;;; my-org-unfill-aware.el --- Org-aware hard line break remover -*- lexical-binding: t; -*-

;; Author: Takahiro Ohkubo / ChatGPT
;; Version: 0.1
;; Keywords: org, text, clipboard
;;
;; This file haves Org-aware functions to remove hard line breaks
;; from normal paragraphs while preserving Org structures such as
;; headings, lists, tables, source/example/export blocks, drawers,
;; comments, horizontal rules, and display math lines.

(require 'cl-lib)
(require 'subr-x)

(defgroup my-org-unfill-aware nil
  "Org-aware hard line break removal."
  :group 'org
  :prefix "my-org-unfill-aware-")

(defcustom my-org-unfill-aware-preserve-list-items t
  "If non-nil, preserve list item lines as structural lines.
Continuation lines after a list item are still treated as normal paragraph
text unless they also match a preserved Org structure."
  :type 'boolean
  :group 'my-org-unfill-aware)

(defun my-org-unfill-aware--region-or-buffer ()
  "Return region bounds if active, otherwise whole buffer bounds."
  (if (use-region-p)
      (list (region-beginning) (region-end))
    (list (point-min) (point-max))))

(defun my-org-unfill-aware--japanese-p (text)
  "Return non-nil if TEXT contains Japanese/CJK characters."
  (string-match-p "[ぁ-んァ-ン一-龯々〆〤ー]" text))

(defun my-org-unfill-aware--ascii-normalize-space (s)
  "Normalize whitespace in ASCII paragraph S."
  (string-trim
   (replace-regexp-in-string "[ \t\n]+" " " s)))

(defun my-org-unfill-aware--join-lines (lines joiner ascii-p)
  "Join paragraph LINES using JOINER.
If ASCII-P is non-nil, normalize multiple spaces into one."
  (let ((s (mapconcat #'string-trim lines joiner)))
    (if ascii-p
        (my-org-unfill-aware--ascii-normalize-space s)
      (string-trim s))))

(defun my-org-unfill-aware--blank-line-p (line)
  "Return non-nil if LINE is blank."
  (string-match-p "\\`[ \t]*\\'" line))

(defun my-org-unfill-aware--heading-p (line)
  "Return non-nil if LINE is an Org heading."
  (string-match-p "\\`[ \t]*\\*+\\s-+" line))

(defun my-org-unfill-aware--keyword-p (line)
  "Return non-nil if LINE is an Org keyword line."
  (string-match-p "\\`[ \t]*#\\+[-A-Za-z0-9_]+:" line))

(defun my-org-unfill-aware--comment-p (line)
  "Return non-nil if LINE is an Org comment line."
  (string-match-p "\\`[ \t]*#\\($\\|[^+]\\)" line))

(defun my-org-unfill-aware--table-p (line)
  "Return non-nil if LINE is an Org table line."
  (string-match-p "\\`[ \t]*|" line))

(defun my-org-unfill-aware--hline-p (line)
  "Return non-nil if LINE is an Org horizontal rule."
  (string-match-p "\\`[ \t]*-----+[ \t]*\\'" line))

(defun my-org-unfill-aware--drawer-begin-p (line)
  "Return non-nil if LINE begins an Org drawer."
  (string-match-p "\\`[ \t]*:[A-Za-z0-9_-]+:[ \t]*\\'" line))

(defun my-org-unfill-aware--drawer-end-p (line)
  "Return non-nil if LINE ends an Org drawer."
  (string-match-p "\\`[ \t]*:END:[ \t]*\\'" line))

(defun my-org-unfill-aware--list-item-p (line)
  "Return non-nil if LINE looks like an Org list item."
  (and my-org-unfill-aware-preserve-list-items
       (string-match-p
        "\\`[ \t]*\\(?:[-+]\\|[0-9]+[.)]\\|[A-Za-z][.)]\\)\\s-+"
        line)))

(defun my-org-unfill-aware--display-math-p (line)
  "Return non-nil if LINE looks like display math delimiter/content."
  (string-match-p
   "\\`[ \t]*\\(?:\\\\\\[\\|\\\\\\]\\|\\\\begin{\\|\\\\end{\\|\\$\\$\\)"
   line))

(defun my-org-unfill-aware--block-begin-p (line)
  "Return non-nil if LINE begins an Org block."
  (string-match-p "\\`[ \t]*#\\+BEGIN_" (upcase line)))

(defun my-org-unfill-aware--block-end-p (line)
  "Return non-nil if LINE ends an Org block."
  (string-match-p "\\`[ \t]*#\\+END_" (upcase line)))

