Hi there,

I use orgmode links extensively. In particular, I have hundreds (our thousands) of links to emails (in notmuch). I would very much like to add a tag to any email that is being referenced from org.

So, what I need to do first is be able to find all matches of a regular expression in all my org files. Non-interactively.

I have been playing around with xref-matches-in-directory but that would not cover when multiple emails are mentioned in the same line…

I have been looking and looking but everything I find missed the non-interactive part.

So: What is the function to “search for this REGEX in FILES in DIR and return the first capture group”? (the regex has a capture group for the msgid)

The regex:

  (rx-to-string
   (rx "notmuch:id:"
       (group (1+ (not "]")))))
  • telenieko@alien.topOPBEnglish
    1·
    11 months ago

    I ask only for the first step. After I have all the referenced message IDs I will start working on the notmuch side to find and tag!

    • yantar92@alien.topBEnglish
      1·
      11 months ago

      You can simply loop over all the org files, open one by one, and search within each file individually using (while (re-seacrch-forward …) …).

      • telenieko@alien.topOPBEnglish
        1·
        11 months ago

        I plan on searching hundreds of files (all notes) so… some way that does not open all of them would be preferable.

        • yantar92@alien.topBEnglish
          1·
          11 months ago

          If opening a file is a problem, you can create temporary buffer and then use insert-file-contents + search + erase-buffer.

        • telenieko@alien.topOPBEnglish
          1·
          11 months ago

          It’s an opportunity to learn more lisp!

          I do currently have a shell script using ripgrep & notmuch-tag, but felt it could be an interesting problem to solve within Emacs :)

          • mmaug@alien.topBEnglish
            1·
            11 months ago

            The lispy/emacs-like solution is to open each file, search it, and close it. Don’t optimize too early. Build the simplest most straightforward implementation and then worry about performance. But understand why the most emacs-like solution fails.

    • terminal_prognosis@alien.topBEnglish
      1·
      11 months ago

      With filenames:

      for file in `find . -name "*.org"`; do sed -nr "s;.*notmuch:id:([^]\n]*).*$;$file: \1;p" "$file"; done