--- trunk/WebPac.pm 2003/03/15 21:34:36 41 +++ branches/fsb/WebPac.pm 2004/04/18 01:03:27 321 @@ -3,29 +3,84 @@ use base 'CGI::Application'; use strict; -use HTML::Pager; use HTML::FillInForm; -use SWISH; +use SWISH::API; use Text::Iconv; use DBI; +use Config::IniFiles; +use Text::Unaccent; +use Data::Pageset; use lib '..'; -use index_DBI; +use index_DBI_cache; use back2html; -# configuration options -# FIX: they really should go in configuration file! -my $TEMPLATE_PATH = '/data/webpac/template_html'; -my $CHARSET = 'ISO-8859-2'; -my $SWISH = '/usr/bin/swish-e'; -my $INDEX = '/data/webpac/index/isis.index'; -my $MAX_HITS = 500; -my $ON_PAGE = 10; + +# read global.conf configuration +my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'"; + +# configuration options from global.conf +my $TEMPLATE_PATH = $cfg_global->val('webpac', 'template_html') || die "need template_html in global.conf, section webpac"; +my $CHARSET = $cfg_global->val('webpac', 'charset') || 'ISO-8859-1'; +my $SWISH = $cfg_global->val('webpac', 'swish') || '/usr/bin/swish-e'; +my $INDEX = $cfg_global->val('webpac', 'index') || die "need index in global.conf, section webpac"; +my $MAX_HITS = $cfg_global->val('webpac', 'max_hits') || 0; +my $ON_PAGE =$cfg_global->val('webpac', 'on_page') || 10; +my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1; +my $TEMPLATE =$cfg_global->val('webpac', 'template'); +my $UNAC_FILTER =$cfg_global->val('global', 'my_unac_filter'); +my $BASE_PATH =$cfg_global->val('webpac', 'base_path'); +# for pager +my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10; Text::Iconv->raise_error(0); # Conversion errors raise exceptions my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); +if ($UNAC_FILTER) { + require $UNAC_FILTER; +} else { + sub WebPac::my_unac_string { + my ($charset, $string) = (@_); + return $string; + } +} + +# use path from cgi script to support templates in subdirs +sub url_ex { + my $q = shift || die "suff2file needs CGI object!"; + my $tpl = shift || die "url_ex needs template name!"; + return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl); +} + +sub suff2file($$$$) { + my ($base_path, $p, $path, $tpl) = @_; + + return $tpl if (! $base_path); + + # strip everything to and including base path, leaving only + # additional (virtual) path + if ($base_path eq "/") { + $p =~ s,/*,,g; + my ($name,$ext) = split(/\./,$tpl); + $p = $name . "-" . $p . "." . $ext; + } elsif ($p =~ s,^.*?$base_path,,) { + $p =~ s,/*,,g; + my ($name,$ext) = split(/\./,$tpl); + $p = $name . $p . "." . $ext; + } else { + # if unable reset it! + $p = $tpl; + } + + if ( -e "$path/$p") { + return $p; + } else { + return $tpl; + } + +} + sub setup { my $self = shift; $self->tmpl_path($TEMPLATE_PATH); @@ -41,19 +96,124 @@ $self->header_props(-charset=>$CHARSET); } +sub in_template { + my $q = shift || die "need CGI object!"; + my $html = shift || die "This page is left unintentionally blank"; + return $html if (! defined($TEMPLATE)); + + my ($dir,$tpl); + if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) { + ($dir,$tpl) = ($1,$2); + } else { + die "can't parse TEMPLATE path"; + } + + my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl); + if (open(T, $master_tpl)) { + my $template_html = join("\n",); + close(T); + $template_html =~ s/##webpac##/$html/gsi; + return $template_html; + } else { + return "Can't read template '$master_tpl'"; + } +} + +#-------------------------------------------------------------------------- + +# +# make pager navigation and fill