Use the date of the last major branch as a base date for expiring histograms.

Bug: 753485
Change-Id: Ieed576a7150cb2c39e728e623645ab2126eff9e3
Reviewed-on: https://chromium-review.googlesource.com/665391Reviewed-by: Jesse Doherty <jwd@chromium.org>
Reviewed-by: Alexei Svitkine <asvitkine@chromium.org>
Commit-Queue: Ira Burak <iburak@google.com>
Cr-Commit-Position: refs/heads/master@{#502902}

chrome/browser/metrics/BUILD.gn

@@ -11,4 +11,5 @@ generate_expired_histograms_array("expired_histograms_array") {
   ]
   namespace = "chrome_metrics"
   header_filename = "expired_histograms_array.h"
+  major_branch_date_filepath = "//chrome/MAJOR_BRANCH_DATE"
 }

components/metrics/generate_expired_histograms_array.gni

@@ -17,6 +17,9 @@
 #   header_filename:
 #     Name of the generated header file.
 #
+#   major_branch_date_filepath:
+#     A path to the file with the base date.
+#
 template("generate_expired_histograms_array") {
   action(target_name) {
     header_filename = "$target_gen_dir/" + invoker.header_filename
@@ -27,6 +30,7 @@ template("generate_expired_histograms_array") {
     ]
 
     inputs = invoker.inputs
+    major_branch_date_filepath = invoker.major_branch_date_filepath
 
     args = []
 
@@ -37,6 +41,7 @@ template("generate_expired_histograms_array") {
     args += [
               "-o" + rebase_path(root_gen_dir, root_build_dir),
               "-H" + rebase_path(header_filename, root_gen_dir),
+              "-d" + rebase_path(major_branch_date_filepath, root_build_dir),
             ] + rebase_path(inputs, root_build_dir)
   }
 }

tools/metrics/histograms/generate_expired_histograms_array.py

@@ -14,6 +14,8 @@ import sys
 import extract_histograms
 import merge_xml
 
+_DATE_FILE_PATTERN = r".*MAJOR_BRANCH_DATE=(.+).*"
+
 _SCRIPT_NAME = "generate_expired_histograms_array.py"
 _HASH_DATATYPE = "uint64_t"
 _HEADER = """// Generated from {script_name}. Do not edit!
@@ -72,6 +74,34 @@ def _GetExpiredHistograms(histograms, base_date):
   return expired_histograms_names
 
 
+def _GetBaseDate(content, pattern):
+  """Fetches base date from |content| to compare expiry dates with.
+
+  Args:
+   content: A string with the base date.
+   pattern(str): A regular expression that matches the base date.
+
+  Returns:
+   A base date as datetime.date object.
+
+  Raises:
+    Error if |content| doesn't match |pattern| or the matched date has invalid
+    format.
+  """
+  match_result = re.search(pattern, content)
+  if not match_result:
+    raise Error("Unable to match {pattern} with provided content: {content}".
+                format(pattern=pattern, content=content))
+  base_date_str = match_result.group(1)
+  try:
+    base_date = datetime.datetime.strptime(
+        base_date_str, extract_histograms.EXPIRY_DATE_PATTERN).date()
+    return base_date
+  except ValueError:
+    raise Error("Unable to parse base date {date} from {content}.".
+                format(date=base_date_str, content=content))
+
+
 def _HashName(name):
   """Returns hash for the given histogram |name|."""
   return "0x" + hashlib.md5(name).hexdigest()[:16]
@@ -134,8 +164,10 @@ def _GenerateFile(arguments):
       extract_histograms.ExtractHistogramsFromDom(descriptions))
   if had_errors:
     raise Error("Error parsing inputs.")
-  today = datetime.datetime.now().date()
-  expired_histograms_names = _GetExpiredHistograms(histograms, today)
+  with open(arguments.major_branch_date_filepath, "r") as date_file:
+    file_content = date_file.read()
+  base_date = _GetBaseDate(file_content, _DATE_FILE_PATTERN)
+  expired_histograms_names = _GetExpiredHistograms(histograms, base_date)
   expired_histograms_map = _GetHashToNameMap(expired_histograms_names)
   header_file_content = _GenerateHeaderFileContent(
       arguments.header_filename, arguments.namespace,
@@ -165,6 +197,11 @@ def _ParseArguments():
       default="",
       help="Namespace of the generated factory function (code will be in "
       "the global namespace if this is omitted).")
+  arg_parser.add_argument(
+      "--major_branch_date_filepath",
+      "-d",
+      default="",
+      help="A path to the file with the base date.")
   arg_parser.add_argument(
       "inputs",
       nargs="+",

tools/metrics/histograms/generate_expired_histograms_array_unittest.py

@@ -82,13 +82,39 @@ class ExpiredHistogramsTest(unittest.TestCase):
     base_date = datetime.date(2000, 10, 01)
 
     with self.assertRaises(generate_expired_histograms_array.Error) as error:
-        _ = generate_expired_histograms_array._GetExpiredHistograms(histograms,
+        generate_expired_histograms_array._GetExpiredHistograms(histograms,
             base_date)
 
     self.assertEqual(
         "Unable to parse expiry date 2000-10-01 in histogram SecondHistogram.",
         str(error.exception))
 
+  def testGetBaseDate(self):
+    pattern = generate_expired_histograms_array._DATE_FILE_PATTERN
+
+    # Does not match the pattern.
+    content = "MAJOR_BRANCH__FAKE_DATE=2017/09/09"
+    with self.assertRaises(generate_expired_histograms_array.Error):
+        generate_expired_histograms_array._GetBaseDate(content, pattern)
+
+    # Has invalid format.
+    content = "MAJOR_BRANCH_DATE=2010-01-01"
+    with self.assertRaises(generate_expired_histograms_array.Error):
+        generate_expired_histograms_array._GetBaseDate(content, pattern)
+
+    # Has invalid format.
+    content = "MAJOR_BRANCH_DATE=2010/20/02"
+    with self.assertRaises(generate_expired_histograms_array.Error):
+        generate_expired_histograms_array._GetBaseDate(content, pattern)
+
+    # Has invalid date.
+    content = "MAJOR_BRANCH_DATE=2017/02/29"
+    with self.assertRaises(generate_expired_histograms_array.Error):
+        generate_expired_histograms_array._GetBaseDate(content, pattern)
+
+    content = "!!FOO!\nMAJOR_BRANCH_DATE=2010/01/01\n!FOO!!"
+    base_date = generate_expired_histograms_array._GetBaseDate(content, pattern)
+    self.assertEqual(base_date, datetime.date(2010, 01, 01))
 
   def testGenerateHeaderFileContent(self):
     header_filename = "test/test.h"