xenodium / company-org-block

👉 Support my work via GitHub Sponsors

Hi, I’m Álvaro 👋

• I do software for a living.
• I live in an Emacs buffer somewhere in London.
  • My Emacs config is available in the dotsies repo.
• I write about all sorts of things at xenodium.com.
• I've written a few Emacs packages you may like:
  • chatgpt-shell
  • ready-player
  • sqlite-mode-extras
  • dwim-shell-command
  • org-block-capf
  • company-org-block
  • ob-swiftui
• Command line utils like macosrec.
• Take care of your eyes on macOS with Fresh Eyes.
• Inspired by Atomic Habits, I wrote Flat Habits for iOS. Now on the App Store.
• I'm an Org mode fan, so I wrote Plain Org for iOS. Also on the App Store.
• I needed an Emacs-inspired *scratch* buffer on iOS (who doesn't?), so I built *scratch*.
• You can find more about me over at xenodium.com or just ping on mastodon or twitter.

company-org-block's Issues

Better handling of src blocks

Sometimes the language I want to create a code block isn't setup by default in org. For example SQL code blocks. In this case I enter <src to create a src code block but company-org-block takes me to the body of the block making me have to navigate back to the definition and then attach the language I want.

It would be nice if:

1. company-org-block started a completing-read session in this case to ask the user the type of code block their creating.
2. Remembered this decision so later attempts to create the same code block aren't as indirect.

Thoughts?

Additional whitespace with tng

Inserting #+begin_center yields additional whitespace when using (company-tng-configure-default).

Gracefully handle lack of special environment to edit

Some code blocks (such as quote) give an error when you try to edit them, however with company-org-block-edit-style set to auto, company-org-block will just let the error propagate.

This diff should gracefully handle it.

--- /home/mohkale/.config/dotfiles/prog/editors/emacs/lisp/straight/repos/company-org-block/company-org-block.el
+++ #<buffer company-org-block.el>
@@ -110,11 +110,16 @@
   (insert (make-string org-edit-src-content-indentation ?\s))
   (save-excursion
     (insert (format "\n#+end_%s" end)))
-  (cond ((eq company-org-block-edit-style 'auto)
-         (org-edit-special))
-        ((and (eq company-org-block-edit-style 'prompt)
-              (yes-or-no-p "Edit now?"))
-         (org-edit-special))))
+  (condition-case err
+      (cond ((eq company-org-block-edit-style 'auto)
+             (org-edit-special))
+            ((and (eq company-org-block-edit-style 'prompt)
+                  (yes-or-no-p "Edit now?"))
+             (org-edit-special)))
+    (user-error
+     (unless (string-equal "No special environment to edit here"
+                           (error-message-string err))
+       (signal (car err) (cdr err))))))
 
 (defun company-org-block--grab-symbol-cons ()
   "Return cons with symbol and t whenever prefix of < is found.

Can't automatically trigger completion

Great package, but I'm having trouble automatically triggering completion. I've set it up using the use-package instructions you provided, but nothing pops up when I insert <. But completion does pop up when I run company-org-block.

Indentation breaks if there's text below

[Question] Completion suggests language's file extension

Is there a reason why the completion is suggesting both the languages and their file extensions?

The extensions are not valid headers for org blocks, so this is a bit annoying.

Errors after "<" invocation.

Running into below errors from opening "<" for org block completion.

