Stabilize WARC headers

In preparation for 1.0 release:

- Correct mime types
- Add X-Crocoite-Type, so logs, scripts, dom-snapshots and screenshots
  can be identified easily
- Remove random WARC headers like X-Chrome-Initiator. We don’t want to
  maintain those.
- Remove non-standard urn-based package URLs. Can’t use them without a
  urn-registration

1 files changed, 2 insertions, 6 deletions

diff --git a/crocoite/util.py b/crocoite/util.py
index ded5e99..da377a3 100644
--- a/crocoite/util.py
+++ b/crocoite/util.py
@@ -26,6 +26,8 @@ import random, sys, platform, os, json, urllib
 from datetime import datetime
 import hashlib, pkg_resources
 
+from yarl import URL
+
 class StrJsonEncoder (json.JSONEncoder):
     """ JSON encoder that turns unknown classes into a string and thus never
     fails """
@@ -39,12 +41,6 @@ class StrJsonEncoder (json.JSONEncoder):
         except TypeError:
             return str (obj)
 
-def packageUrl (path):
-    """
-    Create URL for package data stored into WARC
-    """
-    return 'urn:' + __package__ + ':' + urllib.parse.quote (path)
-
 async def getFormattedViewportMetrics (tab):
     layoutMetrics = await tab.Page.getLayoutMetrics ()
     # XXX: I’m not entirely sure which one we should use here