!~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
!    $Source: pir-ascii.sdl $
!    $Revision: 1.0 $
!    $Date: 1995/04/14 $
!    $Author: d.gilbert $
!
!~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

!! this one is written to allow wais-type searches as nearly as possible
!! no phrases, only words are indexed.

#library /id=%PIR_DB 
    /group=@SEQUENCE_LIBS /name=PIR /logicname=PIR /format=@PIRA_FORMAT  
    /help="pir" /comment="Protein Identification Resource"
    /oddfile="pir-ascii.sdl" /cachesize=32  /maxIndexSizeKb=1500
{
!    #file /name=pironly
    #file /name=pir1 /logicname=PIR1
    #file /name=pir2 /logicname=PIR2
    #file /name=pir3 /logicname=PIR3
}




#libformat /id=%PIRA_FORMAT /parser=@PIR_PARSER
    /file_type=@PIRAREF_FILE, @PIRASEQ_FILE
!    /featureExpression=@EMBL_FEATURE_EXPRESSION
{
    !ID
    #field /gblid=%PIR_ID_FIELD
           /itype=ID /ftype=@DF_ID /idtype=@SRSxSEQID
           /begstr="ENTRY " /find=id /maxlines=1
    
    !all flds
    #field /itype=group /ftype=@DF_ALL /idtype=@SRSxSEQID
           
    !accession
    #field /gblid=%PIR_ACC_FIELD
           /itype=key /ftype=@DF_ACCESSION /find=accession /idtype=@SRSxSEQID
           /begstr="ACCESSIONS " /nextstr="   " 
           
    !definition
    #field /itype=key /ftype=@DF_DEFINITION /idtype=@SRSxSEQID
           /begstr="TITLE "  /nextstr="   "  /find=definition 

    !organism
    #field /itype=key /ftype=@DF_ORGANISM /idtype=@SRSxSEQID
           /find=species /begstr="ORGANISM " /nextstr="       " 
           
    !date
    #field /itype=num /ftype=@DF_DATE /idtype=@SRSxSEQID
           /begstr="DATE "  /nextstr="    "  /find=dates 

    !keywords
    #field /itype=key /ftype=@DF_KEYWORDS /idtype=@SRSxSEQID
           /begstr="KEYWORDS","CLASSIFICATION","GENETICS"
           /nextstr=" " /find=keywords
           
    !!authors
    !!#field /gblid=%PIR_AUT /itype=key /ftype=@DF_AUTHORS  /idtype=@SRSxSEQID
    !!       /begstr="   #authors" /nextstr="    " 
    !!       ! /begleft=4 /begright=4 
    !!       /find=authors
    !title
    !!#field /itype=key /ftype=@DF_TITLE /idtype=@SRSxSEQID
    !!       /begstr="   #title" /nextstr="     " 
    !!       ! /begleft=4 /begright=4 
    !!       /find=title 

    !reference
    #field /itype=key /ftype=@DF_REF /idtype=@SRSxSEQID  
           /begstr = "REFERENCE" /nextstr= "   "
           !/begstr= "#title ","#authors " /begleft=4 /begright=4  /nextstr="    " 
           /find=reference 
           !/suppress_error=yes

    !comment
    #field /itype=key /ftype=@DF_COMMENT /idtype=@SRSxSEQID
           /begstr="COMMENT"  /nextstr=" " /find=keywords

    !features
    #field !/gblid=%PIR_FEATURES_FIELD
           /itype=key /ftype=@DF_FEATURES /idtype=@SRSxSFTID
           /begstr="FEATURE"  /nextstr=" " 
           /find=feature  
           
    !sequence length
    #field /itype=num /ftype=@DF_SEQLENGTH /idtype=@SRSxSEQID /find="" 
}


#library /id=%PIRALN_DB 
    /group=@SEQRELATED_LIBS
    /name=PIRALN /logicname=PIRALN /format=@PIRALN_FORMAT  
    /comment="Database of Protein Sequence Alignments"
    /oddfile="pir.sdl" /cachesize=32 /maxIndexSizeKb=1500
{
    #file /name=aln
}


#libformat /id=%PIRALN_FORMAT /parser=@PIR_PARSER
           /file_type=@PIRAREF_FILE
{
    ! fix later
    ! id
    #field /gblid=%PIRALN_ID_FIELD 
           /itype=ID /ftype=@DF_ID /idtype=@ENTRY_ID
           /begstr=">" /find=id /maxlines=1

    ! definition
    #field /itype=key /ftype=@DF_DEFINITION /idtype=@ENTRY_ID
           /after=@PIRALN_ID_FIELD /find=definition /maxlines=1   

    !member references
    #field /gblid=%PIRALN_LINK_FIELD
           /itype=link /ftype=@DF_LINK /idtype=@SRSxLINKID
           /begstr="C;Members:", "C;MEMBERS:" /maxlines=1 /find=alnmembers
    
    #readlink /id=%PIRALN_PIR_REF /link=@PIRALN_PIR_LINK
} 

#link /id=%PIRALN_PIR_LINK
      /lib1=@PIRALN_DB /lib2=@PIR_DB
      /field1=@PIRALN_ID_FIELD /field2=@PIR_ID_FIELD
      /idtype1=@ENTRY_ID /idtype2=@SRSxSEQID



#library /id=%NRL3D_DB 
    /group=@SEQUENCE_LIBS /name=NRL3D /logicname=NRL /format=@NRL3D_FORMAT  
    /comment=" NRL_3D protein sequence structure database"
    /cachesize=32 /oddfile="pir.sdl"

