Search::Estraier(3) pure perl module to use Hyper Estraier search engine

SYNOPSIS

Simple indexer

        use Search::Estraier;
        # create and configure node
        my $node = new Search::Estraier::Node(
                url => 'http://localhost:1978/node/test',
                user => 'admin',
                passwd => 'admin',
                create => 1,
                label => 'Label for node',
                croak_on_error => 1,
        );
        # create document
        my $doc = new Search::Estraier::Document;
        # add attributes
        $doc->add_attr('@uri', "http://estraier.gov/example.txt");
        $doc->add_attr('@title', "Over the Rainbow");
        # add body text to document
        $doc->add_text("Somewhere over the rainbow.  Way up high.");
        $doc->add_text("There's a land that I heard of once in a lullaby.");
        die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) });

Simple searcher

        use Search::Estraier;
        # create and configure node
        my $node = new Search::Estraier::Node(
                url => 'http://localhost:1978/node/test',
                user => 'admin',
                passwd => 'admin',
                croak_on_error => 1,
        );
        # create condition
        my $cond = new Search::Estraier::Condition;
        # set search phrase
        $cond->set_phrase("rainbow AND lullaby");
        my $nres = $node->search($cond, 0);
        if (defined($nres)) {
                print "Got ", $nres->hits, " results\n";
                # for each document in results
                for my $i ( 0 ... $nres->doc_num - 1 ) {
                        # get result document
                        my $rdoc = $nres->get_doc($i);
                        # display attribte
                        print "URI: ", $rdoc->attr('@uri'),"\n";
                        print "Title: ", $rdoc->attr('@title'),"\n";
                        print $rdoc->snippet,"\n";
                }
        } else {
                die "error: ", $node->status,"\n";
        }

DESCRIPTION

This module is implementation of node API of Hyper Estraier. Since it's perl-only module with dependencies only on standard perl modules, it will run on all platforms on which perl runs. It doesn't require compilation or Hyper Estraier development files on target machine.

It is implemented as multiple packages which closly resamble Ruby implementation. It also includes methods to manage nodes.

There are few examples in "scripts" directory of this distribution.

Inheritable common methods

This methods should really move somewhere else.

_s

Remove multiple whitespaces from string, as well as whitespaces at beginning or end

 my $text = $self->_s(" this  is a text  ");
 $text = 'this is a text';

Search::Estraier::Document

This class implements Document which is single item in Hyper Estraier.

It's is collection of:

attributes
'key' => 'value' pairs which can later be used for filtering of results

You can add common filters to "attrindex" in estmaster's "_conf" file for better performance. See "attrindex" in Hyper Estraier P2P Guide <http://hyperestraier.sourceforge.net/nguide-en.html>.

vectors
also 'key' => 'value' pairs
display text
Text which will be used to create searchable corpus of your index and included in snippet output.
hidden text
Text which will be searchable, but will not be included in snippet.

new

Create new document, empty or from draft.

  my $doc = new Search::HyperEstraier::Document;
  my $doc2 = new Search::HyperEstraier::Document( $draft );

add_attr

Add an attribute.

  $doc->add_attr( name => 'value' );

Delete attribute using

  $doc->add_attr( name => undef );

add_text

Add a sentence of text.

  $doc->add_text('this is example text to display');

add_hidden_text

Add a hidden sentence.

  $doc->add_hidden_text('this is example text just for search');

add_vectors

Add a vectors

  $doc->add_vector(
        'vector_name' => 42,
        'another' => 12345,
  );

set_score

Set the substitute score

  $doc->set_score(12345);

score

Get the substitute score

id

Get the ID number of document. If the object has never been registred, "-1" is returned.

  print $doc->id;

attr_names

Returns array with attribute names from document object.

  my @attrs = $doc->attr_names;

attr

Returns value of an attribute.

  my $value = $doc->attr( 'attribute' );

texts

Returns array with text sentences.

  my @texts = $doc->texts;

cat_texts

Return whole text as single scalar.

 my $text = $doc->cat_texts;

dump_draft

Dump draft data from document object.

  print $doc->dump_draft;

delete

Empty document object

  $doc->delete;

This function is addition to original Ruby API, and since it was included in C wrappers it's here as a convinience. Document objects which go out of scope will be destroyed automatically.

