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

#
# 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_parsing,v 1.4 2004/05/28 13:15:30 lha Exp $
#


# Tests (or should eventually test) the following config attributes:
#	description_meta_tag_names
#	ignore_alt_text
#	max_doc_size
#	max_keywords
#	max_meta_description_length
#	max_description_length
#	max_descriptions
#	max_head_length
#	noindex_end
#	noindex_start
#	external_parsers
#	external_protocols
#	use_meta_description


test_functions_action=--start-apache
. ./test_functions

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

# set up config file with chosen non-default values
cp $testdir/conf/htdig.conf $config

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
}


# Tests (or should eventually test) the following config attributes:
#	description_meta_tag_names
#	ignore_alt_text
#	max_doc_size
#	max_keywords
#	max_meta_description_length
#	max_description_length		(May put in t_templates)
#	max_descriptions		(May put in t_templates)
#	max_head_length
#	noindex_end
#	noindex_start
#	external_parsers		(TODO)
#	external_protocols
#	use_meta_description

$htdig "$@" -t -i -c $config	|| fail "Couldn't do first dig"
$htpurge -c $config		|| fail "Couldn't do first purge"

try "Search for alt text 'earth'" \
    "words=earth" \
    '1 matches' 'site3.html'

try "'claims and collections', unlimited doc size" \
    "words=%22claims+and+collections%22" \
    '1 matches' 'site4.html'

try "Search for keyword 'martial', default max_keywords" \
    "words=martial" \
    '1 matches' 'site2.html'

try "Search for 'service', default noindex_start/end" \
    "words=technical" \
    '1 matches' 'site%201.html'

set_attr use_meta_description true
try "Search for 'call handling' with default max_meta_description_length" \
    "words=%22call+handling%22" \
    '1 matches' 'script.html' 'call handling.*signalling'

set_attr ignore_alt_text true
set_attr max_doc_size 15112
set_attr max_keywords 5
set_attr noindex_start "'Software Distribution'"
set_attr noindex_end "'Contact Information'"
set_attr max_meta_description_length 80
set_attr description_meta_tag_names "description generator"
set_attr max_head_length 30

$htdig "$@" -t -i -c $config	|| fail "Couldn't do second dig"
$htpurge -c $config		|| fail "Couldn't do second purge"

try "Search for alt text 'earth' with ignore_alt_text=true" \
    "words=earth" \
    'No matches'

try "'claims and collections', max_doc_size 15112" \
    "words=%22claims+and+collections%22" \
    '1 matches' 'site4.html'

# (Martial is 6th keyword listed in site 2, but "Fu" is too short and omitted.)
try "Search for keyword 'martial', max_keywords = 5" \
    "words=martial" \
    'No matches'

# Only occurrence of "technical" is between noindex_start and _end in  site 1
try "Search for 'technical', noindex_start=Software Distribution, noindex_end=Contact Information" \
    "words=technical" \
    'No matches'

# Visitor occurs after  noindex_end
try "Search for 'visitor', noindex_start=Software Distribution, noindex_end=Contact Information" \
    "words=visitor" \
    '2 matches' 'site%201.html' 'site3.html'

# Displaying meta description instead of excerpt, check it is truncated
try "Search for 'call handling' with max_meta_description_length=80" \
    "words=%22call+handling%22" \
    '1 matches' 'script.html' 'means of<br>'

# Check <meta name="generator"...> counts as a description
try "Search for 'category', description_meta_tag_names includes 'generator'" \
    "words=category" \
    '1 matches' 'site3.html' 'FrontPage'

# Check that only specified number of bytes of header is stored.
# Header size is rounded up to contain the whole of the last word.
try "Search for 'also', max_head_length=30" \
    "words=also" \
    '4 matches' 'bad_local.htm' 'site2.html' 'script.html' 'site4.html' \
    'WHERE.*Copyright<br>'

set_attr max_doc_size 15042
set_attr max_keywords 6
set_attr noindex_start "'software distribution'"
set_attr noindex_end "'contact information'"

$htdig "$@" -t -i -c $config	|| fail "Couldn't do third dig"
$htpurge -c $config		|| fail "Couldn't do third purge"

try "Search for keyword 'martial', max_keywords = 6" \
    "words=martial" \
    '1 matches' 'site2.html'

try "'claims and collections', max_doc_size 15042" \
    "words=%22claims+and+collections%22" \
    'No matches'

# Check noindex_start/end are case-insensitive
try "Search for 'technical', noindex_start=software distribution, noindex_end=contact information" \
    "words=technical" \
    'No matches'

PROTOCOL=my-protocol
echo '#!/bin/sh
      echo "s	200"
      echo "t	text/html"
      echo
      echo "<html>$2</html>"' > $PROTOCOL
chmod 755 $PROTOCOL
set_attr external_protocols "echo: $PWD/$PROTOCOL"
set_attr start_url "echo:foo.html"
$htdig "$@" -t -i -c $config	|| fail "Couldn't do fourth dig"
try "trying external protocol  echo" \
    "words=foo" \
    "1 matches" "echo:foo.html"


test_functions_action=--stop-apache
. ./test_functions

rm -f $tmp $PROTOCOL

exit 0