From 3215b5a9821d69a2216c2cdf2b64cd3eb27f7370 Mon Sep 17 00:00:00 2001
From: Joey Hess <joey@kodama.kitenet.net>
Date: Wed, 4 Jun 2008 15:24:05 -0400
Subject: [PATCH] finishing touches on the new search plugin

- Add a Help link.
- If the pageterm is too long, hash it.
---
 IkiWiki/Plugin/search.pm   | 55 +++++++++++++++++++++++++++-----------
 debian/NEWS                |  4 +--
 doc/ikiwiki.mdwn           |  1 +
 doc/ikiwiki/searching.mdwn |  2 +-
 doc/plugins/search.mdwn    |  8 +++---
 po/ikiwiki.pot             |  2 +-
 templates/searchquery.tmpl |  8 +-----
 7 files changed, 50 insertions(+), 30 deletions(-)

diff --git a/IkiWiki/Plugin/search.pm b/IkiWiki/Plugin/search.pm
index 249f267b9..3ac435197 100644
--- a/IkiWiki/Plugin/search.pm
+++ b/IkiWiki/Plugin/search.pm
@@ -20,19 +20,10 @@ sub checkconfig () { #{{{
 			error(sprintf(gettext("Must specify %s when using the search plugin"), $required));
 		}
 	}
-
+	
 	if (! exists $config{omega_cgi}) {
 		$config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
 	}
-	
-	if (! -e $config{wikistatedir}."/xapian" || $config{rebuild}) {
-		writefile("omega.conf", $config{wikistatedir}."/xapian",
-			"database_dir .\n".
-			"template_dir ./templates\n");
-		writefile("query", $config{wikistatedir}."/xapian/templates",
-			IkiWiki::misctemplate(gettext("search"),
-				readfile(IkiWiki::template_file("searchquery.tmpl"))));
-	}
 } #}}}
 
 my $form;
@@ -59,6 +50,12 @@ sub index (@) { #{{{
 	my %params=@_;
 	
 	return $params{content} if $IkiWiki::preprocessing{$params{destpage}};
+
+	setupfiles();
+
+	# A unique pageterm is used to identify the document for a page.
+	my $pageterm=pageterm($params{page});
+	return $params{content} unless defined $pageterm;
 	
 	my $db=xapiandb();
 	my $doc=Search::Xapian::Document->new();
@@ -127,8 +124,6 @@ sub index (@) { #{{{
 		$tg->index_text(lc($link), 1, "ZLINK"); # for link:bar
 	}
 
-	# A unique pageterm is used to identify the document for a page.
-	my $pageterm=pageterm($params{page});
 	$doc->add_term($pageterm);
 	$db->replace_document_by_term($pageterm, $doc);
 
@@ -138,7 +133,8 @@ sub index (@) { #{{{
 sub delete (@) { #{{{
 	my $db=xapiandb();
 	foreach my $page (@_) {
-		$db->delete_document_by_term(pageterm(pagename($page)));
+		my $pageterm=pageterm(pagename($page));
+		$db->delete_document_by_term($pageterm) if defined $pageterm;
 	}
 } #}}}
 
@@ -150,6 +146,9 @@ sub cgi ($) { #{{{
 		chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
 		$ENV{OMEGA_CONFIG_FILE}="./omega.conf";
 		$ENV{CGIURL}=$config{cgiurl},
+		IkiWiki::loadindex();
+		$ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
+			noimageinline => 1, linktext => "Help");
 		exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
 	}
 } #}}}
@@ -157,9 +156,22 @@ sub cgi ($) { #{{{
 sub pageterm ($) { #{{{
 	my $page=shift;
 
-	# TODO: check if > 255 char page names overflow term
-	# length; use sha1 if so?
-	return "U:".$page;
+	# 240 is the number used by omindex to decide when to hash an
+	# overlong term. This does not use a compatible hash method though.
+	if (length $page > 240) {
+		eval q{use Digest::SHA1};
+		if ($@) {
+			debug("search: ".sprintf(gettext("need Digest::SHA1 to index %s"), $page)) if $@;
+			return undef;
+		}
+
+		# Note no colon, therefore it's guaranteed to not overlap
+		# with a page with the same name as the hash..
+		return "U".lc(Digest::SHA1::sha1_hex($page));
+	}
+	else {
+		return "U:".$page;
+	}
 } #}}}
 
 my $db;
@@ -176,4 +188,15 @@ sub xapiandb () { #{{{
 	return $db;
 } #}}}
 
+sub setupfiles () { #{{{
+	if (! -e $config{wikistatedir}."/xapian" || $config{rebuild}) {
+		writefile("omega.conf", $config{wikistatedir}."/xapian",
+			"database_dir .\n".
+			"template_dir ./templates\n");
+		writefile("query", $config{wikistatedir}."/xapian/templates",
+			IkiWiki::misctemplate(gettext("search"),
+				readfile(IkiWiki::template_file("searchquery.tmpl"))));
+	}
+} #}}}
+
 1
diff --git a/debian/NEWS b/debian/NEWS
index 201f3f134..249827e7d 100644
--- a/debian/NEWS
+++ b/debian/NEWS
@@ -1,8 +1,8 @@
 ikiwiki (2.49) unstable; urgency=low
 
   The search plugin no longer uses hyperestrair. Instead, to use it you
-  will now need to install xapian-omega, and the Search::Xapian and
-  HTML::Scrubber perl modules.
+  will now need to install xapian-omega, and the Search::Xapian,
+  HTML::Scrubber, and Digest::SHA1 perl modules.
 
