Use JSON-encoded POST body for /raw

3 files changed, 66 insertions, 23 deletions

diff --git a/swayback/__init__.py b/swayback/__init__.py
index 80600f3..f609736 100644
--- a/swayback/__init__.py
+++ b/swayback/__init__.py
@@ -1,11 +1,11 @@
 import os
+from io import BytesIO
+from urllib.parse import urlparse, urlunparse
 
 from flask import Flask, render_template, send_file, request, make_response
 from warcio.archiveiterator import ArchiveIterator
 from warcio.recordloader import ArcWarcRecordLoader
 from warcio.bufferedreaders import DecompressingBufferedReader
-from io import BytesIO
-from urllib.parse import urlparse, urlunparse
 
 app = Flask(__name__)
 app.url_map.host_matching = True
@@ -33,24 +33,40 @@ def index ():
     """ A simple index of all HTML pages inside the WARCs """
     return render_template ('index.html', index=htmlindex)
 
-@app.route('/raw', host='swayback.localhost:5000', methods=['POST'])
-def raw ():
-    """ Retrieve the original response for a given request """
-    print (request.form)
-    url = urlparse (request.form['url'])
+@app.route('/raw', host='swayback.localhost:5000', methods=['OPTIONS'])
+def rawPreflight ():
+    """ CORS preflight request, allow user-defined fetch() headers """
+    resp = make_response ('', 200)
+    resp.headers.add ('Access-Control-Allow-Origin', '*')
+    resp.headers.add ('Access-Control-Allow-Headers', 'Content-Type')
+    resp.headers.add ('Access-Control-Allow-Methods', 'POST')
+    return resp
+
+def lookupRecord (url):
+    """ Look up URL in database. """
     try:
         filename, offset, length = urlmap[url]
         with open(filename, 'rb') as stream:
             stream.seek (offset, 0)
             buf = BytesIO (stream.read (length))
             loader = ArcWarcRecordLoader ()
-            record = loader.parse_record_stream (DecompressingBufferedReader (buf))
-            statuscode = record.http_headers.get_statuscode ()
-            record.http_headers.remove_header ('Content-Security-Policy')
-            record.http_headers.replace_header ('Access-Control-Allow-Origin', '*')
-            headers = record.http_headers.headers
-            return record.content_stream().read(), statuscode, headers
+            return loader.parse_record_stream (DecompressingBufferedReader (buf))
     except KeyError:
+        return None
+
+@app.route('/raw', host='swayback.localhost:5000', methods=['POST'])
+def raw ():
+    """ Retrieve the original response for a given request """
+    data = request.get_json ()
+    url = urlparse (data['url'])
+    record = lookupRecord (url)
+    if record:
+        statuscode = record.http_headers.get_statuscode ()
+        record.http_headers.remove_header ('Content-Security-Policy')
+        record.http_headers.replace_header ('Access-Control-Allow-Origin', '*')
+        headers = record.http_headers.headers
+        return record.content_stream().read(), statuscode, headers
+    else:
         resp = make_response ('', 404)
         resp.headers.add ('Access-Control-Allow-Origin', '*')
         return resp
diff --git a/swayback/static/sw.js b/swayback/static/sw.js
index f3a63db..56889c0 100644
--- a/swayback/static/sw.js
+++ b/swayback/static/sw.js
@@ -3,13 +3,18 @@ self.addEventListener('install', function(event) {
 	self.skipWaiting();
 });
 /* load stuff through service worker immediately? XXX: only debugging? */
-self.addEventListener('activate', event => {
-  event.waitUntil(clients.claim());
+self.addEventListener('activate', async function() {
+    if (self.registration.navigationPreload) {
+      // Enable navigation preloads!
+      await self.registration.navigationPreload.enable();
+    } /*event => {
+  event.waitUntil(clients.claim());*/
 });
 
 self.addEventListener('fetch', function(event) {
-	console.log ('fetch event', event.request.url, event);
-	let url = new URL (event.request.url);
+	let origreq = event.request;
+	console.log ('fetch event', origreq.url, event);
+	let url = new URL (origreq.url);
 	url.protocol = 'https:';
 	url.port = 443;
 	url.hash = '';
@@ -17,10 +22,23 @@ self.addEventListener('fetch', function(event) {
 		url.hostname = url.hostname.slice (0, url.hostname.length-'.swayback.localhost'.length);
 	}
 	console.log ('orig url', url);
-	let body = new FormData ();
-	body.append ('url', url);
-	body.append ('method', event.request.method);
-	let req = new Request ('http://swayback.localhost:5000/raw', {method: 'POST', body: body});
+	/* should contain everything we cannot use in the actual request (i.e. url and method) */
+	let body = {
+		'url': url.href,
+		'method': origreq.method,
+		};
+	let headers = {
+		'Content-Type': 'application/json',
+		};
+	/* add a few well-known request headers */
+	let origheaders = origreq.headers;
+	if (origheaders.has ('accept')) {
+		headers['Accept'] = origreq.headers.get ('accept');
+	}
+	console.log ('sending', body, headers);
+	let req = new Request ('http://swayback.localhost:5000/raw',
+			{method: 'POST', body: JSON.stringify (body), headers: headers,
+			mode: 'cors'});
 
 	event.respondWith (
 		fetch(req)
diff --git a/swayback/templates/sw.html b/swayback/templates/sw.html
index 2abfac2..0735ae9 100644
--- a/swayback/templates/sw.html
+++ b/swayback/templates/sw.html
@@ -24,8 +24,17 @@
 		/* service workers must be hosted in the same origin (i.e. subdomain) */
 		navigator.serviceWorker.register('/static/sw.js', {scope: '/'})
 		.then(function(reg) {
-			/* load actual content using the service worker */
-			window.location.reload ();
+			/* load actual content using the service worker when done installing */
+			if (reg.installing) {
+				reg.installing.addEventListener ('statechange', function (e) {
+					if (e.target.state !== 'installing') {
+						console.log ('reloading');
+						window.location.reload ();
+						return false;
+					}
+					return true;
+				});
+			}
 		}).catch(function(error) {
 			console.log ('sw error', error);
 			document.getElementById ('unsupported').classList.remove ('hidden');