;;;-*- Mode: Lisp; Package: CGP -*-

(in-package :cgp)

#||

1. N [noun]

(i) Type: 1. Common 2. Proper 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Case: 1. Nominative 2. Genitive 3. Dative 4. Accusative 5. Vocative

(v) Countability: 1. Countable 2. Mass

(vi) Definiteness: 1. Definite 2. Indefinite 3. Unmarked [Danish]
(ii) Gender: 4. Common [Danish, Dutch]
(iv) Case: 6. Vocative 7. Indeclinable [Greek]

2. V [verb]

(i) Person: 1. First 2. Second 3. Third 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Finiteness: 1. Finite 2. Non-finite 
(v) Verb form / Mood: 1. Indicative 2. Subjunctive 3. Imperative 4. Conditional 5. Infinitive 6. Participle 7. Gerund 8. Supine 
(vi) Tense: 1. Present 2. Imperfect 3. Future 4. Past 
(vii) Voice: 1. Active 2. Passive 
(viii) Status: 1. Main 2. Auxiliary

(ix) Aspect: 1. Perfective 2. Imperfective
(x) Separability: 1. Non-separable 2. Separable
(xi) Reflexivity: 1. Reflexive 2. Non-reflexive
(xii) Auxiliary: 1. Have 2. Be
(viii) Status: 3. Semi-auxiliary

(xiii) Aux.-function: 1. Primary 2. Modal [English]
(v) Verb-form / Mood: 9. -Ing form [English]

3. AJ [adjective]

(i) Degree: 1. Positive 2. Comparative 3. Superlative 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Case: 1. Nominative 2. Genitive 3. Dative 4. Accusative

(v) Inflection-type: 1. Weak-Flection 2. Strong-Flection 3. Mixed 
(vi) Use: 1. Attributive 2. Predicative 
(vii) NP Function: 1. Premodifying 2. Postmodifying 3. Head-function

(iv) Case: 5. Vocative 6. Indeclinable [Greek]

4. PD [pronoun/determiner]

(i) Person: 1. First 2. Second 3. Third 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Possessive: 1. Singular 2. Plural 
(v) Case: 1. Nominative 2. Genitive 3. Dative 4. Accusative 5. Non-genitive 6. Oblique 
(vi) Category: 1. Pronoun 2. Determiner 3. Both 
(vii) Pron.-Type: 1. Demonstrative 2. Indefinite 3. Possessive 4. Int./Rel. 5. Pers./Refl. 
(viii) Det.-Type: 1. Demonstrative 2. Indefinite 3. Possessive 4. Int./Rel. 5. Partitive

(ix) Special Pronoun Type: 1. Personal 2. Reflexive 3. Reciprocal
(x) Wh-Type: 1. Interrogative 2. Relative 3. Exclamatory
(xi) Politeness: 1. Polite 2. Familiar 

(ii) Gender: 4. Common [Danish]
(v) Case: 7. Prepositional [Spanish]
(xii) Strength 1. Weak 2. Strong [French, Dutch, Greek]

5. AT [article]

(i) Article-Type: 1. Definite 2. Indefinite 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Case: 1. Nominative 2. Genitive 3. Dative 4. Accusative

-

(i) Article-Type: 3. Partitive [French]
(ii) Gender: 4. Common [Danish]
(iv) Case: 5. Vocative 6. Indeclinable [Greek]


6. AV [adverb]

(i) Degree: 1. Positive 2. Comparative 3. Superlative

(ii) Adverb-Type: 1. General 2. Degree 
(iii) Polarity: 1. Wh-type 2. Non-wh-type 
(iv) Wh-type: 1. Interrogative 2. Relative 3. Exclamatory

(ii) Adverb-Type: 3. Particle 4. Pronominal [English, German, Dutch]

7. AP [adposition]

(i) Type: 1. Preposition

(i) Type: 2. Fused prep-art

(i) Type: 3. Postposition 4. Circumposition [German, English]

8. C [conjunction]

(i) Type: 1. Coordinating 2. Subordinating

(ii) Coord-Type: 1. Simple 2. Correlative 3. Initial 4. Non-initial

(iii) Subord.-type: 1. With-finite 2. With-infin. 3. Comparative [German]

