License removed by default. Why?

[toastal]

Describe the bug

SVG files, especially those exported from various editors, usually contain a lot of redundant and useless information.

How is the metadata, especially the license, considered "useless"? Some information might be, but I would assume authors want to retain their licenses.

To Reproduce
Run the default settings on any image with a <metadata> tag

Expected behavior
Some or all of the metadata is preserved or turned into specific options for removeMetadata.

Desktop (please complete the following information):

$ svgo --version
2.7.0
$ node --version
v14.18.0
$ uname -omvprs
Linux 5.14.8-zen1 #1-NixOS ZEN SMP Tue Jan 1 00:00:00 UTC 1980 x86_64 unknown GNU/Linux

[TrySound]

Could you show an example of svg with license?

[toastal]

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- Created with Inkscape (http://www.inkscape.org/) -->

<svg
   width="500"
   height="500"
   viewBox="0 0 132.29167 132.29167"
   version="1.1"
   id="svg5"
   sodipodi:docname="test.svg"
   inkscape:version="1.1 (c68e22c387, 2021-05-23)"
   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
   xmlns:xlink="http://www.w3.org/1999/xlink"
   xmlns="http://www.w3.org/2000/svg"
   xmlns:svg="http://www.w3.org/2000/svg"
   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
   xmlns:cc="http://creativecommons.org/ns#"
   xmlns:dc="http://purl.org/dc/elements/1.1/">
  <title
     id="title1026">MY TITLE</title>
  <sodipodi:namedview
     id="namedview7"
     pagecolor="#ffffff"
     bordercolor="#666666"
     borderopacity="1.0"
     inkscape:pageshadow="2"
     inkscape:pageopacity="0.0"
     inkscape:pagecheckerboard="0"
     inkscape:document-units="px"
     showgrid="true"
     inkscape:zoom="0.838"
     inkscape:cx="-353.22196"
     inkscape:cy="262.52983"
     inkscape:window-width="1918"
     inkscape:window-height="1060"
     inkscape:window-x="0"
     inkscape:window-y="0"
     inkscape:window-maximized="1"
     inkscape:current-layer="svg5">
    <inkscape:grid
       type="axonomgrid"
       id="grid7892"
       color="#3f3fff"
       opacity="0.10196078"
       empcolor="#ff3f5d"
       empopacity="0.2"
       units="px"
       spacingy="2.6458334" />
  </sodipodi:namedview>
  <metadata
     id="metadata1022">
    <rdf:RDF>
      <cc:Work
         rdf:about="">
        <dc:title>MY TITLE</dc:title>
        <dc:date>2021</dc:date>
        <dc:creator>
          <cc:Agent>
            <dc:title>toastal</dc:title>
          </cc:Agent>
        </dc:creator>
        <cc:license
           rdf:resource="http://creativecommons.org/licenses/by-nc-nd/4.0/" />
      </cc:Work>
      <cc:License
         rdf:about="http://creativecommons.org/licenses/by-nc-nd/4.0/">
        <cc:permits
           rdf:resource="http://creativecommons.org/ns#Reproduction" />
        <cc:permits
           rdf:resource="http://creativecommons.org/ns#Distribution" />
        <cc:requires
           rdf:resource="http://creativecommons.org/ns#Notice" />
        <cc:requires
           rdf:resource="http://creativecommons.org/ns#Attribution" />
        <cc:prohibits
           rdf:resource="http://creativecommons.org/ns#CommercialUse" />
      </cc:License>
    </rdf:RDF>
  </metadata>
</svg>

Along with the metadata being stripped, the cc namespace is being stripped related to the editor, but Creative Commons is not limited to Inkscape.

[TrySound]

xmlns:cc attribute is removed as unused after whole metadata is removed. What exactly do you expect to keep in metadata? Only cc:License with its content?

[toastal]

svgo/plugins/_collections.js

Lines 1886 to 1887 in a0bf4f7

	'http://creativecommons.org/ns#',
	'http://www.w3.org/1999/02/22-rdf-syntax-ns#',

module.exports = {
   plugins: [
      {
         name: "preset-default",
         params: {
            overrides: {
               removeEditorsNSData: false,
               removeMetadata: false,
            },
         },
      },
   ],
}

This was the minimal configuration I needed to keep the rdf and cc namespace. But when I keep the Inkscape NSs which I don't need for production uses. However, I want to keep the licensing in tact. I question why L1886-1887 considers Creative Commons and RDF to be part of an editor. It's not.

[toastal]

svgo/plugins/removeEditorsNSData.js

Line 28 in a0bf4f7

namespaces = [...editorNamespaces, ...params.additionalNamespaces];

I suppose if a allowlist was added to the params as well, I could get around my issue, except fundamentally I disagree with rdf and cc being an editor namespace. If anything, these should be in a separate namespace list/plugin.

