sandwich: Use cachetool's batch mode to speed-up cache processing.

tools/android/loading/sandwich_prefetch.py

Index: tools/android/loading/sandwich_prefetch.py
diff --git a/tools/android/loading/sandwich_prefetch.py b/tools/android/loading/sandwich_prefetch.py
index be8f6098136ac8f08da0cab25a1ef291f43d23c1..a76ca7d4b0fff8d6b4f5fef11e758636756f3f39 100644
--- a/tools/android/loading/sandwich_prefetch.py
+++ b/tools/android/loading/sandwich_prefetch.py
@@ -116,7 +116,7 @@ def _FilterOutDataAndIncompleteRequests(requests):
     if request.protocol is None:
       assert not request.HasReceivedResponse()
       continue
-    if request.protocol == 'about':
+    if request.protocol in {'about', 'blob'}:
       continue
     if request.protocol not in {'http/0.9', 'http/1.0', 'http/1.1'}:
       raise RuntimeError('Unknown request protocol {}'.format(request.protocol))
@@ -138,39 +138,49 @@ def _PatchCacheArchive(cache_archive_path, loading_trace_path,
         archive <cache_archive_path>.
     cache_archive_dest_path: Archive destination's path.
   """
+  logging.info('loading trace: %s', loading_trace_path)
   trace = loading_trace.LoadingTrace.FromJsonFile(loading_trace_path)
   with common_util.TemporaryDirectory(prefix='sandwich_tmp') as tmp_path:
     cache_path = os.path.join(tmp_path, 'cache')
     chrome_cache.UnzipDirectoryContent(cache_archive_path, cache_path)
-    cache_backend = chrome_cache.CacheBackend(cache_path, 'simple')
-    cache_entries = set(cache_backend.ListKeys())
-    logging.info('Original cache size: %d bytes' % cache_backend.GetSize())
-    for request in _FilterOutDataAndIncompleteRequests(
-        trace.request_track.GetEvents()):
-      # On requests having an upload data stream such as POST requests,
-      # net::HttpCache::GenerateCacheKey() prefixes the cache entry's key with
-      # the upload data stream's session unique identifier.
-      #
-      # It is fine to not patch these requests since when reopening Chrome,
-      # there is no way the entry can be reused since the upload data stream's
-      # identifier will be different.
-      #
-      # The fact that these entries are kept in the cache after closing Chrome
-      # properly by closing the Chrome tab as the ChromeControler.SetSlowDeath()
-      # do is known chrome bug (crbug.com/610725).
-      if request.url not in cache_entries:
-        continue
-      # Chrome prunes Set-Cookie from response headers before storing them in
-      # disk cache. Also, it adds implicit "Vary: cookie" header to all redirect
-      # response headers. Sandwich manages the cache, but between recording the
-      # cache and benchmarking the cookie jar is invalidated. This leads to
-      # invalidation of all cacheable redirects.
-      raw_headers = request.GetRawResponseHeaders()
-      cache_backend.UpdateRawResponseHeaders(request.url, raw_headers)
-      # NoState-Prefetch would only fetch the resources, but not parse them.
-      cache_backend.DeleteStreamForKey(request.url, 2)
+    with chrome_cache.OnlineCacheBackend(
+        cache_path, chrome_cache.CacheBackendType.SIMPLE) as cache_backend:
+      cache_entries = set(cache_backend.ListKeys())
+      logging.info('Original cache size: %d bytes' % cache_backend.GetSize())
+      for request in _FilterOutDataAndIncompleteRequests(
+          trace.request_track.GetEvents()):
+        # On requests having an upload data stream such as POST requests,
+        # net::HttpCache::GenerateCacheKey() prefixes the cache entry's key with
+        # the upload data stream's session unique identifier.
+        #
+        # It is fine to not patch these requests since when reopening Chrome,
+        # there is no way the entry can be reused since the upload data stream's
+        # identifier will be different.
+        #
+        # The fact that these entries are kept in the cache after closing Chrome
+        # properly by closing the Chrome tab as the
+        # ChromeControler.SetSlowDeath() do is known chrome bug
+        # (crbug.com/610725).
+        if request.url not in cache_entries:
+          continue
+        # Chrome prunes Set-Cookie from response headers before storing them in
+        # disk cache. Also, it adds implicit "Vary: cookie" header to all
+        # redirect response headers. Sandwich manages the cache, but between
+        # recording the # cache and benchmarking the cookie jar is invalidated.
+        # This leads to invalidation of all cacheable redirects.
+        raw_headers = request.GetRawResponseHeaders()
+        try:
+          cache_backend.UpdateRawResponseHeaders(request.url, raw_headers)
+          # NoState-Prefetch would only fetch the resources, but not parse them.
+          cache_backend.DeleteStreamForKey(request.url, 2)
+          # Sync operations to actually catch errors here.
+          cache_backend.Sync()
+        except chrome_cache.CacheBackendError as error:
+          # For some reason, sometimes the cachetool can't find key's entry when
+          # not using the online mode.
+          logging.warning('cachetool error: %s', repr(error))
+      logging.info('Patched cache size: %d bytes' % cache_backend.GetSize())
     chrome_cache.ZipDirectoryContent(cache_path, cache_archive_dest_path)
-    logging.info('Patched cache size: %d bytes' % cache_backend.GetSize())
 
 
 def _DiscoverRequests(dependencies_lens, subresource_discoverer):
@@ -403,8 +413,8 @@ def _ValidateCacheArchiveContent(cache_build_trace_path, cache_archive_path):
   logging.info('lists cached urls from %s' % cache_archive_path)
   with common_util.TemporaryDirectory() as cache_directory:
     chrome_cache.UnzipDirectoryContent(cache_archive_path, cache_directory)
-    cache_keys = set(
-        chrome_cache.CacheBackend(cache_directory, 'simple').ListKeys())
+    cache_keys = set(chrome_cache.CacheBackend(
+        cache_directory, chrome_cache.CacheBackendType.SIMPLE).ListKeys())
   trace = loading_trace.LoadingTrace.FromJsonFile(cache_build_trace_path)
   effective_requests = _ListUrlRequests(trace, _RequestOutcome.All)
   effective_post_requests = _ListUrlRequests(trace, _RequestOutcome.Post)