From c4b80506da3481739083cff39e14794480d46f3a Mon Sep 17 00:00:00 2001
From: Aaron Parecki <aaron@parecki.com>
Date: Mon, 29 Feb 2016 19:33:44 -0800
Subject: [PATCH] support parsing posted HTML

---
 controllers/Parse.php | 56 ++++++++++++++++++++++++-------------------
 lib/Formats/Mf2.php   |  4 +++-
 public/index.php      |  1 +
 3 files changed, 35 insertions(+), 26 deletions(-)

diff --git a/controllers/Parse.php b/controllers/Parse.php
index ed074a8..5cca349 100644
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@@ -45,41 +45,47 @@ class Parse {
     }
 
     $url = $request->get('url');
+    $html = $request->get('html');
 
-    if(!$url) {
+    if(!$url && !$html) {
       return $this->respond($response, 400, [
         'error' => 'missing_url',
-        'error_description' => 'Provide a URL to fetch'
+        'error_description' => 'Provide a URL or HTML to fetch'
       ]);
     }
 
-    // Attempt some basic URL validation
-    $scheme = parse_url($url, PHP_URL_SCHEME);
-    if(!in_array($scheme, ['http','https'])) {
-      return $this->respond($response, 400, [
-        'error' => 'invalid_url',
-        'error_description' => 'Only http and https URLs are supported'
-      ]);
-    }
+    if($html) {
+      // If HTML is provided in the request, parse that, and use the URL provided as the base URL for mf2 resolving
+      $result['body'] = $html;
+    } else {
+      // Attempt some basic URL validation
+      $scheme = parse_url($url, PHP_URL_SCHEME);
+      if(!in_array($scheme, ['http','https'])) {
+        return $this->respond($response, 400, [
+          'error' => 'invalid_url',
+          'error_description' => 'Only http and https URLs are supported'
+        ]);
+      }
 
-    $host = parse_url($url, PHP_URL_HOST);
-    if(!$host) {
-      return $this->respond($response, 400, [
-        'error' => 'invalid_url',
-        'error_description' => 'The URL provided was not valid'
-      ]);
-    }
+      $host = parse_url($url, PHP_URL_HOST);
+      if(!$host) {
+        return $this->respond($response, 400, [
+          'error' => 'invalid_url',
+          'error_description' => 'The URL provided was not valid'
+        ]);
+      }
 
-    $url = \normalize_url($url);
+      $url = \normalize_url($url);
 
-    // Now fetch the URL and check for any curl errors
-    $result = $this->http->get($url);
+      // Now fetch the URL and check for any curl errors
+      $result = $this->http->get($url);
 
-    if($result['error']) {
-      return $this->respond($response, 400, [
-        'error' => $result['error'],
-        'error_description' => $result['error_description']
-      ]);
+      if($result['error']) {
+        return $this->respond($response, 400, [
+          'error' => $result['error'],
+          'error_description' => $result['error_description']
+        ]);
+      }
     }
 
     // attempt to parse the page as HTML
diff --git a/lib/Formats/Mf2.php b/lib/Formats/Mf2.php
index f0137f0..f115b60 100644
--- a/lib/Formats/Mf2.php
+++ b/lib/Formats/Mf2.php
@@ -78,7 +78,7 @@ class Mf2 {
     }
 
     // Always returned as arrays, and may also create external references
-    $properties = ['in-reply-to','like-of','repost-of','bookmark-of','category'];
+    $properties = ['in-reply-to','like-of','repost-of','bookmark-of','category','invitee'];
     foreach($properties as $p) {
       if(array_key_exists($p, $item['properties'])) {
         $data[$p] = [];
@@ -132,6 +132,8 @@ class Mf2 {
     if($name) {
       $data['name'] = $name;
     }
+
+    // If there is content, always return the plaintext content, and return HTML content if it's different
     if($content) {
       $data['content'] = [
         'text' => $textContent
diff --git a/public/index.php b/public/index.php
index 899a7bf..c579bfe 100644
--- a/public/index.php
+++ b/public/index.php
@@ -10,6 +10,7 @@ $templates = new League\Plates\Engine(dirname(__FILE__).'/../views');
 
 $router->addRoute('GET', '/', 'Main::index');
 $router->addRoute('GET', '/parse', 'Parse::parse');
+$router->addRoute('POST', '/parse', 'Parse::parse');
 
 $dispatcher = $router->getDispatcher();
 $request = Request::createFromGlobals();