  -- Joey Hess <joeyh@debian.org>  Wed, 04 Jun 2008 00:29:28 -0400
 
diff --git a/doc/ikiwiki.mdwn b/doc/ikiwiki.mdwn
index 485ac8826..c98a97e59 100644
--- a/doc/ikiwiki.mdwn
+++ b/doc/ikiwiki.mdwn
@@ -14,3 +14,4 @@ Some documentation on using ikiwiki:
 * [[ikiwiki/markdown]]
 * [[ikiwiki/subpage]]
 * [[ikiwiki/openid]]
+* [[ikiwiki/searching]]
diff --git a/doc/ikiwiki/searching.mdwn b/doc/ikiwiki/searching.mdwn
index 0eeb5a6b9..539e7193d 100644
--- a/doc/ikiwiki/searching.mdwn
+++ b/doc/ikiwiki/searching.mdwn
@@ -5,7 +5,7 @@ then="This wiki has searching **enabled**."
 else="This wiki has searching **disabled**."]]
 
 If searching is enabled, you can enter search terms in the search field,
-like you'd expect. There are a few special things you can do to constuct
+as you'd expect. There are a few special things you can do to constuct
 more powerful searches.
 
 * To match a phrase, enclose it in double quotes.
diff --git a/doc/plugins/search.mdwn b/doc/plugins/search.mdwn
index e53c28c7c..9336b2403 100644
--- a/doc/plugins/search.mdwn
+++ b/doc/plugins/search.mdwn
@@ -3,9 +3,11 @@
 
 This plugin adds full text search to ikiwiki, using the
 [xapian](http://xapian.org/) engine, its
-[omega](http://xapian.org/docs/omega/overview.html) frontend,
-and the [[cpan Search::Xapian]] perl module. (The [[cpan HTML::Scrubber]]
-perl module will also be used, if available.)
+[omega](http://xapian.org/docs/omega/overview.html) frontend, and the
+[[cpan Search::Xapian]], [[cpan Digest::SHA1]], and [[cpan HTML::Scrubber]]
+perl modules.
+
+The [[ikiwiki/searching]] page describes how to write search queries.
 
 Ikiwiki will handle indexing new and changed page contents. Note that since
 it only indexes page contents, files copied by the [[rawhtml]] plugin will
diff --git a/po/ikiwiki.pot b/po/ikiwiki.pot
index 00521c441..e4bb7b6f5 100644
--- a/po/ikiwiki.pot
+++ b/po/ikiwiki.pot
@@ -8,7 +8,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version: PACKAGE VERSION\n"
 "Report-Msgid-Bugs-To: \n"
-"POT-Creation-Date: 2008-06-04 01:46-0400\n"
+"POT-Creation-Date: 2008-06-04 14:36-0400\n"
 "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
 "Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
 "Language-Team: LANGUAGE <LL@li.org>\n"
diff --git a/templates/searchquery.tmpl b/templates/searchquery.tmpl
index bf36f3330..0d6eb6398 100644
--- a/templates/searchquery.tmpl
+++ b/templates/searchquery.tmpl
@@ -36,18 +36,12 @@ ACTION="$html{$env{CGIURL}}" TARGET="_top">
 <center>
 <INPUT NAME=P VALUE="$html{$query}" SIZE=65>
 <INPUT TYPE=SUBMIT VALUE="Search">
+$env{HELPLINK}
 <hr>
 <SELECT NAME=DEFAULTOP>
 <OPTION VALUE=or $if{$eq{$defaultop,or},SELECTED}>Matching any words
 <OPTION VALUE=and $if{$eq{$defaultop,and},SELECTED}>Matching all words
 </SELECT>
-$if{$opt{topterms},
- <div title="Suggested terms to add to your query"
- style="text-align:left;background:#cfc;border:1px solid green;padding:2px;font:11px verdana$. arial$. helvetica$. sans-serif;">
- $map{$topterms,<span style="white-space:nowrap"><INPUT TYPE=checkbox NAME=X VALUE="$prettyterm{$_}" onClick="C(this)">$prettyterm{$_}</span> }
- <BR><NOSCRIPT><INPUT TYPE=hidden NAME=ADD VALUE=1></NOSCRIPT>
- </div>
-}
 $or{$html{$error},
 $if{$eq{$msize,0},
 $if{$query,No documents match your query,
-- 
2.26.2