tab/soif.flt

   1 # Crude input-filter for SOIF records -- one record per file.
   2 # Author: Peter Valkenburg / TERENA (valkenburg@terena.nl)
   3 # Version 0.2 (09/09/1998).
   4 # This sort of follows the Nordic Web Index convention of GILS attribute use.
   5 # Modified by Kang-Jin Lee (lee@arco.de)
   6 # 07/10/1999
   7
   8 # We'll use GILS structured records.
   9 BEGIN                                   { begin record gils }
  10
  11 # URL will be GILS' availability/linkage
  12 /^@[A-Za-z](-|[.A-Za-z_])* { / BODY /$/ {
  13                                           begin element availability
  14                                           data -element linkage $1
  15                                           end element
  16                                         }
  17
  18 # Type will be GILS' availability/linkageType
  19 /^[tT]ype{[0-9]+}:\t/ BODY /$/          {
  20                                           begin element availability
  21                                           data -element linkageType $1
  22                                           end element
  23                                         }
  24
  25 # Last modification time will be Bib-1 Use Attribute 1012
  26 /^[lL]ast-[mM]odification-[tT]ime{[0-9]+}:\t/ BODY /$/  {
  27                                           data -element dateOfLastModification $1
  28                                         }
  29
  30 # The MD5 checksum is used as a unique identifier under Bib-1 Use Attribute 1007
  31 /^[mM][dD]5{[0-9]+}:\t/ BODY /$/        { data -element controlIdentifier $1 }
  32
  33 # Description will be Bib-1 Use Attribute 62
  34 /^[dD]escription{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/ {
  35                                           data -element abstract $1
  36                                           unread 2
  37                                         }
  38
  39 # Author will be Bib-1 Use Attribute 1003 (if gils.abs maps originator to it!!)
  40 /^[aA]uthor{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/    {
  41                                           data -element author $1
  42                                           unread 2
  43                                         }
  44
  45 # Keywords will be GILS' localSubjectIndex/localSubjectTerm
  46 /^[kK]eywords{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/  {
  47                                           begin element localSubjectIndex
  48                                           data -element localSubjectTerm $1
  49                                           unread 2
  50                                           end element
  51                                         }
  52
  53 # File-size will be GILS' supplementalInformation/bytes
  54 /^[fF]ile-[sS]ize{[0-9]+}:\t/ BODY /$/  {
  55                                           begin element supplementalInformation
  56                                           data -element bytes $1
  57                                           unread 2
  58                                           end element
  59                                         }
  60
  61 # Update-Time will be GILS' supplementalInformation/lastChecked
  62 /^[uU]pdate-[tT]ime{[0-9]+}:\t/ BODY /$/        {
  63                                           begin element supplementalInformation
  64                                           data -element lastChecked $1
  65                                           unread 2
  66                                           end element
  67                                         }
  68
  69 # url-references will be GILS' crossReference/linkage
  70 /^[uU]rl-[rR]eferences{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/ {
  71                                           begin element crossReference
  72                                           data -element linkage $1
  73                                           unread 2
  74                                           end element
  75                                         }
  76
  77 # Title will be Bib-1 Use Attribute 4
  78 /^[tT]itle{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/     {
  79                                           data -element Title $1
  80                                           unread 2
  81                                         }
  82
  83 # Body and Partial-Text will be Bib-1 Use Attribute 1010
  84 # Is Body really commonly used in SOIF? Anyway, Full-Text is used by Harvest.
  85 #/^[bB]ody{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/     {
  86 #                                         data -element sampleText $1
  87 #                                         unread 2
  88 #                                       }
  89 /^[fF]ull-[tT]ext{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/ {
  90                                           data -element sampleText $1
  91                                           unread 2
  92                                         }
  93 /^[pP]artial-[tT]ext{[0-9]+}:\t/ BODY /^((-|[._A-Za-z0-9])+{[0-9]+}:\t.*|})$/ {
  94                                           data -element sampleText $1
  95                                           unread 2
  96                                         }
  97
  98 /^(-|[a-zA-Z0-9])+{[0-9]+}:\t/  BODY /^((-|[_A-Za-z0-9])+{[0-9]+}:\t.*|})$/ {
  99                                           unread 2
 100                                          }
 101
 102 END                                     { end record }