template variables +# compatibile with HTML::Pager +# + +sub make_pager($$$) { + my ($q,$tmpl,$pager) = @_; + + # + # pager navigation + # + my ($pager_prev,$pager_next, $pager_jump) = ('','',''); + + my $nav_fmt=qq{ %s }; + + if ($pager->current_page() > $pager->first_page) { + $q->param('PAGER_offset', $pager->current_page - 1); + $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'<<'); + } + + if ($pager->previous_set) { + $q->param('PAGER_offset', $pager->previous_set); + $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..'); + } + + + foreach my $p (@{$pager->pages_in_set()}) { + next if ($p < 0); + if($p == $pager->current_page()) { + $pager_jump .= "$p "; + } else { + $q->param('PAGER_offset', $p); + $pager_jump .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),$p); + } + } + + if ($pager->next_set) { + $q->param('PAGER_offset', $pager->next_set); + $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..'); + } + + if ($pager->current_page() < $pager->last_page) { + $q->param('PAGER_offset', $pager->current_page + 1); + $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'>>'); + } + + $tmpl->param('PAGER_PREV', $pager_prev); + $tmpl->param('PAGER_JUMP', $pager_jump); + $tmpl->param('PAGER_NEXT', $pager_next); + +} + +# +# put persisten variables in template +# + +sub make_pager_vars { + my $q = shift @_; + my $tmpl = shift @_; + my @persist_vars = @_; + my $hidden_vars = ''; + foreach my $v (@persist_vars) { + $hidden_vars .= ''."\n"; + } + + $tmpl->param('PAGER_HIDDEN', $hidden_vars); + $tmpl->param('PAGER_JAVASCRIPT', qq{ + + }); +} + +#-------------------------------------------------------------------------- + sub show_search_form { my $self = shift; # Get the CGI.pm query object my $q = $self->query(); - my $tmpl = $self->load_tmpl('search.html'); + my $tmpl = $self->load_tmpl(url_ex($q,'search.html')); my $html = $tmpl->output; my $fif = new HTML::FillInForm; - return $fif->fill(scalarref => \$html, fobject => $q, - target => 'search'); + return in_template($q,$fif->fill(scalarref => \$html, fobject => $q, + target => 'search')); } sub show_results_list { @@ -61,109 +221,187 @@ my $q = $self->query(); - my @swish_results; # results from swish - # load template for this page my @s_arr; # all queries are located here - for(my $i = 1; $i <=10; $i++) { + my @path_arr = $q->param('path'); + my $full = $q->param('full'); + + my @persist_vars = ( 'rm' ); + my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.($q->param('PAGER_offset') || 0) ); + + # support parametars "f" and "v" for start + for(my $i = ""; $i <=30; $i++) { return show_index($self, $i) if ($q->param("f".$i."_index")); - next if (! $q->param("f$i")); + next if (! $q->param("v$i")); + next if (! $q->param("f$i")); - # re-write query from +/- to and/and not - my $s; - my $search = $q->param("v$i"); - while ($search =~ s/\s*("[^"]+")\s*/ /) { - $s .= "$1 "; + push @persist_vars, "f$i"; + push @persist_vars, "v$i"; + push @persist_vars, "e$i" if ($q->param("e$i")); + + push @url_params,"f$i=".$q->url_param("f$i"); + foreach my $v ($q->url_param("v$i")) { + push @url_params,"v$i=$v"; } - $search =~ s/^\s+//; - $search =~ s/\s+$//; + push @url_params,"e$i=".$q->url_param("e$i"); + + my $wc="*"; # swish wildcard + $wc="" if ($i eq ""); # don't apply wildcard on field 0 - foreach (split(/\s+/,$search)) { - if (m/^([+-])(\S+)/) { - $s.= ($s) ? "and " : ""; - $s.="not " if ($1 eq "-"); - $s.="$2* "; - } else { - $s.