Search::Estraier::Condition

new

  my $cond = new Search::HyperEstraier::Condition;

set_phrase

  $cond->set_phrase('search phrase');

add_attr

  $cond->add_attr('@URI STRINC /~dpavlin/');

set_order

  $cond->set_order('@mdate NUMD');

set_max

  $cond->set_max(42);

set_options

  $cond->set_options( 'SURE' );
  $cond->set_options( qw/AGITO NOIDF SIMPLE/ );

Possible options are:

SURE
check every N-gram
USUAL
check every second N-gram
FAST
check every third N-gram
AGITO
check every fourth N-gram
NOIDF
don't perform TF-IDF tuning
SIMPLE
use simplified query phrase

Skipping N-grams will speed up search, but reduce accuracy. Every call to "set_options" will reset previous options;

This option changed in version 0.04 of this module. It's backwards compatibile.

phrase

Return search phrase.

  print $cond->phrase;

order

Return search result order.

  print $cond->order;

attrs

Return search result attrs.

  my @cond_attrs = $cond->attrs;

max

Return maximum number of results.

  print $cond->max;

"-1" is returned for unitialized value, 0 is unlimited.

options

Return options for this condition.

  print $cond->options;

Options are returned in numerical form.

set_skip

Set number of skipped documents from beginning of results

  $cond->set_skip(42);

Similar to "offset" in RDBMS.

skip

Return skip for this condition.

  print $cond->skip;

set_distinct

  $cond->set_distinct('@author');

distinct

Return distinct attribute

  print $cond->distinct;

set_mask

Filter out some links when searching.

Argument array of link numbers, starting with 0 (current node).

  $cond->set_mask(qw/0 1 4/);

Search::Estraier::ResultDocument

new

  my $rdoc = new Search::HyperEstraier::ResultDocument(
        uri => 'http://localhost/document/uri/42',
        attrs => {
                foo => 1,
                bar => 2,
        },
        snippet => 'this is a text of snippet'
        keywords => 'this\tare\tkeywords'
  );

uri

Return URI of result document

  print $rdoc->uri;

attr_names

Returns array with attribute names from result document object.

  my @attrs = $rdoc->attr_names;

attr

Returns value of an attribute.

  my $value = $rdoc->attr( 'attribute' );

snippet

Return snippet from result document

  print $rdoc->snippet;

keywords

Return keywords from result document

  print $rdoc->keywords;

Search::Estraier::NodeResult

new

  my $res = new Search::HyperEstraier::NodeResult(
        docs => @array_of_rdocs,
        hits => %hash_with_hints,
  );

doc_num

Return number of documents

  print $res->doc_num;

This will return real number of documents (limited by "max"). If you want to get total number of hits, see "hits".

get_doc

Return single document

  my $doc = $res->get_doc( 42 );

Returns undef if document doesn't exist.

hint

Return specific hint from results.

  print $res->hint( 'VERSION' );

Possible hints are: "VERSION", "NODE", "HIT", "HINT#n", "DOCNUM", "WORDNUM", "TIME", "LINK#n", "VIEW".

hints

More perlish version of "hint". This one returns hash.

  my %hints = $res->hints;

hits

Syntaxtic sugar for total number of hits for this query

  print $res->hits;

It's same as

  print $res->hint('HIT');

but shorter.

Search::Estraier::Node

new

  my $node = new Search::HyperEstraier::Node;

or optionally with "url" as parametar

  my $node = new Search::HyperEstraier::Node( 'http://localhost:1978/node/test' );

or in more verbose form

  my $node = new Search::HyperEstraier::Node(
        url => 'http://localhost:1978/node/test',
        user => 'admin',
        passwd => 'admin'
        create => 1,
        label => 'optional node label',
        debug => 1,
        croak_on_error => 1
  );

with following arguments:

url
URL to node
user
specify username for node server authentication
passwd
password for authentication
create
create node if it doesn't exists
label
optional label for new node if "create" is used
debug
dumps a lot of debugging output
croak_on_error
very helpful during development. It will croak on all errors instead of silently returning "-1" (which is convention of Hyper Estraier API in other languages).

set_url

Specify URL to node server

  $node->set_url('http://localhost:1978');

set_proxy

Specify proxy server to connect to node server

  $node->set_proxy('proxy.example.com', 8080);

set_timeout

Specify timeout of connection in seconds

  $node->set_timeout( 15 );

set_auth

Specify name and password for authentication to node server.

  $node->set_auth('clint','eastwood');

status

Return status code of last request.

  print $node->status;

"-1" means connection failure.

put_doc

Add a document

  $node->put_doc( $document_draft ) or die "can't add document";

Return true on success or false on failure.

out_doc

Remove a document

  $node->out_doc( document_id ) or "can't remove document";

Return true on success or false on failture.

out_doc_by_uri

Remove a registrated document using it's uri

  $node->out_doc_by_uri( 'file:///document/uri/42' ) or "can't remove document";

Return true on success or false on failture.

edit_doc

Edit attributes of a document

  $node->edit_doc( $document_draft ) or die "can't edit document";

Return true on success or false on failture.

get_doc

Retreive document

  my $doc = $node->get_doc( document_id ) or die "can't get document";