9. NU [numeral]

(i) Type: 1. Cardinal 2. Ordinal 
(ii) Gender: 1. Masculine 2. Feminine 3. Neuter 
(iii) Number: 1. Singular 2. Plural 
(iv) Case: 1. Nominative 2. Genitive 3. Dative 4. Accusative 
(v) Function: 1. Pronoun 2. Determiner 3. Adjective

10. I [interjection]

11. U [unique/unassigned]

-
-

(i) Unique-type:
 1. Infinitive marker [German zu, Danish at, Dutch, English]
 2. Negative particle [English not, n't]
 3. Existential marker [English there, Danish der]
 4. Second negative particle [French pas]
 5. Anticipatory er [Dutch]
 6. Mediopassive voice marker se [Portuguese]
 7. Preverbal particle [Greek]

12. R [residual]

(i) Type: 1. Foreign word 2. Formula 3. Symbol 4. Acronym 5. Abbreviation 6. Unclassified
(ii) Number: 1. Singular 2. Plural
(iii) Gender: 1. Masculine 2. Feminine 3. Neuter

13. PU [punctuation] 

(i) 1. Period 2. Comma 3. Question mark ...etc. ...
(i) 1. Sentence-final 2. Sentence-medial 3. Left-Parenthetical 4. Right-Parenthetical 

||#

(defun write-eagles-feature (stream features plist)
  (loop for (num f) on plist by #'cddr
        do (if (find f features)
             (write-string (string num) stream)
             (write-string "0" stream))))

(defun code-eagles-tagset (stream pos &rest features)
  (labels ((write-feature (plist)
              (let ((foundp nil))
                (loop for (num f) on plist by #'cddr
                      when (and (not foundp) (find f features))
                      do (write-string (string num) stream)
                      (setf foundp t))
                (unless foundp
                  (write-string "0" stream)))))
    (case pos
    ;; 1. N [noun]
      (N
       (write-string "N" stream)
       ;; (i) Type:
       (write-feature '("1" Common "2" Proper))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter "4" Common-gender))
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural))
       ;; (iv) Case:
       (write-feature '("1" Nominative "2" Genitive "3" Dative "4" Accusative "5" Vocative "6" Indeclinable))
       ;; (v) Countability:
       (write-feature '("1" Countable "2" Mass))
       ;; (vi) Definiteness:
       (write-feature '("1" Definite "2" Indefinite "3" Unmarked)))
      ;; 2. V [verb]
      (V
       (write-string "V" stream)
       ;; (i) Person:
       (write-feature '("1" First "2" Second "3" Third))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter)) 
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural))
       ;; (iv) Finiteness:
       (write-feature '("1" Finite "2" Non-finite ))
       ;; (v) Verb form / Mood:
       (write-feature '("1" Indicative "2" Subjunctive "3" Imperative "4" Conditional "5" Infinitive "6" Participle "7" Gerund "8" Supine "9" ing-form))
       ;; (vi) Tense:
       (write-feature '("1" Present "2" Imperfect "3" Future "4" Past ))
       ;; (vii) Voice:
       (write-feature '("1" Active "2" Passive ))
       ;; (viii) Status:
       (write-feature '("1" Main "2" Auxiliary))
       
       ;; (ix) Aspect:
       (write-feature '("1" Perfective "2" Imperfective))
       ;; (x) Separability:
       (write-feature '("1" Non-separable "2" Separable))
       ;; (xi) Reflexivity:
       (write-feature '("1" Reflexive "2" Non-reflexive))
       ;; (xii) Auxiliary:
       (write-feature '("1" Have "2" Be))
       ;; (viii) Status:
       (write-feature '("3" Semi-auxiliary))
       
       ;; (xiii) Aux.-function:
       (write-feature '("1" Primary "2" Modal))) ;; [English]

      ;; 3. AJ [adjective]
      (AJ
       (write-string "AJ" stream)
       ;; (i) Degree:
       (write-feature '("1" Positive "2" Comparative "3" Superlative ))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter ))
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural ))
       ;; (iv) Case:
       (write-feature '("1" Nominative "2" Genitive "3" Dative "4" Accusative "5" Vocative "6" Indeclinable))
       
       ;; (v) Inflection-type:
       (write-feature '("1" Weak-Flection "2" Strong-Flection "3" Mixed ))
       ;; (vi) Use:
       (write-feature '("1" Attributive "2" Predicative ))
       ;; (vii) NP Function:
       (write-feature '("1" Premodifying "2" Postmodifying "3" Head-function)))