(defun my-org-unfill-aware--structural-line-p (line)
  "Return non-nil if LINE should not be joined with surrounding text."
  (or (my-org-unfill-aware--blank-line-p line)
      (my-org-unfill-aware--heading-p line)
      (my-org-unfill-aware--keyword-p line)
      (my-org-unfill-aware--comment-p line)
      (my-org-unfill-aware--table-p line)
      (my-org-unfill-aware--hline-p line)
      (my-org-unfill-aware--drawer-begin-p line)
      (my-org-unfill-aware--drawer-end-p line)
      (my-org-unfill-aware--list-item-p line)
      (my-org-unfill-aware--display-math-p line)))

(defun my-org-unfill-aware--process-text (text joiner ascii-p)
  "Return processed TEXT using JOINER.
If ASCII-P is non-nil, normalize spaces in joined paragraphs."
  (let ((lines (split-string text "\n"))
        (out nil)
        (para nil)
        (in-block nil)
        (in-drawer nil))
    (cl-labels
        ((flush-para
          ()
          (when para
            (push (my-org-unfill-aware--join-lines
                   (nreverse para) joiner ascii-p)
                  out)
            (setq para nil))))
      (dolist (line lines)
        (cond
         ;; Inside source/example/export/special block: preserve literally.
         (in-block
          (flush-para)
          (push line out)
          (when (my-org-unfill-aware--block-end-p line)
            (setq in-block nil)))

         ;; Inside drawer/properties: preserve literally.
         (in-drawer
          (flush-para)
          (push line out)
          (when (my-org-unfill-aware--drawer-end-p line)
            (setq in-drawer nil)))

         ;; Begin block.
         ((my-org-unfill-aware--block-begin-p line)
          (flush-para)
          (push line out)
          (setq in-block t))

         ;; Begin drawer.
         ((my-org-unfill-aware--drawer-begin-p line)
          (flush-para)
          (push line out)
          (unless (my-org-unfill-aware--drawer-end-p line)
            (setq in-drawer t)))

         ;; Other structural line.
         ((my-org-unfill-aware--structural-line-p line)
          (flush-para)
          (push line out))

         ;; Normal paragraph line.
         (t
          (push line para))))
      (flush-para))
    (mapconcat #'identity (nreverse out) "\n")))

(defun my-org-unfill-aware--replace-region (beg end joiner ascii-p)
  "Replace region BEG END with Org-aware unfilled text."
  (let ((new-text
         (my-org-unfill-aware--process-text
          (buffer-substring-no-properties beg end)
          joiner ascii-p)))
    (delete-region beg end)
    (insert new-text)))

;;;###autoload
(defun my-org-unfill-aware-ascii (beg end)
  "Org-aware ASCII mode.
Remove hard line breaks in normal paragraphs by replacing them with one space.
Org structures are preserved."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (my-org-unfill-aware--replace-region beg end " " t))

;;;###autoload
(defun my-org-unfill-aware-japanese (beg end)
  "Org-aware Japanese mode.
Remove hard line breaks in normal paragraphs without inserting spaces.
Org structures are preserved."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (my-org-unfill-aware--replace-region beg end "" nil))

;;;###autoload
(defun my-org-unfill-aware-auto (beg end)
  "Org-aware auto mode.
If the selected text contains Japanese/CJK characters, use Japanese mode.
Otherwise use ASCII mode."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (let ((text (buffer-substring-no-properties beg end)))
    (if (my-org-unfill-aware--japanese-p text)
        (my-org-unfill-aware-japanese beg end)
      (my-org-unfill-aware-ascii beg end))))

(defun my-org-unfill-aware--copy-only (beg end fn)
  "Copy processed text from BEG END using FN without modifying current buffer."
  (let* ((text (buffer-substring-no-properties beg end))
         (buf (generate-new-buffer " *my-org-unfill-aware-copy*")))
    (unwind-protect
        (with-current-buffer buf
          (insert text)
          (funcall fn (point-min) (point-max))
          (kill-new (buffer-string)))
      (kill-buffer buf))
    (message "Org-aware unfilled text copied to kill-ring.")))

;;;###autoload
(defun my-org-unfill-aware-ascii-copy (beg end)
  "Org-aware ASCII mode, copy only.
Do not modify current buffer."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (my-org-unfill-aware--copy-only beg end #'my-org-unfill-aware-ascii))

;;;###autoload
(defun my-org-unfill-aware-japanese-copy (beg end)
  "Org-aware Japanese mode, copy only.
Do not modify current buffer."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (my-org-unfill-aware--copy-only beg end #'my-org-unfill-aware-japanese))

;;;###autoload
(defun my-org-unfill-aware-auto-copy (beg end)
  "Org-aware auto mode, copy only.
Do not modify current buffer."
  (interactive (my-org-unfill-aware--region-or-buffer))
  (my-org-unfill-aware--copy-only beg end #'my-org-unfill-aware-auto))

(provide 'my-org-unfill-aware)
;;; my-org-unfill-aware.el ends here