*.org files in graph view?

[PolidoroFederico2am]

Is it possible to display relationships between *.org files in the graph view? Or is it way outside the scope of the project?

[BBazard]

The graph view displays markdown links, it is not open source and there is no way to interact with it with the obsidian api as far as I know.

I don't know if you are talking about org<->org relationships or org<->markdown relationships but I don't think it is possible to add anything to the graph view that is not markdown.

[kris-chin]

Hi! It's super unfortunate that the Obsidian devs are not planning on opening the graph api to the public. I currently use your plugin alongside obsidian-binary-file-manager-plugin for org<->markdown relationships. I have been just manually tagging and linking the metadata .md files generated by the plugin to model the relationships.

Would it be possible to provide a way to expose the link and tag connections externally to enable a potential automatable workaround to this problem? I'm considering using Templater for this.

[BBazard]

Unfortunate indeed, even though I have to say I never asked the Obsidian devs about opening the graph api.

The integration with obsidian-binary-file-manager-plugin would make sense. Can I know more about what you have in mind? I am guessing you only care about the links pointing to files in the vault (as opposed to https links). For the tags, what do you expect (how do you define the tags in the org files, what should appear in the graph view)? I don't use the graph view.

[kris-chin]

Since Templater allows for templates to execute arbitrary JavaScript, I want to write a template that automatically places the tags and links into the generated .md file. That way, it shows up on the graph properly represented.

Some of my org files have multiple node IDs in one file. Representing that sounds kinda tricky, so I'm willing to compromise and just represent the links to the file itself, rather than the individual IDs within the file. Additionally, it looks like md-roam + org-roam-ui might provide that functionality already if I need it, but I am interested in doing this within Obsidian's graph specifically.

I am guessing you only care about the links pointing to files in the vault (as opposed to https links).

Yep! I only care about the file links in the vault.

For the tags, what do you expect (how do you define the tags in the org files, what should appear in the graph view)? I don't use the graph view.

I try to stay in line with how org-roam treats tags by applying file-level tags with #+FILETAGS. This then applies the tags to all node IDs in the file. My uses for heading-level tags are not for org-roam specifically, so I don't care about linking those.

Regarding the graph view, I use the tags for filtering, so either displaying nodes with a certain tag or displaying the nodes that don't have it. Obsidian's graph also provides the option to represent the tag as a node itself and I use that sometimes.

[BBazard]

Thank you for the precision, here is how I see it:
First an orgmode file bfm-example.org:

#+TITLE: Example file to use with Binary File Manager (BFM) plugin
#+FILETAGS: :dog:cat:unicorn:
* first subject :tagignored:

some text with an [[file1][internal]] link and an [[https://google.com][external]] link

PS: To make it work in the graph view, we also need to create the file file1.md explicitly

The companion file (that could be generated by Binary File Manager, or not) could look like INFO_bfm-example_ORG.md:

---
link: "[[bfm-example.org]]"
created_at: 2025-03-28
file_type: org
---

# Tags
- #dog
- #cat
- #unicorn

# Links
- [[file1|internal]]
- [[https://google.com|external]]

Finally for linking to the orgmode file from other files there is two possibilities, with the previously mentioned file1.md:

linking to the org file: [[bfm-example.org]]
linking to the companion file: [[INFO_bfm-example_org]]

Heading-level tags are just ignored since the graph view show the files.
I added the https link in the companion file since I believe it matches more closely the behavior of other markdown files.
@kris-chin I believe this setup would make the graph work like you intend, anything you would change?

[BBazard]

For the templater generation the frontmatter can be generated by binary-file-manager like so:

---
link: "{{LINK}}"
created_at: "{{CDATE:YYYY-MM-DD}}"
file_type:  "{{EXTENSION}}"
---

Extracting the tags and links from the orgmode file is more tricky since we have to convert the syntax from orgmode to markdown, but at least you can automate some part by opening the devtools (ctrl+shift+I) while having the orgmode file opened and execute:

// capture the codemirror state of the file at that point, doesn't take into account later modifications of the file
const state = app.workspace.getActiveFileView().codeMirror.state
// Iterate on the syntax tree of the orgmode file and extracts #+FILETAGS lines
state.tree.iterate(
  {
    enter(node) {
      if (node.type.name === 'KeywordComment') {
        const text = state.doc.sliceString(node.from, node.to)
        if (text.toUpperCase().contains("#+FILETAGS")) {
          const filetagsText = text
          console.log(filetagsText)
        }
      }
    }
  }
)
// Iterate on the syntax tree of the orgmode file and extracts links
state.tree.iterate(
  {
    enter(node) {
      if (node.type.name === 'PlainLink' || node.type.name === 'AngleLink' || node.type.name === 'RegularLink') {
        const linkText = state.doc.sliceString(node.from, node.to)
        console.log(linkText)
      }
    }
  }
)

(We iterate on the codemirror syntax tree and filter on the grammar tokens)

[kris-chin]

Thank you for the snippets! @BBazard I've modified them into a single Templater template to fully automate the process without the need for devtools!
(cc: @PolidoroFederico2am)

Results:

Auto Generated Metadata File Content:

Resulting full graph. Dark blue is auto-generated org-mode nodes.