      ;; 4. PD [pronoun/determiner]
      (PD
       (write-string "PD" stream)
       ;; (i) Person:
       (write-feature '("1" First "2" Second "3" Third ))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter "4" Common)) 
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural ))
       ;; (iv) Possessive:
       (write-feature '("1" Singular "2" Plural ))
       ;; (v) Case:
       (write-feature '("1" Nominative "2" Genitive "3" Dative "4" Accusative "5" Non-genitive "6" Oblique "7" Prepositional))
       ;; (vi) Category:
       (write-feature '("1" Pronoun "2" Determiner "3" Both))
       ;; (vii) Pron.-Type:
       (write-feature '("1" Demonstrative "2" Indefinite "3" Possessive "4" Int./Rel. "5" Pers/Refl))
       ;; (viii) Det.-Type:
       (write-feature '("1" Demonstrative "2" Indefinite "3" Possessive "4" Int/Rel "5" Partitive))
       
       ;; (ix) Special Pronoun Type:
       (write-feature '("1" Personal "2" Reflexive "3" Reciprocal))
       ;; (x) Wh-Type:
       (write-feature '("1" Interrogative "2" Relative "3" Exclamatory))
       ;; (xi) Politeness:
       (write-feature '("1" Polite "2" Familiar))
       
       ;; (xii) Strength: 
       (write-feature '("1" Weak "2" Strong))) ;; [French, Dutch, Greek]

      ;; 5. AT [article]
      (AT
       (write-string "AT" stream)
       ;; (i) Article-Type:
       (write-feature '("1" Definite "2" Indefinite "3" Partitive))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter "4" Common))
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural ))
       ;; (iv) Case:
       (write-feature '("1" Nominative "2" Genitive "3" Dative "4" Accusative "5" Vocative "6" Indeclinable)))

      ;; 6. AV [adverb]
      (AV
       (write-string "AV" stream)
       ;; (i) Degree:
       (write-feature '("1" Positive "2" Comparative "3" Superlative))
       
       ;; (ii) Adverb-Type:
       (write-feature '("1" General "2" Degree "3" Particle "4" Pronominal))
       ;; (iii) Polarity:
       (write-feature '("1" Wh-type "2" Non-wh-type ))
       ;; (iv) Wh-type:
       (write-feature '("1" Interrogative "2" Relative "3" Exclamatory)))

      ;; 7. AP [adposition]
      (AP
       (write-string "AP" stream)
       ;; (i) Type:
       (write-feature '("1" Preposition "2" Fused-prep-art "3" Postposition "4" Circumposition)))

      ;; 8. C [conjunction]
      (C
       (write-string "C" stream)
       ;; (i) Type:
       (write-feature '("1" Coordinating "2" Subordinating))
       
       ;; (ii) Coord-Type:
       (write-feature '("1" Simple "2" Correlative "3" Initial "4" Non-initial))
       
       ;; (iii) Subord.-type:
       (write-feature '("1" With-finite "2" With-infin. "3" Comparative)))

      ;; 9. NU [numeral]
      (NU
       (write-string "NU" stream)
       ;; (i) Type:
       (write-feature '("1" Cardinal "2" Ordinal))
       ;; (ii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter))
       ;; (iii) Number:
       (write-feature '("1" Singular "2" Plural ))
       ;; (iv) Case:
       (write-feature '("1" Nominative "2" Genitive "3" Dative "4" Accusative ))
       ;; (v) Function:
       (write-feature '("1" Pronoun "2" Determiner "3" Adjective)))

      ;; 10. I [interjection]
      (I
       (write-string "I" stream))

      ;; 11. U [unique/unassigned]
      (U
       (write-string "U" stream)
       ;; (i) Unique-type:
       (write-feature '("1" Infinitive-marker ;; [German zu, Danish at, Dutch, English]
                        "2" Negative-particle ;; [English not, n't]
                        "3" Existential-marker ;; [English there, Danish der]
                        "4" Second-negative-particle ;; [French pas]
                        "5" Anticipatory-er ;; [Dutch]
                        "6" Mediopassive-voice-marker-se ;; [Portuguese]
                        "7" Preverbal-particle))) ;; [Greek]

