diff --git a/tools/quality_tracking/constants.py b/tools/quality_tracking/constants.py index a38c733733..1ca7ac0dea 100644 --- a/tools/quality_tracking/constants.py +++ b/tools/quality_tracking/constants.py @@ -28,7 +28,7 @@ ACCESS_TOKEN_URL = DASHBOARD_SERVER_HTTP + '/_ah/OAuthGetAccessToken' # The build master URL. BUILD_MASTER_SERVER = 'webrtc-cb-linux-master.cbf.corp.google.com:8010' -BUILD_MASTER_LATEST_BUILD_URL = '/one_box_per_builder' +BUILD_MASTER_TRANSPOSED_GRID_URL = '/tgrid' # The build-bot user which runs build bot jobs. BUILD_BOT_USER = 'phoglund' diff --git a/tools/quality_tracking/dashboard/add_build_status_data.py b/tools/quality_tracking/dashboard/add_build_status_data.py index f5dbb729e2..533b2b27dd 100644 --- a/tools/quality_tracking/dashboard/add_build_status_data.py +++ b/tools/quality_tracking/dashboard/add_build_status_data.py @@ -16,15 +16,44 @@ from google.appengine.ext import db import oauth_post_request_handler +VALID_STATUSES = ['OK', 'failed', 'building'] -SUCCESSFUL_STRING_TO_BOOLEAN = {'successful': True, 'failed': False} + +class OrphanedBuildStatusesExistException(Exception): + pass + + +class BuildStatusRoot(db.Model): + """Exists solely to be the root parent for all build status data. + + Since all build status data will refer to this as their parent, + we can run transactions on the build status data as a whole. + """ + pass class BuildStatusData(db.Model): """This represents one build status report from the build bot.""" bot_name = db.StringProperty(required=True) + revision = db.IntegerProperty(required=True) build_number = db.IntegerProperty(required=True) - successful = db.BooleanProperty(required=True) + status = db.StringProperty(required=True) + + +def _ensure_build_status_root_exists(): + root = db.GqlQuery('SELECT * FROM BuildStatusRoot').get() + if not root: + # Create a new root, but ensure we don't have any orphaned build statuses + # (in that case, we would not have a single entity group as we desire). + orphans = db.GqlQuery('SELECT * FROM BuildStatusData').get() + if orphans: + raise OrphanedBuildStatusesExistException('Parent is gone and there are ' + 'orphaned build statuses in ' + 'the database!') + root = BuildStatusRoot() + root.put() + + return root def _filter_oauth_parameters(post_keys): @@ -32,26 +61,96 @@ def _filter_oauth_parameters(post_keys): post_keys) +def _parse_status(build_number_and_status): + parsed_status = build_number_and_status.split('--') + if len(parsed_status) != 2: + raise ValueError('Malformed status string %s.' % build_number_and_status) + + parsed_build_number = int(parsed_status[0]) + status = parsed_status[1] + + if status not in VALID_STATUSES: + raise ValueError('Invalid status in %s.' % build_number_and_status) + + return (parsed_build_number, status) + + +def _parse_name(revision_and_bot_name): + parsed_name = revision_and_bot_name.split('--') + if len(parsed_name) != 2: + raise ValueError('Malformed name string %s.' % revision_and_bot_name) + + revision = parsed_name[0] + bot_name = parsed_name[1] + + return (int(revision), bot_name) + + +def _delete_all_with_revision(revision, build_status_root): + query_result = db.GqlQuery('SELECT * FROM BuildStatusData ' + 'WHERE revision = :1 AND ANCESTOR IS :2', + revision, build_status_root) + for entry in query_result: + entry.delete() + + class AddBuildStatusData(oauth_post_request_handler.OAuthPostRequestHandler): - """Used to report build status data.""" + """Used to report build status data. + + Build status data is reported as a POST request. The POST request, aside + from the required oauth_* parameters should contain name-value entries that + abide by the following rules: + + 1) The name should be on the form --, for instance + 1568--Win32Release. + 2) The value should be on the form --, for instance + 553--OK, 554--building. The status is permitted to be failed, OK or + building. + + Data is keyed by revision. This handler will delete all data from a revision + if data with that revision is present in the current update, since we + assume that more recent data is always better data. We also assume that + an update always has complete information on a revision (e.g. the status + for all the bots are reported in each update). + + In particular the revision arrangement solves the problem when the latest + revision reports 'building' for a bot. Had we not deleted the old revision + we would first store a 'building' status for that bot and revision, and + later store a 'OK' or 'failed' status for that bot and revision. This is + undesirable since we don't want multiple statuses for one bot-revision + combination. Now we will effectively