#!/usr/bin/env icarus
# NCIMB Bacteria 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={
  Strain_number:strain_no
  Other_collection_numbers:other_nos
  Name:name
  Infrasubspecific_names:inf_names
  Organism_type:org_type
  Restrictions:restrictions
  Status:status
  History:history
  Other_names:oth_names
  Isolated_from:isol_from
  Genotype:genotype
  Literature:lit
  Conditions_for_growth:cond
  Form_of_supply:supply
  Further_information:finfo
}

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

# fields
  fields:	~ {$In:entry $Out $Skip:1}
                  (/\/[^\n]+\n/ {$Wrt} |
                  word {$Wrt:$fn.$Ct} 
                  (/[\/A-Z]/ {$Not} ln {$App})+)+ ~
 
#indexing
  strain_no:	~ {$In:[fields c:strain_no] $Out}
		  word ' ' (word2{$Wrt} |punct2)* ~

  name:		~ {$In:[fields c:name] $Out $Skip:1}
		  word / /
                  word {$Wrt} / / {$App} word {$App} (word2 {$Wrt} | punct)*  ~

  strX:		~ {$In:[fields c:{restrictions org_type name inf_names status
                                  history applic
				  cond supply isol_from 
                                  genotype finfo}] $Out}
		  word (str{$Uniq:$Itc}| /./ |punct)* ~ 

  lit:          ~ {$In:[fields c:lit] $Out $Skip:0}
                  word (parola|numero{$Uniq:$Itc})* ~

  other_nos:    ~ {$In:[fields c:other_nos] $Out}
                  word / /? '-'? (word2{$Uniq} | /; ?/)* ~

  concat2: 	~ {$In:[fields c:oth_names] $Out}
                  /[^\n]+/ {$Wrt} '\n' (/[^\n]+/ {$App} '\n')* ~

  oth_names:	~ {$In:concat2 $Out}
		  word / /?
                  (/;/? / /? word3 {$Wrt:[ s:$StrRep:[s:$Ct r:"[\"|\.]" to:""]]}
                   | /,/ /[^;]/+)* ~

#HTML stuff...
 h_top:      ~ {$In:[fields c:strain_no t:html] pre if:$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
  parola:       ~ /[^0-9]+/~
  numero:       ~ /[0-9-]+/ ~
  tag:          ~ word ':'?~
  ln:		~ /[^\n]*\n/ ~
  word:		~ /[0-9a-zA-Z_\/]+/ ~
  word2:        ~ /[ 0-9a-zA-Z_:\\.\\(\\)\/-]+/ ~
  str:		~ /[a-zA-Z0-9&'=:\/\\%_-]+/ ~
  punct:	~ /[ \t,;:\\.\\(\\)\\+\\*]+/ ~  
  punct2:	~ /[\t,;:\/\\+]+/ ~
  word3:	~ /[ 0-9a-zA-Z_\\"\\.\\(\\)\/\\*-]+/ ~ 
}