--- loncom/interface/lonsearchcat.pm 2001/08/26 21:49:49 1.99 +++ loncom/interface/lonsearchcat.pm 2001/11/28 19:26:20 1.106 @@ -3,104 +3,25 @@ # Search Catalog # # YEAR=2001 -# 03/08/2001 Scott Harrison -# Scott Harrison: 03/12/2001, 03/13/2001, 03/14/2001, 03/15/2001, 03/19/2001 -# Scott Harrison: 03/20/2001, 03/21/2001, 03/22/2001, 03/26/2001, 03/27/2001 -# Scott Harrison: 04/02/2001, 08/15/2001, 08/24/2001, 08/25/2001 +# 3/8, 3/12, 3/13, 3/14, 3/15, 3/19 Scott Harrison +# 3/20, 3/21, 3/22, 3/26, 3/27, 4/2, 8/15, 8/24, 8/25 Scott Harrison +# 10/12,10/14,10/15,10/16,11/28 Scott Harrison +# +# $Id: lonsearchcat.pm,v 1.106 2001/11/28 19:26:20 harris41 Exp $ +### ############################################################################### ## ## ## ORGANIZATION OF THIS PERL MODULE ## ## ## -## 1. Description of functions ## -## 2. Modules used by this module ## -## 3. Choices for different output views (detailed, summary, xml, etc) ## -## 4. BEGIN block (to be run once after compilation) ## -## 5. Handling routine called via Apache and mod_perl ## -## 6. Other subroutines ## +## 1. Modules used by this module ## +## 2. Choices for different output views (detailed, summary, xml, etc) ## +## 3. BEGIN block (to be run once after compilation) ## +## 4. Handling routine called via Apache and mod_perl ## +## 5. Other subroutines ## ## ## ############################################################################### -# ---------------------------------------------------- Description of functions -# -# -# === WEB HANDLER FUNCTIONS -# BEGIN() : run once after compilation to initialize values -# handler(server reference) : interacts with the Apache server layer -# (for /adm/searchcat URLs) -# get_unprocessed_cgi() : reads in critical name/value pairs that may have not -# been processed and passed into %ENV by the web server -# make_persistent() : makes a set of hidden HTML fields to make -# SQL search interface information to be persistent -# -# -# === WEB INTERFACE COMPONENT FUNCTIONS -# simpletextfield(name,value) : returns HTML formatted string for simple text -# field -# simplecheckbox(name,value) : returns HTML formatted string for simple -# checkbox -# searchphrasefield(title,name,value) : returns HTML formatted string for -# a search expression phrase field -# dateboxes(name, defaultmonth, defaultday, defaultyear) : returns HTML -# formatted string -# for a calendar date -# selectbox(title,name,value,%HASH=options) : returns HTML formatted string for -# a selection box field -# -# -# === SEARCH FUNCTIONS -# advancedsearch(server reference, environment reference) : perform a complex -# multi-field logical query -# basicsearch(server reference, environment reference) : perform a simple -# single-field logical query -# build_SQL_query(field name, logic) : builds a SQL query string from a -# logical expression with AND/OR keywords -# build_custommetadata_query(field_name, logic_statement) : builds a perl -# regular expression from a logical expression with AND/OR -# keywords -# recursive_SQL_query_build(field name, reverse notation expression) : -# builds a SQL query string from a reverse notation expression -# logical expression with AND/OR keywords -# build_date_queries(cmonth1, cday1, cyear1, cmonth2, cday2, cyear2, -# lmonth1, lday1, lyear1, lmonth2, lday2, lyear2) : -# Builds a SQL logic query to check time/date entries. -# -# -# === OUTPUTTING RESULTS FUNCTION -# output_results(output mode, -# server reference, -# environment reference, -# reply list reference) : outputs results from search -# -# -# === DIFFERENT WAYS TO VIEW METADATA RECORDS -# detailed_citation_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : -# see metadata viewing notes below -# summary_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : -# see metadata viewing notes below -# fielded_format_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : -# see metadata viewing notes below -# xml_sgml_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : -# see metadata viewing notes below -# ___________________________________________________________________________ -# | * Metadata viewing notes | -# | Output is a HTML-ified string. | -# | Input arguments are title, author, subject, url, keywords, version, notes,| -# | short abstract, mime, language, creation date, last revision date, owner, | -# | copyright, hostname, httphost, and extra custom metadata to show. | -# --------------------------------------------------------------------------- -# -# -# === TEST CONDITIONAL FUNCTIONS -# filled(field) : determines whether a given field has been filled -# -# -# === ERROR FUNCTIONS -# output_blank_field_error(server reference) : outputs a message saying that -# more fields need to be filled in -# output_date_error(server reference, error message) : outputs -# an error message specific to bad date format. - package Apache::lonsearchcat; # ------------------------------------------------- modules used by this module @@ -110,6 +31,7 @@ use Apache::lonnet(); use Apache::File(); use CGI qw(:standard); use Text::Query; +use GDBM_File; # ---------------------------------------- variables used throughout the module @@ -127,8 +49,8 @@ my $importbutton; # button to take the s # -- miscellaneous variables my $scrout; # string that holds portions of the screen output -my $fnum; # search result counter my $yourself; # allows for quickly limiting to oneself +my %hash; # ------------------------------------------ choices for different output views # Detailed Citation View ---> sub detailed_citation_view @@ -195,16 +117,35 @@ sub BEGIN { } } +my $diropendb = ""; +my $domain = ""; + # ----------------------------- Handling routine called via Apache and mod_perl sub handler { my $r = shift; - + untie %hash; &get_unprocessed_cgi(); $r->content_type('text/html'); $r->send_http_header; return OK if $r->header_only; + $domain = $r->dir_config('lonDefDomain'); + + $diropendb= "/home/httpd/perl/tmp/$domain\_$ENV{'user.name'}_searchcat.db"; + + if ($ENV{'form.launch'} eq '1') { + if (tie(%hash,'GDBM_File',$diropendb,&GDBM_WRCREAT,0640)) { + &start_fresh_session(); + untie %hash; + } + else { + $r->print('Unable to tie hash to db '. + 'file'); + return OK; + } + } + # ----------------------------------- configure dynamic components of interface my $hidden=''; if ($ENV{'form.catalogmode'} eq 'interactive') { @@ -236,10 +177,10 @@ END # ----------------------------------- See if a search invocation should be done if ($ENV{'form.basicsubmit'} eq 'SEARCH') { - return &basicsearch($r,\%ENV); + untie %hash; return &basicsearch($r,\%ENV); } elsif ($ENV{'form.advancedsubmit'} eq 'SEARCH') { - return &advancedsearch($r,\%ENV); + untie %hash; return &advancedsearch($r,\%ENV); } # ----------------------------- Else, begin building search interface to output @@ -335,7 +276,7 @@ LASTREVISIONDATEEND LIMIT BY SPECIAL METADATA FIELDS: For resource-specific metadata, enter in an expression in the form of -key=value separated by operators such as AND or OR.
+key=value separated by operators such as AND, OR or NOT.
Example: grandmother=75 OR grandfather=85
CUSTOMMETADATA @@ -358,6 +299,13 @@ $scrout.=' initial users of this syst The LearningOnline Network with CAPA + @@ -368,7 +316,7 @@ $hidden

Basic Search

Enter terms or phrases separated by search operators -such as AND or OR then press SEARCH below. Terms should be specific +such as AND, OR, or NOT then press SEARCH below. Terms should be specific to the title, author, subject, notes, or abstract information associated with a resource.
