Bug 292454: [GOOGLE] A module to search Google

Patch By Max Kanat-Alexander <mkanat@bugzilla.org> r=colin, r=Hixie

webtools/mozbot/BotModules/Google.bm

@@ -0,0 +1,117 @@
+################################
+# Google Module                #
+################################
+
+# Author: Max Kanat-Alexander <mkanat@bugzilla.org>
+
+# Uses the Net::Google module to query Google for keywords.
+# Note that Net::Google requires SOAP::Lite.
+
+package BotModules::Google;
+use vars qw(@ISA);
+@ISA = qw(BotModules);
+use Net::Google;
+
+use constant SEPARATOR => ' -- ';
+1;
+
+sub Help {
+    my $self = shift;
+    my ($event) = @_;
+    return {
+        '' =>    q{Queries Google for specified search terms. Note that you}
+               . q{ have to specify your Google API Key (the 'googleApiKey'}
+               . q{ var) before using any commands. To get a Google API Key,}
+               . q{ see: http://www.google.com/apis/ and go to "Create an}
+               . q{ Account."},
+        'google' =>  q{Searches google for the specified terms.}
+                   . q{Syntax: 'google <terms>'},
+    };
+}
+
+# RegisterConfig - Called when initialised, should call registerVariables
+sub RegisterConfig {
+    my $self = shift;
+    $self->SUPER::RegisterConfig(@_);
+    $self->registerVariables(
+      # [ name, save?, settable? ]
+      ['maxResults', 1, 1, 10],
+      ['maxInChannel', 1, 1, 1],
+      # XXX - Not yet implemented.
+      #['backoffTime', 1, 1, 120],
+      ['googleApiKey', 1, 1, ''],
+      ['safeSearch', 1, 1, 1],
+      # Note that maxLineLength will not be respected if the URL
+      # alone is longer than maxLineLength.
+      ['maxLineLength', 1, 1, 256]
+    );
+}
+
+sub Told {
+    my $self = shift;
+    my ($event, $message) = @_;
+    # We take anything that occurs at the end of the line,
+    # because Google will ignore punctuation anyway.
+    if ($message =~ /^(\s*google\s+)(.+)$/osi) {
+        my $terms = $2;
+
+        my @searchResults = $self->doSearch($terms);
+
+        if (!@searchResults) {
+            $self->say($event, "Nothing found.");
+            # XXX - We could use $google->spell here to make suggestions.
+        } 
+        # If we are in a channel, and not a /msg
+        elsif ($event->{'channel'}) {
+            splice(@searchResults, $self->{'maxInChannel'});
+        }
+        # We're in a /msg
+        else {
+            unshift(@searchResults, scalar(@searchResults) . " results found: ");
+        }
+
+        foreach my $result (@searchResults) {
+            $self->say($event, $event->{'from'} . ': ' . $result);
+        }
+    } else {
+        return $self->SUPER::Told(@_);
+    }
+    return 0; # we've dealt with it, no need to do anything else.
+}
+
+
+# Performs the actual Google search and returns the
+# result as an array of lines to say.
+sub doSearch {
+    my $self = shift;
+    my ($terms) = @_;
+
+    my $google = Net::Google->new(key => $self->{'googleApiKey'});
+    my $search = $google->search(safe => $self->{'safeSearch'},
+                                 filter => $self->{'safeSearch'},
+                                 max_results => $self->{'maxResults'});
+    $search->query($terms);
+    my $results = $search->results();
+
+    my @searchLines = ();
+
+    foreach my $result (@$results) {
+        my $title = $result->title();
+        # The Google API puts <b></b> tags into the title if the search
+        # terms appear in the title.
+        $title =~ s|</?b>||g;
+        $title = $self->unescapeXML($title);
+        my $url = $result->URL();
+        my $line_size = (length($title) + length($result) + length(SEPARATOR));
+        if ($line_size  > $self->{'maxLineLength'} ) {
+            # The 3 is for the '...'
+            my $new_title_size = ($line_size - $self->{'maxLineLength'}) - 3;
+            my $title = substr($title, 0, $new_title_size)
+                        . '...';
+        }
+        my $resultLine = $title . SEPARATOR . $url;
+        push(@searchLines, $resultLine);
+    }
+
+    return @searchLines;
+}