;;;-*- Mode: Lisp; Package: CONSTRAINT-GRAMMAR-PARSER -*-

(in-package :cgp)

(setf *cg* *nbo-cg* *tagger* *nbo-tagger*)

(print (get-features "fisk"))

(print (tag-compound "fisk"))

(print (analyse-compound "fiskeskjær"))

(defun classify-words (file out-file &key pos-only-p)
  (with-open-file (stream out-file :direction :output :if-exists :supersede)
    (with-file-lines (word file)
      (let* ((ordbank (get-features word))
	     (scarrie (unless ordbank (get-scarrie-features word))))
	(multiple-value-bind (segmentation features)
	    (unless (or ordbank scarrie) (most-probable-compound word))
	  (format stream "~a~c~a~c" (or segmentation word) #\Tab
		  (cond (ordbank "ordbank")
			(scarrie "scarrie")
			(segmentation "samset")
			(t "ukjent"))
		  #\Tab)
	  (let ((features (remove-feature-inclusions (or features scarrie ordbank)))
		(seen-lfs ()))
	    (if pos-only-p
		(loop for fl in features
		    when (car fl)
		    do (let* ((cf (car (code-features (cdr fl))))
			      (lf (cons (car fl) cf)))
			 (unless (find lf seen-lfs :test #'equal)
			   (push lf seen-lfs)
			   (write-string "[\"" stream)
			   (write-string (car fl) stream)
			   (write-string "\" " stream)
			   (write-string (stringify cf) stream)
			   (write-string "]" stream))))
	      (loop for fl in features
		  when (car fl)
		  do
		    (write-string "[\"" stream)
		    (write-string (car fl) stream)
		    (write-char #\" stream)
		    (mapc (lambda (f)
			    (write-string " " stream)
			    (write-string (stringify f) stream))
			  (code-features (cdr fl)))
		    (write-string "]" stream))))
	  (terpri stream))))))

;; adlyde+t ??

(classify-words "projects:cgp;texts;nyord.txt" "projects:cgp;texts;nyord.ann")
(progn
  (classify-words "projects:cgp;texts;hyphen-unix.txt" "projects:cgp;texts;hyphen-unix.ann")
  (classify-words "projects:cgp;texts;remaining-unix.txt" "projects:cgp;texts;remaining-unix.ann"))

(print (analyse-compound "abonnementsbibliotek"))
(print (analyse-compound "ablegøyemakeri"))

(print (analyse-compound "ablegøye"))

(print (analyse-compound "makeri"))

("bibliotek" "811:1" "810:3" "810:1" "801:1" "800:3" "800:1")

(print (most-probable-compound-last-chunk "abonnementsbibliotek" t))
(print (most-probable-compound "abonnementsbibliotek" t))

(print (grammar-code-to-features "bibliotek" "811:1" :features-as-list-p t))

(print (get-features "fisk"))

(print (tag-compound "überhjelperne"))

:eof