Merge pull request #89 from LR-POR/issue-85

Added punctuation support

Version.lsp

@@ -1,4 +1,4 @@
 ;;; -*- Mode: TDL; Coding: utf-8 -*-
 (in-package :common-lisp-user)
 
-(defparameter *grammar-version* "Portuguese (2022-08-10)")
+(defparameter *grammar-version* "PorGram (2023-02-13)")

lexicon/punct-lex.tdl

@@ -0,0 +1,12 @@
+period-pct := pt-period-le &
+ [ STEM < "." > ].
+
+qmark-pct := pt-qmark-le &
+ [ STEM < "?" > ].
+
+bang-pct := pt-bang-le &
+ [ STEM < "!" > ].
+
+comma-pct := pt-comma-le &
+ [ STEM < "," > ].
+

my-portuguese-pet.tdl

@@ -43,6 +43,7 @@
 :include "lexicon/random-sample-209-entries".
 :include "lexicon/new-ditrans".
 :include "lexicon/my-adjectives".
+:include "lexicon/punct-lex".
 :end :instance.
 
 ;;

my-portuguese.tdl

@@ -911,5 +911,97 @@ a-que-cl-verb-lex := subj-cl-verb-lex & clausal-nom-verb-lex & clausal-second-ar
 ;     NH-DTR.SYNSEM.NON-LOCAL.QUE.LIST #que ].
 
 basic-adjective-lex :+
-   [ SYNSEM.NON-LOCAL.QUE.LIST < > ].
-
+   [ SYNSEM.NON-LOCAL [ SLASH.LIST < >,
+			QUE.LIST < >,
+			REL.LIST < > ] ].
+
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; Additions for punctuation
+;;
+
+synsem-min :+
+  [ PUNCT punctuation-min ].
+
+punct-hd := head.
+
+punct-word := word &
+  [ INFLECTED infl-satisfied,
+    SYNSEM [ LOCAL [ CAT [ HEAD punct-hd,
+                           VAL [ SUBJ < >, SPR < >, SPEC < >,
+                                 COMPS < expressed-synsem &
+                                         [ OPT - ] > ] ],
+                     CONT [ HOOK.INDEX event-or-ref-index,
+			    RELS.LIST < >,
+			    HCONS.LIST < >,
+			    ICONS.LIST < > ] ],
+             NON-LOCAL non-local-none ] ].
+
+punct-mark := avm &
+  [ PSF iforce ].
+
+punctuation-min := avm.
+punctuation := punctuation-min &
+  [ LPUNCT punct-mark,
+    RPUNCT punct-mark ].
+
+comma-or-clause := punct-mark.
+comma-or-no-punct := punct-mark.
+comma-punct := comma-or-clause & comma-or-no-punct.
+clause-or-no-punct := punct-mark.
+clause-punct := comma-or-clause & clause-or-no-punct.
+no-punct := comma-or-no-punct & clause-or-no-punct.
+
+punct-right-word := punct-word &
+  [ SYNSEM [ LOCAL.CAT.POSTHEAD +,
+             PUNCT [ LPUNCT no-punct,
+                     RPUNCT punct-mark ] ] ].
+
+punct-left-word := punct-word &
+  [ SYNSEM [ LOCAL.CAT.POSTHEAD -,
+             PUNCT [ LPUNCT punct-mark,
+                     RPUNCT no-punct ] ] ].
+
+punct-right-clause-word := punct-right-word &
+  [ SYNSEM [ LOCAL.CAT.VAL.COMPS < [ PUNCT.RPUNCT clause-or-no-punct ] >,
+             PUNCT.RPUNCT clause-punct ] ].
+
+pt-comma-le := punct-right-word &
+  [ SYNSEM [ LOCAL.CAT.VAL.COMPS < [ PUNCT.RPUNCT no-punct ] >,
+             PUNCT [ LPUNCT no-punct,
+		     RPUNCT comma-punct ] ] ].
+
+prop-or-comm := iforce.
+prop := prop-or-ques & prop-or-comm.
+comm := prop-or-comm.
+
+pt-period-le := punct-right-clause-word &
+  [ SYNSEM.PUNCT.RPUNCT.PSF prop-or-comm ].
+
+pt-qmark-le := punct-right-clause-word &
+  [ SYNSEM.PUNCT.RPUNCT.PSF ques ].
+
+pt-bang-le := punct-right-clause-word &
+  [ SYNSEM.PUNCT.RPUNCT.PSF prop-or-comm ].
+
+punct-phrase := phrase &
+  [ SYNSEM [ LOCAL #local,
+	     NON-LOCAL #nonlocal,
+	     LIGHT #light,
+             MODIFIED #modif ],
+    PT-DTR.SYNSEM.LOCAL.CAT [ HEAD punct-hd,
+			      VAL.COMPS < #synsem > ],
+    NONPT-DTR [ INFLECTED #infl,
+	        SYNSEM #synsem & [ LOCAL #local,
+				   NON-LOCAL #nonlocal,
+				   LIGHT #light & +,
+				   MODIFIED #modif ] ],
+    INFLECTED #infl ].
+
+punct-hinit-rule := punct-phrase &
+ [ PT-DTR #ptdtr,
+   NONPT-DTR #nonptdtr,
+   ARGS < #nonptdtr, #ptdtr > ].
+
+;;
+;; End additions for punctuation
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