Return true on success or false on failture.

get_doc_by_uri

Retreive document

  my $doc = $node->get_doc_by_uri( 'file:///document/uri/42' ) or die "can't get document";

Return true on success or false on failture.

get_doc_attr

Retrieve the value of an atribute from object

  my $val = $node->get_doc_attr( document_id, 'attribute_name' ) or
        die "can't get document attribute";

get_doc_attr_by_uri

Retrieve the value of an atribute from object

  my $val = $node->get_doc_attr_by_uri( document_id, 'attribute_name' ) or
        die "can't get document attribute";

etch_doc

Exctract document keywords

  my $keywords = $node->etch_doc( document_id ) or die "can't etch document";

etch_doc_by_uri

Retreive document

  my $keywords = $node->etch_doc_by_uri( 'file:///document/uri/42' ) or die "can't etch document";

Return true on success or false on failture.

uri_to_id

Get ID of document specified by URI

  my $id = $node->uri_to_id( 'file:///document/uri/42' );

This method won't croak, even if using "croak_on_error".

_fetch_doc

Private function used for implementing of "get_doc", "get_doc_by_uri", "etch_doc", "etch_doc_by_uri".

 # this will decode received draft into Search::Estraier::Document object
 my $doc = $node->_fetch_doc( id => 42 );
 my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42' );
 # to extract keywords, add etch
 my $doc = $node->_fetch_doc( id => 42, etch => 1 );
 my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42', etch => 1 );
 # to get document attrubute add attr
 my $doc = $node->_fetch_doc( id => 42, attr => '@mdate' );
 my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42', attr => '@mdate' );
 # more general form which allows implementation of
 # uri_to_id
 my $id = $node->_fetch_doc(
        uri => 'file:///document/uri/42',
        path => '/uri_to_id',
        chomp_resbody => 1
 );

name

  my $node_name = $node->name;

label

  my $node_label = $node->label;

doc_num

  my $documents_in_node = $node->doc_num;

word_num

  my $words_in_node = $node->word_num;

size

  my $node_size = $node->size;

search

Search documents which match condition

  my $nres = $node->search( $cond, $depth );

$cond is "Search::Estraier::Condition" object, while <$depth> specifies depth for meta search.

Function results "Search::Estraier::NodeResult" object.

cond_to_query

Return URI encoded string generated from Search::Estraier::Condition

  my $args = $node->cond_to_query( $cond, $depth );

shuttle_url

This is method which uses "LWP::UserAgent" to communicate with Hyper Estraier node master.

  my $rv = shuttle_url( $url, $content_type, $req_body, \$resbody );

$resheads and $resbody booleans controll if response headers and/or response body will be saved within object.

set_snippet_width

Set width of snippets in results

  $node->set_snippet_width( $wwidth, $hwidth, $awidth );

$wwidth specifies whole width of snippet. It's 480 by default. If it's 0 snippet is not sent with results. If it is negative, whole document text is sent instead of snippet.

$hwidth specified width of strings from beginning of string. Default value is 96. Negative or zero value keep previous value.

$awidth specifies width of strings around each highlighted word. It's 96 by default. If negative of zero value is provided previous value is kept unchanged.

set_user

Manage users of node

  $node->set_user( 'name', $mode );

$mode can be one of:

0
delete account
1
set administrative right for user
2
set user account as guest

Return true on success, otherwise false.

set_link

Manage node links

  $node->set_link('http://localhost:1978/node/another', 'another node label', $credit);

If $credit is negative, link is removed.

admins

 my @admins = @{ $node->admins };

Return array of users with admin rights on node

guests

 my @guests = @{ $node->guests };

Return array of users with guest rights on node

links

 my $links = @{ $node->links };

Return array of links for this node

cacheusage

Return cache usage for a node

  my $cache = $node->cacheusage;

master

Set actions on Hyper Estraier node master ("estmaster" process)

  $node->master(
        action => 'sync'
  );

All available actions are documented in http://hyperestraier.sourceforge.net/nguide-en.html#protocol <http://hyperestraier.sourceforge.net/nguide-en.html#protocol>

PRIVATE METHODS

You could call those directly, but you don't have to. I hope.

_set_info

Set information for node

  $node->_set_info;

_clear_info

Clear information for node

  $node->_clear_info;

On next call to "name", "label", "doc_num", "word_num" or "size" node info will be fetch again from Hyper Estraier.

EXPORT

Nothing.

AUTHOR

Dobrica Pavlinusic, <[email protected]>

Robert Klep <[email protected]> contributed refactored search code

COPYRIGHT AND LICENSE

Copyright (C) 2005-2006 by Dobrica Pavlinusic

This library is free software; you can redistribute it and/or modify it under the GPL v2 or later.