;;; md4tj_parse --- Summary

;;; Commentary:

;;; Code:

(require 'cl-lib)

(defun getline ()
  "Get current line from loaded buffer."
  (buffer-substring-no-properties (line-beginning-position) (line-end-position)))

(defun md4tj-begin-tag (tag &optional attrs)
  "Return beginning html tag for TAG with optional ATTRS."
  (concat "<" tag
		  (mapconcat 'identity (cl-map 'listp (lambda (l) (concat " " (nth 0 l) "=" "\"" (nth 1 l) "\"")) attrs) "")
		  ">"))

(defun md4tj-end-tag (tag)
  "Return end html tag for TAG."
  (concat "</" tag ">"))

(defun md4tj-process-header (line)
  "Process LINE known to be header, return HTML."
  (let ((level (length (nth 0 (split-string line)))))
	(if (or (< level 0) (> level 6))
		(error (concat "Error parsing: " line "\n"))
	  (concat (md4tj-begin-tag (concat "h" (number-to-string level)))
			  (mapconcat 'identity (cdr (split-string line)) " ")
			  (md4tj-end-tag (concat "h" (number-to-string level)))))))

(defun md4tj-process-paragraph (line)
  "Process LINE that is paragraph, return HTML."
  (concat (md4tj-begin-tag "p") line (md4tj-end-tag "p")))

(defun md4tj-process-line (line)
  "Process all inline elements of the LINE, return HTML."
  ;; Finally links
  (replace-regexp-in-string
   "\\[\\(.*\\)\\](\\(.*\\))"
   (concat (md4tj-begin-tag "a" (list '("href" "\\2"))) "\\1" (md4tj-end-tag "a"))
   ;; Then images
   (replace-regexp-in-string
	"!\\[\\(.*\\)\\](\\(.*\\))"
	(md4tj-begin-tag "img" (list '("src" "\\2") '("alt" "\\1")))
	;; Then emphasis
	(replace-regexp-in-string
	 "\\*\\(.*\\)\\*"
	 "<em>\\1</em>"
	 ;; Then strong
	 (replace-regexp-in-string
	  "\\*\\*\\(.*\\)\\*\\*"
	  "<strong>\\1</strong>"
	  ;; First code
	  (replace-regexp-in-string
	   "`\\(.*\\)`"
	   "<code>\\1</code>" line))))))

(defun md4tj-clean-multiline (line)
  "Clean LINE of markdown syntax for ul."
  (replace-regexp-in-string
   "^```" ""
   (replace-regexp-in-string
	"^[0-9]+\\. " ""
	(replace-regexp-in-string "^- " "" line))))

(defun md4tj-clean-code-for-html (line)
  "Escape characters in LINE that would be misinterpreted by the browser."
  (string-replace
   "<" "&lt;"
   (string-replace ">" "&gt;" line)))
				  
(defun md4tj-convert-line-to-html (line state)
  "Process LINE with STATE and return html."
  (let ((cleanline (md4tj-clean-multiline line)))
	(concat
	 ;; Beginning of multiline block
	 (cond ((eq state 'beginul) "<ul>\n<li>")
		   ((eq state 'beginol) "<ol>\n<li>")
		   ((eq state 'begincode) "<pre>\n<code>")
		   ((eq state 'ul) "<li>")
		   ((eq state 'ol) "<li>")
		   ((eq state 'code) "")
		   ((eq state 'endul) "</ul>\n")
		   ((eq state 'endol) "</ol>\n")
		   ((eq state 'endcode) "</code>\n</pre>\n")
		   (t ""))
	 
	 ;; Body
	 (cond ((or (eq state 'code) (eq state 'begincode)) (md4tj-clean-code-for-html cleanline))
		   ((string-match "^#+ " cleanline) (md4tj-process-header (md4tj-process-line cleanline)))
		   ((string= "---" cleanline) "<hr>") ;; horizontal line
		   ((= (length cleanline) 0) "<br/>") ;; blank line
		   (t (md4tj-process-paragraph (md4tj-process-line cleanline))))
	 
	 ;; End of multiline block
	 (cond ((or (eq state 'ul) (eq state 'beginul)) "</li>")
		   ((or (eq state 'ol) (eq state 'beginol)) "</li>")
		   ((eq state 'code) "")
		   (t "")))))

(defun md4tj-next-state (currline prevstate)
  "Return the state based on CURRLINE and PREVSTATE."
  (cond ((and (string-match "^- " currline) (not (or (eq prevstate 'beginul) (eq prevstate 'ul))) 'beginul))
		((and (string-match "^- " currline) (or (eq prevstate 'beginul) (eq prevstate 'ul)) 'ul))
		((and (not (string-match "^- " currline)) (or (eq prevstate 'ul) (eq prevstate 'beginul))) 'endul)
		((and (string-match "^[0-9]+\\. " currline) (not (or (eq prevstate 'beginol) (eq prevstate 'ol))) 'beginol))
		((and (string-match "^[0-9]+\\. " currline) (or (eq prevstate 'beginol) (eq prevstate 'ol)) 'ol))
		((and (not (string-match "^[0-9]+\\. " currline)) (or (eq prevstate 'ol) (eq prevstate 'beginol))) 'endol)
		((and (string-match "^```" currline) (not (or (eq prevstate 'begincode) (eq prevstate 'code))) 'begincode))
		((and (not (string-match "```$" currline)) (or (eq prevstate 'begincode) (eq prevstate 'code)) 'code))
		((and (string-match "```$" currline) (or (eq prevstate 'code) (eq prevstate 'begincode))) 'endcode)
		(t 'normal)))

(defun md4tj-finalize (state)
  "Finalizes HTML document by inserting missing end tags based on STATE."
  (concat
   (cond ((or (eq state 'beginul) (eq state 'ul)) "</ul>")
		 ((or (eq state 'beginol) (eq state 'ol)) "</ol>")
		 ((or (eq state 'begincode) (eq state 'code)) "</code>")
		 (t ""))
   "</body>\n"
   "</html>"))

(defun md4tj-parse (mdfile outfile)
  "Entry point to parse MDFILE and output to OUTFILE."
  (let ((inbuf (generate-new-buffer " in"))
		(outbuf (generate-new-buffer " out"))
		(state 'normal)
		(line nil))
	(set-buffer inbuf)
	(insert-file-contents mdfile)
	(goto-char (point-min))
	(set-buffer outbuf)
	(insert "<body>\n")
	(while (with-current-buffer inbuf (< (point) (point-max)))
	  (setq line (with-current-buffer inbuf (getline)))
	  (setq state (md4tj-next-state line state))
	  ;; Insert next line(s) into output file
	  (insert (concat (md4tj-convert-line-to-html line state) "\n"))
	  ;; Advance input file by a line
	  (with-current-buffer inbuf (forward-line)))
	(insert (md4tj-finalize state))
	;; Write outbuf to outfile
	(write-region nil nil outfile nil)))

(provide 'md4tj_parse)
;;; md4tj_parse.el ends here