--- SearchHelper.pm 2006/09/27 16:55:38 1.2 +++ SearchHelper.pm 2007/02/04 13:07:24 1.26 @@ -17,6 +17,8 @@ use HTML; use BasicLocation; use FeatureQuery; + use URI::Escape; + use PageBuilder; =head1 Search Helper Base Class @@ -73,6 +75,19 @@ List of JavaScript statements to be executed after the form is closed. +=item genomeHash + +Cache of the genome group hash used to build genome selection controls. + +=item genomeParms + +List of the parameters that are used to select multiple genomes. + +=item filtered + +TRUE if this is a feature-filtered search, else FALSE. B that this +field is updated by the B object. + =back =head2 Adding a new Search Tool @@ -98,8 +113,7 @@ =item 4 -In the C script, add a C statement for your search tool -and then put the class name in the C<@advancedClasses> list. +In the C script and add a C statement for your search tool. =back @@ -139,7 +153,11 @@ =item 1 -L generates a control for selecting one or more genomes. +L generates a control for selecting one or more genomes. Use +L to retrieve all the genomes passed in for a specified parameter +name. Note that as an assist to people working with GET-style links, if no +genomes are specified and the incoming request style is GET, all genomes will +be returned. =item 2 @@ -158,6 +176,10 @@ =back +If you are doing a feature search, you can also change the list of feature +columns displayed and their display order by overriding +L. + Finally, when generating the code for your controls, be sure to use any incoming query parameters as default values so that the search request is persistent. @@ -195,15 +217,15 @@ } } } + # Close the session file. + $self->CloseSession(); } - # Close the session file. - $self->CloseSession(); # Return the result count. return $retVal; } A Find method is of course much more complicated than generating a form, and there -are variations on the above them. For example, you could eschew feature filtering +are variations on the above theme. For example, you could eschew feature filtering entirely in favor of your own custom filtering, you could include extra columns in the output, or you could search for something that's not a feature at all. The above code is just a loose framework. @@ -218,35 +240,12 @@ by calling L. If the parameters are valid, then the method must return the number of items found. -=head2 Virtual Methods - -=head3 Form - -C<< my $html = $shelp->Form(); >> - -Generate the HTML for a form to request a new search. - -=head3 Find - -C<< my $resultCount = $shelp->Find(); >> - -Conduct a search based on the current CGI query parameters. The search results will -be written to the session cache file and the number of results will be -returned. If the search parameters are invalid, a result count of C will be -returned and a result message will be stored in this object describing the problem. - -=head3 Description - -C<< my $htmlText = $shelp->Description(); >> - -Return a description of this search. The description is used for the table of contents -on the main search tools page. It may contain HTML, but it should be character-level, -not block-level, since the description is going to appear in a list. - =cut # This counter is used to insure every form on the page has a unique name. my $formCount = 0; +# This counter is used to generate unique DIV IDs. +my $divCount = 0; =head2 Public Methods @@ -258,7 +257,7 @@ =over 4 -=item query +=item cgi The CGI query object for the current script. @@ -268,22 +267,32 @@ sub new { # Get the parameters. - my ($class, $query) = @_; + my ($class, $cgi) = @_; # Check for a session ID. - my $session_id = $query->param("SessionID"); + my $session_id = $cgi->param("SessionID"); my $type = "old"; if (! $session_id) { + Trace("No session ID found.") if T(3); # Here we're starting a new session. We create the session ID and # store it in the query object. $session_id = NewSessionID(); $type = "new"; - $query->param(-name => 'SessionID', -value => $session_id); + $cgi->param(-name => 'SessionID', -value => $session_id); + } else { + Trace("Session ID is $session_id.") if T(3); } # Compute the subclass name. - $class =~ /SH(.+)$/; - my $subClass = $1; + my $subClass; + if ($class =~ /SH(.+)$/) { + # Here we have a real search class. + $subClass = $1; + } else { + # Here we have a bare class. The bare class cannot search, but it can + # process search results. + $subClass = 'SearchHelper'; + } # Insure everybody knows we're in Sprout mode. - $query->param(-name => 'SPROUT', -value => 1); + $cgi->param(-name => 'SPROUT', -value => 1); # Generate the form name. my $formName = "$class$formCount"; $formCount++; @@ -291,13 +300,16 @@ # as well as an indicator as to whether or not the session is new, plus the # class name and a placeholder for the Sprout object. my $retVal = { - query => $query, + query => $cgi, type => $type, class => $subClass, sprout => undef, orgs => {}, name => $formName, scriptQueue => [], + genomeList => undef, + genomeParms => [], + filtered => 0, }; # Bless and return it. bless $retVal, $class; @@ -319,6 +331,8 @@ return $self->{query}; } + + =head3 DB C<< my $sprout = $shelp->DB(); >> @@ -450,13 +464,15 @@ my ($self, $title) = @_; # Get the CGI object. my $cgi = $self->Q(); - # Start the form. + # Start the form. Note we use the override option on the Class value, in + # case the Advanced button was used. my $retVal = "
\n" . $cgi->start_form(-method => 'POST', -action => $cgi->url(-relative => 1), -name => $self->FormName()) . $cgi->hidden(-name => 'Class', - -value => $self->{class}) . + -value => $self->{class}, + -override => 1) . $cgi->hidden(-name => 'SPROUT', -value => 1) . $cgi->h3($title); @@ -610,7 +626,7 @@ =head3 PutFeature -C<< $shelp->PutFeature($fquery); >> +C<< $shelp->PutFeature($fdata); >> Store a feature in the result cache. This is the workhorse method for most searches, since the primary data item in the database is features. @@ -621,8 +637,8 @@ the feature query object using the B method. For example, the following code adds columns for essentiality and virulence. - $fq->AddExtraColumns(essential => $essentialFlag, virulence => $vfactor); - $shelp->PutFeature($fq); + $fd->AddExtraColumns(essential => $essentialFlag, virulence => $vfactor); + $shelp->PutFeature($fd); For correct results, all values should be specified for all extra columns in all calls to B. (In particular, the column header names are computed on the first @@ -632,14 +648,14 @@ if (! $essentialFlag) { $essentialFlag = undef; } - $fq->AddExtraColumns(essential => $essentialFlag, virulence => $vfactor); - $shelp->PutFeature($fq); + $fd->AddExtraColumns(essential => $essentialFlag, virulence => $vfactor); + $shelp->PutFeature($fd); =over 4 -=item fquery +=item fdata -FeatureQuery object containing the current feature data. +B object containing the current feature data. =back @@ -647,33 +663,44 @@ sub PutFeature { # Get the parameters. - my ($self, $fq) = @_; + my ($self, $fd) = @_; + # Get the CGI query object. + my $cgi = $self->Q(); # Get the feature data. - my $record = $fq->Feature(); - my $extraCols = $fq->ExtraCols(); + my $record = $fd->Feature(); + my $extraCols = $fd->ExtraCols(); # Check for a first-call situation. if (! defined $self->{cols}) { - # Here we need to set up the column information. Start with the defaults. - $self->{cols} = $self->DefaultFeatureColumns(); - # Append the extras, sorted by column name. + Trace("Setting up the columns.") if T(3); + # Here we need to set up the column information. Start with the extras, + # sorted by column name. + my @colNames = (); for my $col (sort keys %{$extraCols}) { - push @{$self->{cols}}, "X=$col"; + push @colNames, "X=$col"; } + # Add the default columns. + push @colNames, $self->DefaultFeatureColumns(); + # Add any additional columns requested by the feature filter. + push @colNames, FeatureQuery::AdditionalColumns($self); + Trace("Full column list determined.") if T(3); + # Save the full list. + $self->{cols} = \@colNames; # Write out the column headers. This also prepares the cache file to receive # output. + Trace("Writing column headers.") if T(3); $self->WriteColumnHeaders(map { $self->FeatureColumnTitle($_) } @{$self->{cols}}); + Trace("Column headers written.") if T(3); } # Get the feature ID. - my ($fid) = $record->Value('Feature(id)'); + my $fid = $fd->FID(); # Loop through the column headers, producing the desired data. my @output = (); for my $colName (@{$self->{cols}}) { push @output, $self->FeatureColumnValue($colName, $record, $extraCols); } - # Compute the sort key. The sort key floats NMPDR organism features to the + # Compute the sort key. The sort key usually floats NMPDR organism features to the # top of the return list. - my $group = $self->FeatureGroup($fid); - my $key = ($group ? "A$group" : "ZZ"); + my $key = $self->SortKey($fd); # Write the feature data. $self->WriteColumnData($key, @output); } @@ -754,6 +781,7 @@ # Check