="$_* "; + # re-write query from +/- to and/and not + my @param_vals = $q->param("v$i"); + my @swish_q; + my ($pre,$post,$exact) = ('','',''); + while (my $search = shift @param_vals) { + my $s; + # remove accents + $search = my_unac_string($CHARSET,$search); + while ($search =~ s/\s*("[^"]+")\s*/ /) { + $s .= "$1 "; } + $search =~ s/^\s+//; + $search =~ s/\s+$//; + + # filed e[nr] is exact match bitmask + # 1 = beginning, 2=end, 3=both + $pre = '"xxbxx ' if ($q->param("e$i") & 1); + $post = ' xxexx"' if ($q->param("e$i") & 2); + # add qotes on other side + if ($q->param("e$i")) { + $pre = '"' if (! $pre); + $post = '"' if (! $post); + # what about wildcards? + $wc = ''; + $wc = '*' if ($q->param("e$i") & 4); + $exact = '_exact'; + } + + foreach (split(/\s+/,$search)) { + if (m/^([+-])(\S+)/) { + $s.= ($s) ? "and " : ""; + $s.="not " if ($1 eq "-"); + $s.=$2.$wc." "; + } elsif (m/^\s*(and|or|not)\s*$/i) { + $s.=$_." "; + # don't add * to words with less than x chars + } elsif (length($_) <= $MIN_WILDCARD) { + $s.=$_." "; + } else { + $s.=$_.$wc." "; + } + } + $s =~ s/\*+/*/g; + $s = $pre.$s.$post if ($q->param("e$i")); + push @swish_q,$s; } - $s =~ s/\*+/*/g; + # FIXME default operator for multi-value fields is or. There is + # no way to change it, except here for now. Is there need? + push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")"; + } - push @s_arr,$q->param("f$i")."_swish=($s)"; + my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1); + + $tmpl->param('url_params',"?".join("&",@url_params)); + + sub esc_html { + my $html = shift; + $html =~ s//>/g; + return $html; } - my $tmpl = $self->load_tmpl('results.html'); + my $sort = 'swishrank'; + if ($q->param("sort")) { + $sort = 'headline'; + push @persist_vars, "sort"; + } - # call swish - my $sh = SWISH->connect('Fork', - prog => $SWISH, - indexes => $INDEX, - properties => [qw/swishdocpath swishrank swishtitle headline html/], - results => sub { - my ($sh,$hit) = @_; - - push @swish_results, { - nr => ($#swish_results + 2), - path => $hit->swishdocpath, - headline => $from_utf8->convert($hit->headline), - html => back2html($from_utf8->convert($hit->html)), - rank => $hit->swishrank }; - - }, - #startnum => 0, - maxhits => $MAX_HITS, - ); + # construct swish query + my $sw_q = join(" and ",@s_arr); + if (@path_arr && $q->param('show_full')) { + $sw_q .= "and (swishdocpath=\""; + $sw_q .= join("\" or swishdocpath=\"",@path_arr); + $sw_q .= "\")"; + $tmpl->param('full',1); # show full records + } elsif ($q->param('show_full')) { + # just show full path, no path defined + $tmpl->param('full',1); + } else { + $tmpl->param('full',0); + } - die $SWISH::errstr unless $sh; + # create new swish instance + my $swish = SWISH::API->new($INDEX); + $swish->AbortLastError if $swish->Error; - my $hits = $sh->query(join(" and ",@s_arr)) || 0; # FIX: and/or + # execute query and get number of results from SWISH-E + my $search = $swish->New_Search_Object; + + $search->SetSort($sort); + + my $results = $search->Execute($sw_q); + $swish->AbortLastError if $swish->Error; + + my $hits = $results->Hits; $tmpl->param('hits',$hits); - $tmpl->param('search',join(" and ",@s_arr)); + $tmpl->param('search',$sw_q); - # create a Pager object - my $pager = HTML::Pager->new( - # required parameters - query => $q, - get_data_callback => sub { - my ($offset, $rows) = @_; - - my @result; - for (my $i=0; $i<$rows; $i++) { - push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; - } - return \@result; - }, - rows => $hits, - page_size => $ON_PAGE, - # some optional parameters - persist_vars => [ - 