#!/bin/sh 

# standard sh program for building queryable MG collections

# set no_tags to -G to exclude SGML tags from index
no_tags=

# change stem_case to change stemming/casefold:
# -s 0 = no casefold or stemming
# -s 1 = casefold, no stemming
# -s 2 = stemming, no casefold
# -s 3 = stemming and casefold
stem_case=1

# if MG 1.2 is located somewhere else, change this appropriately
PATH=/usr/local/mg-1.2/bin:$PATH

# set source to collection text generation command + arguments
source=

# set text to name of collection
text=

# change dir to explicit directory of collection
# if script is not in the same dir as the collection
# or if script may not be executed from the collection directory
dir=.

# make generated files have 777 permission so
# future users can regenerate the collection w/o 
# having to chown or chmod

umask 000 

set -ex  # exit if any of the following fail, print commands

cd ${dir} 
touch ${text}.rebuilding 

${source} | mg_passes ${no_tags} -s ${stem_case} -T1 -I1 -t 1 -d . -f ${text} 
mg_compression_dict -d . -f ${text} 
mg_perf_hash_build -d . -f ${text} 
${source} | mg_passes ${no_tags} -s ${stem_case} -T2 -I2 -t 1 -d . -f ${text} 
mg_weights_build -d . -f ${text} 
mg_invf_dict -d . -f ${text} 
mg_fast_comp_dict -d . -f ${text} 

rm ${text}.rebuilding 

# remove files not needed by mgquery
rm ${text}.text.dict
rm ${text}.invf.dict
rm ${text}.invf.idx
rm ${text}.text.stats
rm ${text}.text.idx
rm ${text}.weight
rm ${text}.invf.chunk
rm ${text}.invf.chunk.trans
rm ${text}.invf.dict.hash
rm ${text}.trace

# create .lastbuild file for future reference
echo $0 $@ > ${text}.lastbuild