Provided by: openguides_0.76-1_all
NAME
OpenGuides::Search - Search form generation and processing for OpenGuides.
DESCRIPTION
Does search stuff for OpenGuides. Distributed and installed as part of the OpenGuides project, not intended for independent installation. This documentation is probably only useful to OpenGuides developers.
SYNOPSIS
use CGI; use OpenGuides::Config; use OpenGuides::Search; my $config = OpenGuides::Config->new( file => "wiki.conf" ); my $search = OpenGuides::Search->new( config => $config ); my %vars = CGI::Vars(); $search->run( vars => \%vars );
METHODS
new my $config = OpenGuides::Config->new( file => "wiki.conf" ); my $search = OpenGuides::Search->new( config => $config ); wiki my $wiki = $search->wiki; An accessor; returns the underlying Wiki::Toolkit object. config my $config = $search->config; An accessor; returns the underlying OpenGuides::Config object. run my %vars = CGI::Vars(); $search->run( vars => \%vars, return_output => 1, # defaults to 0 return_tt_vars => 1, # defaults to 0 ); The "return_output" parameter is optional. If supplied and true, the stuff that would normally be printed to STDOUT will be returned as a string instead. The "return_tt_vars" parameter is also optional. If supplied and true, the template is not processed and the variables that would have been passed to it are returned as a hash. This parameter takes precedence over "return_output". These two parameters exist to make testing easier; you probably don't want to use them in production. You can also request just the raw search results: my %results = $search->run( os_x => 528864, os_y => 180797, os_dist => 750, format => "raw", ); Results are returned as a hash, keyed on the page name. All results are returned, not just the first "page". The values in the hash are hashes themselves, with the following key/value pairs: • name • wgs84_lat - WGS-84 latitude • wgs84_long - WGS-84 longitude • summary • distance - distance (in metres) from origin, if origin exists • score - relevance to search string, if search string exists; higher score means more relevance In case you're struggling to follow the code, it does the following: 1) Processes the parameters, and bails out if it hit a problem with them 2) If a search string was given, do a text search 3) If distance search paramaters were given, do a distance search 4) If no search has occured, print out the search form 5) If an error occured, bail out 6) If we got a single hit on a string search, redirect to it 7) If no results were found, give an empty search results page 8) Sort the results by either score or distance 9) Decide which results to show, based on paging 10) Display the appropriate page of the results
INPUT
word a single word will be matched as-is. For example, a search on escalator will return all pages containing the word "escalator". AND searches A list of words with no punctuation will be ANDed, for example: restaurant vegetarian will return all pages containing both the word "restaurant" and the word "vegetarian". OR searches A list of words separated by commas (and optional spaces) will be ORed, for example: restaurant, cafe will return all pages containing either the word "restaurant" or the word "cafe". phrase searches Enclose phrases in double quotes, for example: "meat pie" will return all pages that contain the exact phrase "meat pie" - not pages that only contain, for example, "apple pie and meat sausage".
SEARCHING BY DISTANCE
To perform a distance search, you need to supply one of the following sets of criteria to specify the distance to search within, and the origin (centre) of the search: os_dist, os_x, and os_y Only works if you chose to use British National Grid in wiki.conf osie_dist, osie_x, and osie_y Only works if you chose to use Irish National Grid in wiki.conf latlong_dist, latitude, and longitude Should always work, but has a habit of "finding" things a couple of metres away from themselves. You can perform both pure distance searches and distance searches in combination with text searches.
OUTPUT
Results will be put into some form of relevance ordering. These are the rules we have tests for so far (and hence the only rules that can be relied on): • A match on page title will score higher than a match on page category or locale. • A match on page category or locale will score higher than a match on page content. • Two matches in the title beats one match in the title and one in the content.
AUTHOR
The OpenGuides Project (openguides-dev@lists.openguides.org)
COPYRIGHT
Copyright (C) 2003-2013 The OpenGuides Project. All Rights Reserved. The OpenGuides distribution is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
SEE ALSO
OpenGuides