path: root/debian/htdig/htdig-3.2.0b6/test/t_fuzzy

#!/bin/sh
# Part of the ht://Dig package   <http://www.htdig.org/>
# Copyright (c) 1999-2004 The ht://Dig Group
# For copyright details, see the file COPYING in your distribution
# or the GNU Library General Public License (LGPL) version 2 or later
# <http://www.gnu.org/copyleft/lgpl.html>
#
# $Id: t_fuzzy,v 1.2 2004/05/28 13:15:30 lha Exp $
#

try() {
    comment="$1"
    shift
    query="$1"
    shift
    $htsearch -c $config "$query" > $tmp
    for pattern
    do
	if grep "$pattern" $tmp > /dev/null 
	then :
	else
	    $htsearch -vv -c $config "$query" > /dev/null
	    echo "Output doesn't match \"$pattern\""
	    fail "$htsearch -c $config '$query' >> $tmp --
		  $comment"
	fi
    done
}




test_functions_action=--start-apache
. ./test_functions

config=$testdir/conf/htdig.conf.tmp
tmp=/tmp/t_htsearch$$

databases="accents metaphone soundex synonym endings_root2word endings_word2root"

# set up config file with chosen non-default values
cp $testdir/conf/htdig.conf $config
for database in $databases; do
    set_attr ${database}_db "$testdir/var/htdig/test_${database}.db"
    rm -f $testdir/var/htdig/test_${database}.db
done
set_attr synonym_dictionary "$testdir/synonym_dict"
set_attr endings_affix_file "$testdir/dummy.affixes"
set_attr endings_dictionary "$testdir/dummy.stems"

# db.words.db needed by  htfuzzy, so dig before generating fuzzy databases
#set_attr locale fr
set_attr extra_word_characters çé
$htdig "$@" -t -i -c $config
$htpurge -c $config

# Check that databases can be generated
$htfuzzy -c $config accents soundex metaphone endings synonyms || fail "Error generating fuzzy database"

# Make sure databases put in correct locations
for database in $databases; do
    if [ ! -f $testdir/var/htdig/test_${database}.db ] ; then
	fail "htfuzzy didn't recognise ${database}_db attribute"
    fi
done

try "Search for '&aacute;lso' without search_algorithm=accents" \
    "words=álso;search_algorithm=exact" \
    'No matches'

try "Search for '&aacute;lso' with search_algorithm not in  allow_in_form" \
    "words=álso;search_algorithm=accents" \
    'No matches'

set_attr allow_in_form search_algorithm

try "Search for '&aacute;lso' with search_algorithm=accents" \
    "words=álso;search_algorithm=accents:0.1" \
    '4 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html' '<strong>also</strong>'

try "Search for 'francais' without search_algorithm=accents" \
    "words=francais;search_algorithm=exact:0.1" \
    'No matches'

#try "Search for '\"fran ais\"' with search_algorithm=accents" \
#    "words=%22fran+ais%22;search_algorithm=accents:1" \
#    '1 matches' 'site4.html'

try "Search for 'francais' with search_algorithm=accents" \
    "words=francais;search_algorithm=accents:0.1" \
    '1 matches' 'site4.html'

try "Search for 'quebec' without search_algorithm=accents" \
    "words=quebec;search_algorithm=exact:0.1" \
    'No matches'

try "Search for 'quebec' with search_algorithm=accents" \
    "words=quebec;search_algorithm=accents:0.1" \
    '1 matches' 'site4.html'




try "Search for 'accownt' without search_algorithm=soundex" \
    "words=accownt;search_algorithm=exact:1" \
    'No matches'

try "Search for 'accownt' with search_algorithm=soundex" \
    "words=accownt;search_algorithm=soundex:1" \
    '2 matches' 'script.html' 'site4.html' '<strong>account</strong>'

try "Search for 'accownt' with search_algorithm=metaphone" \
    "words=accownt;search_algorithm=metaphone:1" \
    '2 matches' 'script.html' 'site4.html' '<strong>account</strong>'




try "Search for 'zzyyxx' without search_algorithm=synonyms" \
    "words=zzyyxx;search_algorithm=exact:1" \
    'No matches'

try "Checking synonyms of the form   word-in-doc word-in-query" \
    "words=zzyyxx;search_algorithm=synonyms:1" \
    '1 matches' 'site3.html'

try "Search for 'xxyyzz' without search_algorithm=synonyms" \
    "words=xxyyzz;search_algorithm=exact:1" \
    'No matches'

try "Checking synonyms of the form   word-in-query word-in-doc" \
    "words=xxyyzz;search_algorithm=synonyms:1" \
    '1 matches' 'site3.html'

try "Search for 'xyzxyz' without search_algorithm=synonyms" \
    "words=xyzxyz;search_algorithm=exact:1" \
    'No matches'

try "Checking synonyms of the form   common-word word-in-query word-in-doc" \
    "words=xyzxyz;search_algorithm=synonyms:1" \
    '1 matches' 'site3.html'

try "Checking exact is used as a default if fuzzy rules yield no word" \
    "words=road;search_algorithm=synonyms:1" \
    '1 matches' 'site4.html' '<strong>Road</strong>'