for an open session file. if (defined $self->{fileHandle}) { # We found one, so close it. + Trace("Closing session file.") if T(2); close $self->{fileHandle}; } } @@ -771,19 +799,13 @@ my $retVal; # Get a digest encoder. my $md5 = Digest::MD5->new(); - # If we have a randomization file, use it to seed the digester. - if (open(R, "/dev/urandom")) { - my $b; - read(R, $b, 1024); - $md5->add($b); - } - # Add the PID and the time stamp. - $md5->add($$, gettimeofday); - # Hash it up and clean the result so that it works as a file name. - $retVal = $md5->b64digest(); - $retVal =~ s,/,\$,g; - $retVal =~ s,\+,@,g; - # Return it. + # Add the PID, the IP, and the time stamp. Note that the time stamp is + # actually two numbers, and we get them both because we're in list + # context. + $md5->add($$, $ENV{REMOTE_ADDR}, $ENV{REMOTE_PORT}, gettimeofday()); + # Hash up all this identifying data. + $retVal = $md5->hexdigest(); + # Return the result. return $retVal; } @@ -827,20 +849,9 @@ ['Genome(genus)', 'Genome(species)', 'Genome(unique-characterization)', 'Genome(primary-group)']); - # Null out the supporting group. - $group = "" if ($group eq $FIG_Config::otherGroup); - # If the organism does not exist, format an unknown name. - if (! defined($genus)) { - $orgName = "Unknown Genome $genomeID"; - } else { - # It does exist, so format the organism name. - $orgName = "$genus $species"; - if ($strain) { - $orgName .= " $strain"; - } - } - # Save this organism in the cache. - $cache->{$genomeID} = [$orgName, $group]; + # Format and cache the name and display group. + ($orgName, $group) = $self->SaveOrganismData($group, $genomeID, $genus, $species, + $strain); } # Return the result. return ($orgName, $group); @@ -942,8 +953,8 @@ } else { # Here we can get its genome data. $retVal = $self->Organism($genomeID); - # Append the type and number. - $retVal .= " [$type $num]"; + # Append the FIG ID. + $retVal .= " [$fid]"; } # Return the result. return $retVal; @@ -951,22 +962,15 @@ =head3 ComputeFASTA -C<< my $fasta = $shelp->ComputeFASTA($incomingType, $desiredType, $sequence); >> +C<< my $fasta = $shelp->ComputeFASTA($desiredType, $sequence); >> -Parse a sequence input and convert it into a FASTA string of the desired type. Note -that it is possible to convert a DNA sequence into a protein sequence, but the reverse -is not possible. +Parse a sequence input and convert it into a FASTA string of the desired type. =over 4 -=item incomingType - -C if this is a DNA sequence, C if this is a protein sequence. - =item desiredType -C to return a DNA sequence, C to return a protein sequence. If the -I<$incomingType> is C and this value is C, an error will be thrown. +C to return a DNA sequence, C to return a protein sequence. =item sequence @@ -988,78 +992,235 @@ sub ComputeFASTA { # Get the parameters. - my ($self, $incomingType, $desiredType, $sequence) = @_; + my ($self, $desiredType, $sequence) = @_; # Declare the return variable. If an error occurs, it will remain undefined. my $retVal; + # This variable will be cleared if an error is detected. + my $okFlag = 1; # Create variables to hold the FASTA label and data. my ($fastaLabel, $fastaData); + Trace("FASTA desired type is $desiredType.") if T(4); # Check for a feature specification. if ($sequence =~ /^\s*(\w+\|\S+)\s*$/) { # Here we have a feature ID in $1. We'll need the Sprout object to process # it. my $fid = $1; + Trace("Feature ID for fasta is $fid.") if T(3); my $sprout = $self->DB(); # Get the FIG ID. Note that we only use the first feature found. We are not # supposed to have redundant aliases, though we may have an ID that doesn't # exist. my ($figID) = $sprout->FeaturesByAlias($fid); if (! $figID) { - $self->SetMessage("No feature found with the ID \"$fid\"."); + $self->SetMessage("No gene found with the ID \"$fid\"."); + $okFlag = 0; } else { # Set the FASTA label. my $fastaLabel = $fid; # Now proceed according to the sequence type. - if ($desiredType =~ /prot/i) { + if ($desiredType eq 'prot') { # We want protein, so get the translation. $fastaData = $sprout->FeatureTranslation($figID); + Trace(length $fastaData . " characters returned for translation of $fastaLabel.") if T(3); } else { # We want DNA, so get the DNA sequence. This is a two-step process. my @locList = $sprout->FeatureLocation($figID); $fastaData = $sprout->DNASeq(\@locList); + Trace(length $fastaData . " characters returned for DNA of $fastaLabel.") if T(3); } } - } elsif ($incomingType =~ /prot/ && $desiredType =~ /dna/) { - # Here we're being asked to do an impossible conversion. - $self->SetMessage("Cannot convert a protein sequence to DNA."); } else { + Trace("Analyzing FASTA sequence.") if T(4); # Here we are expecting a FASTA. We need to see if there's a label. - if ($sequence =~ /^>\s*(\S.*)\s*\n(.+)$/) { + if ($sequence =~ /^>[\n\s]*(\S[^\n]*)\n(.+)$/s) { + Trace("Label \"$1\" found in match to sequence:\n$sequence") if T(4); # Here we have a label, so we split it from the data. $fastaLabel = $1; $fastaData = $2; } else { + Trace("No label found in match to sequence:\n$sequence") if T(4); # Here we have no label, so we create one and use the entire sequence # as data. - $fastaLabel = "User-specified $incomingType sequence"; + $fastaLabel = "User-specified $desiredType sequence"; $fastaData = $sequence; } # The next step is to clean the junk out of the sequence. $fastaData =~ s/\n//g; $fastaData =~ s/\s+//g; - # Finally, if the user wants to convert to protein, we do it here. Note that - # we've already prevented a conversion from protein to DNA. - if ($incomingType ne $desiredType) { - $fastaData = Sprout::Protein($fastaData); + # Finally, verify that it's DNA if we're doing DNA stuff. + if ($desiredType eq 'dna' && $fastaData =~ /[^agctxn]/i) { + $self->SetMessage("Invalid characters detected. Is the input really a DNA sequence?"); + $okFlag = 0; } } - # At this point, either "$fastaLabel" and "$fastaData" have values or an error is - # in progress. - if (defined $fastaLabel) { + Trace("FASTA data sequence: $fastaData") if T(4); + # Only proceed if no error was detected. + if ($okFlag) { # We need to format the sequence into 60-byte chunks. We use the infamous # grep-split trick. The split, because of the presence of the parentheses, # includes the matched delimiters in the output list. The grep strips out # the empty list items that appear between the so-called delimiters, since # the delimiters are what we want. my @chunks = grep { $_ } split /(.{1,60})/, $fastaData; - my $retVal = join("\n", ">$fastaLabel", @chunks, ""); + $retVal = join("\n", ">$fastaLabel", @chunks, ""); } # Return the result. return $retVal; } +=head3 SubsystemTree + +C<< my $tree = SearchHelper::SubsystemTree($sprout, %options); >> + +This method creates a subsystem selection tree suitable for passing to +L. Each leaf node in the tree will have a link to the +subsystem display page. In addition, each node can have a radio button. The +radio button alue is either CI, where I is +a classification string, or CI, where I is a subsystem ID. +Thus, it can either be used to filter by a group of related subsystems or a +single subsystem. + +=over 4 + +=item sprout + +Sprout database object used to get the list of subsystems. + +=item options + +Hash containing options for building the tree. + +=item RETURN + +Returns a reference to a tree list suitable for passing to L. + +=back + +The supported options are as follows. + +=over 4 + +=item radio + +TRUE if the tree should be configured for radio buttons. The default is FALSE. + +=item links + +TRUE if the tree should be configured for links. The default is TRUE. + +=back + +=cut + +sub SubsystemTree { + # Get the parameters. + my ($sprout, %options) = @_; + # Process the options. + my $optionThing = Tracer::GetOptions({ radio => 0, links => 1 }, \%options); + # Read in the subsystems. + my @subs = $sprout->GetAll(['Subsystem'], "ORDER BY Subsystem(classification), Subsystem(id)", [], + ['Subsystem(classification)', 'Subsystem(id)']); + # Put any unclassified subsystems at the end. They will always be at the beginning, so if one + # is at the end, ALL subsystems are unclassified and we don't bother. + if ($#subs >= 0 && $subs[$#subs]->[0] ne '') { + while ($subs[0]->[0] eq '') { + my $classLess = shift @subs; + push @subs, $classLess; + } + } + # Declare the return variable. + my @retVal = (); + # Each element in @subs represents a leaf node, so as we loop through it we will be + # producing one leaf node at a time. The leaf node is represented as a 2-tuple. The + # first element is a semi-colon-delimited list of the classifications for the + # subsystem. There will be a stack of currently-active classifications, which we will + # compare to the incoming classifications from the end backward. A new classification + # requires starting a new branch. A different classification requires closing an old + # branch and starting