'rm', - 'f1', 'v1', - 'f2', 'v2', - 'f3', 'v3', - 'f4', 'v4', - 'f5', 'v5', - 'f6', 'v6', - 'f7', 'v7', - 'f8', 'v8', - 'f9', 'v9', - ], - #cell_space_color => '#000000', - #cell_background_color => '#ffffff', - #nav_background_color => '#dddddd', - #javascript_presubmit => 'last_minute_javascript()', - debug => 1, - template => $tmpl, - ); + $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); + $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); - my $html = $pager->output; + # + # build pager + # - return $html; + my $current_page = $q->param('PAGER_offset') || 1; + + my $pager = Data::Pageset->new({ + 'total_entries' => $hits, + 'entries_per_page' => $ON_PAGE, + 'current_page' => $current_page, + 'pages_per_set' => $pages_per_set, + }); + + $results->SeekResult( $pager->first - 1 ); + + # get number of entries on this page + my $i = $pager->entries_on_this_page; + + # results from swish for template + my @pager_data_list; + + for(my $i=$pager->first; $i<=$pager->last; $i++) { + + my $result = $results->NextResult; + last if (! $result); + + my $r = { + nr => $i, + path => $result->Property('swishdocpath'), + headline => esc_html($from_utf8->convert($result->Property('headline'))), + rank => $result->Property('swishrank') + }; + + $r->{html} = back2html($from_utf8->convert($result->Property('html'))) if ($q->param('show_full')); + + push @pager_data_list, $r; + } + + + + # put something in template + make_pager($q, $tmpl, $pager); + make_pager_vars($q, $tmpl, @persist_vars); + $tmpl->param('PAGER_DATA_LIST', \@pager_data_list); + + my $html = $tmpl->output; + + return in_template($q,$html); } sub show_index { @@ -177,46 +415,52 @@ my $html; - my $index = new index_DBI(); + my $index = new index_DBI( + $cfg_global->val('global', 'dbi_dbd'), + $cfg_global->val('global', 'dbi_dsn'), + $cfg_global->val('global', 'dbi_user'), + $cfg_global->val('global', 'dbi_passwd') || '' + ); + + my $total = $index->count($field,$limit); - my $total = $index->check($field); if (! $total) { - my $tmpl = $self->load_tmpl('no_index.html'); + my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html')); $tmpl->param('field',$field); $html = $tmpl->output; return $html; } - my $tmpl = $self->load_tmpl('index_res.html'); + my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1); $tmpl->param('field',$field); $tmpl->param('limit',$limit); $tmpl->param('total',$total); -# FIX: I should set offset and leave out limit from fetch!! +# FIXME I should set offset and leave out limit from fetch!! # if (! $q->param("PAGER_offset") { # $q->param("Pager_offet) # } - my $pager = HTML::Pager->new( - query => $q, - get_data_callback => sub { - my ($offset, $rows) = @_; - - my @result = $index->fetch($field,'item',$limit, $offset, $rows); - return \@result; - }, - rows => $total, - page_size => $ON_PAGE, - persist_vars => [ - 'rm', - "f$i", "v$i", "f".$i."_index", - 'offset', - ], - debug => 1, - template => $tmpl, - ); - return $pager->output; + # + # build pager + # + my $pager = Data::Pageset->new({ + 'total_entries' => $total, + 'entries_per_page' => $ON_PAGE, + 'current_page' => $q->param('PAGER_offset') || 1, + 'pages_per_set' => $pages_per_set + }); + + my @persist_vars = qw{rm f$i v$i f$i_index offset}; + + make_pager($q, $tmpl, $pager); + make_pager_vars($q, $tmpl, @persist_vars); + + my @pager_data_list = $index->fetch($field,$limit, $pager->first - 1, $pager->entries_on_this_page); + $tmpl->param('PAGER_DATA_LIST', \@pager_data_list); + + return in_template($q,$tmpl->output); } 1;