content/test/gpu/unexpected_passes/gpu_queries.py

# Copyright 2021 The Chromium Authors
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.
"""GPU-specific implementation of the unexpected passes' queries module."""

from typing import Iterable, Optional

from unexpected_passes_common import queries as queries_module

from gpu_tests import gpu_integration_test

# This query gets us the most recent |num_builds| CI builds from the past month
# for each builder.
CI_BUILDS_SUBQUERY = """\
  builds AS (
    WITH
      all_builds AS (
        SELECT
          DISTINCT exported.id AS build_inv_id,
          variant.*,
          partition_time
        FROM
          `chrome-luci-data.{project}.gpu_ci_test_results` AS tr,
          UNNEST(variant) AS variant
        WHERE
          DATE(partition_time) > DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)
          AND exported.realm = "{project}:ci"
          AND key = "builder"
      ),
      grouped_builds AS (
        SELECT
          build_inv_id,
          value AS builder,
          partition_time,
          RANK() OVER (
            PARTITION BY value ORDER BY partition_time DESC
          ) AS rank_idx,
        FROM all_builds
      )
    SELECT
      build_inv_id,
      builder,
      partition_time
    FROM grouped_builds
    WHERE rank_idx <= {num_builds}
  )"""

# The same as CI_BUILDS_SUBQUERY, but takes into account submitted builds for
# tryjobs.
TRY_BUILDS_SUBQUERY = """\
  builds AS (
    WITH
      all_builds AS (
        SELECT
          DISTINCT exported.id AS build_inv_id,
          variant.*,
          partition_time
        FROM
          `chrome-luci-data.{project}.gpu_try_test_results` AS tr,
          UNNEST(variant) AS variant,
          submitted_builds sb
        WHERE
          DATE(partition_time) > DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)
          AND exported.realm = "{project}:try"
          AND key = "builder"
          AND exported.id = sb.id
      ),
      grouped_builds AS (
        SELECT
          build_inv_id,
          value AS builder,
          partition_time,
          RANK() OVER (
            PARTITION BY value ORDER BY partition_time DESC
          ) AS rank_idx,
        FROM all_builds
      )
    SELECT
      build_inv_id,
      builder,
      partition_time
    FROM grouped_builds
    WHERE rank_idx <= {num_builds}
  )"""

# step_name can be either the step_name tag or test_suite variant because of the
# way Skylab builders work. In normal Chromium test tasks, the step name is
# reported to the task-level RDB invocation, which then gets applied to every
# result that task reports. Skylab test steps are actually separate
# Buildbucket builds that report results a bit differently. They do not report
# a step_name tag, but do put the same information in for the test_suite
# variant. So, we will look for step_name first to cover most builders and fall
# back to test_suite for Skylab builders.
RESULTS_SUBQUERY = """\
  results AS (
    SELECT
      exported.id,
      test_id,
      test_metadata.name as test_name,
      status,
      (
        SELECT value
        FROM tr.variant
        WHERE key = "builder"
      ) as builder_name,
      IFNULL(
        (
          SELECT value
          FROM tr.tags
          WHERE key = "step_name"),
        (
          SELECT value
          FROM tr.variant
          WHERE key = "test_suite")) as step_name,
      ARRAY(
        SELECT value
        FROM tr.tags
        WHERE key = "typ_tag") as typ_tags,
      ARRAY(
        SELECT value
        FROM tr.tags
        WHERE key = "raw_typ_expectation") as typ_expectations
    FROM
      `chrome-luci-data.{project}.gpu_{ci_or_try}_test_results` tr,
      builds b
    WHERE
      DATE(tr.partition_time) > DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)
      AND exported.id = build_inv_id
      AND status != "SKIP"
      AND STRUCT("gpu_test_class", "{suite_class}") IN UNNEST(tags)
  )"""

# Selects the relevant columns from results that had either a Failure or a
# RetryOnFailure expectation when they were run, ordered by builder name.
FINAL_SELECTOR_QUERY = """\
SELECT id, test_id, test_name, builder_name, status, step_name, typ_tags
FROM results
WHERE
  "Failure" IN UNNEST(typ_expectations)
  OR "RetryOnFailure" IN UNNEST(typ_expectations)
ORDER BY builder_name DESC"""


def _PartitionedSubmittedBuildsFor(project_view: str) -> str:
  return queries_module.PARTITIONED_SUBMITTED_BUILDS_TEMPLATE.format(
    project_view=project_view
  )


# Gets the Buildbucket IDs for all the public trybots that:
#   1. Run GPU tests
#   2. Were used for CL submission (i.e. weren't for intermediate patchsets)
PUBLIC_TRY_SUBMITTED_BUILDS_SUBQUERY = f"""\
  submitted_builds AS (
{_PartitionedSubmittedBuildsFor('chromium')}
    UNION ALL
{_PartitionedSubmittedBuildsFor('angle')}
  )"""

# The same as PUBLIC_TRY_SUBMITTED_BUILDS_SUBQUERY, but for internal trybots.
# There are no internal ANGLE tryjobs, so no need to look for attempts there.
INTERNAL_TRY_SUBMITTED_BUILDS_SUBQUERY = f"""\
  submitted_builds AS (
{_PartitionedSubmittedBuildsFor('chrome')}
  )"""


class GpuBigQueryQuerier(queries_module.BigQueryQuerier):
  def __init__(self, *args, **kwargs):
    super().__init__(*args, **kwargs)

    name_mapping = gpu_integration_test.GenerateTestNameMapping()
    _suite_class = name_mapping[self._suite]
    # __qualname__ returns the same value as __name__, so we need to manually
    # construct the qualified name.
    self._qualified_class_name = (
      f'{_suite_class.__module__}.{_suite_class.__name__}'
    )

  def _CiBuildsFor(self, project: str) -> str:
    """Helper function to generate a CI builds subquery."""
    return CI_BUILDS_SUBQUERY.format(
      project=project, num_builds=self._num_samples
    )

  def _TryBuildsFor(self, project: str) -> str:
    """Helper function to generate a try builds subquery."""
    return TRY_BUILDS_SUBQUERY.format(
      project=project, num_builds=self._num_samples
    )

  def _ResultsFor(self, project: str, ci_or_try: str) -> str:
    """Helper function to generate a results subquery."""
    return RESULTS_SUBQUERY.format(
      project=project,
      ci_or_try=ci_or_try,
      suite_class=self._qualified_class_name,
    )

  def _GetPublicCiQuery(self) -> str:
    return f"""\
WITH
{self._CiBuildsFor('chromium')},
{self._ResultsFor('chromium', 'ci')}
{FINAL_SELECTOR_QUERY}
"""

  def _GetInternalCiQuery(self) -> str:
    return f"""\
WITH
{self._CiBuildsFor('chrome')},
{self._ResultsFor('chrome', 'ci')}
{FINAL_SELECTOR_QUERY}
"""

  def _GetPublicTryQuery(self) -> str:
    return f"""\
WITH
{PUBLIC_TRY_SUBMITTED_BUILDS_SUBQUERY},
{self._TryBuildsFor('chromium')},
{self._ResultsFor('chromium', 'try')}
{FINAL_SELECTOR_QUERY}
"""

  def _GetInternalTryQuery(self) -> str:
    return f"""\
WITH
{INTERNAL_TRY_SUBMITTED_BUILDS_SUBQUERY},
{self._TryBuildsFor('chrome')},
{self._ResultsFor('chrome', 'try')}
{FINAL_SELECTOR_QUERY}
"""

  def _GetRelevantExpectationFilesForQueryResult(
    self, _: queries_module.QueryResult
  ) -> Optional[Iterable[str]]:
    # Only one expectation file is ever used for the GPU tests, so just use
    # whichever one we've read in.
    return None