From 10f40661a2f517393331b554c2fec295c8c160e8 Mon Sep 17 00:00:00 2001
From: Brion Vibber <brion@pobox.com>
Date: Mon, 30 Nov 2009 08:09:30 -0800
Subject: [PATCH] Ticket 1870: drop unnecessary Tidy module installation
 requirement.

Tidy was only being used by a couple of non-default URL shortener plugins,
PtitUrl and TightUrl. Both were easily changed to load the tag-soup HTML
via DOMDocument (using the default DOM module which is already used by
other dependencies).

Added xml, dom, and simplexml modules to the requirements check
in install.php, as they were being used but not checked for.

Also cleaned up LilUrl, PtitUrl, and TightUrl to return URL as a string
instead of as a SimpleXML node object.
---
 README                                      | 2 --
 install.php                                 | 2 +-
 plugins/LilUrl/LilUrlPlugin.php             | 2 +-
 plugins/PtitUrl/PtitUrlPlugin.php           | 9 ++++++---
 plugins/TightUrl/TightUrlPlugin.php         | 9 ++++++---
 plugins/UrlShortener/UrlShortenerPlugin.php | 8 --------
 6 files changed, 14 insertions(+), 18 deletions(-)

diff --git a/README b/README
index eb1fb8cd70..51a8e91cc5 100644
--- a/README
+++ b/README
@@ -98,7 +98,6 @@ released Aug 26 2009. Notable changes this version:
 - Better error handling in Twitter posting.
 - Show oEmbed data for XHTML files as well as plain HTML.
 - Updated bug database link in README.
-- require HTML tidy extension.
 - add support for HTTP Basic Auth in PHP CGI or FastCGI (e.g. GoDaddy).
 - autofocus input to selected entry elements depending on page.
 - updated layout for filter-by-tag form.
@@ -179,7 +178,6 @@ Your PHP installation must include the following PHP extensions:
 - GD. For scaling down avatar images.
 - mbstring. For handling Unicode (UTF-8) encoded strings.
 - gettext. For multiple languages. Default on many PHP installs.
-- tidy. Used to clean up HTML/URLs for the URL shortener to consume.
 
 For some functionality, you will also need the following extensions:
 
diff --git a/install.php b/install.php
index e7f7cf3187..1c62bb2b21 100644
--- a/install.php
+++ b/install.php
@@ -301,7 +301,7 @@ function checkPrereqs()
     }
 
     $reqs = array('gd', 'curl',
-                  'xmlwriter', 'mbstring','tidy');
+                  'xmlwriter', 'mbstring', 'xml', 'dom', 'simplexml');
 
     foreach ($reqs as $req) {
         if (!checkExtension($req)) {
diff --git a/plugins/LilUrl/LilUrlPlugin.php b/plugins/LilUrl/LilUrlPlugin.php
index e906751e84..4a6f1cdc79 100644
--- a/plugins/LilUrl/LilUrlPlugin.php
+++ b/plugins/LilUrl/LilUrlPlugin.php
@@ -54,7 +54,7 @@ class LilUrlPlugin extends UrlShortenerPlugin
         if (!isset($y->body)) return;
         $x = $y->body->p[0]->a->attributes();
         if (isset($x['href'])) {
-            return $x['href'];
+            return strval($x['href']);
         }
     }
 }
diff --git a/plugins/PtitUrl/PtitUrlPlugin.php b/plugins/PtitUrl/PtitUrlPlugin.php
index ef453e96d8..76a438dd5e 100644
--- a/plugins/PtitUrl/PtitUrlPlugin.php
+++ b/plugins/PtitUrl/PtitUrlPlugin.php
@@ -47,11 +47,14 @@ class PtitUrlPlugin extends UrlShortenerPlugin
     {
         $response = $this->http_get(sprintf($this->serviceUrl,urlencode($url)));
         if (!$response) return;
-        $response = $this->tidy($response);
-        $y = @simplexml_load_string($response);
+        $dom = new DOMDocument();
+        @$dom->loadHTML($response);
+        $y = @simplexml_import_dom($dom);
         if (!isset($y->body)) return;
         $xml = $y->body->center->table->tr->td->pre->a->attributes();
-        if (isset($xml['href'])) return $xml['href'];
+        if (isset($xml['href'])) {
+            return strval($xml['href']);
+        }
     }
 }
 
diff --git a/plugins/TightUrl/TightUrlPlugin.php b/plugins/TightUrl/TightUrlPlugin.php
index 56414c8c8d..6ced9afdc5 100644
--- a/plugins/TightUrl/TightUrlPlugin.php
+++ b/plugins/TightUrl/TightUrlPlugin.php
@@ -48,10 +48,13 @@ class TightUrlPlugin extends UrlShortenerPlugin
     {
         $response = $this->http_get(sprintf($this->serviceUrl,urlencode($url)));
         if (!$response) return;
-        $response = $this->tidy($response);
-        $y = @simplexml_load_string($response);
+        $dom = new DOMDocument();
+        @$dom->loadHTML($response);
+        $y = @simplexml_import_dom($dom);
         if (!isset($y->body)) return;
         $xml = $y->body->p[0]->code[0]->a->attributes();
-        if (isset($xml['href'])) return $xml['href'];
+        if (isset($xml['href'])) {
+            return strval($xml['href']);
+        }
     }
 }
diff --git a/plugins/UrlShortener/UrlShortenerPlugin.php b/plugins/UrlShortener/UrlShortenerPlugin.php
index 37206aa896..027624b7ae 100644
--- a/plugins/UrlShortener/UrlShortenerPlugin.php
+++ b/plugins/UrlShortener/UrlShortenerPlugin.php
@@ -68,14 +68,6 @@ abstract class UrlShortenerPlugin extends Plugin
         return $response->getBody();
     }
 
-    protected function tidy($response) {
-        $response = str_replace('&nbsp;', ' ', $response);
-        $config = array('output-xhtml' => true);
-        $tidy = new tidy;
-        $tidy->parseString($response, $config, 'utf8');
-        $tidy->cleanRepair();
-        return (string)$tidy;
-    }
     //------------Below are the methods that connect StatusNet to the implementing Url Shortener plugin------------\\
 
     function onInitializePlugin(){