repp/vanilla.rpp

@@ -3,7 +3,19 @@
 ;;; to your choices file.
 ;;;
 ;;; 
-:[ \t!"#$%&'()\*\+,\./:;<=>?@\[\]\^_`{|}~\\]
+;;;:[ \t!"#$%&'()\*\+,\./:;<=>?@\[\]\^_`{|}~\\]
+:[ \t]+
 !^(.+)$								 \1 
 !\t								 
 !  +								 
+
+#1
+! ([^ (]+)(\)) ([^ ]|$)						 \1 \2 \3
+! ([^ \[]+)(]) ([^ ]|$)						 \1 \2 \3
+!([^ ])([\]}?!,，、;:@#$€¢£¥%&”"’'_]) ([^ ]|$)			\1 \2 \3
+!([^ ])([\]}?!,，、;:@#$€¢£¥%&”"’'_])([–-])([^ ]|$)		\1 \2 \3 \4
+!([^ ])(\.) ([\])}”"’'… ]*)(⌊.⌋)*$				\1 \2 \3\4
+!(^|[^ ]) ([\[({:@#$€¢£¥%&\*“‘])([^ ])				\1 \2 \3
+#
+
+>1

rules.tdl

@@ -33,3 +33,5 @@ ex-subj := extracted-subj-phrase.
 wh-ques := wh-ques-phrase.
 
 in-situ-ques := insitu-int-cl.
+
+hd-pct := punct-hinit-rule.

tsdb/gold/core/decision

@@ -0,0 +1,35 @@
+110@-1@1@7@head-spec@@3@5@23-6-2013 14:28:24
+160@-1@1@7@normadp1-norm-adposition-lex@@3@4@23-6-2013 14:28:24
+160@-1@1@7@subj-head@@0@6@23-6-2013 14:28:24
+170@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@10@11@23-6-2013 14:28:24
+170@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\sde-ditrans-second-arg-raising-verb-lex@@6@7@23-6-2013 14:28:24
+180@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+210@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@7@8@23-6-2013 14:28:24
+280@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@7@8@23-6-2013 14:28:24
+340@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+350@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+360@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+370@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+380@-1@1@7@zero-refl-mark-lex\sinfl-inf-3sg-suffix\strans-verb-lex@@6@7@23-6-2013 14:28:24
+610@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+790@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+800@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+810@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+820@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+900@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\snonrefl-int-verb-lex@@3@4@23-6-2013 14:28:24
+910@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\snonrefl-int-verb-lex@@3@4@23-6-2013 14:28:24
+960@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\snonrefl-int-verb-lex@@3@4@23-6-2013 14:28:24
+1010@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+1020@-1@1@7@decl-head-opt-subj\szero-refl-mark-lex\spres-ind-1pl-suffix\snonrefl-int-verb-lex@@0@1@23-6-2013 14:28:24
+1090@-1@1@7@refl-mark-3-suffix\sperf-ind-3pl-suffix\srefl-com-obj-verb-lex@@0@1@23-6-2013 14:28:24
+1100@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+1230@-1@1@7@obl-case-marking-adp-lex@@4@5@23-6-2013 14:28:24
+1250@-1@1@7@zero-refl-mark-lex\spres-ind-3sg-suffix\sinf-trans-first-arg-control-verb-lex@@1@2@23-6-2013 14:28:24
+1310@-1@1@7@obl-case-marking-adp-lex@@5@6@23-6-2013 14:28:24
+1330@-1@1@7@zero-refl-mark-lex\spres-ind-3pl-suffix\sinf-trans-first-arg-control-verb-lex@@1@2@23-6-2013 14:28:24
+1340@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\sinf-trans-first-arg-control-verb-lex@@2@3@23-6-2013 14:28:24
+1350@-1@1@7@refl-mark-3-suffix\sperf-ind-3pl-suffix\srefl-int-verb-lex@@1@2@23-6-2013 14:28:24
+1400@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+1440@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+1450@-1@1@7@zero-refl-mark-lex\sperf-ind-3pl-suffix\strans-verb-lex@@2@3@23-6-2013 14:28:24
+1470@-1@1@7@decl-head-opt-subj\szero-refl-mark-lex\sinfl-inf-1sg-suffix\snonrefl-int-verb-lex@@2@3@23-6-2013 14:28:24