOffice packages?
Thanks,
Ludo’.