strip_tags_long: use htmlpurifier to properly reformat html content

6 files changed, 144 insertions, 0 deletions

diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/ftp.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/ftp.php
new file mode 100755
index 000000000..5849bf7ff
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/ftp.php
@@ -0,0 +1,43 @@
+<?php
+
+/**
+ * Validates ftp (File Transfer Protocol) URIs as defined by generic RFC 1738.
+ */
+class HTMLPurifier_URIScheme_ftp extends HTMLPurifier_URIScheme {
+
+    public $default_port = 21;
+    public $browsable = true; // usually
+    public $hierarchical = true;
+
+    public function validate(&$uri, $config, $context) {
+        parent::validate($uri, $config, $context);
+        $uri->query    = null;
+
+        // typecode check
+        $semicolon_pos = strrpos($uri->path, ';'); // reverse
+        if ($semicolon_pos !== false) {
+            $type = substr($uri->path, $semicolon_pos + 1); // no semicolon
+            $uri->path = substr($uri->path, 0, $semicolon_pos);
+            $type_ret = '';
+            if (strpos($type, '=') !== false) {
+                // figure out whether or not the declaration is correct
+                list($key, $typecode) = explode('=', $type, 2);
+                if ($key !== 'type') {
+                    // invalid key, tack it back on encoded
+                    $uri->path .= '%3B' . $type;
+                } elseif ($typecode === 'a' || $typecode === 'i' || $typecode === 'd') {
+                    $type_ret = ";type=$typecode";
+                }
+            } else {
+                $uri->path .= '%3B' . $type;
+            }
+            $uri->path = str_replace(';', '%3B', $uri->path);
+            $uri->path .= $type_ret;
+        }
+
+        return true;
+    }
+
+}
+
+// vim: et sw=4 sts=4
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/http.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/http.php
new file mode 100755
index 000000000..b097a31d6
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/http.php
@@ -0,0 +1,20 @@
+<?php
+
+/**
+ * Validates http (HyperText Transfer Protocol) as defined by RFC 2616
+ */
+class HTMLPurifier_URIScheme_http extends HTMLPurifier_URIScheme {
+
+    public $default_port = 80;
+    public $browsable = true;
+    public $hierarchical = true;
+
+    public function validate(&$uri, $config, $context) {
+        parent::validate($uri, $config, $context);
+        $uri->userinfo = null;
+        return true;
+    }
+
+}
+
+// vim: et sw=4 sts=4
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/https.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/https.php
new file mode 100755
index 000000000..29e380919
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/https.php
@@ -0,0 +1,12 @@
+<?php
+
+/**
+ * Validates https (Secure HTTP) according to http scheme.
+ */
+class HTMLPurifier_URIScheme_https extends HTMLPurifier_URIScheme_http {
+
+    public $default_port = 443;
+
+}
+
+// vim: et sw=4 sts=4
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/mailto.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/mailto.php
new file mode 100755
index 000000000..c1e2cd5aa
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/mailto.php
@@ -0,0 +1,27 @@
+<?php
+
+// VERY RELAXED! Shouldn't cause problems, not even Firefox checks if the
+// email is valid, but be careful!
+
+/**
+ * Validates mailto (for E-mail) according to RFC 2368
+ * @todo Validate the email address
+ * @todo Filter allowed query parameters
+ */
+
+class HTMLPurifier_URIScheme_mailto extends HTMLPurifier_URIScheme {
+
+    public $browsable = false;
+
+    public function validate(&$uri, $config, $context) {
+        parent::validate($uri, $config, $context);
+        $uri->userinfo = null;
+        $uri->host     = null;
+        $uri->port     = null;
+        // we need to validate path against RFC 2368's addr-spec
+        return true;
+    }
+
+}
+
+// vim: et sw=4 sts=4
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/news.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/news.php
new file mode 100755
index 000000000..f5f54f4f5
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/news.php
@@ -0,0 +1,22 @@
+<?php
+
+/**
+ * Validates news (Usenet) as defined by generic RFC 1738
+ */
+class HTMLPurifier_URIScheme_news extends HTMLPurifier_URIScheme {
+
+    public $browsable = false;
+
+    public function validate(&$uri, $config, $context) {
+        parent::validate($uri, $config, $context);
+        $uri->userinfo = null;
+        $uri->host     = null;
+        $uri->port     = null;
+        $uri->query    = null;
+        // typecode check needed on path
+        return true;
+    }
+
+}
+
+// vim: et sw=4 sts=4
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIScheme/nntp.php b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/nntp.php
new file mode 100755
index 000000000..5bf93ea78
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIScheme/nntp.php
@@ -0,0 +1,20 @@
+<?php
+
+/**
+ * Validates nntp (Network News Transfer Protocol) as defined by generic RFC 1738
+ */
+class HTMLPurifier_URIScheme_nntp extends HTMLPurifier_URIScheme {
+
+    public $default_port = 119;
+    public $browsable = false;
+
+    public function validate(&$uri, $config, $context) {
+        parent::validate($uri, $config, $context);
+        $uri->userinfo = null;
+        $uri->query    = null;
+        return true;
+    }
+
+}
+
+// vim: et sw=4 sts=4