;;;-*- Mode: Lisp; Package: CL-USER -*-

(in-package "CL-USER")

#+ignore
(eval-when (:execute :compile-toplevel :load-toplevel)
  (um:use-module :text-utlis "projects:utilities;text-utils"))

(mk :sql)
(use-package :sql)
(initialize-database-type)
(enable-sql-reader-syntax)

(connect "Oslotagger" :user-id "paulm" :password "gvprckvnis" :if-exists :warn-old)

#+test
(do-query ((line)
           [select [conc [oppslag] ":" [paradigme-id] ":" [boy-nummer] ":" [grammatikk]]
                   :from [taggeadm-nn v-tagger-tag]
                   :where [apply 'or 
                              (loop for tag
                                    in '("<" "<ditrans5/forpart2over>" "<reefl12/for>" "<reefl12/over>" "<trans3s>"
                                         "<trans11/etterpart1ut>" "<trans11på)>" "<part1/trans11på>" "<trans11/uttrans11på>" 
                                         "<intrans1trans11/etter>" "<intrans1trans11/om>" "<.trans11/med>" "<inrans1>"
                                         "<tranns1>" "<intrans1intrans2>" "<trans1part1/av>" "<part1(opp>" "<tran11/til>"
                                         "<intrans1part3/med>" "/opp>" "<intrams1>" "<>" "<part4.adv12/fram>" 
                                         "<intrans1.part4/opp>" "<ref9/til>" "<part2/utovert>" "</på>" "<rans1>"
                                         "<refl16/overpart1opp>" "<adv4" "<refll14/med>" "<trans1trans11/i>"
                                         "<part5/i" "<intrans2.>" "<trans1part1/opp>" "<part1/utpart2opp>" "<intrans1part3/i>" 
                                         "veg>" "<intrans1part3/til>" "<intrans1trans11/til>" "<refll4>" "<intran1>"
                                         "<nullv1nullv2/frå>" "<intrans1><adv3>" "<ditranms5/av>" "<inttans1>" "<refll9/i>" 
                                         "<refll9/på>" "refl4>" "<refl9(fram+til>" "<part3/tilpart4på>" "<tran3>" "<part5/5>"
                                         "*verb" "<inrans4>" "<auv1/infinitiv>" "v+v" "perf-part." "nøut" "eint/fl"
                                         "subst+v+subst" "susbt" "refl")
                                    collect [like [grammatikk] (u:concat "% " tag " %")]
                                    collect [like [grammatikk] (u:concat tag " %")]
                                    collect [like [grammatikk] (u:concat "% " tag)]
                                    collect [like [grammatikk] ?tag])]] )
          (print line))

(select [table-name] :from [all-tables] :where [= [owner] "TAGGEADM_NN"] :flatp t)

(select [conc [oppslag] ":" [grammatikk]]
        :from [taggeadm-nn v-tagger-fullform]
        :where [like [grammatikk] "%eint/fl%"] :flatp t)

(let ((count 0))
  (with-open-file (stream "projects:cgp;nets;fullforms+tags-nn.txt"
                          :direction :output :if-exists :supersede :if-does-not-exist :create)
    (do-query ((line)
               [select [conc [oppslag] ":" [grammatikk]]
                       :from [taggeadm-nn v-tagger-fullform]])
      (when (zerop (mod (incf count) 1000)) (print count))
      (write-line line stream))))

(let ((count 0))
  (with-open-file (stream "projects:cgp;multitagger;fullforms+tags.txt"
                          :direction :output :if-exists :supersede :if-does-not-exist :create)
    (do-query ((line)
               [select [conc [oppslag] ":" [grammatikk]]
                       :from [tagger v-tagger-fullform]])
      (when (zerop (mod (incf count) 1000)) (print count))
      (write-line line stream))))

(let ((count 0))
  (with-open-file (stream "projects:cgp;multitagger;nn-lexicon.txt"
                          :direction :output :if-exists :supersede :if-does-not-exist :create)
    (do-query ((line)
               [select [conc [oppslag] ":" [paradigme-id] ":" [boy-nummer] ":" [grammatikk]]
                       :from [tagger v-tagger-tag]])
      (when (zerop (mod (incf count) 1000)) (print count))
      (write-line line stream))))

(let ((count 0))
  (with-open-file (stream "projects:cgp;nets;tags-nn.txt"
                          :direction :output :if-exists :supersede :if-does-not-exist :create)
    (do-query ((line)
               [select [conc [oppslag] ":" [paradigme-id] ":" [boy-nummer] ":" [grammatikk]]
                       :from [taggeadm-nn v-tagger-tag]])
      (when (zerop (mod (incf count) 1000)) (print count))
      (write-line line stream))))

(block test
  (let ((count 0))
    (with-open-file (stream "projects:cgp;bm-lexicon.txt"
                            :direction :output :if-exists :supersede :if-does-not-exist :create)
      (u::with-file-lines (line "projects:cgp;tags.txt")
        (format stream "~a~%" line)
        #+ignore
        (when (= (incf count) 100)
          (return-from test))))))