update the bot's status instead. + """ def post(self): - for bot_name in _filter_oauth_parameters(self.request.arguments()): - status = self.request.get(bot_name) - parsed_status = status.split('-') - if len(parsed_status) != 2: - raise ValueError('Malformed status string %s for bot %s.' % - (status, bot_name)) + build_status_root = _ensure_build_status_root_exists() + build_status_data = _filter_oauth_parameters(self.request.arguments()) - parsed_build_number = int(parsed_status[0]) - successful = parsed_status[1] + db.run_in_transaction(self._parse_and_store_data_in_transaction, + build_status_root, build_status_data) - if successful not in SUCCESSFUL_STRING_TO_BOOLEAN: - raise ValueError('Malformed status string %s for bot %s.' % (status, - bot_name)) - parsed_successful = SUCCESSFUL_STRING_TO_BOOLEAN[successful] + def _parse_and_store_data_in_transaction(self, build_status_root, + build_status_data): + encountered_revisions = set() + for revision_and_bot_name in build_status_data: + build_number_and_status = self.request.get(revision_and_bot_name) - item = BuildStatusData(bot_name=bot_name, - build_number=parsed_build_number, - successful=parsed_successful) + try: + (build_number, status) = _parse_status(build_number_and_status) + (revision, bot_name) = _parse_name(revision_and_bot_name) + except ValueError as error: + self._show_error_page('Invalid parameter in request: %s.' % error) + + if revision not in encountered_revisions: + # There's new data on this revision in this update, so clear all status + # entries with that revision. Only do this once when we first encounter + # the revision. + _delete_all_with_revision(revision, build_status_root) + encountered_revisions.add(revision) + + # Finally, write the item. + item = BuildStatusData(parent=build_status_root, + bot_name=bot_name, + revision=revision, + build_number=build_number, + status=status) item.put() diff --git a/tools/quality_tracking/tgrid_parser.py b/tools/quality_tracking/tgrid_parser.py new file mode 100644 index 0000000000..35468f4027 --- /dev/null +++ b/tools/quality_tracking/tgrid_parser.py @@ -0,0 +1,72 @@ +#!/usr/bin/env python +#-*- coding: utf-8 -*- +# Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. +# +# Use of this source code is governed by a BSD-style license +# that can be found in the LICENSE file in the root of the source +# tree. An additional intellectual property rights grant can be found +# in the file PATENTS. All contributing project authors may +# be found in the AUTHORS file in the root of the source tree. + +"""Contains functions for parsing the build master's transposed grid page.""" + +__author__ = 'phoglund@webrtc.org (Patrik Höglund)' + +import re + + +class FailedToParseBuildStatus(Exception): + pass + + +def _parse_builds(revision, html): + """Parses the bot list, which is a sequence of lines. + + Example input: + OK + The first regular expression group captures Android, second 119, third OK. + """ + result = {} + + for match in re.finditer('.*?' + '(OK|failed|building).*?', html, re.DOTALL): + revision_and_bot_name = revision + "--" + match.group(1) + build_number_and_status = match.group(2) + "--" + match.group(3) + + result[revision_and_bot_name] = build_number_and_status + + return result + + +def parse_tgrid_page(html): + """Parses the build master's tgrid page. + + Example input: + + 1568 + LIST OF BOTS + + The first regular expression group captures 1568, second group captures + everything in LIST OF BOTS. The list of bots is then passed into a + separate function for parsing. + + Args: + html: The raw HTML from the tgrid page. + + Returns: A dictionary with -- mapped to + --, where status is either OK, failed or + building. + """ + result = {} + + for match in re.finditer('(\d+)(.*?)', + html, re.DOTALL): + revision = match.group(1) + builds_for_revision_html = match.group(2) + result.update(_parse_builds(revision, builds_for_revision_html)) + + if not result: + raise FailedToParseBuildStatus('Could not find any build statuses in %s.' % + html) + + return result diff --git a/tools/quality_tracking/tgrid_parser_test.py b/tools/quality_tracking/tgrid_parser_test.py new file mode 100755 index 0000000000..4fc7c910f8 --- /dev/null +++ b/tools/quality_tracking/tgrid_parser_test.py @@ -0,0 +1,180 @@ +#!/usr/bin/env python +#-*- coding: utf-8 -*- +# Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. +# +# Use of this source code is governed by a BSD-style license +# that can be found in the LICENSE file in the root of the source +# tree. An additional intellectual property rights grant can be found +# in the file PATENTS. All contributing project authors may +# be found in the AUTHORS file in the root of the source tree. + +"""Contains functions for parsing the build master's transposed grid page.""" + +__author__ = 'phoglund@webrtc.org (Patrik Höglund)' + +import unittest + +import tgrid_parser + + +SAMPLE_FILE = """ + + + + + Buildbot + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
