#                               -*- Mode: Sh -*- 
# test.fmt -- 
# ITIID           : $ITI$ $Header $__Header$
# Author          : Ulrich Pfeifer
# Created On      : Fri Oct  6 13:27:35 1995
# Last Modified By: Ulrich Pfeifer
# Last Modified On: Sun Apr 21 21:52:13 1996
# Language        : FMW
# Update Count    : 26
# Status          : Unknown, Use with caution!
# 
# (C) Copyright 1995, Universität Dortmund, all rights reserved.
# 
# $Locker:  $
# $Log: testg.fmt,v $
# Revision 1.1  1997/02/14 15:04:30  pfeifer
# Sources for the two test databases.
#
# Revision 2.2  1997/02/04 17:10:42  pfeifer
# Switched to CVS
#
# Revision 2.0.1.5  1996/04/29 15:41:12  pfeifer
# patch66: Fixed region end for numeric indexing.
#
# Revision 2.0.1.4  1995/11/17  14:50:38  pfeifer
# patch52: Indexing for py was erroneous.
#
# Revision 2.0.1.3  1995/11/03  09:46:15  pfeifer
# patch43: Fixed off-by-one bug in headlines.
# patch43: Fixed off-by-one bug in skip expressions.
#
# Revision 2.0.1.2  1995/10/16  12:29:24  pfeifer
# patch38:
#
# Revision 2.0.1.1  1995/10/06  17:19:55  pfeifer
# patch24: The format description demonstrated now the use of the date
# patch24: field. Also a new syntax without the nasty '<...>' is
# patch24: shown. Added some hopefully useful comments.
#
# 

# Remember: indexing and matching is done line by line. Each regular
# expression can match one line maximum.

# This regular expression should match the line separating documents.
# The matching line is NOT indexed!  Use /\n\n/ if each file contains
# only one record.

record-sep: /\f/    # formfeed (alias for \L)

# Now we specify how the headline is constructed.  The date is part of
# the headline and not searchable.
layout:

# First me map the regions starting with /^PY: / up to (and excluding)
# the next /^[A-Z][A-Z]:/ skipping everything from the beginning up after
# the end of /^PY: */ to the first 4 characters of the headline.

  headline: /^PY: / /^[A-Z][A-Z]:/ 4 /^PY: */
  headline: /^AU: / /^[A-Z][A-Z]:/ 20 /AU: */
  headline: /^TI: / /^[A-Z][A-Z]:/ 40 /TI: */

# The date is found on lines matching /^ED: /. The line is kipped up
# to the end of /^ED: *./. The line is handled to scanf with format
# string "%d.%d.%d". Day, month, and year are expected in this
# order. Month may be a string containing the 3 letter abreviation for
# the name of the month. In this case the keyword "string" must follow
# "month"

  date: /^ED: / /%d.%d.%d/ day month year /^ED: *./

# Now we define the first region. It starts at /^CK:/. everything up
# after the end of /^CK: */ is skipped. Region ends at
# /^[A-Z][A-Z]:/. The match for end is search BEHIND the match for the
# begin.
region: /^CK:/ /^CK: */

# All words are transfered in the 'ck' index and the global
# index. Description of 'ck' is overwritten by "Citation Code".

ck "Citation Code" TEXT BOTH
end: /^[A-Z][A-Z]:/

# All words in this region will be numeric. They are merge in the 'py'
# index which will be tagged as nummeric from here on (thats not how
# it should be - I know). Make sure that you do not pollute 'ck' in
# the following with non-numeric data.  The "numeric" keyword has it's
# own skip expression /^PY: [^ ]/ which overwrites the region skip
# expression. This way you can index the same region nonnumeric (This
# not ...).

region: /^PY: / /^PY: */
py "Publication Year" numeric /^[^0-9]*/ 4 TEXT LOCAL
end: /[^0-9]/

# This region will be indexed plain for the global index and the 'au'
# index. The 'au' index will also contain the soundex codes for the
# words.

region: /^AU: / /^AU: */
au "Author" SOUNDEX LOCAL TEXT BOTH
end: /^[A-Z][A-Z]:/

# All words in theis reagion will be stemmend before insertion in the
# global and the 'ti' index'

region: /^TI: / /^TI: */
ti "Title" stemming TEXT BOTH
end: /^[A-Z][A-Z]:/

# This region is indexed for multiple local and the global index.
# Stemming applies for the 'ti' index and the global index.

region: /^JT: / /^JT: */
jt "Journal Title" ti stemming TEXT BOTH
end: /^[A-Z][A-Z]:/

# Here we make the data searchable. Format is merely the same as in
# the layout section. There is no local skip expression. That wasnt a
# good idea for numerics anyway. The date can be search with the
# yymmdd format. See test DATE.

region: /^ED: / /^ED: */
ed "Date of insertion" <date> /%d.%d.%d/ day month year TEXT LOCAL
end: /$/

region: /^JT: / /^JT: */
TEXT GLOBAL
end: /^[A-Z][A-Z]:/