a new one. Each classification in the stack will also contain + # that classification's current branch. We'll add a fake classification at the + # beginning that we can use to represent the tree as a whole. + my $rootName = ''; + # Create the classification stack. Note the stack is a pair of parallel lists, + # one containing names and the other containing content. + my @stackNames = ($rootName); + my @stackContents = (\@retVal); + # Add a null entry at the end of the subsystem list to force an unrolling. + push @subs, [' ', undef]; + # Loop through the subsystems. + for my $sub (@subs) { + # Pull out the classification list and the subsystem ID. + my ($classString, $id) = @{$sub}; + Trace("Processing class \"$classString\" and subsystem $id.") if T(4); + # Convert the classification string to a list with the root classification in + # the front. + my @classList = ($rootName, split($FIG_Config::splitter, $classString)); + # Find the leftmost point at which the class list differs from the stack. + my $matchPoint = 0; + while ($matchPoint <= $#stackNames && $matchPoint <= $#classList && + $stackNames[$matchPoint] eq $classList[$matchPoint]) { + $matchPoint++; + } + Trace("Match point is $matchPoint. Stack length is " . scalar(@stackNames) . + ". Class List length is " . scalar(@classList) . ".") if T(4); + # Unroll the stack to the matchpoint. + while ($#stackNames >= $matchPoint) { + my $popped = pop @stackNames; + pop @stackContents; + Trace("\"$popped\" popped from stack.") if T(4); + } + # Start branches for any new classifications. + while ($#stackNames < $#classList) { + # The branch for a new classification contains its radio button + # data and then a list of children. So, at this point, if radio buttons + # are desired, we put them into the content. + my $newLevel = scalar(@stackNames); + my @newClassContent = (); + if ($optionThing->{radio}) { + my $newClassString = join($FIG_Config::splitter, @classList[1..$newLevel]); + push @newClassContent, { value => "classification=$newClassString%" }; + } + # The new classification node is appended to its parent's content + # and then pushed onto the stack. First, we need the node name. + my $nodeName = $classList[$newLevel]; + # Add the classification to its parent. This makes it part of the + # tree we'll be returning to the user. + push @{$stackContents[$#stackNames]}, $nodeName, \@newClassContent; + # Push the classification onto the stack. + push @stackContents, \@newClassContent; + push @stackNames, $nodeName; + Trace("\"$nodeName\" pushed onto stack.") if T(4); + } + # Now the stack contains all our parent branches. We add the subsystem to + # the branch at the top of the stack, but only if it's NOT the dummy node. + if (defined $id) { + # Compute the node name from the ID. + my $nodeName = $id; + $nodeName =~ s/_/ /g; + # Create the node's leaf hash. This depends on the value of the radio + # and link options. + my $nodeContent = {}; + if ($optionThing->{links}) { + # Compute the link value. + my $linkable = uri_escape($id); + $nodeContent->{link} = "../FIG/display_subsys.cgi?ssa_name=$linkable;request=show_ssa;sort=by_phylo;SPROUT=1"; + } + if ($optionThing->{radio}) { + # Compute the radio value. + $nodeContent->{value} = "id=$id"; + } + # Push the node into its parent branch. + Trace("\"$nodeName\" added to node list.") if T(4); + push @{$stackContents[$#stackNames]}, $nodeName, $nodeContent; + } + } + # Return the result. + return \@retVal; +} + + =head3 NmpdrGenomeMenu -C<< my $htmlText = $shelp->NmpdrGenomeMenu($menuName, \%options, \@selected); >> +C<< my $htmlText = $shelp->NmpdrGenomeMenu($menuName, $multiple, \@selected, $rows); >> This method creates a hierarchical HTML menu for NMPDR genomes organized by category. The category indicates the low-level NMPDR group. Organizing the genomes in this way makes it @@ -1071,12 +1232,9 @@ Name to give to the menu. -=item options +=item multiple -Reference to a hash containing the options to be applied to the C menu inside a form. @@ -1094,42 +1264,65 @@ sub NmpdrGenomeMenu { # Get the parameters. - my ($self, $menuName, $options, $selected) = @_; + my ($self, $menuName, $multiple, $selected, $rows, $cross) = @_; # Get the Sprout and CGI objects. my $sprout = $self->DB(); my $cgi = $self->Q(); + # Compute the row count. + if (! defined $rows) { + $rows = ($multiple ? 10 : 1); + } + # Create the multiple tag. + my $multipleTag = ($multiple ? " multiple" : ""); # Get the form name. my $formName = $self->FormName(); - # Get a list of all the genomes in group order. In fact, we only need them ordered - # by name (genus,species,strain), but putting primary-group in front enables us to - # take advantage of an existing index. - my @genomeList = $sprout->GetAll(['Genome'], - "ORDER BY Genome(primary-group), Genome(genus), Genome(species), Genome(unique-characterization)", - [], ['Genome(primary-group)', 'Genome(id)', - 'Genome(genus)', 'Genome(species)', - 'Genome(unique-characterization)']); - # Create a hash to organize the genomes by group. Each group will contain a list of - # 2-tuples, the first element being the genome ID and the second being the genome - # name. - my %groupHash = (); - for my $genome (@genomeList) { - # Get the genome data. - my ($group, $genomeID, $genus, $species, $strain) = @{$genome}; - # Form the genome name. - my $name = "$genus $species"; - if ($strain) { - $name .= " $strain"; + # Check to see if we already have a genome list in memory. + my $genomes = $self->{genomeList}; + my $groupHash; + if (defined $genomes) { + # We have a list ready to use. + $groupHash = $genomes; + } else { + # Get a list of all the genomes in group order. In fact, we only need them ordered + # by name (genus,species,strain), but putting primary-group in front enables us to + # take advantage of an existing index. + my @genomeList = $sprout->GetAll(['Genome'], + "ORDER BY Genome(primary-group), Genome(genus), Genome(species), Genome(unique-characterization)", + [], ['Genome(primary-group)', 'Genome(id)', + 'Genome(genus)', 'Genome(species)', + 'Genome(unique-characterization)']); + # Create a hash to organize the genomes by group. Each group will contain a list of + # 2-tuples, the first element being the genome ID and the second being the genome + # name. + my %gHash = (); + for my $genome (@genomeList) { + # Get the genome data. + my ($group, $genomeID, $genus, $species, $strain) = @{$genome}; + # Compute and cache its name and display group. + my ($name, $displayGroup) = $self->SaveOrganismData($group, $genomeID, $genus, $species, + $strain); + # Push the genome into the group's list. Note that we use the real group + # name here, not the display group name. + push @{$gHash{$group}}, [$genomeID, $name]; } - # Push the genome into the group's list. - push @{$groupHash{$group}}, [$genomeID, $name]; + # Save the genome list for future use. + $self->{genomeList} = \%gHash; + $groupHash = \%gHash; } # Now we are ready to unroll the menu out of the group hash. First, we sort the groups, putting # the supporting-genome group last. - my @groups = sort grep { $_ ne $FIG_Config::otherGroup } keys %groupHash; + my @groups = sort grep { $_ ne $FIG_Config::otherGroup } keys %{$groupHash}; push @groups, $FIG_Config::otherGroup; - # Next, create a hash that specifies the pre-selected entries. - my %selectedHash = map { $_ => 1 } @{$selected}; - # Now it gets complicated. We need a way to mark all the NMPDR genomes. + # Next, create a hash that specifies the pre-selected entries. Note that we need to deal + # with the possibility of undefined values in the incoming list. + my %selectedHash = (); + if (defined $selected) { + %selectedHash = map { $_ => 1 } grep { defined($_) } @{$selected}; + } + # Now it gets complicated. We need a way to mark all the NMPDR genomes. We take advantage + # of the fact they come first in the list. We'll accumulate a count of the NMPDR genomes + # and use that to make the selections. + my $nmpdrCount = 0; # Create the type counters. my $groupCount = 1; # Compute the ID for the status display. @@ -1138,30 +1331,33 @@ my $showSelect = "showSelected($menuName, '$divID', 1000);"; # If multiple selection is supported, create an onChange event. my $onChange = ""; - if ($options->{multiple}) { + if ($cross) { + # Here we have a paired menu. Selecting something in our menu unselects it in the + # other and redisplays the status of both. + $onChange = " onChange=\"crossUnSelect($menuName, '$divID', $cross, '${formName}_${cross}_status', 1000)\""; + } elsif ($multiple) { + # This is an unpaired menu, so all we do is redisplay our status. $onChange = " onChange=\"$showSelect\""; } # Create the SELECT tag and stuff it into the output array. - my $select = "<" . join(" ", "SELECT name=\"$menuName\"$onChange", map { " $_=\"$options->{$_}\"" } keys %{$options}) . ">"; - my @lines = ($select); + my @lines = (""; # Check for multiple selection. - if ($options->{multiple}) { - # Since multi-select is on, we can set up some buttons to set and clear selections. + if ($multiple) { + # Multi-select is on, so we need to add some selection helpers. First is + # the search box. This allows the user to type text and have all genomes containing + # the text selected automatically. + my $searchThingName = "${menuName}_SearchThing"; + push @lines, "