      ;; 12. R [residual]
      (R
       (write-string "R" stream)
       ;; (i) Type:
       (write-feature '("1" Foreign-word "2" Formula "3" Symbol "4" Acronym "5" Abbreviation "6" Unclassified))
       ;; (ii) Number:
       (write-feature '("1" Singular "2" Plural))
       ;; (iii) Gender:
       (write-feature '("1" Masculine "2" Feminine "3" Neuter)))
      
      ;; 13. PU [punctuation] 
      (PU
       (write-string "PU" stream)
       ;; (i) "1" Period "2" Comma "3" Question mark ...etc. ...
       ;; (i) 
       (write-feature '("1" Sentence-final "2" Sentence-medial "3" Left-Parenthetical "4" Right-Parenthetical))))))

;;    *  A common noun, feminine, plural, countable, is represented: N122010
#+test
(print
 (with-output-to-string (stream)
  (code-eagles-tagset stream 'N 'common 'feminine 'plural 'countable)))

(defun cg-to-eagles-features (pos cg-features)
  (collecting
    (labels ((translate (plist)
               (loop for (cg-f eagles-f) on plist by #'cddr
                     when (find cg-f cg-features)
                     do (if (consp eagles-f)
                          (collect-append eagles-f)
                          (collect eagles-f)))))
      (case pos
        (subst
         (collect 'N)
         (translate '(ent singular
                      fl plural
                      mask masculine
                      fem feminine
                      nøyt neuter
                      ub indefinite
                      be definite
                      appell common
                      prop proper)))
	(adj
	 (cond ((find '<ordenstall> cg-features)
		(collect 'NU)
		(collect 'ordinal)
		(translate '(ent singular
			     fl plural
			     mask masculine
			     fem feminine
			     nøyt neuter)))
	       (t
		(collect 'AJ)
		(translate '(pos positive
			     komp comparative
			     sup superlative
			     ent singular
			     fl plural
			     mask masculine
			     fem feminine
			     nøyt neuter
			     ub indefinite
			     be definite)))))
        (verb
         (collect 'V)
         (translate '(perf-part (participle past)
                      inf infinitive
                      pass passive
                      pres present
                      pret past
                      imp imperative
		      <aux1/infinitiv> auxiliary
		      <aux1/perf_part> auxiliary
		      ))
	 (unless (or (find '<aux1/infinitiv> cg-features)
		     (find '<aux1/perf_part> cg-features))
	   (collect 'main)))
	(det
	 (cond ((find 'kvant cg-features)
		(collect 'NU)
		(collect 'cardinal))
	       (t
		(collect 'PD)
		(collect 'determiner)))
	 (translate '(dem demonstrative
		      mask masculine
                      fem feminine
                      nøyt neuter
                      ent singular
                      fl plural
		      sp int/rel
		      poss possessive)))
	(pron
	 (collect 'PD)
	 (translate '(pers (pers/refl personal)
		      refl (pers/refl reflexive)
		      res (pers/refl reciprocal)
		      sp int/rel
		      ;; hum
		      \1 first
		      \2 second
		      \3 third
		      nom nominative
		      akk accusative
                      mask masculine
                      fem feminine
                      nøyt neuter
                      ent singular
                      fl plural
		      høflig polite)))
	(konj
	 (collect 'C)
	 (collect 'coordinating)
	 ;;(translate '(clb ...))
	 )
	(prep
	 (collect 'AP)
	 (collect 'preposition))
	(adv
	 (collect 'AV))
	(clb
	 (collect 'PU)
	 (translate '(<punkt> sentence-final
		      <utrop> sentence-final
		      <spm> sentence-final
		      <colon> sentence-final
		      <ellipse> sentence-final
		      ;; ??
		      <komma> sentence-medial)))
	(<komma>
	 (collect 'PU)
	 (collect 'sentence-medial))
	(<strek>
	 (collect 'PU)
	 (collect 'sentence-medial))
	(inf-merke
	 (collect 'U)
	 (collect 'inifinitive-marker))
	(sbu
	 (collect 'C)
	 (collect 'subordinating))
	(fork
	 (collect 'R)
	 (collect 'abbreviation)
	 (translate '(ent singular
                      fl plural
                      mask masculine
                      fem feminine
                      nøyt neuter)))
	(symb
	 (collect 'R)
	 (collect 'symbol))
	(ukjent
	 (collect 'R)
	 (collect 'foreign-word))
	(interj
	 (collect 'I))))))

