fixes for windows: convert url to file with pathname2url, use 'b' for

reading warcs, don't use %s for timestamp conversion (not portable)
(#56)

pywb/warc/cdxindexer.py

@@ -116,7 +116,7 @@ def write_multi_cdx_index(output, inputs, **options):
             outpath = os.path.join(output, outpath)
 
             with open(outpath, 'w') as outfile:
-                with open(fullpath, 'r') as infile:
+                with open(fullpath, 'rb') as infile:
                     write_cdx_index(outfile, infile, filename, **options)
 
     # write to one cdx file
@@ -133,7 +133,7 @@ def write_multi_cdx_index(output, inputs, **options):
 
         with writer_cls(outfile, options.get('cdx09')) as writer:
             for fullpath, filename in iter_file_or_dir(inputs):
-                with open(fullpath, 'r') as infile:
+                with open(fullpath, 'rb') as infile:
                     entry_iter = create_index_iter(infile, **options)
 
                     for entry in entry_iter:

pywb/warc/pathresolvers.py

@@ -3,6 +3,7 @@ import redis
 from pywb.utils.binsearch import iter_exact
 
 import urlparse
+import urllib
 import os
 import logging
 
@@ -92,6 +93,7 @@ def make_best_resolver(param):
 
     if url_parts.scheme == 'file':
         path = url_parts.path
+        path = urllib.url2pathname(path)
 
     if os.path.isfile(path):
         logging.debug('Adding Path Index: ' + path)

pywb/warc/test/test_pathresolvers.py

@@ -33,13 +33,13 @@ PrefixResolver('http://myhost.example.com/warcs/', contains = '/')
 RedisResolver('redis://myhost.example.com:1234/1')
 
 # a file
->>> r = make_best_resolver('file://' + os.path.realpath(__file__))
+>>> r = make_best_resolver(to_local_url(os.path.realpath(__file__)))
 >>> r.__class__.__name__
 'PathIndexResolver'
 
 # a dir
 >>> path = os.path.realpath(__file__)
->>> r = make_best_resolver('file://' + os.path.dirname(path))
+>>> r = make_best_resolver(to_local_url(os.path.dirname(path)))
 >>> r.__class__.__name__
 'PrefixResolver'
 
@@ -56,6 +56,7 @@ from pywb.warc.pathresolvers import PrefixResolver, PathIndexResolver, RedisReso
 from pywb.warc.pathresolvers import make_best_resolver, make_best_resolvers
 import os
 
+from urllib import pathname2url
 
 from fakeredis import FakeStrictRedis
 from mock import patch
@@ -68,6 +69,11 @@ def init_redis_resolver():
 def hset_path(filename, path):
     redis_resolver.redis.hset(redis_resolver.key_prefix + filename, 'path', path)
 
+def to_local_url(filename):
+    filename = os.path.abspath(filename)
+    res = 'file:' + pathname2url(filename)
+    #print(res)
+    return res
 
 redis_resolver = init_redis_resolver()
 

pywb/webapp/views.py

@@ -4,6 +4,7 @@ from pywb.framework.memento import make_timemap, LINK_FORMAT
 
 import urlparse
 import logging
+import time
 
 from os import path
 from itertools import imap
@@ -42,7 +43,10 @@ class template_filter(object):
 @template_filter
 def format_ts(value, format_='%a, %b %d %Y %H:%M:%S'):
     value = timestamp_to_datetime(value)
-    return value.strftime(format_)
+    if format_ == '%s':
+        return int(time.mktime(value.timetuple()) * 1000)
+    else:
+        return value.strftime(format_)
 
 
 @template_filter('urlsplit')