;;;-*- Mode: Lisp; Package: (REGEXP) -*-


#+mcl(ccl:require :regexp-fsa)


#|

(cfg-chart-parse "(d)" :chart *regexp-chart*)

(cfg-chart-parse "(woll){3}" :chart *regexp-chart*)

(cfg-chart-parse "(((woll){3,5}|<rigus>)?[1a-rt]+o\\+ho)*" :chart *regexp-chart*
                 :goal 'expression)
(cfg-chart-parse "((wollwutz)|<rigus>oho)*" :chart *regexp-chart*)

(cfg-chart-parse "(((woll)|rigus)?lull+oho)*" :chart *regexp-chart*)

(tokenize-string "(((woll){3,5}|<@rigus>)?[^1a-rt]+o\\+ho)*" *regexp-chart*)

(tokenize-string "(<@num>)?" *regexp-chart*)
(tokenize-string "(num)?" *regexp-chart*)

(defparameter *fsa*
  (compile-regexp "(((woll)|rigus)?lull+oho)*" :chart *regexp-chart*))
(defparameter *fsa*
  (compile-regexp "(((woll){3,5}|<rigus>)?[1a-rt]+o\\+ho)*" :chart *regexp-chart*))

|#

#|
(dfa-compile-parsed-regexp
 '(:seq
   (:or :<stem>
        (:seq :<root>
              (:? "n")
              "ing"))
   (:* (:seq (:? (:or "s" "e"))
             (:or "<stem>"
                  (:seq "<root>"
                        (:? "n")
                        "ing")))))
 (make-instance 'regexp-fsa))
|#

#|
(time (defparameter *fsa*
  (compile-regexp " [01]?[0-9]\\.([01]?[0-9]\\.|(jan|feb|mar|apr|mai|jun|jul|aug|sep|okt|nov|des)\\.?)(19)?[0-9][0-9][ \\.\\,\\;\\:]")))
|#

#|

(defparameter *fsa* (any-determinize (compile-regexp "(.*|wu)tz")))

(fsa::fsa-print *fsa*)

(in-language-p "atz" *fsa*)

(in-language-p "wollalulflloho" *fsa*)


(defparameter *fsa*
  (compile-regexp "(jaja|trull)+"))

; 600, 378, 367, 355, 450, class state: 493, more than 1 final-state: 645
; dfa: 273; 340; lw: 120
(time 
 (dotimes ( i 100)
   (substring-match *fsa* "wollwutzriguswollwutzwollwutzwollwutzwollwutzrigusrigus jaja "
                    :minimal nil)))
|#

(tokenize-string "(a-[bjjjjjjc-d])" *regexp-chart*)
(compile-regexp "(a-[bc-d])")

(fsa::fsa-print (compile-regexp "(a-[csadsf])"))

(fsa::fsa-print (compile-regexp "\\<tr\\>*\\<"))

\>(-[\<])*\((-[\<])*/(-[\<])+/(-[\<])*\)\.*\<

\((-[\)])*/(-[\)])+/.*\)

\<tr\>-[<]*(Q|W|R|Y|U|I|O|A|D|F|J|L|V|B|N|M|f|ä|ö|ü|ß|Ä|Ö|Ü|h)-[<]*\<

\((-[/\(\)])*/(-[/\(\)])+/(-[/\(\)])*\)
\<tr\>-[<]*\(G -[<]*\<

(fsa::fsa-print *fsa*)

(setf *fsa* (compile-regexp "\\((-[\\)])*/(-[\\)])*/.*\\)"))
(setf *fsa* (compile-regexp "\\((-[/()])*/(-[/()])+/(-[/()])*\\)"))

(substring-match *fsa* "eba (Hmarobs; /mo/iHmars)," :minimal nil)

#|
; 54 ms with big grammar and intermediate tree build
; 31 ms with smaller grammar and no tree build
; 17 ms without extended-lists
; 11 ms in reversed order
; 20 ms with one-char-strings
(cfg-chart-parse "(((woll){3,5}|<r>)?[1a-rt]+o\\+ho)*" :chart *regexp-chart*
                 :goal 'expression)

(tokenize-string "(((woll){3,5}|<rigus>)?[1a-rt]+o\\+ho)*" *regexp-chart*)

; 27 ms
(print-fsa 
 (compile-regexp "(((woll){3,5}|<r>)?[1a-rt]+o\\+ho)*"))

(fsa::fsa-print 
 (compile-regexp "(((woll)|rigus)?.1a\\-rt+o\\+ho)*"))

(print-fsa 
 (compile-regexp "((woll|<rigus>)?rt+o\\+ho)*"))

(fsa::fsa-print (compile-regexp "woll*" :chart *regexp-chart*))

(defparameter *fsa* (compile-regexp "(woll.?)+"))

(fsa::fsa-print *fsa*)


(defparameter *fsa*
  (compile-regexp "[a-zA-Z]+.3\\*.?"))


(time (dotimes (i 100) (string-match *fsa* "aaaaaahadddddd43*r")))
(get-matching-substring *fsa* "aaaaaahadddddd3*")

(string-match *fsa* "aaaaaahadddddd43*r")
|#


;(fsa::fsa-print *fsa*)
;(fsa::fsa-print (any-determinize *fsa*))

#|
(defparameter *fsa* (compile-regexp "\\((~|¢).*; ?(~|¢)[a-z]*\\)"))

(string-match *fsa* "(~a;~x)" :exact nil :minimal nil)

(substring-match *fsa* "(~a;~b)" :minimal nil)

(inspect *regexp-chart*)

(defparameter *fsa*
  (compile-regexp "(jaja|trull)+"))

(fsa::fsa-print *fsa*)

; 600, 378, 367, 355, 392, 480; dfa: 311; lw: 140
(time 
 (dotimes ( i 1000)
   (substring-match *fsa* "wollwutzriguswollwutzwollwutzwollwutzwollwutzrigusrigus jaja "
                    :minimal nil)))

(defparameter *fsa*
  (compile-regexp "(woll.?|wutz[i4-9o])+"))

(substring-match *fsa* "ja diese wollowollwollwutz5rigus wolllllllllllllelein oho")

(time (defparameter *fsa*
 (compile-regexp " [01]?[0-9]\\.([01]?[0-9]\\.|(jan|feb|mar|apr|mai|jun|jul|aug|sep|okt|nov|des)\\.?)(19)?[0-9][0-9][ \\.\\,\\;\\:]")))


(tokenize-string "[a-zA-Z]+.3\\*" *regexp-chart*)
(tokenize-string " [01]?[0-9]\\.([01]?[0-9]\\.|(jan|feb|mar|apr|mai|jun|jul|aug|sep|okt|nov|des)\\.?)(19)?[0-9][0-9][ \\.\\,\\;\\:]" *regexp-chart*)

(get-matching-substring *fsa* "aaaaaahadddddd3*45*")
(substring-match *fsa* "aaaaaahadddddd3*")

(fsa::fsa-print *fsa*)

(get-matching-substring *fsa* "am 1.feb.1998 da hat das Wollwutzerl Geburtstag.")
|#