" . + " " . + ""; + # Next are the buttons to set and clear selections. push @lines, "
"; - push @lines, ""; push @lines, ""; - push @lines, ""; - push @lines, ""; + push @lines, ""; + push @lines, ""; + push @lines, ""; # Add the status display, too. push @lines, "
"; # Queue to update the status display when the form loads. We need to modify the show statement @@ -1185,6 +1388,9 @@ # in case we decide to twiddle the parameters. $showSelect =~ s/showSelected\(/showSelected\(thisForm\./; $self->QueueFormScript($showSelect); + # Finally, add this parameter to the list of genome parameters. This enables us to + # easily find all the parameters used to select one or more genomes. + push @{$self->{genomeParms}}, $menuName; } # Assemble all the lines into a string. my $retVal = join("\n", @lines, ""); @@ -1192,6 +1398,56 @@ return $retVal; } +=head3 PropertyMenu + +C<< my $htmlText = $shelp->PropertyMenu($menuName, $selected, $force); >> + +Generate a property name dropdown menu. + +=over 4 + +=item menuName + +Name to give to the menu. + +=item selected + +Value of the property name to pre-select. + +=item force (optional) + +If TRUE, then the user will be forced to choose a property name. If FALSE, +then an additional menu choice will be provided to select nothing. + +=item RETURN + +Returns a dropdown menu box that allows the user to select a property name. An additional +selection entry will be provided for selecting no property name + +=back + +=cut + +sub PropertyMenu { + # Get the parameters. + my ($self, $menuName, $selected, $force) = @_; + # Get the CGI and Sprout objects. + my $sprout = $self->DB(); + my $cgi = $self->Q(); + # Create the property name list. + my @propNames = (); + if (! $force) { + push @propNames, ""; + } + # Get all the property names, putting them after the null choice if one exists. + push @propNames, $sprout->GetChoices('Property', 'property-name'); + # Create a menu from them. + my $retVal = $cgi->popup_menu(-name=> $menuName, -values => \@propNames, + -default => $selected); + # Return the result. + return $retVal; +} + =head3 MakeTable C<< my $htmlText = $shelp->MakeTable(\@rows); >> @@ -1242,26 +1498,45 @@ =head3 SubmitRow -C<< my $htmlText = $shelp->SubmitRow(); >> +C<< my $htmlText = $shelp->SubmitRow($caption); >> Returns the HTML text for the row containing the page size control and the submit button. All searches should have this row somewhere near the top of the form. +=over 4 + +=item caption (optional) + +Caption to be put on the search button. The default is C. + +=item RETURN + +Returns a table row containing the controls for submitting the search +and tuning the results. + +=back + =cut sub SubmitRow { # Get the parameters. - my ($self) = @_; + my ($self, $caption) = @_; my $cgi = $self->Q(); - # Declare the return variable. + # Compute the button caption. + my $realCaption = (defined $caption ? $caption : 'Go'); + # Get the current page size. + my $pageSize = $cgi->param('PageSize'); + # Get the incoming external-link flag. + my $aliases = ($cgi->param('ShowAliases') ? 1 : 0); + # Create the row. my $retVal = $cgi->Tr($cgi->td("Results/Page"), $cgi->td($cgi->popup_menu(-name => 'PageSize', - -values => [10, 25, 45, 100, 1000], - -default => $cgi->param('PageSize'))), + -values => [10, 25, 50, 100, 1000], + -default => $pageSize)), $cgi->td($cgi->submit(-class => 'goButton', -name => 'Search', - -value => 'Go'))); + -value => $realCaption))); # Return the result. return $retVal; } @@ -1270,9 +1545,9 @@ C<< my $htmlText = $shelp->FeatureFilterRows(); >> -This method creates table rows that can be used to filter features. There are -two rows returned, and the values can be used to select features by genome -using the B object. +This method creates table rows that can be used to filter features. The form +values can be used to select features by genome using the B +object. =cut @@ -1322,7 +1597,8 @@ # Get the feature location string. my $loc = $sprout->FeatureLocation($feat); # Compute the contig, start, and stop points. - my($start, $stop, $contig) = BasicLocation::Parse($loc); + my($contig, $start, $stop) = BasicLocation::Parse($loc); + Trace("Start and stop are ($start,$stop) on contig $contig.") if T(3); # Now we need to do some goofiness to insure that the location is not too # big and that we get some surrounding stuff. my $mid = int(($start + $stop) / 2); @@ -1352,181 +1628,767 @@ } my $seg_id = $contig; $seg_id =~ s/:/--/g; + Trace("Show limits are ($show_start,$show_stop) in genome $genomeID with ref $seg_id.") if T(3); # Assemble all the pieces. - $retVal = "gbrowse.cgi/GB_$genomeID?ref=$seg_id&start=$show_start&stop=$show_stop"; + $retVal = "gbrowse.cgi/GB_$genomeID?ref=$seg_id;start=$show_start;stop=$show_stop"; } # Return the result. return $retVal; } -=head2 Feature Column Methods +=head3 GetGenomes -The methods in this column manage feature column data. If you want to provide the -capability to include new types of data in feature columns, then all the changes -are made to this section of the source file. Technically, this should be implemented -using object-oriented methods, but this is simpler for non-programmers to maintain. -To add a new column of feature data, you must first give it a name. For example, -the name for the protein page link column is C. If the column is to appear -in the default list of feature columns, add it to the list returned by -L. Then add code to produce the column title to -L and code to produce its value to L, and -everything else will happen automatically. +C<< my @genomeList = $shelp->GetGenomes($parmName); >> -There is one special column name syntax for extra columns (that is, nonstandard -feature columns). If the column name begins with C, then it is presumed to be -an extra column. The column title is the text after the C, and its value is -pulled from the extra column hash. +Return the list of genomes specified by the specified CGI query parameter. +If the request method is POST, then the list of genome IDs is returned +without preamble. If the request method is GET and the parameter is not +specified, then it is treated as a request for all genomes. This makes it +easier for web pages to link to a search that wants to specify all genomes. -=head3 DefaultFeatureColumns +=over 4 + +=item parmName + +Name of the parameter containing the list of genomes. This will be the +first parameter passed to the L call that created the +genome selection control on the form. -C<< my $colNames = $shelp->DefaultFeatureColumns(); >> +=item RETURN + +Returns a list of the genomes to process. -Return a reference to a list of the default feature column identifiers. These -identifiers can be passed to L and L in -order to produce the column titles and row values. +=back =cut -sub DefaultFeatureColumns { +sub GetGenomes { + # Get the parameters. + my ($self, $parmName) = @_; + # Get the CGI query object. + my $cgi = $self->Q(); + # Get the list of genome IDs in the request header. + my @retVal = $cgi->param($parmName); + Trace("Genome list for $parmName is (" . join(", ", @retVal) . ") with method " . $cgi->request_method() . ".") if T(3); + # Check for the special GET case. + if ($cgi->request_method() eq "GET" && ! @retVal) { + # Here the caller wants all the genomes. + my $sprout = $self->DB(); + @retVal = $sprout->Genomes(); + } + # Return the result. + return @retVal; +} + +=head3 GetHelpText + +C<< my $htmlText = $shelp->GetHelpText(); >> + +Get the help text for this search. The help text is stored in files on the template +server. The help text for a specific search is taken from a file named +CIC<.inc> in the template directory C<$FIG_Config::template_url>. +There are also three standard help files: C describes the +feature filtering performed by the B object, C +describes how to use a multiple-selection genome control, and C +describes the standard controls for a search, such as page size, URL display, and +external alias display. + +=cut + +sub GetHelpText { # Get the parameters. my ($self) = @_; + # Create a list to hold the pieces of the help. + my @helps = (); + # Get the template directory URL. + my $urlBase = $FIG_Config::template_url; + # Start with the specific help. + my $class = $self->{class}; + push @helps, PageBuilder::GetPage("$urlBase/SearchHelp_$class.inc"); + # Add the genome control help if needed. + if (scalar @{$self->{genomeParms}}) { + push @helps, PageBuilder::GetPage("$urlBase/SearchHelp1_GenomeControl.inc"); + } + # Next the filter help. + if ($self->{filtered}) { + push @helps, PageBuilder::GetPage("$urlBase/SearchHelp1_Filtering.inc"); + } + # Finally, the standard help. + push @helps, PageBuilder::GetPage("$urlBase/SearchHelp1_Standard.inc"); + # Assemble the pieces. + my $retVal = join("\n

 

