From 09981cc8d27de8cda1d8fe93454f56ffd11fdf01 Mon Sep 17 00:00:00 2001
From: Aaron Parecki <aaron@parecki.com>
Date: Thu, 6 Apr 2017 16:24:35 -0700
Subject: [PATCH] add new endpoint to parse all rel values from a URL

Grabs rel values from both HTTP headers and link tags and merges them together, and resolves any relative values. Also returns the final URL requested and the HTTP response.
---
 composer.json        |  1 +
 controllers/Rels.php | 93 ++++++++++++++++++++++++++++++++++++++++++++
 public/index.php     |  2 +
 3 files changed, 96 insertions(+)
 create mode 100644 controllers/Rels.php

diff --git a/composer.json b/composer.json
index dac355b..0f60c3b 100644
--- a/composer.json
+++ b/composer.json
@@ -14,6 +14,7 @@
       "controllers/Main.php",
       "controllers/Parse.php",
       "controllers/Token.php",
+      "controllers/Rels.php",
       "controllers/Certbot.php",
       "lib/HTTPCurl.php",
       "lib/HTTPStream.php",
diff --git a/controllers/Rels.php b/controllers/Rels.php
new file mode 100644
index 0000000..9932513
--- /dev/null
+++ b/controllers/Rels.php
@@ -0,0 +1,93 @@
+<?php
+use Symfony\Component\HttpFoundation\Request;
+use Symfony\Component\HttpFoundation\Response;
+
+class Rels {
+
+  public $http;
+  private $_pretty = false;
+
+  public function __construct() {
+    $this->http = new p3k\HTTP();
+  }
+
+  private function respond(Response $response, $code, $params, $headers=[]) {
+    $response->setStatusCode($code);
+    foreach($headers as $k=>$v) {
+      $response->headers->set($k, $v);
+    }
+    $response->headers->set('Content-Type', 'application/json');
+    $opts = JSON_UNESCAPED_SLASHES + JSON_FORCE_OBJECT;
+    if($this->_pretty) $opts += JSON_PRETTY_PRINT;
+    $response->setContent(json_encode($params, $opts)."\n");
+    return $response;
+  }
+
+  public function fetch(Request $request, Response $response) {
+    if($request->get('timeout')) {
+      // We might make 2 HTTP requests, so each request gets half the desired timeout
+      $this->http->timeout = $request->get('timeout') / 2;
+    }
+
+    if($request->get('max_redirects')) {
+      $this->http->max_redirects = (int)$request->get('max_redirects');
+    }
+
+    if($request->get('pretty')) {
+      $this->_pretty = true;
+    }
+
+    $url = $request->get('url');
+
+    if(!$url) {
+      return $this->respond($response, 400, [
+        'error' => 'missing_url',
+        'error_description' => 'Provide a URL 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'
+      ]);
+    }
+
+    $host = parse_url($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);
+
+    $result = $this->http->get($url);
+
+    $html = $result['body'];
+    $mf2 = mf2\Parse($html, $result['url']);
+
+    $rels = p3k\HTTP::link_rels($result['headers']);
+    if(isset($mf2['rels'])) {
+      $rels = array_merge($rels, $mf2['rels']);
+    }
+
+    // Resolve all relative URLs
+    foreach($rels as $rel=>$values) {
+      foreach($values as $i=>$value) {
+        $value = \mf2\resolveUrl($result['url'], $value);
+        $rels[$rel][$i] = $value;
+      }
+    }
+
+    return $this->respond($response, 200, [
+      'url' => $result['url'],
+      'code' => $result['code'],
+      'rels' => $rels
+    ]);
+  }
+
+}
diff --git a/public/index.php b/public/index.php
index b0cb02a..d6e4b5d 100644
--- a/public/index.php
+++ b/public/index.php
@@ -24,6 +24,8 @@ $router->addRoute('GET', '/parse', 'Parse::parse');
 $router->addRoute('POST', '/parse', 'Parse::parse');
 $router->addRoute('POST', '/token', 'Token::token');
 
+$router->addRoute('GET', '/rels', 'Rels::fetch');
+
 $router->addRoute('GET', '/cert', 'Certbot::index');
 $router->addRoute('GET', '/cert/auth', 'Certbot::start_auth');
 $router->addRoute('GET', '/cert/logout', 'Certbot::logout');