-
- #!/usr/bin/env bash
- # -*- mode: scheme -*-
- exec guile $0
- # !#
-
- (import (ice-9 rdelim)
- (ice-9 hash-table))
-
- (define (count-or-unicode>? a b)
- (let ((Na (car a))(Nb (car b)))
- (or (> Na Nb)
- (and (= Na Nb)
- (string<? (cdr a) (cdr b))))))
-
- (define (count-words port)
- (define delimiters " \t\n")
- (define delimiters-set (string->char-set delimiters))
- (define buf (make-string 10000))
- (define (read-word port)
- (let ((res (%read-delimited! delimiters buf #t port)))
- (or (and (eof-object? (car res))
- (car res))
- (substring buf 0 (cdr res)))))
- (define count (make-hash-table (inexact->exact 1e5)))
- (define (add-word next)
- (hash-set! count next
- (1+ (hash-ref count next 0))))
- (let loop ((next (read-word port)))
- (unless (eof-object? next)
- (unless (string-every delimiters-set next)
- (add-word next))
- (loop (read-word port))))
- (sort! ;; destructive sort to save memory
- (hash-map->list (lambda (key val) (cons val key)) count)
- count-or-unicode>?))
-
- (for-each
- (lambda (x) (format #t "~a\t~a\n" (cdr x) (car x)))
- (count-words (current-input-port)))
-