[toastal]

https://stackoverflow.com/questions/38775825/semantically-correct-way-to-add-a-copyright-notice-into-a-svg-file#answer-38882847

While not exactly the same as the CC namespace, this was the top web search result for how to properly put a license in an SVG file. I think users would be surprised to see their license stripped after setting it in the file not related to their editor.

[timegrinder]

I've also just noticed this on a bunch of SVG's I've been making (unfortunately I hadn't double-checked during the process as I thought it would be considered metadata and not editor data) and found this reported issue.

The attributes in dc fields would also be an optional candidate for keeping if requested as there's already 'Keep title' and 'keep description' options which unfortunately Illustrator and Inkscape don't use directly, all of that goes into what ends up considered 'editor data'.

The allow list would definitely be appreciated as personally the option to keep some accessibility / informational data is entirely necessary, even within an optimised image. This also covers issues that can arise when people are trying to honour creative commons licenses when editing SVG's that they still want to optimise.

Having the ability to 'keep x but optimise where appropriate' the rdf data would be great.

[TrySound]

I see we have a lot of not standartized attributes. Im not familiar with any of them. It would help it you specify a set of rules when its safe and when its not to delete attributes.

[toastal]

not [standardized] attributes

<svg
   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
   xmlns:cc="http://creativecommons.org/ns#"
   xmlns:dc="http://purl.org/dc/elements/1.1/"
   ></svg>

Those namespace links link to the standards. I would say 99% of the time it's unsafe to delete them by default and could maybe sometimes be manually removed (for instance, removing all of the attribution info but putting a link to that information elsewhere on a website to minimize the size of the file).

[timegrinder]

There's a list of broad standards for what attributes and elements are acceptable to use but both programs list the standards in a different structure, and the standards aren't applied evenly from one program to the next. Inkscape and Illustrator both have the same information stored in an entirely different tree and element structures (even to the point the same field is in two different element types) so we'll have to take a list of both of them at least.

The Inkscape version looks more bare and simplified vs the Illustrator one, so that's a start at least.

[GreLI]

@toastal XML is eXtensible format. What you refer is just identifiers of that extensions that someone did. It's not necessarily standard though, anyone involved can extend it in a such way. Well, RDF is known standard, but it has nothing specific with the license, at least that's how it represented here.

See also #431. You words “unsafe” have nothing in support. Please, don't mix your opinion with facts. SVGO just a tool, like knife or axe, and it is within users responsibility how to use it.

[timegrinder]

My mistake for not checking closed issues, and this is more of a feature request than a bug.

While the RDF extension isn't a small fixed standard, and the relevant data isn't stored the same way from one program to the next I'm sure we can break down the editor data cleaner into parts to choose to remove and keep+optimise the portions we want. The data definitely isn't stored optimally anyway, and the title/desc naturally have their own elements that don't need to be inside an rdf area and can be moved out.

I can definitely see why rdf gets cleaned out with the EditorNS - an example is Illustrator nests the rdf space inside the x:xmpmeta area. It places the details we're talking about inside an unneccesarily deep tree. It also doesn't even use the same RDF conventions as Inkscape does.

Amusingly I just noticed that 'keep metadata' will still keep the <metadata/> tag when empty despite it being useless (although, admittedly, I use SVGO through the SVGOMG which might not be a perfect replication of SVGO's workings).

What I would suggest is those who are interested in working on it bring together the list of rules that TrySound requested and also work out a more efficient method of keeping the details we want to keep that's compatible with various editor formats as there is definitely room to optimise that information if keeping it.

An example being with license information - we do not need to keep anything more than the actual license name+link itself. Inkscape includes all of the extra license resource web links that are not needed since a user could just open the initial link and see all of those details for themselves. But this could end up included if we just blanketly permitted cc spaces.

[toastal]

I can’t remark much on that. I’ve not opened Illustrator in like 10 years; my typical workflow is bouncing between Inkscape & Neovim. The Inkscape output seems reasonable though.

[toastal]

Please, don't mix your opinion with facts. SVGO just a tool [...]

@GreLI Flipped, how is it a “fact” that metadata is useless? As in the context of

SVG files, especially those exported from various editors, usually contain a lot of redundant and useless information.

I must’ve searched the wrong words to not have bumped #431 instead of opening a new bug, but I agree with that post that the this feels like opt-in instead of opt-out behavior.

Is it not actionable to separate license information from the actual editor metadata? Currently opting out of the overly-broad removeMetadata to retain license information, means you get all of the editor metadata which is one of the primary reasons for using this tool. My use case was to retain the metadata for Inkscape in my source control for future editing while stripping the editor metadata but leaving the license information for using on the web.

SVGO just a tool [...] and it is within users responsibility how to use it.

svgo is included in a lot of front-end tooling with just the default settings--and in the case of like react-scripts users don't know what they are using & often can't even configure it if they did. This why I feel the current default settings are problematic.