Add ability to linkify hashes on isolate server

appengine/isolate/handlers_frontend.py

 # Copyright 2012 The LUCI Authors. All rights reserved.
 # Use of this source code is governed by the Apache v2.0 license that can be
 # found in the LICENSE file.
 
 """This module defines Isolate Server frontend url handlers."""
 
 import datetime
 import json
+import re
 
 import webapp2
 
 from google.appengine.api import memcache
 from google.appengine.api import users
 
 import acl
 import config
 import gcs
 import mapreduce_jobs
@@ skipping 125 matching lines @@
 ### Non-restricted handlers
 
 
 class BrowseHandler(auth.AuthenticatingHandler):
   @auth.autologin
   @auth.require(acl.isolate_readable)
   def get(self):
     namespace = self.request.get('namespace', 'default-gzip')
     # Support 'hash' for compatibility with old links. To remove eventually.
     digest = self.request.get('digest', '') or self.request.get('hash', '')
+    params = {
+      u'digest': unicode(digest),
+      u'namespace': unicode(namespace),
+      # TODO(maruel): Add back once Web UI authentication is switched to OAuth2.

[M-A Ruel, 2016/04/13 20:32:09]

Remove this while at it, not needed.

[kjlubick, 2016/04/14 13:18:43]

Done.  

[kjlubick, 2016/04/14 13:18:43]

Done.

+      #'onload': 'update()' if digest else '',
+      u'onload': '',
+    }
+    # Check for existence of element, so we can 400/404
+    if digest and namespace:
+      # TODO(maruel): Refactor into a function.

[M-A Ruel, 2016/04/13 20:32:09]

Can you make it a function now that it is used at 3 different places (including
handlers_endpoints_v1). Put it in model.py.

[kjlubick, 2016/04/14 13:18:43]

I put it in model.py and used it twice here.  I couldn't figure out a nice way
to stick it in handlers_endpoints_v1 because of how the various pieces of the
lookups were entangled.

+      memcache_entry = memcache.get(digest, namespace='table_%s' % namespace)
+      if memcache_entry is None:
+        try:
+          key = model.get_entry_key(namespace, digest)
+        except ValueError:
+          self.abort(400, 'Invalid key')
+        entity = key.get()
+        if entity is None:
+          self.abort(404, 'Unable to retrieve the entry')
+    self.response.write(template.render('isolate/browse.html', params))
+

[M-A Ruel, 2016/04/13 20:32:09]

2 empty lines between file level symbols

[kjlubick, 2016/04/14 13:18:43]

Done.

+class ContentHandler(auth.AuthenticatingHandler):
+  @auth.autologin
+  @auth.require(acl.isolate_readable)
+  def get(self):
+    namespace = self.request.get('namespace', 'default-gzip')
+    digest = self.request.get('digest', '')
+
     content = None
     if digest and namespace:
       # TODO(maruel): Refactor into a function.
       memcache_entry = memcache.get(digest, namespace='table_%s' % namespace)
       if memcache_entry is not None:
         raw_data = memcache_entry
       else:
         try:
           key = model.get_entry_key(namespace, digest)
         except ValueError:
           self.abort(400, 'Invalid key')
         entity = key.get()
         if entity is None:
           self.abort(404, 'Unable to retrieve the entry')
         raw_data = entity.content
       if not raw_data:
         stream = gcs.read_file(config.settings().gs_bucket, key.id())
       else:
         stream = [raw_data]
       content = ''.join(model.expand_content(namespace, stream))
+
+      self.response.headers['X-Frame-Options'] = 'SAMEORIGIN'
+      del self.response.headers['Content-Type']
+      # Apparently, setting the content type to text/plain encourages the
+      # browser (Chrome, at least) to sniff the mime type and display
+      # things like images.  Images are autowrapped in <img> and text is
+      # wrapped in <pre>.
+      self.response.headers['Content-Type'] = 'text/plain; charset=utf-8'
+      self.response.headers['Content-Disposition'] = str('filename=%s' % digest)
       if content.startswith('{'):
         # Try to format as JSON.
         try:
           content = json.dumps(
               json.loads(content), sort_keys=True, indent=2,
               separators=(',', ': '))
+          content = (
+            '<div style="font-family:monospace;white-space:pre-wrap;">%s</div>'

[M-A Ruel, 2016/04/13 20:32:09]

Looks like this is not strictly necessary from what I could see by browsing a
bash script.

[kjlubick, 2016/04/14 13:18:43]

I'll add a comment that mirrors this, but I don't add this for the styling, but
to use a div instead of a pre.  I can't put anchor tags inside of a pre.

+             % content)
+          # Linkify things that look like hashes
+          content = re.sub(r'([0-9a-f]{40})',
+            r"""<a target="_blank" href="browse?namespace=default-gzip
+              &digest=\1">\1</a>""",
+            content)
+          self.response.headers['Content-Type'] = 'text/html; charset=utf-8'
         except ValueError:
           pass
-      content = content.decode('utf8', 'replace')
-    params = {
-      u'content': content,
-      u'digest': unicode(digest),
-      u'namespace': unicode(namespace),
-      # TODO(maruel): Add back once Web UI authentication is switched to OAuth2.
-      #'onload': 'update()' if digest else '',
-      u'onload': '',
-    }
-    self.response.write(template.render('isolate/browse.html', params))
+
+    self.response.write(content)
 
 
 class StatsHandler(webapp2.RequestHandler):
   """Returns the statistics web page."""
   def get(self):
     """Presents nice recent statistics.
 
     It fetches data from the 'JSON' API.
     """
     # Preloads the data to save a complete request.
@@ skipping 89 matching lines @@
       # Administrative urls.
       webapp2.Route(r'/restricted/config', RestrictedConfigHandler),
 
       # Mapreduce related urls.
       webapp2.Route(
           r'/restricted/launch_mapreduce',
           RestrictedLaunchMapReduceJob),
 
       # User web pages.
       webapp2.Route(r'/browse', BrowseHandler),
+      webapp2.Route(r'/content', ContentHandler),
       webapp2.Route(r'/stats', StatsHandler),
       webapp2.Route(r'/isolate/api/v1/stats/days', StatsGvizDaysHandler),
       webapp2.Route(r'/isolate/api/v1/stats/hours', StatsGvizHoursHandler),
       webapp2.Route(r'/isolate/api/v1/stats/minutes', StatsGvizMinutesHandler),
       webapp2.Route(r'/', RootHandler),
 
       # AppEngine-specific urls:
       webapp2.Route(r'/_ah/mail/<to:.+>', EmailHandler),
       webapp2.Route(r'/_ah/warmup', WarmupHandler),
   ]
 
 
 def create_application(debug):
   """Creates the url router.
 
   The basic layouts is as follow:
   - /restricted/.* requires being an instance administrator.
   - /stats/.* has statistics.
   """
   acl.bootstrap()
   template.bootstrap()
   return webapp2.WSGIApplication(get_routes(), debug=debug)