Added support for ols, uls, and multiline codes

md4tj_parse.el

@@ -3,6 +3,7 @@
 ;;; Commentary:
 
 ;;; Code:
+
 (require 'cl-lib)
 
 (defun getline ()
@@ -55,39 +56,97 @@
 	   "`\\(.*\\)`"
 	   "<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)))
+				  
+
 ;; Note: a "block" is the smallest unit of parsing
 ;; It is normally a line of the code, but can be
 ;; multiple lines in the case of a block (NI)
-(defun md4tj-process-block (codeblock)
+(defun md4tj-process-block (codeblock state)
-  "Process CODEBLOCK and return html."
+  "Process CODEBLOCK with STATE and return html."
-  (cond ((string-match "^#+ " codeblock) (md4tj-process-header (md4tj-process-line codeblock)))
+  (let ((cleanline (md4tj-clean-multiline codeblock)))
-		((string= "---" codeblock) "<hr>")
+	(concat
-		((= (length codeblock) 0) "<br/>")
+	 ;; Beginning of multiline block
-		(t (md4tj-process-paragraph (md4tj-process-line codeblock)))))
+	 (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-block ()
+(defun md4tj-next-state (currline prevstate)
-  "Retrieve the next block in the open file."
+  "Return the state based on CURRLINE and PREVSTATE."
-  (let ((currline (getline)))
+  (cond ((and (string-match "^- " currline) (not (or (eq prevstate 'beginul) (eq prevstate 'ul))) 'beginul))
-	(cond ((string-match "^- " currline) nil) ;; TODO ul's
+		((and (string-match "^- " currline) (or (eq prevstate 'beginul) (eq prevstate 'ul)) 'ul))
-		  ((string-match "^[0-9]+\\. " currline) nil) ;; TODO ol's
+		((and (not (string-match "^- " currline)) (or (eq prevstate 'ul) (eq prevstate 'beginul))) 'endul)
-		  (t currline))))
+		((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")))
+		(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))
+	  (print line)
 	  ;; Insert next line(s) into output file
-	  (insert (concat (md4tj-process-block (with-current-buffer
+	  (insert (concat (md4tj-process-block line state) "\n"))
-									   inbuf
-									   (md4tj-next-block)))
-					  "\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)

test_file.html

@@ -1,7 +1,32 @@
+<body>
 <h1>Hello <code>world</code>!</h1>
 <h2>Hello world!</h2>
 <h3>Hello world!</h3>
 <h4>Hello <strong><em>world</em></strong>!</h4>
 <br/>
 <p><a href=https://example.com>Example website</a></p>
-<p><a href=image.jpg>Example image</a></p>
+<p><img src=usbs.png alt=Example image></p>
+<br/>
+<ul>
+<li><p>unordered item 1</p></li>
+<li><p>unordered item 2</p></li>
+<li><p>unordered item 3</p></li>
+</ul>
+<br/>
+<ol>
+<li><p>ordered item 1</p></li>
+<li><p>ordered item 2</p></li>
+</ol>
+<br/>
+<pre>
+<code>
+#include &lt;stdio.h&gt;
+int main() {
+	printf("%s\n", "Hello World!");
+	return 0;
+}
+</code>
+</pre>
+<br/>
+</body>
+</html>

test_file.md4tj

@@ -4,4 +4,19 @@
 #### Hello ***world***!
 
 [Example website](https://example.com)
-[Example image](image.jpg)
+![Example image](usbs.png)
+
+- unordered item 1
+- unordered item 2
+- unordered item 3
+
+1. ordered item 1
+2. ordered item 2
+
+```
+#include <stdio.h>
+int main() {
+	printf("%s\n", "Hello World!");
+	return 0;
+}
+```