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

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

# Tests the following config attributes:
#	add_anchors_to_excerpt
#	allow_in_form
#	anchor_target
#	any_keywords
#	build_select_lists
#	date_format
#	end_ellipses
#	end_highlight
#	excerpt_show_top
#	image_url_prefix
#	iso_8601
#	matches_per_page
#	matches_per_page_list
#	max_excerpts
#	max_descriptions
#	max_description_length
#	max_stars
#	maximum_page_buttons
#	maximum_pages
#	method_names
#	next_page_text
#	no_excerpt_show_top
#	no_excerpt_text
#	no_page_list_header
#	no_page_number_text
#	no_next_page_text
#	no_prev_page_text
#	no_title_text
#	nothing_found_file
#	page_list_header
#	page_number_separator
#	page_number_text
#	prev_page_text
#	plural_suffix
#	search_results_contenttype
#	search_results_wrapper
#	sort_names
#	star_blank	
#	star_image	
#	star_patterns
#	start_ellipses
#	start_highlight
#	syntax_error_file
#	use_star_image


try() {
    comment="$1"
    shift
    query="$1"
    shift
    $htsearch -c $config "$query" > $tmp 2> /dev/null
    for pattern
    do
	if grep "$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$$

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

$htdig "$@" -t -i -c $config	|| fail "Couldn't dig"

$htpurge -vv -c $config > tmp1		|| fail "Couldn't purge"

# How can I check that unretrieved urls have been removed, but bad ones haven't?


set_attr search_results_wrapper "$testdir/conf/main-template"
set_attr any_keywords "true"
set_attr image_url_prefix "image-prefix"

try ".  Single page of results" \
    "words=also;config=hello;exclude=site2;format=builtin-short;keywords=subject+please+node;matchesperpage=10;method=and;page=1;restrict=http;sort=score;startyear=1999;startmonth=8;startday=1;endyear=2005;endmonth=9;endday=30" \
    'MATCHES *=3' 'bad_local.htm' 'script.html' 'site4.html' \
    'CGI *=-' 'CONFIG *=hello' 'EXCLUDE *=site2' 'FIRSTDISPLAYED *=1' \
    'builtin-short" selected>Short' 'KEYWORDS *=subject please node' \
    'LASTDISPLAYED *=3' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=10' 'MATCH_MESSAGE *=All' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>All' 'MODIFIED *=-' \
    'NEXTPAGE *=-' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=-' \
    'PAGELIST *=-' 'PAGES *=1' 'PLURAL_MATCHES *=s' \
    'PREVPAGE *=-' 'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-short' \
    'SELECTED_METHOD *=and' 'SELECTED_SORT *=score' \
    'selected>Score' 'VERSION *=3.2' 'WORDS *=also' \
    'image-prefix/star.gif'


set_attr plural_suffix "es"
set_attr allow_in_form "script_name"
set_attr page_list_header "head"
set_attr add_anchors_to_excerpt "true"
# make sure "script" is the page returned...
set_attr search_results_order	"bad_local script *"
# set_attr iso_8601 FALSE	# no point -- format set by locale

# (Ensure keyword 'moderate' for script.html occurs after 'also', as anchors
# are broken if keyword preceeds anchor, but search term follows it.)
try "2nd page, 1 match per page, no excludes" \
    "words=also;script_name=qtest;format=builtin-long;keywords=subject+please+moderate;matchesperpage=1;method=or;page=2;restrict=http;sort=revtime;startyear=1999;startmonth=8;startday=1;endyear=2005;endmonth=9;endday=30" \
    'MATCHES *=3' 'script.html' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' 'FIRSTDISPLAYED *=2' \
    'builtin-long" selected>Long' 'KEYWORDS *=subject please moderate' \
    'LASTDISPLAYED *=2' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=1' 'MATCH_MESSAGE *=Any' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>Any' 'MODIFIED *=-' \
    'NEXTPAGE *=.*;page=3' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=head' \
    'PAGELIST *=.*;page=1.*;page=3' 'PAGES *=3' 'PLURAL_MATCHES *=es' \
    'PREVPAGE *=.*;page=1' 'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revtime' \
    'selected>Reverse Time' 'VERSION *=3.2' 'WORDS *=also' \
    '<strong><code>\.\.\. </code></strong>' '<strong><a href="[^"]*">also</a></strong>' \

set_attr method_names "or Or and And boolean Logical"
set_attr    page_list_header 'multiple'
set_attr no_page_list_header 'single'
set_attr anchor_target "body"
set_attr maximum_page_buttons "2"
set_attr no_next_page_text "none"
set_attr no_prev_page_text "none"
set_attr next_page_text "following"
set_attr prev_page_text "preceeding"
set_attr iso_8601 true