@@ -385,6 +333,7 @@ ENDDOCUMENT $closebutton $basicviewselect +

Advanced Search

@@ -394,6 +343,7 @@ $scrout $closebutton $advancedviewselect +

@@ -408,7 +358,7 @@ sub get_unprocessed_cgi { my ($name, $value) = split(/=/,$_); $value =~ tr/+/ /; $value =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; - if ($name eq 'catalogmode') { + if ($name eq 'catalogmode' or $name eq 'launch' or $name eq 'acts') { $ENV{'form.'.$name}=$value; } } (split(/&/,$ENV{'QUERY_STRING'})); @@ -451,7 +401,7 @@ sub searchphrasefield { my ($title,$name,$value)=@_; my $instruction=<$uctitle:". @@ -632,7 +582,7 @@ sub advancedsearch { 'lastrevisiondateend_day','lastrevisiondateend_year', 'notes','abstract','mime','language','owner', 'custommetadata','customshow') { - $ENV{"form.$field"}=~s/[^\w\s\(\)\=\-\"\']//g; + $ENV{"form.$field"}=~s/[^\w\/\s\(\)\=\-\"\']//g; } # Check to see if enough information was filled in @@ -882,11 +832,12 @@ sub build_date_queries { # gives the status of the network search (time elapsed, number of machines # contacted, etc.) sub output_results { + my $fnum; # search result counter my ($mode,$r,$envhash,$query,$replyref)=@_; my %ENV=%{$envhash}; my %rhash=%{$replyref}; my $compiledresult=''; - my $timeremain=30; + my $timeremain=300; my $elapsetime=0; my $resultflag=0; my $tflag=1; @@ -904,7 +855,7 @@ BEGINNING # conditional output of script functions dependent on the mode in # which the search was invoked $r->print(< + SCRIPT $r->print(< + SCRIPT - $r->print(< + $r->print(< function displayinfo(val) { popwin.document.forms.popremain.sdetails.value=val; } + function openhelp(val) { + openhelpwin=open('/adm/help/searchcat.html','helpscreen', + 'scrollbars=1,width=400,height=300'); + openhelpwin.focus(); + } + function abortsearch(val) { + openhelpwin=open('/adm/help/searchcat.html','helpscreen', + 'scrollbars=1,width=400,height=300'); + openhelpwin.focus(); + } SCRIPT $r->rflush(); @@ -1019,7 +981,7 @@ RESULTS $sn++; } $r->print(< +'. "\n"); $r->rflush(); + $bloop--; + $orkey{$rkey}=1; } else { $reply=~/^([\.\w]+)$/; # must do since 'use strict' checks for tainting @@ -1093,22 +1062,32 @@ ENDPOP my $temp=0; WLOOP: while (1) { if (-e $replyfile && $tflag) { - $r->print(''. "\n"); $r->rflush(); - $r->print(''. "\n"); $r->rflush(); $tflag=0; } - last WLOOP if $temp>5; + if ($temp>1) { + sleep 1; + $timeremain--; + $elapsetime++; + last WLOOP; + } if (-e "$replyfile.end") { + $bloop--; + $orkey{$rkey}=1; if (-s $replyfile) { - $r->print(''. - "\n"); + $r->print(''."\n"); $r->rflush(); my $fh=Apache::File->new($replyfile) or ($r->print('ERROR: file '. @@ -1116,23 +1095,28 @@ ENDPOP return OK); @results=<$fh> if $fh; $hitcount{$rkey}=@results+0; - $r->print(''. "\n"); $r->rflush(); $hitcountsum+=$hitcount{$rkey}; - $r->print(''. "\n"); $r->rflush(); } else { - $r->print(''. - "\n"); + $r->print(''. + "\n"); $r->rflush(); - $r->print(''. "\n"); $r->rflush(); @@ -1143,12 +1127,18 @@ ENDPOP sleep 1; $timeremain--; $elapsetime++; - $r->print(''."\n"); $r->rflush(); $temp++; } } + $r->print(''."\n"); + $r->rflush(); } my $customshow=''; my $extrashow=''; @@ -1174,80 +1164,115 @@ ENDPOP $customhash{$k}=$v; } } - foreach my $result (@results) { - next if $result=~/^custom\=/; - chomp $result; - next unless $result; - my @fields=map - {&Apache::lonnet::unescape($_)} - (split(/\,/,$result)); - my ($title,$author,$subject,$url,$keywords,$version, - $notes,$abstract,$mime,$lang, - $creationdate,$lastrevisiondate,$owner,$copyright)=@fields; - my $shortabstract=$abstract; - $shortabstract=substr($abstract,0,200) if length($abstract)>200; - $fields[7]=$shortabstract; - my $extrashow2=$extrashow; - if ($extrashow) { - foreach my $field (@customfields) { - my $value=''; - if ($customhash{$url}=~/\<${field}[^\>]*\>(.*?)\<\/${field}[^\>]*\>/s) { - $value=$1; - } - $extrashow2=~s/\<\!