telephone numbers

Author: flammie

src/fst/Makefile.am

@@ -22,6 +22,7 @@ pkgsrcfst_DATA=url.lexc
 
 if CAN_HFST_LEXC
 check-local:
-	$(HFST_LEXC) -v url.lexc
+	$(HFST_LEXC) --Werror -v url.lexc -o /dev/null
+	$(HFST_LEXC) --Werror -v testroot.lexc stems/telephone.lexc -o /dev/null
 endif
 

src/fst/stems/telephone.lexc

@@ -0,0 +1,38 @@
+!! Lexica for recognising telephone numbers.
+!! Language independent, if you need to compound or inflect telephone
+!! numbers, create a language specific version.
+
+
+LEXICON telephones
+
++       countrycodes    ;   ! +358 050 ...
+%0%0    countrycodes    ;   ! 00358 050 ...
+%0%0%   countrycodes    ;   ! 00 358 ...
+0       areacodes       ;   ! 050 ...
+tel%:   tel-URL         ;   ! tel:+358-50-55555555 in URL standard RFC 3966
+
+
+LEXICON tel-URL
+
+< [%0|1|2|3|4|5|6|7|8|9] [0|1|2|3|4|5|6|7|8|9|%-]* > telephoneURLTags ;
+
+LEXICON countrycodes
+
+< [1|2|3|4|5|6|7|8|9] ([%0|1|2|3|4|5|6|7|8|9]) ([%0|1|2|3|4|5|6|7|8|9]) ([% | %-]) > areacodes ; ! 1-999 (kind of)
+
+LEXICON areacodes
+
+< %( %0 %) [%0|1|2|3|4|5|6|7|8|9]^2,4 ([% | %-]) > localcodes   ;   ! (0)50
+< [%0|1|2|3|4|5|6|7|8|9]^2,5 ([% | %-]) > localcodes ;   ! 050, 921
+
+LEXICON localcodes
+
+< [%0|1|2|3|4|5|6|7|8|9] [%0|1|2|3|4|5|6|7|8|9|% |%-]* [%0|1|2|3|4|5|6|7|8|9]  > telephoneNumTags ;  ! e.g. 12 34 56 78, 123-45-678
+
+LEXICON telephoneNumTags
+
++Num+Arab+Use/Circ+TEL:0    telephoneNumInflections    ;
+
+LEXICON telephoneURLTags
+
++URL+TEL:0      telephoneURLInflections ;

src/fst/testroot.lexc

@@ -0,0 +1,25 @@
+!! this lexicon is not part of the shared lexcies but an example of what
+!! you need in language specific stuff for the shared parts to work and
+!! it is also automatically tested for in case of someone breaks the shared
+!! stuffs.
+
+Multichar_Symbols
+
++TEL +URL +Gen +Sg +URL +Num +Arab +Use/Circ
+
+
+LEXICON Root
+
+0    telephones ;
+
+
+
+LEXICON telephoneNumInflections
+!! for example allow 's possessive for telephone numbers like this:
+0   #   ;
++Gen+Sg:'s  #   ;   !!= `@CODE@`
+
+LEXICON telephoneURLInflections
+
+0   #   ;
+