try "2nd page, 1 match per page, no excludes, no keywords" \
    "words=also;script_name=qtest;format=builtin-long;matchesperpage=1;method=or;page=2;restrict=http;sort=revtime;startyear=1999;startmonth=8;startday=1;endyear=2005;endmonth=9;endday=30" \
    'MATCHES *=4' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' 'FIRSTDISPLAYED *=2' \
    'builtin-long" selected>Long' 'KEYWORDS *=-' \
    'LASTDISPLAYED *=2' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=1' 'MATCH_MESSAGE *=Or' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>Or' 'MODIFIED *=-' \
    'NEXTPAGE *=.*;page=3.*following' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=multiple' \
    'PAGELIST *=.*;page=1.*2[^0-9]*-' 'PAGES *=4' 'PLURAL_MATCHES *=es' \
    'PREVPAGE *=.*;page=1.*preceeding' 'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revtime' \
    'selected>Reverse Time' 'VERSION *=3.2' 'WORDS *=also' \
    '<strong><code>\.\.\. </code></strong>' '<strong><a target="body" href="[^"]*">also</a></strong>' \
    '20[0-9][0-9]-[0-9]*-[0-9]* [0-9]*:[0-9]*:[0-9]* [a-zA-Z]*' # iso date

# ('\\\\' because original line in htdig.conf has a continuation '\' )
set_attr page_number_text "first second third fourth \\\\"
set_attr no_page_number_text "FIRST SECOND THIRD FOURTH \\\\"
set_attr page_number_separator "PAGE_SEP"
set_attr maximum_pages 2
set_attr date_format "%Y-date-%m-date-%d-"	# overrides iso_8601

try "2nd page, 1 match per page, changed page numbers" \
    "words=also;script_name=qtest;format=builtin-long;matchesperpage=1;method=or;page=2;restrict=http;sort=revtime;startyear=1999;startmonth=8;startday=1;endyear=2005;endmonth=9;endday=30" \
    'MATCHES *=4' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' 'FIRSTDISPLAYED *=2' \
    'builtin-long" selected>Long' 'KEYWORDS *=-' \
    'LASTDISPLAYED *=2' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=1' 'MATCH_MESSAGE *=Or' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>Or' 'MODIFIED *=-' \
    'NEXTPAGE *=none' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=multiple' \
    'PAGELIST *=.*first.*PAGE_SEP.*SECOND' 'PAGES *=2' 'PLURAL_MATCHES *=es' \
    'PREVPAGE *=.*;page=1' 'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revtime' \
    'selected>Reverse Time' 'VERSION *=3.2' 'WORDS *=also' \
    '<strong><code>\.\.\. </code></strong>' '<strong><a target="body" href="[^"]*">also</a></strong>' \
    '20[0-9][0-9]-date-[0-9]*-date-[0-9]*-'

#'MATCHES *=4' 'bad_local.htm' 'script.html' 'site2.html' 'site4.html'

set_attr any_keywords "false"
set_attr nothing_found_file "$testdir/conf/main-template"
set_attr add_anchors_to_excerpt "false"
try "fails keyword match" \
    "words=also;script_name=qtest;format=builtin-long;keywords=subject+please+node;matchesperpage=1;method=or;page=2;restrict=http;sort=revtime;startyear=1999;startmonth=8;startday=1;endyear=2005;endmonth=9;endday=30" \
    'MATCHES *=0' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' \
    'builtin-long" selected>Long' 'KEYWORDS *=subject please node' \
    'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=1' 'MATCH_MESSAGE *=Or' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>Or' 'MODIFIED *=-' \
    'NSTARS *=-' 'PAGELIST *=-' 'PAGES *=1' \
    'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revtime' \
    'selected>Reverse Time' 'VERSION *=3.2' 'WORDS *=also' \
    'starting date   =1999/8/1-' \
    'ending date     =2005/9/30-'

set_attr start_ellipses  "START_ELLIPSES"
set_attr end_ellipses    "END_ELLIPSES"
set_attr star_blank	 "NIL"
set_attr star_image	 "STAR"
set_attr matches_per_page 8

try "test highlighting" \
    "words=also;script_name=qtest;format=builtin-long;method=or;restrict=http;sort=revscore" \
    'MATCHES *=4' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' 'FIRSTDISPLAYED *=1' \
    'LASTDISPLAYED *=4' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=8' 'MATCH_MESSAGE *=Or' 'MAX_STARS *=4' \
    'METADESCRIPTION *=-' 'selected>Or' 'MODIFIED *=-' \
    'NEXTPAGE *=-' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=single' \
    'PAGELIST *=-' 'PAGES *=1' 'PLURAL_MATCHES *=es' \
    'PREVPAGE *=-' 'RESTRICT *=http' 'SELECTED_FORMAT *=builtin-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revscore' \
    'selected>Reverse Score' 'VERSION *=3.2' 'WORDS *=also' \
    'START_ELLIPSES' 'END_ELLIPSES' '<strong>also</strong>' \
    '^[^S]*"STAR".*"NIL".*"NIL".*"NIL"'