\-\- $field \-\-\>/ $value/g; - } - } + if (keys %hash) { + untie %hash; + } + if (tie(%hash,'GDBM_File',$diropendb,&GDBM_WRCREAT,0640)) { + if ($ENV{'form.launch'} eq '1') { + &start_fresh_session(); + } + foreach my $result (@results) { + next if $result=~/^custom\=/; + chomp $result; + next unless $result; + my @fields=map + {&Apache::lonnet::unescape($_)} + (split(/\,/,$result)); + my ($title,$author,$subject,$url,$keywords,$version, + $notes,$abstract,$mime,$lang, + $creationdate,$lastrevisiondate,$owner,$copyright)=@fields; + + unless ($ENV{'user.adv'}) { + $keywords='- not displayed -'; + $fields[4]=$keywords; + $notes='- not displayed -'; + $fields[6]=$notes; + $abstract='- not displayed -'; + $fields[7]=$abstract; + $subject='- not displayed -'; + $fields[2]=$subject; + } + + my $shortabstract=$abstract; + $shortabstract=substr($abstract,0,200).'...' if length($abstract)>200; + $fields[7]=$shortabstract; + my $shortkeywords=$keywords; + $shortkeywords=substr($keywords,0,200).'...' if length($keywords)>200; + $fields[4]=$shortkeywords; + + my $extrashow2=$extrashow; + if ($extrashow) { + foreach my $field (@customfields) { + my $value=''; + if ($customhash{$url}=~/\<${field}[^\>]*\>(.*?)\<\/${field}[^\>]*\>/s) { + $value=$1; + } + $extrashow2=~s/\<\!\-\- $field \-\-\>/ $value/g; + } + } - $compiledresult.=< END - $compiledresult.=< END - $compiledresult.=<
END - $compiledresult.=< - -
END - my $httphost=$ENV{'HTTP_HOST'}; - - my $viewselect; - if ($mode eq 'Basic') { - $viewselect=$ENV{'form.basicviewselect'}; - } - elsif ($mode eq 'Advanced') { - $viewselect=$ENV{'form.advancedviewselect'}; - } +# +# + $fnum++; + } + my $httphost=$ENV{'HTTP_HOST'}; + + my $viewselect; + if ($mode eq 'Basic') { + $viewselect=$ENV{'form.basicviewselect'}; + } + elsif ($mode eq 'Advanced') { + $viewselect=$ENV{'form.advancedviewselect'}; + } - if ($viewselect eq 'Detailed Citation View') { - $compiledresult.=&detailed_citation_view(@fields, + if ($viewselect eq 'Detailed Citation View') { + $compiledresult.=&detailed_citation_view(@fields, $hostname,$httphost, $extrashow2); - } - elsif ($viewselect eq 'Summary View') { - $compiledresult.=&summary_view(@fields,$hostname,$httphost, + } + elsif ($viewselect eq 'Summary View') { + $compiledresult.=&summary_view(@fields,$hostname,$httphost, $extrashow2); - } - elsif ($viewselect eq 'Fielded Format') { - $compiledresult.=&fielded_format_view(@fields,$hostname, + } + elsif ($viewselect eq 'Fielded Format') { + $compiledresult.=&fielded_format_view(@fields,$hostname, $httphost,$extrashow2); - } - elsif ($viewselect eq 'XML/SGML') { - $compiledresult.=&xml_sgml_view(@fields,$hostname,$httphost, + } + elsif ($viewselect eq 'XML/SGML') { + $compiledresult.=&xml_sgml_view(@fields,$hostname,$httphost, $extrashow2); - } + } - } + } + untie %hash; + } + else { + $r->print('Unable to tie hash to db '. + 'file'); + } if ($compiledresult) { $resultflag=1; } @@ -1257,10 +1282,11 @@ $compiledresult RESULTS my $percent=sprintf('%3.0f',($servercount/$servernum*100)); } + } unless ($resultflag) { $r->print("\nThere were no results that matched your query\n"); } -# $r->print(''."\n"); $r->rflush(); +# $r->print(''."\n"); $r->rflush(); $r->print(< @@ -1454,6 +1480,261 @@ $message RESULTS } +# --------- settings whenever the user causes the search window to be launched +sub start_fresh_session { + delete $hash{'mode_catalog'}; + map { + if ($_ =~ /^pre_/) { + delete $hash{$_}; + } + if ($_ =~ /^store/) { + delete $hash{$_}; + } + } keys %hash; +} + 1; __END__ + +=head1 NAME + +Apache::lonsearchcat - mod_perl module for handling a searchable catalog + +=head1 SYNOPSIS + +Invoked by /etc/httpd/conf/srm.conf: + + + PerlAccessHandler Apache::lonacc + SetHandler perl-script + PerlHandler Apache::lonsearchcat + ErrorDocument 403 /adm/login + ErrorDocument 500 /adm/errorhandler + + +=head1 INTRODUCTION + +This module enables searching for a distributed browseable catalog. + +This is part of the LearningOnline Network with CAPA project +described at http://www.lon-capa.org. + +=head1 BEGIN SUBROUTINE + +This routine is only run once after compilation. + +=over 4 + +=item * + +Initializes %language hash table. + +=item * + +Initializes %cprtag hash table (for copyright.tab). + +=item * + +Initializes %mimetag hash table (for filetypes.tab). + +=item * + +Initializes %hostdomains and hostips hash table (for hosts.tab). + +=back + +=head1 HANDLER SUBROUTINE + +This routine is called by Apache and mod_perl. + +=over 4 + +=item * + +configure dynamic components of interface + +=item * + +determine current user + +=item * + +see if a search invocation should be done + +=item * + +else, begin building search interface to output + +=item * + +compute date selection boxes + +=item * + +compute customized metadata field + +=item * + +print screen + +=back + +=head1 OTHER SUBROUTINES + +=over 4 + +=item * + +get_unprocessed_cgi() : reads in critical name/value pairs that may have not +been processed and passed into %ENV by the web server + +=item * + +make_persistent() : makes a set of hidden HTML fields to make +SQL search interface information to be persistent + +=back + +WEB INTERFACE COMPONENT FUNCTIONS + +=over 4 + +=item * + +simpletextfield(name,value) : returns HTML formatted string for simple text +field + +=item * + +simplecheckbox(name,value) : returns HTML formatted string for simple +checkbox + +=item * + +searchphrasefield(title,name,value) : returns HTML formatted string for +a search expression phrase field + +=item * + +dateboxes(name, defaultmonth, defaultday, defaultyear) : returns HTML +formatted string for a calendar date + +=item * + +selectbox(title,name,value,%HASH=options) : returns HTML formatted string for +a selection box field + +=back + +SEARCH FUNCTIONS + +=over 4 + +=item * + +advancedsearch(server reference, environment reference) : perform a complex +multi-field logical query + +=item * + +basicsearch(server reference, environment reference) : perform a simple +single-field logical query + +=item * + +build_SQL_query(field name, logic) : builds a SQL query string from a +logical expression with AND/OR keywords + +=item * + +build_custommetadata_query(field_name, logic_statement) : builds a perl +regular expression from a logical expression with AND/OR keywords + +=item * + +recursive_SQL_query_build(field name, reverse notation expression) : +builds a SQL query string from a reverse notation expression +logical expression with AND/OR keywords + +=item * + +build_date_queries(cmonth1, cday1, cyear1, cmonth2, cday2, cyear2, +lmonth1, lday1, lyear1, lmonth2, lday2, lyear2) : +Builds a SQL logic query to check time/date entries. + +=back + +OUTPUTTING RESULTS FUNCTION + +=over 4 + +=item * + +output_results(output mode, server reference, environment reference, +reply list reference) : outputs results from search + +=back + +DIFFERENT WAYS TO VIEW METADATA RECORDS + +=over 4 + +=item * + +detailed_citation_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : +see metadata viewing notes below + +=item * + +summary_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : +see metadata viewing notes below + +=item * + +fielded_format_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : +see metadata viewing notes below + +=item * + +xml_sgml_view(ORDERED METADATA LIST FOR A RESULT OBJECT INSTANCE) : +see metadata viewing notes below + +=back + + _____________________________________________________________________ + | * Metadata viewing notes | + | Output is a HTML-ified string. | + | Input arguments are title, author, subject, url, keywords, version, | + | notes, short abstract, mime, language, creation date, | + | last revision date, owner, copyright, hostname, httphost, and | + | extra custom metadata to show. | + --------------------------------------------------------------------- + +TEST CONDITIONAL FUNCTIONS + +=over 4 + +=item * + +filled(field) : determines whether a given field has been filled + +=back + +ERROR FUNCTIONS + +=over 4 + +=item * + +output_blank_field_error(server reference) : outputs a message saying that +more fields need to be filled in + +=item * + +output_date_error(server reference, error message) : outputs +an error message specific to bad date format. + +=back + +=cut