#!/usr/bin/env icarus
# NCIMB Phages Catalogue - Syntax file
# Version 2004.12.17.01
# Last Updated December 17, 2004
# Author: Domenico Marra & Francesca Piersigilli & Paolo Romano
# (c) CABRI 1999-2023

$fn={
  Collection_number:coll_no
  Other_collection_numbers:other_nos 
  Name:name
  Type:type
  Restricted_distribution:rest_distrib  
  Literature:literature
  History_of_deposit:history
  Host_for_propagation:host_for_prop
  Host_used_for_propagation:host_str
  Lysogenicity:lyso
  Virus_used_for:virus_for
}

$rules={
  entry:	~ {$In:[file:text] $Out pre $Skip:0}
		  ('Collection_number' {$Not} ln)*
		  ('Collection_number' {$entryFip=$Fip $Wrt} ln {$App}
		  ('Collection_number' {$Not} ln {$App})*)?
		~

# fields
  fields:	~ {$In:entry $Out $Skip:1}
                  (/\/[^\n]+\n/ {$Wrt} |
                  word {$Wrt:$fn.$Ct} 
                  (/[\/A-Z]/ {$Not} ln {$App})+)+ ~

#indexing
  coll_no:      ~ {$In:[fields c:coll_no] $Out}
                  word ' '(word3{$Wrt} |punct4) ~

  other_nos:      ~ {$In:[fields c:other_nos] $Out}
                  word ' '(word3{$Wrt} |punct4) ~

  name:	         ~ {$In:[fields c:name] $Out}
		  word ' '
                  (word {$Wrt:[s:$StrRep:[s:$Ct r:"[\/|\.|-]" to:""]]}
                   | punct)* ~

  strX:		~ {$In:[fields c:{type class cons rest_distrib host_for_sel
                        vimor parsi plafo clepla cell_surf_rec minpac maxpac
                        remap cloge trans_el
                        start_codon further_inf rest_sites price_code medium
                        literature history host_for_prop lyso cell_surf_rec
                        host_str na_type virus_for }] $Out}
		  word (str{$Uniq:$Itc}| /./ |punct)* ~ 

#HTML stuff...
 h_top:      ~ {$In:[fields c:coll_no t:html] pre if:$ParInt:isTable $Fail}
               word {$Rep:
                     |</TR><TR VALIGN=TOP>
                     |<TD colspan=2 bgcolor=\"#ffffff\">
                     |<font color=\"#000066\">
                     |<B>CABRI:($entry.libName)</B></TD>
                     |</TR>
		     |<TR VALIGN=TOP><TD bgcolor=\"#ffffff\">
                     |<font color=\"#000066\"><b><i>$Ct</i></b></font></TD>
                     |<TD bgcolor=\"#ffffff\">
		    } 
		    ':'? {$Rep:""} /.*/ {$Rep:"$Ct</TD></TR>"}  ~

 h_fields:      ~ {$In:[fields t:html] pre if:$ParInt:isTable $Fail}
                   word {if:$isTable==0
                     $Rep:
                      |<TR VALIGN=TOP><TD bgcolor=\"#ffffff\">
                      |<font color=\"#000066\"><b><i>$Ct</i></b></font></TD>
                      |<TD bgcolor=\"#ffffff\">
                     } ':'? {$Rep:""} /.*/ {$Rep:"$Ct</TD></TR>"}  ~

 t_fields:     ~ {$In:[fields] $Out} tag /.*/ {$Wrt:$Itc} ~

# definitions
  tag:          ~ word ':'?~
  ln:		~/[^\n]*\n/ ~
  word:		~/[0-9a-zA-Z_\/]+/ ~
  str:		~/[a-zA-Z0-9&.'=\/\\%_-]+/ ~
  punct:	~ /[ \t,;:\\.\\(\\)\\+\\*]+/ ~  
  punct4:       ~ /[\t,;:\\.\\(\\)\\+\\*]+/ ~  
  word3:        ~/[ 0-9a-zA-Z_\/\\(\\)-]+/ ~
}