www: override info.cern.ch and cernvax.cern.ch to allow for loading d…

…efault.html from proxy server!

proxy serves default.html and rewritten content from pywb, filtering out most headers to avoid breaking browser
add icon for www
pywb: use blank header, /all/ endpoint
app: set default dims to 1024x768, add per-browser override configurable in config.yaml
addresses oldweb-today#20

app/config.yaml

@@ -66,10 +66,12 @@ browsers:
     - id: www
       name: 'WWW'
       os: 'NextSTEP'
-      icon: 'firefox.png'
+      icon: 'next-www.png'
       path: www
       version: '1.0'
-      about: ''
+      about: 'https://en.wikipedia.org/wiki/WorldWideWeb'
+      req_width: 1120
+      req_height: 856
 
     # mosaic
     - id: mosaic

app/main.py

@@ -124,6 +124,12 @@ def new_container(self, browser_id, env=None, default_host=None):
         if not browser:
             browser = self.browser_paths.get(self.default_browser)
 
+        if browser.get('req_width'):
+            env['SCREEN_WIDTH'] = browser.get('req_width')
+
+        if browser.get('req_height'):
+            env['SCREEN_HEIGHT'] = browser.get('req_height')
+
         container = self.cli.create_container(image=self.image_prefix + '/' + browser['id'],
                                               ports=[self.VNC_PORT, self.CMD_PORT],
                                               environment=env,

app/static/dropdown.css

@@ -94,6 +94,11 @@
                 line-height: 0;
             }
 
+            #browser-icon {
+                width: 32px;
+                height: 32px;
+            }
+
             #browser-text {
                 line-height: 58px;
             }

app/static/main.css

@@ -140,7 +140,7 @@ iframe {
 
 /* replay page */
 
-#page-replay {overflow: hidden;}
+#page-replay {overflow: inherit;}
 
     /* fonts */
 

browsers/www/Dockerfile

@@ -21,6 +21,8 @@ COPY tars.iso.dmg /home/browser/
 
 COPY previous.cfg /home/browser/.previous/previous.cfg
 
+COPY proxy.py /home/browser/proxy.py
+
 COPY run.sh /app/run.sh
 RUN sudo chmod a+x /app/run.sh
 

browsers/www/proxy.py

@@ -0,0 +1,66 @@
+from bottle import request, route, HTTPResponse, run
+from argparse import ArgumentParser
+
+import requests
+
+pywb_prefix = 'localhost'
+proxy_host = '10.0.2.2'
+start_ts = '1990'
+start_url = ''
+
+ALLOWED_HEADERS = ('content-type', 'content-length', 'location', 'date',
+                   'last-modified', 'set-cookie', 'server', 'content-encoding')
+
+@route('/default.html')
+def do_default():
+    return do_proxy(start_ts + '/' + start_url)
+
+
+@route('/all/<url:re:.*>')
+def do_proxy(url):
+    headers = {'User-Agent': request.environ.get('HTTP_USER_AGENT'),
+               'Host': proxy_host,
+               'Accept-Encoding': 'identity',
+              }
+
+    r = requests.get(url=pywb_prefix + url, headers=headers,
+                     stream=True,
+                     allow_redirects=False)
+
+    resp_headers = []
+    for n, v in r.headers.iteritems():
+        if n == 'content-type':
+            v = v.split(';')[0].strip()
+
+        if n.lower() in ALLOWED_HEADERS:
+            resp_headers.append((n, v))
+
+    resp = HTTPResponse(body=r.iter_content(8192),
+                        status=str(r.status_code) + ' ' + r.reason,
+                        headers=resp_headers)
+
+    return resp
+
+
+if __name__ == "__main__":
+    parser = ArgumentParser('netcapsule http1.0 proxy')
+    parser.add_argument('--pywb-prefix')
+    parser.add_argument('--start-url')
+    parser.add_argument('--start-ts')
+    parser.add_argument('--port', type=int)
+
+    r = parser.parse_args()
+
+    global pywb_prefix
+    pywb_prefix = r.pywb_prefix
+
+    global start_ts
+    start_ts = r.start_ts
+
+    global start_url
+    start_url = r.start_url
+
+    port = r.port or 8081
+
+    run(host='0.0.0.0', port=port)
+

browsers/www/run.sh

@@ -14,5 +14,8 @@ sudo chown browser ./.previous/previous.cfg
 #fluxbox -display $DISPLAY -log /tmp/fluxbox.log &
 #fvwm -d $DISPLAY &
 
+PYWB_IP=$(grep netcapsule_pywb_1 /etc/hosts | cut -f 1 | head -n 1)
+sudo python proxy.py --start-url $URL --pywb-prefix http://$PYWB_IP:8080/all/ --start-ts $TS --port 80 &
+
 run_browser Previous
 

docker-compose.yml

@@ -59,8 +59,8 @@ app:
         - ${RANDOM_URL_LIST}:/app/urls.txt
 
     environment:
-        - SCREEN_WIDTH=1120
-        - SCREEN_HEIGHT=856
+        - SCREEN_WIDTH=1024
+        - SCREEN_HEIGHT=768
 
     ports:
         - 9020:9020

pywb/config.yaml

@@ -11,7 +11,7 @@ collections:
 
 
     # Memento Aggregator Collection: Specify paths to Timegate, Timemap
-    memento_reconstruct:
+    all:
         index_paths:
             # for LANL aggregator
             #timegate: http://timetravel.mementoweb.org/api/json/
@@ -25,6 +25,8 @@ collections:
 
         wb_handler_class: !!python/name:archivereplayview.MementoHandler
 
+        head_insert_html: ./templates/blank.html
+
 #        fallback: live
 
     live:
@@ -55,7 +57,7 @@ proxy_options:
 
     # Select one of these
     #use_default_coll: 'single_archive'
-    use_default_coll: 'memento_reconstruct'
+    use_default_coll: 'all'
 
     cookie_resolver: ip
     use_client_rewrite: false