#+test
(print (cg-to-eagles-features 'verb '(perf-part)))
#+test
(print (cg-to-eagles-features 'subst '(mask appell ub ent)))

#+old ;; used for tree-tagger pre-3.1
(defun penn-to-eagles-features (penn-tag)
  (case penn-tag
    (JJ '(AJ positive))
    (JJR '(AJ comparative))
    (JJS '(AJ superlative))
    (RB '(AV))
    (RBR '(AV comparative))
    (RBS '(AV superlative))
    (CD '(NU cardinal))
    (CC '(C coordinating))
    (DT '(PD determiner))
    (EX '(U existential-marker))
    (FW '(R foreign-word))
    (UH '(I))
    (LS '(R unclassified))
    (MD '(V auxiliary modal))
    (NN '(N singular common))
    (NNS '(N plural common))
    (NP '(N singular proper))
    (NPS '(N plural proper))
    (RP '(AV)) ;; ??
    (PP '(PD pronoun pers/refl personal))
    (PP$ '(PD pronoun possessive))
    (WP$ '(PD pronoun int/rel))
    (PDT '(PD determiner)) ;; predeterminer
    (IN '((AP preposition) (C subordinating)))
    (SYM '(R symbol))
    (TO '((U inifinitive-marker) (AP preposition)))
    (VB '((V non-finite infinitive) (V finite imperative) (V finite subjunctive)))
    (VBD '(V finite past))
    (VBG '((V non-finite gerund) (V non-finite participle present)))
    (VBN '(V non-finite participle past))
    (VBP '(V finite present)) ;; non-3.sg
    (VBZ '(V finite present third singular))
    (WDT '(PD determiner int/rel))
    (WRB '(AV wh-type))))


;; for tree-tagger 3.1:

;; From: http://www.ims.uni-stuttgart.de/projekte/corplex/TreeTagger/DecisionTreeTagger.html
;;   The tagset used by the TreeTagger is a refinement of this tagset [sic: the Penn-Treebank tagset]
;;   where the second letter of the verb part-of-speech tags
;;   distinguishes between "be" verbs (B), "have" verbs (H) and other
;;   verbs (V).
;; These verbs will in eagles be marked with auxilliary
(defun penn-to-eagles-features (penn-tag)
  (case penn-tag
    (JJ '(AJ positive))
    (JJR '(AJ comparative))
    (JJS '(AJ superlative))
    (RB '(AV))
    (RBR '(AV comparative))
    (RBS '(AV superlative))
    (CD '(NU cardinal))
    (CC '(C coordinating))
    (DT '(PD determiner))
    (EX '(U existential-marker))
    (FW '(R foreign-word))
    (UH '(I))
    (LS '(R unclassified))
    (MD '(V auxiliary modal))
    (NN '(N singular common))
    (NNS '(N plural common))
    (NP '(N singular proper))
    (NPS '(N plural proper))
    (RP '(AV)) ;; ??
    (PP '(PD pronoun pers/refl personal))
    (PP$ '(PD pronoun possessive))
    (WP$ '(PD pronoun int/rel))
    (PDT '(PD determiner)) ;; predeterminer
    (IN '((AP preposition) (C subordinating)))
    (SYM '(R symbol))
    (TO '((U inifinitive-marker) (AP preposition)))

    ;; Other verbs:
    (VV '((V non-finite infinitive) (V finite imperative) (V finite subjunctive)))
    (VVD '(V finite past))
    (VVG '((V non-finite gerund) (V non-finite participle present)))
    (VVN '(V non-finite participle past))
    (VVP '(V finite present)) ;; non-3.sg
    (VVZ '(V finite present third singular))
    
    ;; 'Be' verbs and 'have' verbs (same as above, but includes 'auxiliary' attribute)
    ((VB VH) '((V auxiliary non-finite infinitive) (V auxiliary finite imperative) (V auxiliary finite subjunctive)))
    ((VBD VHD) '(V auxiliary finite past))
    ((VBG VHG) '((V auxiliary non-finite gerund) (V auxiliary non-finite participle present)))
    ((VBN VHN) '(V auxiliary non-finite participle past))
    ((VBP VHP) '(V auxiliary finite present)) ;; non-3.sg
    ((VBZ VHZ) '(V auxiliary finite present third singular))
    
    (WDT '(PD determiner int/rel))
    (WRB '(AV wh-type))))

;;; EOF