extra-dependencies/debian/htdig/htdig-3.2.0b6/test/t_htsearch

#
# 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_htsearch,v 1.14 2004/05/28 13:15:30 lha Exp $
#

test_functions_action=--start-apache
. ./test_functions

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

$htdig "$@" -t -i -c $config
$htpurge -c $config

test_functions_action=--stop-apache
. ./test_functions

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

try "Simple search for 'also'" \
    "words=also" \
    '4 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html'

try "Implicit and search with two words 'also movies'" \
    "method=and&words=also+movies" \
    '1 match' 'site2.html'

try "Explicit and search with two words 'also movies'" \
    "method=boolean&words=also+and+movies" \
    '1 match' 'site2.html'

try "Implicit or search for 'also distribution'" \
    "method=or&words=also+distribution" \
    '5 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html' 'site%201.html'

try "Explicit or search for 'also distribution'" \
    "method=boolean&words=also+or+distribution" \
    '5 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html' 'site%201.html'

try "Boolean and/or mixed 'also or distribution and ltd'" \
    "method=boolean&words=also+or+distribution+and+ltd" \
    '5 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html' 'site%201.html'

try "Boolean explicit priority '(also or distribution) and ltd'" \
    "method=boolean&words=(also+or+distribution)+and+ltd" \
    '2 matches' 'script.html' 'site%201.html'

try "Wildcard search for '*'" \
    "method=and&words=*" \
    '10 matches' 'set1/bad_local.htm' 'set1/title.html' 'set1/site2.html' 'set1/script.html' 'set1/"' 'set1/site3.html' 'set1/site4.html' 'set1/site%201.html' 'set1/sub%2520dir/' 'set1/sub%2520dir/empty%20file.html'

try "Phrase search for '\"who offer\" and \"loans to graduate\"'" \
    "method=boolean&words=%22who+offer%22+and+%22loans+to+graduate%22" \
    '1 match' 'site4.html'

try "Phrase search for 'may be deferred' -- stop words at start" \
    "method=boolean&words=%22may+be+deferred%22" \
    '1 match' 'site4.html' '<strong>may be deferred</strong>'

try "Phrase search for 'Repayment may be' -- stop words at end" \
    "method=boolean&words=%22Repayment+may+be%22" \
    '1 match' 'site4.html' '<strong>Repayment may be</strong>'

# This test fails
#
#try "Phrase search for 'may be' -- all stop words" \
#    "method=boolean&words=%22may+be%22" \
#    '1 match' 'site4.html'

try "Phrase search for 'Repayment may be deferred' -- stop words in middle" \
    "method=boolean&words=%22Repayment+may+be+deferred%22" \
    '1 match' 'site4.html' '<strong>Repayment may be deferred</strong>'

try "Phrase search for 'Repayment deferred' -- phrase interrupted by stop words" \
    "method=boolean&words=%22Repayment+deferred%22" \
    'No match'

try "Syntax error: finishing with 'and'" \
    "method=boolean&words=also+and" \
    'Expected a search word, a quoted phrase or a boolean expression between () at the end'

try "Syntax error: excess open brackets" \
    "method=boolean&words=(also+or+distribution" \
    "Expected ')' at the end"

try "Syntax error: excess close brackets" \
    "method=boolean&words=also+or+distribution)" \
    "Expected end of expression instead of ')'"

try "Syntax error: missing end quote" \
    "method=boolean&words=also+or+%22distribution" \
    'Expected quotes at the end'

try "Unrestricted search for 'group'" \
    "method=and&words=group" \
    '4 matches' 'script.html' 'bad_local.htm' 'site3.html' 'site4.html'

try "Field-restricted search for 'author:group'" \
    "method=and&words=author:group" \
    '1 match' 'script.html'

try "Field-restricted search for 'text:group'" \
    "method=and&words=text:group" \
    '3 matches' 'bad_local.htm' 'site3.html' 'site4.html'

try "Checking prefix parsing using 'text: group'" \
    "method=and&words=text:%20group" \
    '1 match' 'script.html'

try "Checking prefix parsing using 'text::group'" \
    "method=and&words=text::group" \
    '1 match' 'script.html'

try "Checking prefix parsing using 'unknown:group'" \
    "method=any&words=unknown:group" \
    '5 matches' 'script.html' 'bad_local.htm' 'site3.html' 'site4.html' 'set1/"'

try "Field-restricted search for 'descr:cost'" \
    "method=and&words=descr:cost" \
    '1 match' 'script.html'

config=$testdir/conf/htdig.conf3

try "Testing boolean_keywords and search_rewrite_urls" \
    "method=boolean&words=also+ou+distribution+et+ltd" \
    '5 matches' 'bad_local.htm' 'place2.html' 'script.html' 'place4.html' 'place%201.html'

try "Testing boolean_syntax_errors" \
    "method=boolean&words=ou+distribution" \
    "Attendait un mot au lieu de '|' ou 'ou'"

try "Testing htdig's noindex_start/end overlapping" \
    "words=considered" \
    '1 match' 'script.html'

try "Testing htdig's noindex_start/end nested" \
    "words=neglected" \
    'No match'

rm -f $tmp

exit 0