\n", @helps); # Return the result. - return ['orgName', 'function', 'gblink', 'protlink']; + return $retVal; } -=head3 FeatureColumnTitle +=head3 ComputeSearchURL -C<< my $title = $shelp->FeatureColumnTitle($colName); >> +C<< my $url = $shelp->ComputeSearchURL(%overrides); >> -Return the column heading title to be used for the specified feature column. +Compute the GET-style URL for the current search. In order for this to work, there +must be a copy of the search form on the current page. This will always be the +case if the search is coming from C. + +A little expense is involved in order to make the URL as smart as possible. The +main complication is that if the user specified all genomes, we'll want to +remove the parameter entirely from a get-style URL. =over 4 -=item name +=item overrides -Name of the desired feature column. +Hash containing override values for the parameters, where the parameter name is +the key and the parameter value is the override value. If the override value is +C, the parameter will be deleted from the result. =item RETURN -Returns the title to be used as the column header for the named feature column. +Returns a GET-style URL for invoking the search with the specified overrides. =back =cut -sub FeatureColumnTitle { +sub ComputeSearchURL { # Get the parameters. - my ($self, $colName) = @_; - # Declare the return variable. We default to a blank column name. - my $retVal = " "; - # Process the column name. - if ($colName =~ /^X=(.+)$/) { - # Here we have an extra column. - $retVal = $1; - } elsif ($colName eq 'orgName') { - $retVal = "Name"; - } elsif ($colName eq 'fid') { - $retVal = "FIG ID"; - } elsif ($colName eq 'alias') { - $retVal = "External Aliases"; - } elsif ($colName eq 'function') { - $retVal = "Functional Assignment"; - } elsif ($colName eq 'gblink') { - $retVal = "GBrowse"; - } elsif ($colName eq 'protlink') { - $retVal = "NMPDR Protein Page"; - } elsif ($colName eq 'group') { - $retVal = "NMDPR Group"; + my ($self, %overrides) = @_; + # Get the database and CGI query object. + my $cgi = $self->Q(); + my $sprout = $self->DB(); + # Start with the full URL. + my $retVal = $cgi->url(-full => 1); + # Get all the query parameters in a hash. + my %parms = $cgi->Vars(); + # Now we need to do some fixing. Each multi-valued parameter is encoded as a string with null + # characters separating the individual values. We have to convert those to lists. In addition, + # the multiple-selection genome parameters and the feature type parameter must be checked to + # determine whether or not they can be removed from the URL. First, we get a list of the + # genome parameters and a list of all genomes. Note that we only need the list if a + # multiple-selection genome parameter has been found on the form. + my %genomeParms = map { $_ => 1 } @{$self->{genomeParms}}; + my @genomeList; + if (keys %genomeParms) { + @genomeList = $sprout->Genomes(); + } + # Create a list to hold the URL parameters we find. + my @urlList = (); + # Now loop through the parameters in the hash, putting them into the output URL. + for my $parmKey (keys %parms) { + # Get a list of the parameter values. If there's only one, we'll end up with + # a singleton list, but that's okay. + my @values = split (/\0/, $parms{$parmKey}); + # Check for special cases. + if (grep { $_ eq $parmKey } qw(SessionID ResultCount Page PageSize Trace TF)) { + # These are bookkeeping parameters we don't need to start a search. + @values = (); + } elsif ($parmKey =~ /_SearchThing$/) { + # Here the value coming in is from a genome control's search thing. It does + # not affect the results of the search, so we clear it. + @values = (); + } elsif ($genomeParms{$parmKey}) { + # Here we need to see if the user wants all the genomes. If he does, + # we erase all the values just like with features. + my $allFlag = $sprout->IsAllGenomes(\@values, \@genomeList); + if ($allFlag) { + @values = (); + } + } elsif (exists $overrides{$parmKey}) { + # Here the value is being overridden, so we skip it for now. + @values = (); + } + # If we still have values, create the URL parameters. + if (@values) { + push @urlList, map { "$parmKey=" . uri_escape($_) } @values; + } + } + # Now do the overrides. + for my $overKey (keys %overrides) { + # Only use this override if it's not a delete marker. + if (defined $overrides{$overKey}) { + push @urlList, "$overKey=" . uri_escape($overrides{$overKey}); + } } + # Add the parameters to the URL. + $retVal .= "?" . join(";", @urlList); # Return the result. return $retVal; } -=head3 FeatureColumnValue +=head3 GetRunTimeValue -C<< my $value = $shelp->FeatureColumnValue($colName, $fid, \%extraCols); >> +C<< my $htmlText = $shelp->GetRunTimeValue($text); >> -Return the value to be displayed in the specified feature column. +Compute a run-time column value. =over 4 -=item colName +=item text -Name of the column to be displayed. +The run-time column text. It consists of 2 percent signs, a column type, an equal +sign, and the data for the current row. -=item record +=item RETURN -DBObject record for the feature being displayed in the current row. +Returns the fully-formatted HTML text to go into the current column of the current row. -=item extraCols +=back -Reference to a hash of extra column names to values. If the incoming column name -begins with C, its value will be taken from this hash. +=cut -=item RETURN +sub GetRunTimeValue { + # Get the parameters. + my ($self, $text) = @_; + # Declare the return variable. + my $retVal; + # Parse the incoming text. + if ($text =~ /^%%([^=]+)=(.*)$/) { + $retVal = $self->RunTimeColumns($1, $2); + } else { + Confess("Invalid run-time column string \"$text\" encountered in session file."); + } + # Return the result. + return $retVal; +} -Returns the HTML to be displayed in the named column for the specified feature. +=head3 AdvancedClassList -=back +C<< my @classes = SearchHelper::AdvancedClassList(); >> + +Return a list of advanced class names. This list is used to generate the directory +of available searches on the search page. + +We use the %INC variable to accomplish this. =cut -sub FeatureColumnValue { - # Get the parameters. - my ($self, $colName, $record, $extraCols) = @_; - # Get the sprout and CGI objects. - my $cgi = $self->Q(); - my $sprout = $self->DB(); - # Get the feature ID. - my ($fid) = $record->Value('Feature(id)'); - # Declare the return variable. Denote that we default to a non-breaking space, - # which will translate to an empty table cell (rather than a table cell with no - # interior, which is what you get for a null string). - my $retVal = " "; - # Process according to the column name. - if ($colName =~ /^X=(.+)$/) { - # Here we have an extra column. Only update if the value exists. Note that - # a value of C is treated as a non-existent value, because the - # caller may have put "colName => undef" in the "PutFeature" call in order - # to insure we know the extra column exists. +sub AdvancedClassList { + my @retVal = map { $_ =~ /^SH(\w+)\.pm/; $1 } grep { $_ =~ /^SH/ } keys %INC; + return @retVal; +} + +=head3 SelectionTree + +C<< my $htmlText = SearchHelper::SelectionTree($cgi, \%tree, %options); >> + +Display a selection tree. + +This method creates the HTML for a tree selection control. The tree is implemented as a set of +nested HTML unordered lists. Each selectable element of the tree will contain a radio button. In +addition, some of the tree nodes can contain hyperlinks. + +The tree itself is passed in as a multi-level list containing node names followed by +contents. Each content element is a reference to a similar list. The first element of +each list may be a hash reference. If so, it should contain one or both of the following +keys. + +=over 4 + +=item link + +The navigation URL to be popped up if the user clicks on the node name. + +=item value + +The form value to be returned if the user selects the tree node. + +=back + +The presence of a C key indicates the node name will be hyperlinked. The presence of +a C key indicates the node name will have a radio button. If a node has no children, +you may pass it a hash reference instead of a list reference. + +The following example shows the hash for a three-level tree with links on the second level and +radio buttons on the third. + + [ Objects => [ + Entities => [ + {link => "../docs/WhatIsAnEntity.html"}, + Genome => {value => 'GenomeData'}, + Feature => {value => 'FeatureData'}, + Contig => {value => 'ContigData'}, + ], + Relationships => [ + {link => "../docs/WhatIsARelationShip.html"}, + HasFeature => {value => 'GenomeToFeature'}, + IsOnContig => {value => 'FeatureToContig'}, + ] + ] + ] + +Note