try "Searching for 'copy' without 'endings'" \
    "words=copy;search_algorithm=exact:1" \
    'No matches'

try "Searching for 'copy' with 'endings'" \
    "words=copy;search_algorithm=endings:1" \
    '2 matches' 'bad_local.htm' 'site3.html' '<strong>Copyright</strong>'



try "Searching for 'univers*' without 'prefix'" \
    "words=univers*;search_algorithm=exact:1" \
    'No matches'

try "Searching for 'univers*' with 'prefix'" \
    "words=univers*;search_algorithm=prefix:1" \
    '2 matches' 'script.html' 'site4.html'

set_attr allow_in_form "search_algorithm max_prefix_matches min_prefix_length"
set_attr prefix_match_character "?"

try "Searching for 'res?' with 'prefix'" \
    "words=res?;search_algorithm=prefix:1;max_prefix_matches=100" \
    '3 matches' 'script.html' 'site3.html' 'site4.html' 'reserved' 'residency' 'residents' 'resources' 'respectively' 'response' 'restricted'

try "Searching for 'res?' with 'prefix', at most 2 prefixes" \
    "words=res?;search_algorithm=prefix:1;max_prefix_matches=2" \
    '2 matches' 'site3.html' 'site4.html' 'reserved' 'residency)'

set_attr prefix_match_character "etc"

try "Searching for 'resetc' with 'prefix', prefix_match_character=etc" \
    "words=resetc;search_algorithm=prefix:1;max_prefix_matches=2" \
    '2 matches' 'site3.html' 'site4.html' 'reserved' 'residency)'

set_attr minimum_prefix_length 4

# Check it doesn't count the prefix_match "character" as part of the
# "minimum_prefix_length"
try "Searching for 'resetc' with 'prefix', minimum prefix length 4" \
    "words=resetc;search_algorithm=prefix:1;max_prefix_matches=4" \
    'No matches'

set_attr prefix_match_character "?"

try "Searching for 'res?' with 'prefix', minimum prefix length 4" \
    "words=res?;search_algorithm=prefix:1;max_prefix_matches=100" \
    'No matches'






try "Searching for 'ili' without 'substring'" \
    "words=ili;search_algorithm=exact:1" \
    'No matches'

try "Searching for 'ili' with 'substring'" \
    "words=ili;search_algorithm=substring:1" \
    '2 matches' 'script.html' 'site4.html' '(affiliated' 'utilised)'

set_attr allow_in_form "search_algorithm substring_max_words"

try "Searching for 'ili' with 'substring', substring_max_words=3" \
    "words=ili;search_algorithm=substring:1;substring_max_words=3" \
    '2 matches' 'script.html' 'site4.html' 'eligibility)'



try "Searching for 'acccount' without 'speling'" \
    "words=acccount;search_algorithm=exact:1" \
    'No matches'

try "Searching for 'acccount' with 'speling'" \
    "words=acccount;search_algorithm=speling:1" \
    '2 matches' 'script.html' 'site4.html' '<strong>account</strong>'

try "Searching for 'accountx' with 'speling'" \
    "words=accountx;search_algorithm=speling:1" \
    '2 matches' 'script.html' 'site4.html' '<strong>account</strong>'


set_attr allow_in_form "search_algorithm minimum_speling_length"

try "Searching for 'accountx' with minimum_speling_length=9" \
    "words=accountx;search_algorithm=speling:1;minimum_speling_length=9" \
    'No matches'

try "Searching for 'accountx' with minimum_speling_length=8" \
    "words=accountx;search_algorithm=speling:1;minimum_speling_length=8" \
    '2 matches' 'script.html' 'site4.html' '<strong>account</strong>'





try "Searching for '.*vers[^a].*' without 'regex'" \
    "words=.*vers[^a].*;search_algorithm=exact:1" \
    'No matches'

set_attr extra_word_characters ".*[^]\\$"

try "Searching for '.*vers[^a].*' with 'regex'" \
    "words=.*vers[^a].*;search_algorithm=regex:1" \
    '3 matches' '(universities' 'versions)'

set_attr allow_in_form "search_algorithm regex_max_words"

try "Searching for '.*vers[^a].*' with 'regex', regex_max_words=3" \
    "words=.*vers[^a].*;search_algorithm=regex:1;regex_max_words=3" \
    '2 matches' '(universities' 'version)'




try "Searching for 'versi' without 'substring'" \
    "words=versi;search_algorithm=exact:1" \
    'No matches'

try "Searching for 'versi' with 'substring'" \
    "words=versi;search_algorithm=substring:1" \
    '3 matches' '(universities' 'versions)'

set_attr allow_in_form "search_algorithm substring_max_words"

try "Searching for 'versi' with 'substring', substring_max_words=3" \
    "words=versi;search_algorithm=substring:1;substring_max_words=3" \
    '2 matches' '(universities' 'version)'





for database in $databases; do
    rm -f $testdir/var/htdig/test_${database}.db
    try "Searching width $database database missing" \
        "words=account" \
	'2 matches'
done

test_functions_action=--stop-apache
. ./test_functions