1570 + OK + OK + OK + OK + OK + OK + OK + OK + OK + OK + OK + OK
1571 + OK + OK + OK + OK + OK + OK + OK + OK + failed
+voe_auto_test		 + OK + building + OK
+ + +""" + +MINIMAL_OK = """ + +1570 + +OK + +""" + +MINIMAL_FAIL = """ + +1573 + +failed
+voe_auto_test + +""" + +MINIMAL_BUILDING = """ + +1576 + +building +voe_auto_test + +""" + +class TGridParserTest(unittest.TestCase): + def test_parser_throws_exception_on_empty_html(self): + self.assertRaises(tgrid_parser.FailedToParseBuildStatus, + tgrid_parser.parse_tgrid_page, ''); + + def test_parser_finds_successful_bot(self): + result = tgrid_parser.parse_tgrid_page(MINIMAL_OK) + + self.assertEqual(1, len(result), 'There is only one bot in the sample.') + first_mapping = result.items()[0] + + self.assertEqual('1570--Android', first_mapping[0]) + self.assertEqual('121--OK', first_mapping[1]) + + def test_parser_finds_failed_bot(self): + result = tgrid_parser.parse_tgrid_page(MINIMAL_FAIL) + + self.assertEqual(1, len(result), 'There is only one bot in the sample.') + first_mapping = result.items()[0] + + self.assertEqual('1573--LinuxVideoTest', first_mapping[0]) + self.assertEqual('347--failed', first_mapping[1]) + + def test_parser_finds_building_bot(self): + result = tgrid_parser.parse_tgrid_page(MINIMAL_BUILDING) + + self.assertEqual(1, len(result), 'There is only one bot in the sample.') + first_mapping = result.items()[0] + + self.assertEqual('1576--Win32Debug', first_mapping[0]) + self.assertEqual('434--building', first_mapping[1]) + + def test_parser_finds_all_bots_and_revisions(self): + result = tgrid_parser.parse_tgrid_page(SAMPLE_FILE) + + # 2 * 12 = 24 bots in sample + self.assertEqual(24, len(result)) + + # Make some samples + self.assertTrue(result.has_key('1570--ChromeOS')) + self.assertEquals('578--OK', result['1570--ChromeOS']) + + self.assertTrue(result.has_key('1570--LinuxCLANG')) + self.assertEquals('259--OK', result['1570--LinuxCLANG']) + + self.assertTrue(result.has_key('1570--Win32Release')) + self.assertEquals('440--OK', result['1570--Win32Release']) + + self.assertTrue(result.has_key('1571--ChromeOS')) + self.assertEquals('579--OK', result['1571--ChromeOS']) + + self.assertTrue(result.has_key('1571--LinuxVideoTest')) + self.assertEquals('346--failed', result['1571--LinuxVideoTest']) + + self.assertTrue(result.has_key('1571--Win32Debug')) + self.assertEquals('441--building', result['1571--Win32Debug']) + + +if __name__ == '__main__': + unittest.main() diff --git a/tools/quality_tracking/track_build_status.py b/tools/quality_tracking/track_build_status.py index 53f5301418..068691f841 100755 --- a/tools/quality_tracking/track_build_status.py +++ b/tools/quality_tracking/track_build_status.py @@ -20,47 +20,16 @@ import re import constants import dashboard_connection +import tgrid_parser class FailedToGetStatusFromMaster(Exception): pass -class FailedToParseBuildStatus(Exception): - pass - - -def _parse_status_page(html): - """Parses the build master's one_box_per_builder page. - - Args: - html: The raw HTML from the one_box_per_builder page. - - Returns: the bot name mapped to a string with the build number and the - build status separated by a dash (e.g. 456-successful, 114-failed). - """ - result = {} - - # Example target string: #430 - #
build
successful - # Group 1 captures 'Win32Debug', Group 2 captures '430', group 3 'successful'. - # Implementation note: We match non-greedily (.*?) between the link and - # successful / failed, otherwise we would only find the first status. - for match in re.finditer('' - '.*?(successful|failed)', - html, re.DOTALL): - result[match.group(1)] = match.group(2) + '-' + match.group(3) - - if not result: - raise FailedToParseBuildStatus('Could not find any build statuses in %s.' % - html) - - return result - - def _download_and_parse_build_status(): connection = httplib.HTTPConnection(constants.BUILD_MASTER_SERVER) - connection.request('GET', constants.BUILD_MASTER_LATEST_BUILD_URL) + connection.request('GET', constants.BUILD_MASTER_TRANSPOSED_GRID_URL) response = connection.getresponse() if response.status != 200: @@ -71,7 +40,7 @@ def _download_and_parse_build_status(): full_response = response.read() connection.close() - return _parse_status_page(full_response) + return tgrid_parser.parse_tgrid_page(full_response) def _main():