Telemetry: add a metadata layer between page set and .wpr.

tools/telemetry/telemetry/page_runner.py

Index: tools/telemetry/telemetry/page_runner.py
diff --git a/tools/telemetry/telemetry/page_runner.py b/tools/telemetry/telemetry/page_runner.py
index 6600ce664e0245fe0777dd38a5aff32cca5140d1..6bc4f70504284c5cd88af890d3797aeac984ba7a 100644
--- a/tools/telemetry/telemetry/page_runner.py
+++ b/tools/telemetry/telemetry/page_runner.py
@@ -3,7 +3,6 @@
 # found in the LICENSE file.
 import logging
 import os
-import re
 import time
 import traceback
 import urlparse
@@ -35,7 +34,7 @@ class _RunState(object):
       self.browser.Close()
       self.browser = None
 
-def _ShuffleAndFilterPageSet(page_set, options):
+def _ShufflePageSet(page_set, options):
   if options.test_shuffle_order_file and not options.test_shuffle:
     raise Exception('--test-shuffle-order-file requires --test-shuffle.')
 
@@ -43,13 +42,6 @@ def _ShuffleAndFilterPageSet(page_set, options):
     return page_set.ReorderPageSet(options.test_shuffle_order_file)
 
   pages = page_set.pages[:]
-  if options.page_filter:
-    try:
-      page_regex = re.compile(options.page_filter)
-    except re.error:
-      raise Exception('--page-filter: invalid regex')
-    pages = [page for page in pages if page_regex.search(page.url)]
-
   if options.test_shuffle:
     random.Random().shuffle(pages)
   return [page
@@ -69,30 +61,26 @@ class PageRunner(object):
     self.Close()
 
   def Run(self, options, possible_browser, test, results):
-    # Set up WPR mode.
-    if not self.page_set.archive_path:
-      archive_path = ''
-      if not self.page_set.ContainsOnlyFileURLs():
+    # Check if we can run against WPR.
+    for page in self.page_set.pages:
+      parsed_url = urlparse.urlparse(page.url)
+      if parsed_url.scheme == 'file':
+        continue
+      if not page.archive_path:
         logging.warning("""
-  No page set archive provided for the chosen page set. Benchmarking against
-  live sites! Results won't be repeatable or comparable.
-""")
-    else:
-      archive_path = os.path.abspath(os.path.join(self.page_set.base_dir,
-                                                  self.page_set.archive_path))
-      if options.wpr_mode == wpr_modes.WPR_OFF:
-        if os.path.isfile(archive_path):
-          possible_browser.options.wpr_mode = wpr_modes.WPR_REPLAY
-        else:
-          possible_browser.options.wpr_mode = wpr_modes.WPR_OFF
-          if not self.page_set.ContainsOnlyFileURLs():
-            logging.warning("""
-  The page set archive %s does not exist, benchmarking against live sites!
+  No page set archive provided for the page %s. Benchmarking against live sites!
   Results won't be repeatable or comparable.
+""", page.url)
+      elif options.wpr_mode == wpr_modes.WPR_OFF:
+        # The page has an archive, and we're not recording.
+        if not os.path.isfile(page.archive_path):
+          logging.warning("""
+  The page set archive %s for page %s does not exist, benchmarking against live
+  sites! Results won't be repeatable or comparable.
 
   To fix this, either add svn-internal to your .gclient using
   http://goto/read-src-internal, or create a new archive using record_wpr.
-  """, os.path.relpath(archive_path))
+  """, os.path.relpath(page.archive_path), page.url)
 
     # Verify credentials path.
     credentials_path = None
@@ -120,17 +108,27 @@ class PageRunner(object):
         raise Exception('Trace directory isn\'t empty: %s' % options.trace_dir)
 
     # Reorder page set based on options.
-    pages = _ShuffleAndFilterPageSet(self.page_set, options)
+    pages = _ShufflePageSet(self.page_set, options)
 
     state = _RunState()
+    last_archive_path = None
     try:
       for page in pages:
+        if options.wpr_mode == wpr_modes.WPR_OFF:

[dtu, 2013/01/23 21:07:02]

I think "!= wpr_modes.WPR_RECORD" is clearer

[marja, 2013/01/24 16:03:33]

Done.

+          if os.path.isfile(page.archive_path):
+            possible_browser.options.wpr_mode = wpr_modes.WPR_REPLAY
+          else:
+            possible_browser.options.wpr_mode = wpr_modes.WPR_OFF
+        if last_archive_path != page.archive_path:
+          state.Close()
+          state = _RunState()
+          last_archive_path = page.archive_path
         tries = 3
         while tries:
           try:
             if not state.browser:
               self._SetupBrowser(state, test, possible_browser,
-                                 credentials_path, archive_path)
+                                 credentials_path, page.archive_path)
             if not state.tab:
               if len(state.browser.tabs) == 0:
                 state.browser.tabs.New()