get_tag_and_class.scm

#!/opt/local/bin/gosh
;; -*- coding: utf-8 -*-

;; get_tag_and_class.scm: extract tag and class names from a HTML document.

(define (match-tag-and-class line)
  (rxmatch-case line
		[#/<(\w+)[^<>]*class=[\'\"]([^\'\"]+)[\'\"][^<>]*>(.*)/
		    (all tag class rest)
		    (if (> (string-length rest) 0)
			(append (list (cons tag class))
				(match-tag-and-class rest))
			(list (cons tag class)))
		    ]
		[else ()]))

(define (print-tag-and-class-with-number-inner lst num)
  (if (not (null? lst))
      (begin
	(let ((top (car lst)))
	  (if top
	      (print num ": " (car top) "." (cdr top))))
	(print-tag-and-class-with-number-inner (cdr lst) num))))

(define (print-tag-and-class-with-number lst num)
  (if (not (null? lst))
      (begin
	(let ((result (car lst)))
	  (if (not (null? result))
	      (print-tag-and-class-with-number-inner result num)))
	(print-tag-and-class-with-number (cdr lst) (+ num 1)))))

(define (print-tag-and-class lst)
  (print-tag-and-class-with-number lst 1))

(print-tag-and-class
 (map match-tag-and-class (port->string-list (standard-input-port))))