Astro sitemap allow opting out of dashes in sitemap.xml

[Hacksore]

Summary

The following places there is a hardcoded dash - so we should allow a way to opt-out of that for the SEO best practices.

https://github.com/withastro/astro/blob/ff8004f6a7b2aab4c6ac367f13744a341c3c5462/packages/integrations/sitemap/src/index.ts#L47
https://github.com/withastro/astro/blob/ff8004f6a7b2aab4c6ac367f13744a341c3c5462/packages/integrations/sitemap/src/write-sitemap.ts#L42

Background & Motivation

Someone told me so it's bad practice to have hyphens/dashes and it can yield a site map to no load.
https://www.seroundtable.com/google-sitemaps-error-with-hyphenated-file-names-36829.html#:~:text=Google%20might%20be%20having%20some,will%20not%20trigger%20any%20errors.

Goals

A concise, bulleted-list outlining the intended goals of this RFC.

• SEO best practices

Example

pseudo codeish for how It could work?

import sitemap from "@astrojs/sitemap";

export default defineConfig({
  integrations: [
    sitemap({
      // NOTE: new support for no dashes
      // omit the extension, it will always be .xml
      baseFilename: "sitemap",
      // base name, we omit .xml as they always have to be that
      filename: async ({ index, baseFilename }}) => {
        return baseFilename + index; // sitemap1.xml, sitemap2.xml, etc
      },
      // // NOTE: backwards compat which we can keep as default so no breaking changes?
      // // omit the extension, it will always be .xml
      // baseFilename: "sitemap-index",
      // // base name, we omit .xml as they always have to be that
      // filename: async ({ index, baseFilename }}) => {
      //   return `sitemap-${index}`; // this will be sitemap-1.xml, sitemap-2.xml, etc
      // },
    }),
  ],
});

[RhysSullivan]

Here's an example:

[florian-lefebvre]

I think it could make sense to have a separator property defaulting to '-' in v4. Could make sense to default it to '_' in v5 if it's better for SEO:

export default defineConfig({
	integrations: [sitemap({
		separator: '_' // or ''
	})]
})

[adaliszk]

Why not provide a way to specify the entire pattern or define a pure function that generates the filename?

export default defineConfig({
    integrations: [
        sitemap({
            filename: "sitemap{index}.xml",
        }),
    ],
});

or

export default defineConfig({
    integrations: [
        sitemap({
            filename({ index }) {
                return `sitemap${index}.xml`;
            }, 
        }),
    ],
});

In both cases, the integration can expect a type of ${string}.xml as an output, so people would be forced to provide a valid filename.

[Hacksore]

@adaliszk Should we allow the main sitemap file to be named differently than the index generated files?

base file: sitemap-index.xml
index files: sitemap-0.xml ... sitemap-13.xml

[adaliszk]

well, there are also other patterns that should be supported, like https://developers.google.com/search/docs/crawling-indexing/sitemaps/build-sitemap#cross-submit: https://www.example.com/sub1-example-com-sitemap.xml

[Hacksore]

So this is what I am thinking now which should cover all use-cases. Let me know your thoughts @adaliszk @florian-lefebvre.

import sitemap from "@astrojs/sitemap";

export default defineConfig({
  integrations: [
    sitemap({
      // NOTE: new support for no dashes
      // omit the extension, it will always be .xml
      baseFilename: "sitemap",
      // base name, we omit .xml as they always have to be that
      filename: async ({ index, baseFilename }}) => {
        return baseFilename + index; // sitemap1.xml, sitemap2.xml, etc
      },
      // // NOTE: backwards compat which we can keep as default so no breaking changes?
      // // omit the extension, it will always be .xml
      // baseFilename: "sitemap-index",
      // // base name, we omit .xml as they always have to be that
      // filename: async ({ index, baseFilename }}) => {
      //   return `sitemap-${index}`; // this will be sitemap-1.xml, sitemap-2.xml, etc
      // },
    }),
  ],
});

[adaliszk]

I quite like having hooks so that you can pull in any logic to resolve the name you'd like. One thing that should be supported is async-await just in case some logic is promise-based.

[Hacksore]

@adaliszk SGTM, I've updated the original post as well as this one.

[Hacksore]

Another thing that might be relevant here is how to handle two domains and one static app.

ie. x.app and x.com

@RhysSullivan do you know what’s the typical way to do this?

my guess is dynamic robots.txt and sitemap 😳.

[vin-e]

Good work on this thus far @Hacksore.

I am currently rebuilding a site from Eleventy to Astro. The existing site uses sitemap.xml as the base file. I would appreciate a way to customize the name or remove the -index from the file. I currently do not have access to the google search console to make the appropriate updates.

Thanks

[Hacksore]

yeah then I think we'd want something like this before we give a better programitic api?

sitemap({
  separator: "", // default: -
  indexFilename: "sitemap" // default: sitemap-index
})

If that seems agreeable I can update my PR withastro/astro#11898