how each leaf of the tree has a hash reference for its value, while the branch nodes +all have list references. + +This next example shows how to set up a taxonomy selection field. The value returned +by the tree control will be the taxonomy string for the selected node ready for use +in a LIKE-style SQL filter. Only the single branch ending in campylobacter is shown for +reasons of space. + + [ All => [ + {value => "%"}, + Bacteria => [ + {value => "Bacteria%"}, + Proteobacteria => [ + {value => "Bacteria; Proteobacteria%"}, + Epsilonproteobacteria => [ + {value => "Bacteria; Proteobacteria;Epsilonproteobacteria%"}, + Campylobacterales => [ + {value => "Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales%"}, + Campylobacteraceae => + {value => "Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales; Campylobacteraceae%"}, + ... + ] + ... + ] + ... + ] + ... + ] + ... + ] + ] + + +This method of tree storage allows the caller to control the order in which the tree nodes +are displayed and to completely control value selection and use of hyperlinks. It is, however +a bit complicated. Eventually, tree-building classes will be provided to simplify things. + +The parameters to this method are as follows. + +=over 4 + +=item cgi + +CGI object used to generate the HTML. + +=item tree + +Reference to a hash describing a tree. See the description above. + +=item options + +Hash containing options for the tree display. + +=back + +The allowable options are as follows + +=over 4 + +=item nodeImageClosed + +URL of the image to display next to the tree nodes when they are collapsed. Clicking +on the image will expand a section of the tree. The default is C<../FIG/Html/plus.gif>. + +=item nodeImageOpen + +URL of the image to display next to the tree nodes when they are expanded. Clicking +on the image will collapse a section of the tree. The default is C<../FIG/Html/minus.gif>. + +=item style + +Style to use for the tree. The default is C. Because the tree style is implemented +as nested lists, the key components of this style are the definitions for the C
    and +C
  • tags. The default style file contains the following definitions. + + .tree ul { + margin-left: 0; padding-left: 22px + } + .tree li { + list-style-type: none; + } + +The default image is 22 pixels wide, so in the above scheme each tree level is indented from its +parent by the width of the node image. This use of styles limits the things we can do in formatting +the tree, but it has the advantage of vastly simplifying the tree creation. + +=item name + +Field name to give to the radio buttons in the tree. The default is C. + +=item target + +Frame target for links. The default is C<_self>. + +=item selected + +If specified, the value of the radio button to be pre-selected. + +=back + +=cut + +sub SelectionTree { + # Get the parameters. + my ($cgi, $tree, %options) = @_; + # Get the options. + my $optionThing = Tracer::GetOptions({ name => 'selection', + nodeImageClosed => '../FIG/Html/plus.gif', + nodeImageOpen => '../FIG/Html/minus.gif', + style => 'tree', + target => '_self', + selected => undef}, + \%options); + # Declare the return variable. We'll do the standard thing with creating a list + # of HTML lines and rolling them together at the end. + my @retVal = (); + # Only proceed if the tree is present. + if (defined($tree)) { + # Validate the tree. + if (ref $tree ne 'ARRAY') { + Confess("Selection tree is not a list reference."); + } elsif (scalar @{$tree} == 0) { + # The tree is empty, so we do nothing. + } elsif ($tree->[0] eq 'HASH') { + Confess("Hash reference found at start of selection tree. The tree as a whole cannot have attributes, only tree nodes."); + } else { + # Here we have a real tree. Apply the tree style. + push @retVal, $cgi->start_div({ class => $optionThing->{style} }); + # Give us a DIV ID. + my $divID = GetDivID($optionThing->{name}); + # Show the tree. + push @retVal, ShowBranch($cgi, "(root)", $divID, $tree, $optionThing, 'block'); + # Close the DIV block. + push @retVal, $cgi->end_div(); + } + } + # Return the result. + return join("\n", @retVal, ""); +} + +=head3 ShowBranch + +C<< my @htmlLines = SearchHelper::ShowBranch($cgi, $label, $id, $branch, $options, $displayType); >> + +This is a recursive method that displays a branch of the tree. + +=over 4 + +=item cgi + +CGI object used to format HTML. + +=item label + +Label of this tree branch. It is only used in error messages. + +=item id + +ID to be given to this tree branch. The ID is used in the code that expands and collapses +tree nodes. + +=item branch + +Reference to a list containing the content of the tree branch. The list contains an optional +hash reference that is ignored and the list of children, each child represented by a name +and then its contents. The contents could by a hash reference (indicating the attributes +of a leaf node), or another tree branch. + +=item options + +Options from the original call to L. + +=item displayType + +C if the contents of this list are to be displayed, C if they are to be +hidden. + +=item RETURN + +Returns one or more HTML lines that can be used to display the tree branch. + +=back + +=cut + +sub ShowBranch { + # Get the parameters. + my ($cgi, $label, $id, $branch, $options, $displayType) = @_; + # Declare the return variable. + my @retVal = (); + # Start the branch. + push @retVal, $cgi->start_ul({ id => $id, style => "display:$displayType" }); + # Check for the hash and choose the start location accordingly. + my $i0 = (ref $branch->[0] eq 'HASH' ? 1 : 0); + # Get the list length. + my $i1 = scalar(@{$branch}); + # Verify we have an even number of elements. + if (($i1 - $i0) % 2 != 0) { + Trace("Branch elements are from $i0 to $i1.") if T(3); + Confess("Odd number of elements in tree branch $label."); + } else { + # Loop through the elements. + for (my $i = $i0; $i < $i1; $i += 2) { + # Get this node's label and contents. + my ($myLabel, $myContent) = ($branch->[$i], $branch->[$i+1]); + # Get an ID for this node's children (if any). + my $myID = GetDivID($options->{name}); + # Now we need to find the list of children and the options hash. + # This is a bit ugly because we allow the shortcut of a hash without an + # enclosing list. First, we need some variables. + my $attrHash = {}; + my @childHtml = (); + my $hasChildren = 0; + if (! ref $myContent) { + Confess("Invalid tree definition. Scalar found as content of node \"$myLabel\"."); + } elsif (ref $myContent eq 'HASH') { + # Here the node is a leaf and its content contains the link/value hash. + $attrHash = $myContent; + } elsif (ref $myContent eq 'ARRAY') { + # Here the node may be a branch. Its content is a list. + my $len = scalar @{$myContent}; + if ($len >= 1) { + # Here the first element of the list could by the link/value hash. + if (ref $myContent->[0] eq 'HASH') { + $attrHash = $myContent->[0]; + # If there's data in the list besides the hash, it's our child list. + # We can pass the entire thing as the child list, because the hash + # is ignored. + if ($len > 1) { + $hasChildren = 1; + } + } else { + $hasChildren = 1; + } + # If we have children, create the child list with a recursive call. + if ($hasChildren) { + Trace("Processing children of $myLabel.") if T(4); + push @childHtml, ShowBranch($cgi, $myLabel, $myID, $myContent, $options, 'none'); + } + } + } + # Okay, it's time to pause and take stock. We have the label of the current node + # in $myLabel, its attributes in $attrHash, and if it is NOT a leaf node, we + # have a child list in @childHtml. If it IS a leaf node, $hasChildren is 0. + # Compute the image HTML. It's tricky, because we have to deal with the open and + # closed images. + my @images = ($options->{nodeImageOpen}, $options->{nodeImageClosed}); + my $image = $images[$hasChildren]; + my $prefixHtml = $cgi->img({src => $image, id => "${myID}img"}); + if ($hasChildren) { + # If there are children, we wrap the image in a toggle hyperlink. + $prefixHtml = $cgi->a({ onClick => "javascript:treeToggle('$myID','$images[0]', '$images[1]')" }, + $prefixHtml); + } + # Now the radio button, if any. Note we use "defined" in case the user wants the + # value to be 0. + if (defined $attrHash->{value}) { + # Due to a glitchiness in the CGI stuff, we have to build the attribute + # hash for the "input" method. If the item is pre-selected, we add + # "checked => undef" to the hash. Otherwise, we can't have "checked" + # at all. + my $radioParms = { type => 'radio', + name => $options->{name}, + value => $attrHash->{value}, + }; + if (defined $options->{selected} && $options->{selected} eq $attrHash->{value}) { + $radioParms->{checked} = undef; + } + $prefixHtml .