{
    #file /name=nrl_3d
}


#libformat /id=%NRL3D_FORMAT /parser=@PIR_PARSER
           /file_type=@PIRAREF_FILE, @PIRASEQ_FILE
{
    ! id
    #field /gblid=%NRL3D_ID_FIELD 
           /itype=ID /ftype=@DF_ID /idtype=@SRSxSEQID
           /begstr="ENTRY" /find=id /maxlines=1
    
    !all flds
    #field /itype=group /ftype=@DF_ALL /idtype=@SRSxSEQID
           
    !definition
    #field /itype=key /ftype=@DF_DEFINITION /idtype=@SRSxSEQID
           /begstr="TITLE "  /nextstr="   "  /find=definition 

    !organism
    #field /itype=key /ftype=@DF_ORGANISM /idtype=@SRSxSEQID
           /find=species /begstr="ORGANISM " /nextstr="       " 
           
    !keywords
    #field /itype=key /ftype=@DF_KEYWORDS /idtype=@SRSxSEQID
           /begstr="KEYWORDS" /nextstr=" " /find=keywords

    !reference
    #field /itype=key /ftype=@DF_REF /idtype=@SRSxSEQID  
           /begstr = "REFERENCE" /nextstr= "   "
           !/begstr= "#title ","#authors " /begleft=4 /begright=4  /nextstr="    " 
           /find=reference 
           !/suppress_error=yes

    !comment
    #field /itype=key /ftype=@DF_COMMENT /idtype=@SRSxSEQID
           /begstr="COMMENT"  /nextstr=" " /find=keywords

    !features
    #field !/gblid=%PIR_FEATURES_FIELD
           /itype=key /ftype=@DF_FEATURES /idtype=@SRSxSFTID
           /begstr="FEATURE"  /nextstr=" " 
           /find=feature  
           
    #readlink /id=%NRL3D_PDB_REF /link=@NRL3D_PDB_LINK
} 

#link /id=%NRL3D_PDB_LINK
      /lib1=@NRL3D_DB /lib2=@PDB_DB
      /field1=@NRL3D_ID_FIELD /field2=@PDB_ID_FIELD
      /idtype1=@SRSxSEQID /idtype2=@STRUCTENT_ID


ASSIGN %%PIR_SFT_KEY		10
ASSIGN %%PIR_SFT_FLD		11
ASSIGN %%PIR_SFT_LOC		12
ASSIGN %%PIR_SFT_LOCEND	13


#PARSER /id=%PIR_PARSER /newline="\n" /ignore=" \r"
        /convert=upper /#BNF = 
 noword     = ~#*\\\'\"\-;.:()[]=,/~;
 word       = ~#*\\\'\"\-;.:()[]=,/ ~ <not>;    
 name       = ^A-Za-z0-9$^ ~A-Za-z0-9$_\-~ <noskip>;
 id         = name <wrt>;
 key        = word;
 accnr      = ~A-Z0-9~;
 accession  = accnr <new c=1> [ ';' accession];
 definition = { noword | word <new> };
 fldname    = ~#~ name;

 organizm   = ~A-Za-z0-9_$\-~ <new>;
 org        = [ fldname ] { organizm } [ fldname { organizm } ]; 
 species    = org <c=1>;

 !keyword    = ~;.~ <new not>;
 !keywords   = [ fldname ] keyword { ';' [ fldname ] keyword };
 nokword     = ~*\\\'\"\-,.:()[]=/~;
 kword       = ~#*\\\'\"\-,.:()[]=/ ;~ <new not>;    
 keywords    = [fldname] {kword | nokword} { ';' [fldname] {kword | nokword} };

!
! date
!
 adate      = ~0-9~ <wrt> '-' <app> ~A-Za-z~ <app> '-' <app> ~0-9~ <app>;
 dates      = [ fldname ] adate { fldname adate };

!
!literature reference
!
 refwords   = { (fldname | noword | word <new>) };
 reference  = refwords;
  
!
!parsing feature tables
!
 nofword    = ~*\'\"\-\!;.:()[]=,\/ ~;
 fword      = ^A-Za-z^ [~A-Za-z0-9_\-~ <noskip>];
 !fword      = ~#*\\\'\"\-,;.:()[]=/ ~ <not>;    
 number     = ~0-9~;
 ftend      <wrt c=@PIR_SFT_LOCEND> = number;
 ftbeg      <wrt c=@PIR_SFT_LOC> = number;
 ftkey      <wrt c=@PIR_SFT_KEY> = fword;
 loc        = ftbeg [ '-' ftend ];
 ftstring   = ~#~ name <noskip wrt c=@PIR_SFT_FLD> {ftkey | nofword} ;
 !!feature    = loc { ',' loc } { ftstring } [ ~\\~ loc { ',' loc } { ftstring }] ;
 ftline     = loc { ',' loc } { ftstring };
 feature    = ftline [ '\\' ftline ] ;

!
! other things
!
 pdblink    = ~0-9A-Z~ [';''PDB:' ~0-9A-Z~ <l=4 wrt c=@NRL3D_PDB_REF>];
 alnmembers = alnmember {';' alnmember};
 alnmember  = ~A-Za-z0-9~ <wrt c=@PIRALN_PIR_REF> ['(' ~)~ <not> ')'].    
!end parser



#filetype /ID=%PIRAREF_FILE 
          /typename="dat" /maxline=200  /exitstr="SEQUENCE"
#filetype /ID=%PIRASEQ_FILE 
          /typename="dat" /maxline=200 
          /seqtype=codata ! SLBxCODATA, dgg addition
          /begstr="SEQUENCE" /advance=1 /exitstr="///"