Closeup of org nodes in obsidian graph:

Template File

Usage

Create a new .md file with the contents of this template file. In the settings of binary-file-manager-plugin, set this template file as your metadata template.

Header

---
link: "{{LINK}}"
name: "{{NAME}}"
extension: "{{EXTENSION}}"
file Path: "{{PATH}}"

Body

<%*
/**

This workaround script will grab all tags and valid internal links for org files. 
It does this by opening a new tab, pulling the links+tags, then closing the tab.
TODO:
    - watch for link+tag updates on .org files, 
    - create additional metadata files for multi-id org files.

This may create some lag when initially creating metadata files.

This script assumes all metadata files are formatted as such
eg. TEST.org <--> "_metadata/TEST.org.md"
If your naming scheme is different, please modify it below.
**/
const METADATA_FILE_FORMAT = (file_basename) => {
  return `_metadata/${file_basename}.org.md`
}

const write_tags_and_links = async () => {
  let output = "";
  const PATH = `{{PATH}}`
  if (PATH.match(/\{\{PATH\}\}/) != null) {
    return output += `---\nPlease use this template with binary-file-manager`;
  }
  if (PATH.match(/\.org$/) == null) {
    return output += "---";
  }
  //open a new tab with the .org file, and get the codeMirror state of it.
  const linkedFile = app.vault.getFileByPath(`${PATH}`)
  const leaf = app.workspace.getLeaf('tab')
  await leaf.openFile(linkedFile);
  const state = leaf.view.codeMirror.state

  // Iterate on the syntax tree of the orgmode file and extract #+FILETAGS lines
  output += `tags: `
  state.tree.iterate(
    {
      enter(node) {
        if (!(node.type.name === 'KeywordComment')) {
          return;
        }
        const text = state.doc.sliceString(node.from, node.to)
        if (!(text.toUpperCase().contains("#+FILETAGS"))) {
          return;
        }
        //pull the filetags from our fileTags text
        const FILETAGS_REGEX = /^#\+FILETAGS:\s+(.+)$/
        const match = text.trim().match(FILETAGS_REGEX)
        if (match == null) {
          return;
        }
        //write our filetags to our metadata file
        match[1].split(/\s+/).forEach((fileTag) => {
          output += `\n - ${fileTag} ` //frontmatter tags
        })
      }
    }
  )
  output += `\n`
  // Iterate on the syntax tree of the orgmode file and extract links. 
  // reading file content is asynchronous, so we need to wait until iteration is done before we append our links to the file
  const outgoing_link_promises = []
  state.tree.iterate(
    {
      enter(node) {
        const node_promise = new Promise((resolve, reject) => {
          if (!(node.type.name === 'PlainLink' || node.type.name === 'AngleLink' || node.type.name === 'RegularLink')) {
            resolve(null)
          }
          const linkText = state.doc.sliceString(node.from, node.to)
          const LINK_REGEX = /\[\[id:(.*?)\]\[(.*?)\]\]/
          const match = linkText.trim().match(LINK_REGEX)
          if (match == null) {
            resolve(null)
          }
          const outgoingLink = {
            id: match[1].trim(),
            text: match[2].trim()
          }
          //find the file in the vault with our link's id.
          const PROPERTY_DRAWER_ID_REGEX = /:PROPERTIES:[\s\S]*?:ID:\s*(\S+)[\s\S]*?:END:/g
          app.vault.getFiles().forEach(async (file) => {
            if (file.extension != "org") {
              return;
            }
            const fileContent = await app.vault.cachedRead(file);
            const idMatches = [...fileContent.matchAll(PROPERTY_DRAWER_ID_REGEX)].map(match => match[1].trim())
            if (idMatches == null) {
              return;
            }
            if (idMatches.contains(outgoingLink.id)) {
              resolve(`[[${METADATA_FILE_FORMAT(file.basename)}|${outgoingLink.text}]]`)
            }
          })
        })
        outgoing_link_promises.push(node_promise)
      }
    }
  )
  output += `---\n`
  //add our outgoing links
  const outgoing_links = await Promise.all(outgoing_link_promises)
  const links = outgoing_links.filter(value => value != null)
  if (outgoing_links.length > 0) {
    output += `**Outgoing Links**:\n - ${links.join("\n - ")}`
  }
  //finally close the leaf
  app.workspace.getLeafById(leaf.id).detach()
  return output;
}

tR += await write_tags_and_links()
%>

Room for Improvement

• Currently, the script does not account for watching for tag and link updates on the source org files. I figured that could utilize Templater's dynamic command syntax, but it's very wonky and might require user scripts, which isn't mobile-friendly. Alternatively, a non-obsidian background script might be another option.
• Using Obsidian's Vault.createBinary() method, we could modify the script to create additional metadata files to track nodes within multi-ID files. This would be a great improvement over tracking just the filenames themselves, and would clean up the graph significantly. However, I'm okay with this solution for now, so I'm not gonna work on it until I need it.

---

Overall, this does the job pretty well! Thanks again!

[BBazard]

Great job @kris-chin 👏
There is probably some logic that my plugin can do to simplify the setup, I will have to think about it.