= $cgi->input($radioParms); + } + # Next, we format the label. + my $labelHtml = $myLabel; + Trace("Formatting tree node for $myLabel.") if T(4); + # Apply a hyperlink if necessary. + if (defined $attrHash->{link}) { + $labelHtml = $cgi->a({ href => $attrHash->{link}, target => $options->{target} }, + $labelHtml); + } + # Finally, roll up the child HTML. If there are no children, we'll get a null string + # here. + my $childHtml = join("\n", @childHtml); + # Now we have all the pieces, so we can put them together. + push @retVal, $cgi->li("$prefixHtml$labelHtml$childHtml"); + } + } + # Close the tree branch. + push @retVal, $cgi->end_ul(); + # Return the result. + return @retVal; +} + +=head3 GetDivID + +C<< my $idString = SearchHelper::GetDivID($name); >> + +Return a new HTML ID string. + +=over 4 + +=item name + +Name to be prefixed to the ID string. + +=item RETURN + +Returns a hopefully-unique ID string. + +=back + +=cut + +sub GetDivID { + # Get the parameters. + my ($name) = @_; + # Compute the ID. + my $retVal = "elt_$name$divCount"; + # Increment the counter to make sure this ID is not re-used. + $divCount++; + # Return the result. + return $retVal; +} + +=head2 Feature Column Methods + +The methods in this section manage feature column data. If you want to provide the +capability to include new types of data in feature columns, then all the changes +are made to this section of the source file. Technically, this should be implemented +using object-oriented methods, but this is simpler for non-programmers to maintain. +To add a new column of feature data, you must first give it a name. For example, +the name for the protein page link column is C. If the column is to appear +in the default list of feature columns, add it to the list returned by +L. Then add code to produce the column title to +L and code to produce its value to L, and +everything else will happen automatically. + +There is one special column name syntax for extra columns (that is, nonstandard +feature columns). If the column name begins with C, then it is presumed to be +an extra column. The column title is the text after the C, and its value is +pulled from the extra column hash. + +=head3 DefaultFeatureColumns + +C<< my @colNames = $shelp->DefaultFeatureColumns(); >> + +Return a list of the default feature column identifiers. These identifiers can +be passed to L and L in order to +produce the column titles and row values. + +=cut + +sub DefaultFeatureColumns { + # Get the parameters. + my ($self) = @_; + # Return the result. + return qw(orgName function gblink protlink); +} + +=head3 FeatureColumnTitle + +C<< my $title = $shelp->FeatureColumnTitle($colName); >> + +Return the column heading title to be used for the specified feature column. + +=over 4 + +=item name + +Name of the desired feature column. + +=item RETURN + +Returns the title to be used as the column header for the named feature column. + +=back + +=cut + +sub FeatureColumnTitle { + # Get the parameters. + my ($self, $colName) = @_; + # Declare the return variable. We default to a blank column name. + my $retVal = " "; + # Process the column name. + if ($colName =~ /^X=(.+)$/) { + # Here we have an extra column. + $retVal = $1; + } elsif ($colName eq 'alias') { + $retVal = "External Aliases"; + } elsif ($colName eq 'fid') { + $retVal = "FIG ID"; + } elsif ($colName eq 'function') { + $retVal = "Functional Assignment"; + } elsif ($colName eq 'gblink') { + $retVal = "GBrowse"; + } elsif ($colName eq 'group') { + $retVal = "NMDPR Group"; + } elsif ($colName =~ /^keyword:(.+)$/) { + $retVal = ucfirst $1; + } elsif ($colName eq 'orgName') { + $retVal = "Organism and Gene ID"; + } elsif ($colName eq 'protlink') { + $retVal = "NMPDR Protein Page"; + } elsif ($colName eq 'subsystem') { + $retVal = "Subsystems"; + } + # Return the result. + return $retVal; +} + + +=head3 FeatureColumnValue + +C<< my $value = $shelp->FeatureColumnValue($colName, $fid, \%extraCols); >> + +Return the value to be displayed in the specified feature column. + +=over 4 + +=item colName + +Name of the column to be displayed. + +=item record + +DBObject record for the feature being displayed in the current row. + +=item extraCols + +Reference to a hash of extra column names to values. If the incoming column name +begins with C, its value will be taken from this hash. + +=item RETURN + +Returns the HTML to be displayed in the named column for the specified feature. + +=back + +=cut + +sub FeatureColumnValue { + # Get the parameters. + my ($self, $colName, $record, $extraCols) = @_; + # Get the sprout and CGI objects. + my $cgi = $self->Q(); + my $sprout = $self->DB(); + # Get the feature ID. + my ($fid) = $record->Value('Feature(id)'); + # Declare the return variable. Denote that we default to a non-breaking space, + # which will translate to an empty table cell (rather than a table cell with no + # interior, which is what you get for a null string). + my $retVal = " "; + # Process according to the column name. + if ($colName =~ /^X=(.+)$/) { + # Here we have an extra column. Only update if the value exists. Note that + # a value of C is treated as a non-existent value, because the + # caller may have put "colName => undef" in the "PutFeature" call in order + # to insure we know the extra column exists. if (defined $extraCols->{$1}) { $retVal = $extraCols->{$1}; } - } elsif ($colName eq 'orgName') { - # Here we want the formatted organism name and feature number. - $retVal = $self->FeatureName($fid); + } elsif ($colName eq 'alias') { + # In this case, the user wants a list of external aliases for the feature. + # These are very expensive, so we compute them when the row is displayed. + $retVal = "%%alias=$fid"; } elsif ($colName eq 'fid') { # Here we have the raw feature ID. We hyperlink it to the protein page. $retVal = HTML::set_prot_links($fid); - } elsif ($colName eq 'alias') { - # In this case, the user wants a list of external aliases for the feature. - # The complicated part is we have to hyperlink them. First, get the - # aliases. - my @aliases = $sprout->FeatureAliases($fid); - # Only proceed if we found some. - if (@aliases) { - # Join the aliases into a comma-delimited list. - my $aliasList = join(", ", @aliases); - # Ask the HTML processor to hyperlink them. - $retVal = HTML::set_prot_links($aliasList); - } } elsif ($colName eq 'function') { # The functional assignment is just a matter of getting some text. ($retVal) = $record->Value('Feature(assignment)'); } elsif ($colName eq 'gblink') { # Here we want a link to the GBrowse page using the official GBrowse button. - my $gurl = "GetGBrowse.cgi?fid=$fid"; - $retVal = $cgi->a({ href => $gurl, title => "GBrowse for $fid" }, - $cgi->img({ src => "../images/button-gbrowse.png", - border => 0 }) - ); - } elsif ($colName eq 'protlink') { - # Here we want a link to the protein page using the official NMPDR button. - my $hurl = HTML::fid_link($cgi, $fid, 0, 1); - $retVal = $cgi->a({ href => $hurl, title => "Protein page for $fid" }, - $cgi->img({ src => "../images/button-nmpdr.png", - border => 0 }) - ); + $retVal = FakeButton('GBrowse', "GetGBrowse.cgi", undef, + fid => $fid); } elsif ($colName eq 'group') { # Get the NMPDR group name. my (undef, $group) = $self->OrganismData($fid); @@ -1534,7 +2396,381 @@ my $nurl = $sprout->GroupPageName($group); $retVal = $cgi->a({ href => $nurl, title => "$group summary" }, $group); + } elsif ($colName =~ /^keyword:(.+)$/) { + # Here we want keyword-related values. This is also expensive, so + # we compute them when the row is displayed. + $retVal = "%%$colName=$fid"; + } elsif ($colName eq 'orgName') { + # Here we want the formatted organism name and feature number. + $retVal = $self->FeatureName($fid); + } elsif ($colName eq 'protlink') { + # Here we want a link to the protein page using the official NMPDR button. + $retVal = FakeButton('NMPDR', "protein.cgi", undef, + prot => $fid, SPROUT => 1, new_framework => 0, + user => ''); + }elsif ($colName eq 'subsystem') { + # Another run-time column: subsystem list. + $retVal = "%%subsystem=$fid"; + } + # Return the result. + return $retVal; +} + +=head3 RunTimeColumns + +C<< my $htmlText = $shelp->RunTimeColumns($type, $text); >> + +Return the HTML text for a run-time column. Run-time columns are evaluated when the +list is displayed, rather than when it is generated. + +=over 4 + +=item type + +Type of column. + +=item text + +Data relevant to this row of the column. + +=item RETURN + +Returns the fully-formatted HTML text to go in the specified column. + +=back + +=cut + +sub RunTimeColumns { + # Get the parameters. + my ($self, $type, $text) = @_; + # Declare the return variable. + my $retVal = ""; + # Get the Sprout and CGI objects. + my $sprout = $self->DB(); + my $cgi = $self->Q(); + Trace("Runtime column $type with text \"$text\" found.") if T(4); + # Separate the text into a type and data. + if ($type eq 'alias') { + # Here the caller wants external alias links for a feature. The text + # is the feature ID. + my $fid = $text; + # The complicated part is we have to hyperlink them. First, get the + # aliases. + Trace("Generating aliases for feature $fid.") if T(4); + my @aliases = $sprout->FeatureAliases($fid); + # Only proceed if we found some. + if (@aliases) { + # Join the aliases into a comma-delimited list. + my $aliasList = join(", ", @aliases); + # Ask the HTML processor to hyperlink them. + $retVal = HTML::set_prot_links($cgi, $aliasList); + } + } elsif ($type eq 'subsystem') { + # Here the caller wants the subsystems in which this feature participates. + # The text is the feature ID. We will list the subsystem names with links + # to the subsystem's summary page. + my $fid = $text; + # Get the subsystems. + Trace("Generating subsystems for feature $fid.") if T(4); + my %subs = $sprout->SubsystemsOf($fid); + # Extract the subsystem names. + my @names = map { HTML::sub_link($cgi, $_) } sort keys %subs; + # String them into a list. + $retVal = join(", ", @names); + } elsif ($type =~ /^keyword:(.