Debugger entered--Lisp error: (wrong-type-argument stringp ("#+BEGIN_SRC lisp\n\n#+END_SRC")) string-lessp(("#+BEGIN_SRC lisp\n\n#+END_SRC") "lisp") sort(("lisp") string-lessp) #f(compiled-function (pred list) #<bytecode 0x1022ee5>)(string-lessp ("lisp" ("#+BEGIN_SRC lisp\n\n#+END_SRC") ("#+BEGIN_SRC ?\n\n#+END_SRC") ("#+BEGIN_EXAMPLE\n?\n#+END_EXAMPLE") ("#+BEGIN_QUOTE\n?\n#+END_QUOTE") ("#+BEGIN_VERSE\n?\n#+END_VERSE") ("#+BEGIN_VERBATIM\n?\n#+END_VERBATIM") ("#+BEGIN_CENTER\n?\n#+END_CENTER") ("#+BEGIN_COMMENT\n?\n#+END_COMMENT") ("#+BEGIN_EXPORT latex\n?\n#+END_EXPORT") ("#+LaTeX: ") ("#+BEGIN_EXPORT html\n?\n#+END_EXPORT") ("#+HTML: ") ("#+BEGIN_EXPORT ascii\n?\n#+END_EXPORT") ("#+ASCII: ") ("#+INDEX: ?") ("#+INCLUDE: %file ?") "lisp" "el" "el")) apply(#f(compiled-function (pred list) #<bytecode 0x1022ee5>) string-lessp ("lisp" ("#+BEGIN_SRC lisp\n\n#+END_SRC") ("#+BEGIN_SRC ?\n\n#+END_SRC") ("#+BEGIN_EXAMPLE\n?\n#+END_EXAMPLE") ("#+BEGIN_QUOTE\n?\n#+END_QUOTE") ("#+BEGIN_VERSE\n?\n#+END_VERSE") ("#+BEGIN_VERBATIM\n?\n#+END_VERBATIM") ("#+BEGIN_CENTER\n?\n#+END_CENTER") ("#+BEGIN_COMMENT\n?\n#+END_COMMENT") ("#+BEGIN_EXPORT latex\n?\n#+END_EXPORT") ("#+LaTeX: ") ("#+BEGIN_EXPORT html\n?\n#+END_EXPORT") ("#+HTML: ") ("#+BEGIN_EXPORT ascii\n?\n#+END_EXPORT") ("#+ASCII: ") ("#+INDEX: ?") ("#+INCLUDE: %file ?") "lisp" "el" "el") nil) seq-sort(string-lessp ("lisp" ("#+BEGIN_SRC lisp\n\n#+END_SRC") ("#+BEGIN_SRC ?\n\n#+END_SRC") ("#+BEGIN_EXAMPLE\n?\n#+END_EXAMPLE") ("#+BEGIN_QUOTE\n?\n#+END_QUOTE") ("#+BEGIN_VERSE\n?\n#+END_VERSE") ("#+BEGIN_VERBATIM\n?\n#+END_VERBATIM") ("#+BEGIN_CENTER\n?\n#+END_CENTER") ("#+BEGIN_COMMENT\n?\n#+END_COMMENT") ("#+BEGIN_EXPORT latex\n?\n#+END_EXPORT") ("#+LaTeX: ") ("#+BEGIN_EXPORT html\n?\n#+END_EXPORT") ("#+HTML: ") ("#+BEGIN_EXPORT ascii\n?\n#+END_EXPORT") ("#+ASCII: ") ("#+INDEX: ?") ("#+INCLUDE: %file ?") "lisp" "el" "el")) company-org-block--candidates("") company-org-block(candidates "") apply(company-org-block (candidates "")) company-call-backend-raw(candidates "") company--fetch-candidates("") company-calculate-candidates("") company--begin-new() company--perform() company-auto-begin() company-idle-begin(#<buffer configuration.org> #<window 3 on configuration.org> 1578 11869) apply(company-idle-begin (#<buffer configuration.org> #<window 3 on configuration.org> 1578 11869)) timer-event-handler([t 24766 31847 402309 nil company-idle-begin (#<buffer configuration.org> #<window 3 on configuration.org> 1578 11869) nil 455000])

Here is my company-mode configuration.

(use-package company
   :ensure t
   :init (add-hook 'after-init-hook 'global-company-mode))

 (setq company-idle-delay 0)
 (setq company-minimum-prefix-length 3)
 (setq company-selection-wrap-around t)
 (company-tng-configure-default)

 (use-package company-org-block
   :ensure t
   :custom
   (company-org-block-edit-style 'prompt)
   :hook ((org-mode . (lambda ()
		    (setq-local company-backends '(company-org-block))
		    (company-mode +1)))))

Support leading white-space with org-adapt-indentation

Hi, love the package.

I agree you'd probably only ever want org-block completions when at the beginning of a line but for those of us that have org-adapt-indentation set to t, this package won't offer to complete any lines that're in a sub-heading because the lines are indented to match the heading level. I'd suggest allowing company-org-block-complete-at-bol to disregard leading whitespace.

This diff fixes the issue for me:

--- /home/mohkale/.config/dotfiles/prog/editors/emacs/lisp/straight/repos/company-org-block/company-org-block.el
+++ #<buffer company-org-block.el>
@@ -106,10 +106,12 @@
 #+begin_BEGIN
   |
 #+end_END"
+  (org-indent-line)
   (insert (format "#+begin_%s\n" begin))
   (insert (make-string org-edit-src-content-indentation ?\s))
   (save-excursion
-    (insert (format "\n#+end_%s" end)))
+    (insert (format "\n#+end_%s" end))
+    (org-indent-line))
   (cond ((eq company-org-block-edit-style 'auto)
          (org-edit-special))
         ((and (eq company-org-block-edit-style 'prompt)
@@ -120,7 +122,7 @@
   "Return cons with symbol and t whenever prefix of < is found.
 For example: \"<e\" -> (\"e\" . t)"
   (when (looking-back (if company-org-block-complete-at-bol
-                          (concat "^" company-org-block--regexp)
+                          (concat "^[[:space:]]*" company-org-block--regexp)
                         company-org-block--regexp)
                       (line-beginning-position))
     (cons (match-string-no-properties 1) t)))

Also I'd like to ask whether there's a reason behind the lack of a trailing newline on the end_src? In my emacs not having one leads to the font-lock at that point getting cut off.

Trigger key `<` and `electric-pair-mode`

This looks like a very useful package, I just have one little nit: in an Org buffer, pressing < inserts an additional >, because I have electric-pair-mode turned on. Completion and editing of the source block still work perfectly, but the < character is left in the buffer, which means I need to delete it manually.

Is there something that could be done about this? I know nothing about the internals of company-mode or of electric-pair-mode, so I don't know if there is a way to inhibit the insertion of the closing >. Perhaps you could check for > following < and delete it if electric-pair-mode is turned on? Or perhaps less brittle, make the trigger key customisable?