set_attr max_stars 5
set_attr start_highlight "START"
set_attr end_highlight   "STOP"
set_attr max_excerpts 2
set_attr template_map "Long builtin-long builtin-long Short builtin-short builtin-short Very-long user-long $PWD/conf/entry-template"
try "test max_stars. Also needs max_descriptions >= 2" \
    "words=also;script_name=qtest;format=user-long;method=or;restrict=http;sort=revscore" \
    'MATCHES *=4' \
    'CGI *=qtest' 'CONFIG *=-' 'EXCLUDE *=-' 'FIRSTDISPLAYED *=1' \
    'LASTDISPLAYED *=4' 'LOGICAL_WORDS *=also' \
    'MATCHES_PER_PAGE *=8' 'MATCH_MESSAGE *=Or' 'MAX_STARS *=5' \
    'METADESCRIPTION *=-' 'selected>Or' 'MODIFIED *=-' \
    'NEXTPAGE *=-' 'NSTARS *=-' 'PAGE *=1' 'PAGEHEADER *=single' \
    'PAGELIST *=-' 'PAGES *=1' 'PLURAL_MATCHES *=es' \
    'PREVPAGE *=-' 'RESTRICT *=http' 'SELECTED_FORMAT *=user-long' \
    'SELECTED_METHOD *=or' 'SELECTED_SORT *=revscore' \
    'selected>Reverse Score' 'VERSION *=3.2' 'WORDS *=also' \
    'START_ELLIPSES.*START_ELLIPSES.*END_ELLIPSES' 'STARTalsoSTOP' \
    '^[^N]*"STAR".*"NIL".*"NIL".*"NIL".*"NIL"' \
    'DESCRIPTIONS=JavaScript test<br>top<br><br>'

try "Search for 'empty'" \
    "words=empty;format=builtin-long" \
    'MATCHES *=2' \
    'empty%20file.html' 'empty file.html' 'sub%2520dir' 'INDEX OF /SET1/SUB%20DIR'
set_attr no_title_text "Empty-Title"
try "Search for 'empty'" \
    "words=empty;format=builtin-long" \
    'MATCHES *=2' \
    'empty%20file.html' 'Empty-Title' 'sub%2520dir'

set_attr star_patterns "site foo bad_local bar"
try "test star_patterns" \
    "words=also;format=builtin-long" \
    'MATCHES *=4' '"foo"' '"bar"'

set_attr star_patterns "site foo bad_local"
try "test star_patterns syntax error" \
    "words=also;format=builtin-long" \
    'MATCHES *=4' '"foo"' '""'

set_attr search_results_contenttype "foo"
set_attr use_star_image "false"
try "Checking use_star_image" \
    "words=bad_local;format=builtin-short" \
    "^[^*]*bad_local" "Content-type: foo"

set_attr search_results_contenttype ""
set_attr no_excerpt_text "No-excerpt-found"
try "Checking no_excerpt_show_top true" \
    "words=Yuki" \
    "a#bcd"

set_attr no_excerpt_show_top "false"
try "Checking no_excerpt_show_top false" \
    "words=Yuki" \
    "No-excerpt-found"

try "Checking excerpt_show_top false" \
    "words=also" \
    "STARTalsoSTOP"

set_attr build_select_lists 'MATCH_LIST,multiple matchesperpage matches_per_page_list 2 2 1 matches_per_page "Previous Amount"'
set_attr matches_per_page_list 'one 1 five 5 ten 10 twenty 20 "one hundred" 100 "two hundred" 200'
set_attr sort_names "score 'Best Match' time Newest title A-Z revscore 'Worst Match' revtime Oldest revtitle Z-A"
set_attr excerpt_show_top "true"

try "Checking excerpt_show_top true" \
    "words=also;matchesperpage=40" \
    "Copyright" \
    "MATCH_LIST *=<" "selected>Previous Amount" '<option value="1">one' \
    'SORT *=<select name="sort">' \
    '<option value="score" selected>Best Match' \
    '<option value="time">Newest' \
    '<option value="title">A-Z' \
    '<option value="revscore">Worst Match' \
    '<option value="revtime">Oldest' \
    '<option value="revtitle">Z-A' \
    '</select>'

set_attr syntax_error_file "$testdir/conf/main-template"
try "Checking syntax error file" \
    "words=also+and;method=boolean" \
    'MATCHES *=0' 'LOGICAL_WORDS *=also and' \
    'MATCH_MESSAGE *=Logical' 'METADESCRIPTION *=-' 'selected>Logical' \
    'PAGELIST *=-' 'PAGES *=1' 'SELECTED_METHOD *=boolean' \
    'VERSION *=3.2' 'WORDS *=also and'


set_attr max_descriptions 1
set_attr max_description_length 5
$htdig "$@" -t -i -c $config	|| fail "Couldn't dig"
$htpurge -c $config 		|| fail "Couldn't purge"
try "Check anchor descriptions" \
    "words=also;script_name=qtest;format=user-long;method=or;restrict=http;sort=revscore" \
    'DESCRIPTIONS=JavaScript ...<br><br>'
# Why does it end with "..." when START/END_ELLIPSES aren't ...?

test_functions_action=--stop-apache
. ./test_functions