;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; This file contains utility macros and functions for helping read
;;  strings 


;;; A simple macro for reading in lines from a file.  It
;;; is designed to look like (dotimes) and (dolist)
;;; Example:
;;;   (dolines (line "testfile")
;;;     (format t "~a~%" line))
(defmacro dolines ((var filename &optional ret) &body body)
  (let ((stream (gensym)))
    `(with-open-file (,stream ,filename 
			      :direction :input)
	 (while (setf ,var (read-line ,stream nil nil))
	   ,@body)
	 ,ret)))

;;; A simple function for getting all the keys from a hashtable
(defun get-hash-keys (hashtable)
  (let ((ret nil))
    (maphash #'(lambda (key val) (push key ret)) hashtable)
    ret))

;;; A simple macro for groveling over the contents of a hash table. It
;;; is designed to look like (dotimes) and (dolist)
;;; Example:
;;;   (dohash (key val hashtable)
;;;     (format t "~a -> ~a~%" key val))
(defmacro dohash ((key val ht &optional ret) &body body)
  (let ((keys (gensym)))
    `(let ((,keys (get-hash-keys ,ht)))
       (dolist (,key ,keys ,ret)
	 (let ((,val (gethash ,key ,ht)))
	   ,@body)))))


;;; A simple function that takes a character and returns a value from
;;; 0-27, where 1-26 maps to A-Z (case insensitive) and everything else
;;;  maps to 0
(defun character-to-index (char)
  (if (alpha-char-p char)
      (+ (- (char-int (char-downcase char)) (char-int #\a)) 1)
    0))

;;; A simple function that takes a string and returns an array 
;;; of the character-to-index values
(defun string-to-index (string)
  (let* ((stop (length string))
	 (ret (make-array stop)))
    (dotimes (i stop ret)
      (setf (aref ret i) (character-to-index (aref string i))))))

;;;; Some sample runs I did:
;;;; Notice that the sentences begin with a " ".
;;;; It actually makes a difference.  Check out the difference between 
;;;; "hello" and " hello".  Also, notice that these 'sentences' are just
;;;; words.  You'd really want to test with sentences.
#|

(predict " hola")
    (("spanish" 0.6765656) ("english" 0.1993123) ("hiphop" 0.118951924) ("french" 0.0051645134) ("lisp" 5.6704903e-6))

(predict " hello")
    (("hiphop" 0.5174973) ("english" 0.45761773) ("spanish" 0.02424061) ("french" 6.437539e-4) ("lisp" 5.318387e-7))
(predict "hello")
    (("english" 0.50267637) ("hiphop" 0.44940448) ("spanish" 0.039834227) ("french" 0.007994719) ("lisp" 9.016024e-5))

(predict " death")
    (("english" 0.6923835) ("hiphop" 0.30310503) ("french" 0.003657153) ("lisp" 5.40979e-4) ("spanish" 3.1325946e-4))

(predict " rhyme")
    (("hiphop" 0.9995462) ("english" 3.4074156e-4) ("lisp" 9.9734716e-5) ("french" 1.3035273e-5) ("spanish" 4.2282585e-7))

(predict " defun")
    (("lisp" 0.99575996) ("french" 0.003839893) ("english" 2.776178e-4) ("spanish" 8.14848e-5) ("hiphop" 4.0968436e-5))

(predict " charles")
    (("french" 0.5482738) ("spanish" 0.22672346) ("english" 0.22445887) ("hiphop" 5.437327e-4) ("lisp" 1.1742197e-7))

(predict " ashok")
    (("english" 0.7762349) ("hiphop" 0.223395) ("spanish" 2.2333201e-4) ("french" 1.4665788e-4) ("lisp" 4.945446e-8))

(predict "charles")
  (("english" 0.43924308) ("french" 0.40395173) ("spanish" 0.15613052) ("hiphop" 6.744762e-4) ("lisp" 1.7011958e-7))
(predict "ashok")
   (("english" 0.75808936) ("hiphop" 0.24133347) ("spanish" 4.655121e-4) ("french" 1.1154039e-4) ("lisp" 5.4603685e-8))

(predict " french")
    (("english" 0.62157315) ("french" 0.19474714) ("spanish" 0.11906079) ("hiphop" 0.064609304) ("lisp" 9.547511e-6))

(predict " fries")
    (("french" 0.5804049) ("spanish" 0.23300804) ("english" 0.16763599) ("hiphop" 0.01885327) ("lisp" 9.777243e-5))

|#