+)$/) { + # Here the caller wants the value of the named keyword. The text is the + # feature ID. + my $keywordName = $1; + my $fid = $text; + # Get the attribute values. + Trace("Getting $keywordName values for feature $fid.") if T(4); + my @values = $sprout->GetFlat(['Feature'], "Feature(id) = ?", [$fid], + "Feature($keywordName)"); + # String them into a list. + $retVal = join(", ", @values); + } + # Return the result. + return $retVal; +} + +=head3 SaveOrganismData + +C<< my ($name, $displayGroup) = $shelp->SaveOrganismData($group, $genomeID, $genus, $species, $strain); >> + +Format the name of an organism and the display version of its group name. The incoming +data should be the relevant fields from the B record in the database. The +data will also be stored in the genome cache for later use in posting search results. + +=over 4 + +=item group + +Name of the genome's group as it appears in the database. + +=item genomeID + +ID of the relevant genome. + +=item genus + +Genus of the genome's organism. If undefined or null, it will be assumed the genome is not +in the database. In this case, the organism name is derived from the genomeID and the group +is automatically the supporting-genomes group. + +=item species + +Species of the genome's organism. + +=item strain + +Strain of the species represented by the genome. + +=item RETURN + +Returns a two-element list. The first element is the formatted genome name. The second +element is the display name of the genome's group. + +=back + +=cut + +sub SaveOrganismData { + # Get the parameters. + my ($self, $group, $genomeID, $genus, $species, $strain) = @_; + # Declare the return values. + my ($name, $displayGroup); + # If the organism does not exist, format an unknown name and a blank group. + if (! defined($genus)) { + $name = "Unknown Genome $genomeID"; + $displayGroup = ""; + } else { + # It does exist, so format the organism name. + $name = "$genus $species"; + if ($strain) { + $name .= " $strain"; + } + # Compute the display group. This is currently the same as the incoming group + # name unless it's the supporting group, which is nulled out. + $displayGroup = ($group eq $FIG_Config::otherGroup ? "" : $group); } + # Cache the group and organism data. + my $cache = $self->{orgs}; + $cache->{$genomeID} = [$name, $displayGroup]; + # Return the result. + return ($name, $displayGroup); +} + +=head3 ValidateKeywords + +C<< my $okFlag = $shelp->ValidateKeywords($keywordString, $required); >> + +Insure that a keyword string is reasonably valid. If it is invalid, a message will be +set. + +=over 4 + +=item keywordString + +Keyword string specified as a parameter to the current search. + +=item required + +TRUE if there must be at least one keyword specified, else FALSE. + +=item RETURN + +Returns TRUE if the keyword string is valid, else FALSE. Note that a null keyword string +is acceptable if the I<$required> parameter is not specified. + +=back + +=cut + +sub ValidateKeywords { + # Get the parameters. + my ($self, $keywordString, $required) = @_; + # Declare the return variable. + my $retVal = 0; + my @wordList = split /\s+/, $keywordString; + # Right now our only real worry is a list of all minus words. The problem with it is that + # it will return an incorrect result. + my @plusWords = grep { $_ =~ /^[^\-]/ } @wordList; + if (! @wordList) { + if ($required) { + $self->SetMessage("No search words specified."); + } else { + $retVal = 1; + } + } elsif (! @plusWords) { + $self->SetMessage("At least one keyword must be positive. All the keywords entered are preceded by minus signs."); + } else { + $retVal = 1; + } + # Return the result. + return $retVal; +} + +=head3 FakeButton + +C<< my $html = SearchHelper::FakeButton($caption, $url, $target, %parms); >> + +Create a fake button that hyperlinks to the specified URL with the specified parameters. +Unlike a real button, this one won't visibly click, but it will take the user to the +correct place. + +The parameters of this method are deliberately identical to L so that we +can switch easily from real buttons to fake ones in the code. + +=over 4 + +=item caption + +Caption to be put on the button. + +=item url + +URL for the target page or script. + +=item target + +Frame or target in which the new page should appear. If C is specified, +the default target will be used. + +=item parms + +Hash containing the parameter names as keys and the parameter values as values. +These will be appended to the URL. + +=back + +=cut + +sub FakeButton { + # Get the parameters. + my ($caption, $url, $target, %parms) = @_; + # Declare the return variable. + my $retVal; + # Compute the target URL. + my $targetUrl = "$url?" . join(";", map { "$_=" . uri_escape($parms{$_}) } keys %parms); + # Compute the target-frame HTML. + my $targetHtml = ($target ? " target=\"$target\"" : ""); + # Assemble the result. + return "
    $caption
    "; +} + +=head3 Formlet + +C<< my $html = SearchHelper::Formlet($caption, $url, $target, %parms); >> + +Create a mini-form that posts to the specified URL with the specified parameters. The +parameters will be stored in hidden fields, and the form's only visible control will +be a submit button with the specified caption. + +Note that we don't use B services here because they generate forms with extra characters +and tags that we don't want to deal with. + +=over 4 + +=item caption + +Caption to be put on the form button. + +=item url + +URL to be put in the form's action parameter. + +=item target + +Frame or target in which the form results should appear. If C is specified, +the default target will be used. + +=item parms + +Hash containing the parameter names as keys and the parameter values as values. + +=back + +=cut + +sub Formlet { + # Get the parameters. + my ($caption, $url, $target, %parms) = @_; + # Compute the target HTML. + my $targetHtml = ($target ? " target=\"$target\"" : ""); + # Start the form. + my $retVal = "
    "; + # Add the parameters. + for my $parm (keys %parms) { + $retVal .= ""; + } + # Put in the button. + $retVal .= ""; + # Close the form. + $retVal .= "
    "; + # Return the result. + return $retVal; +} + +=head2 Virtual Methods + +=head3 Form + +C<< my $html = $shelp->Form(); >> + +Generate the HTML for a form to request a new search. + +=head3 Find + +C<< my $resultCount = $shelp->Find(); >> + +Conduct a search based on the current CGI query parameters. The search results will +be written to the session cache file and the number of results will be +returned. If the search parameters are invalid, a result count of C will be +returned and a result message will be stored in this object describing the problem. + +=head3 Description + +C<< my $htmlText = $shelp->Description(); >> + +Return a description of this search. The description is used for the table of contents +on the main search tools page. It may contain HTML, but it should be character-level, +not block-level, since the description is going to appear in a list. + +=head3 SortKey + +C<< my $key = $shelp->SortKey($fdata); >> + +Return the sort key for the specified feature data. The default is to sort by feature name, +floating NMPDR organisms to the top. If a full-text search is used, then the default +sort is by relevance followed by feature name. This sort may be overridden by the +search class to provide fancier functionality. This method is called by +B, so it is only used for feature searches. A non-feature search +would presumably have its own sort logic. + +=over 4 + +=item record + +The C containing the current feature. + +=item RETURN + +Returns a key field that can be used to sort this row in among the results. + +=back + +=cut + +sub SortKey { + # Get the parameters. + my ($self, $fdata) = @_; + # Get the feature ID from the record. + my $fid = $fdata->FID(); + # Get the group from the feature ID. + my $group = $self->FeatureGroup($fid); + # Ask the feature query object to form the sort key. + my $retVal = $fdata->SortKey($self, $group); # Return the result. return $retVal; }