gatsby-transformer-remark - unable to filter allMarkdownRemark nodes by parent File node sourceInstanceName.

[kimbaudi]

Description

MarkdownRemark nodes already provide fileAbsolutePath, but it would be really useful to also provide parent File node sourceInstanceName, relativeDirectory, name in MarkdownRemark nodes as fileSourceInstanceName, fileRelativeDirectory, and fileName.

Gatsby provides gatsby-source-filesystem and gatsby-transformer-remark plugins to source markdown files and generate html pages.

gatsby-config.js

module.exports = {
  plugins: [
    {
      resolve: `gatsby-source-filesystem`,
      options: {
        name: `blog-posts`,
        path: path.resolve(__dirname, `./src/contents/blog-posts`),
      },
    },
    {
      resolve: `gatsby-transformer-remark`,
      options: {
        excerpt_separator: `<!--more-->`,
      },
    },
  ],
}

Right now, I am filtering MarkdownRemark nodes using glob on fileAbsolutePath field to get all markdowns under src/contents/blog-posts folder. However, I want to filter MarkdownRemark nodes using parent File node sourceInstanceName field.

{
  allBlogPost: allMarkdownRemark(
    filter: {fileAbsolutePath: {glob: "**/blog-posts/**"}}
    sort: {fields: frontmatter___date, order: DESC}
  ) {
    blogPosts: nodes {
      fileAbsolutePath
      fields {
        slug
      }
      parent {
        ... on File {
          sourceInstanceName
        }
      }
      frontmatter {
        title
        author
        date(formatString: "MMMM DD, YYYY")
        tags
      }
    }
  }
}

Unfortunately, I cannot filter on parent other than id so I am unable to filter MarkdownRemark nodes by sourceInstanceName.

{
  allBlogPost: allMarkdownRemark(
    filter: {
      parent: {sourceInstanceName: "blog-posts"}
    }
    sort: {fields: frontmatter___date, order: DESC}
  ) {
    blogPosts: nodes {
      fileAbsolutePath
      fields {
        slug
      }
      parent {
        ... on File {
          sourceInstanceName
        }
      }
      frontmatter {
        title
        author
        date(formatString: "MMMM DD, YYYY")
        tags
      }
    }
  }
}

It would be great if MarkdownRemark provided fileSourceInstanceName as well as fileRelativeDirectory and fileName so we can filter allMarkdownRemark using those fields:

{
  allBlogPost: allMarkdownRemark(
    filter: {fileSourceInstanceName: {eq: "blog-posts"}}
    sort: {fields: frontmatter___date, order: DESC}
  ) {
    blogPosts: nodes {
      fileAbsolutePath
      fileSourceInstanceName
      fileRelativeDirectory
      fileName
    }
  }
}

or provide a file node with those fields:

{
  allBlogPost: allMarkdownRemark(
    filter: {file: sourceInstanceName: {eq: "blog-posts"}}
    sort: {fields: frontmatter___date, order: DESC}
  ) {
    blogPosts: nodes {
      file {
        absolutePath
        sourceInstanceName
        relativeDirectory
        name
      }
    }
  }
}

I think having the ability to filter allMarkdownRemark nodes by File sourceInstanceName would be a cleaner alternative to globbing fileAbsolutePath.

Steps to reproduce

I've made a minimal repo at https://github.com/kimbaudi/gatsby-group-query.

Expected result

Have ability to filter allMarkdownRemark nodes by file sourceInstanceName field

Actual result

Only have ability to filter allMarkdownRemark nodes by fileAbsolutePath field and not sourceInstanceName field (or file relativeDirectory, name fields)

{
  allBlogPost: allMarkdownRemark(
    filter: { fileAbsolutePath: { glob: "**/blog-posts/**" } }
  ) {
    blogPosts: nodes {
      fileAbsolutePath
    }
  }
}

[kimbaudi]

linking to discussion #30393

[LekoArts]

You can use https://www.gatsbyjs.com/docs/reference/config-files/actions/#createNodeField to add additional fields

[kimbaudi]

@LekoArts - thank you for your suggestion... it works perfectly!

For anyone else interested, here is my onCreateNode using createNodeField to add absolutePath, name and sourceInstanceName fields:

gatsby-node.js

exports.onCreateNode = ({ node, actions: { createNodeField }, getNode }) => {
  if (node.internal.type === `MarkdownRemark`) {
    const { absolutePath, name, sourceInstanceName } = getNode(node.parent)
    const relativeFilePath = createFilePath({
      node,
      getNode,
    })
    createNodeField({
      name: `absolutePath`,
      node,
      value: absolutePath,
    })
    createNodeField({
      name: `name`,
      node,
      value: name,
    })
    createNodeField({
      name: `sourceInstanceName`,
      node,
      value: sourceInstanceName,
    })
    createNodeField({
      name: `slug`,
      node,
      value: `/${sourceInstanceName}${relativeFilePath}`,
    })
  }
}

And now I can filter those fields: