cmd/autogardener/suspect.go

// Copyright 2022 The Fuchsia Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

package main

import (
	"fmt"
	"path"
	"strings"

	buildbucketpb "go.chromium.org/luci/buildbucket/proto"
	gerritpb "go.chromium.org/luci/common/proto/gerrit"
	"go.chromium.org/luci/common/proto/git"
	"golang.org/x/exp/maps"
)

// suspectCommit represents a commit that is being considered as a potential
// culprit.
type suspectCommit struct {
	// Signature of the failure mode for which this commit is a suspect. This is
	// only attached to each suspectCommit so that `features()` can access it
	// features that depend on the details of the failure mode.
	signature failureSignature `json:"-"`

	CommitInfo   *git.Commit                 `json:"-"`
	GerritChange *buildbucketpb.GerritChange `json:"-"`
	// ChangeInfo is only set after filtering down the list of commits to
	// high-probability suspects, so it cannot be assumed to be set.
	ChangeInfo     *gerritpb.ChangeInfo `json:"-"`
	CommitPosition int                  `json:"commit_position"`
	AffectedTest   bool                 `json:"affected_test"`
	// TODO(olivernewman): Only consider tags that appear in a small subset of
	// test names. Tags like "test" or "fuchsia" should not be considered.
	TagMatchesTest bool `json:"tag_matches_test"`
	// For each CI builder where the test is failing, the number of builds
	// containing this commit that passed before the first failed build.
	// If the first failure was *before* the commit landed, the value will be
	// negative.
	BlamelistDistances map[string]int `json:"blamelist_distances"`
}

func (c *suspectCommit) changedFiles() []string {
	if c.ChangeInfo == nil {
		return nil
	}

	// TODO(olivernewman): Get the path to each project in the checkout using a
	// roller commit message footer. We shouldn't be hardcoding this mapping
	// here.
	projectMap := map[string]string{
		"fuchsia":     "",
		"experiences": "src/experiences",
	}
	projectDir, ok := projectMap[c.ChangeInfo.Project]
	if !ok {
		projectDir = c.ChangeInfo.Project
	}

	var files []string
	revision := c.ChangeInfo.Revisions[c.ChangeInfo.CurrentRevision]
	for file := range revision.Files {
		if projectDir != "" {
			file = path.Join(projectDir, file)
		}
		files = append(files, file)
	}
	return files
}

func (c *suspectCommit) gerritURL() string {
	return fmt.Sprintf(
		"https://%s/c/%s/+/%d",
		c.GerritChange.Host,
		c.GerritChange.Project,
		c.GerritChange.Change,
	)
}

func (c *suspectCommit) commitSummary() string {
	return strings.Split(c.CommitInfo.Message, "\n")[0]
}

// score computes the 0-100 culprit score of the change, where a higher number
// means the commit is more likely to be the cause of the failure mode in
// question.
func (c *suspectCommit) score() int {
	var total, possible int
	for _, feature := range c.features() {
		if feature.Score < 0 || feature.Score > 100 {
			fmt.Printf("warning: score is out of range for change %s: %+v\n", c.gerritURL(), feature)
			if feature.Score < 0 {
				feature.Score = 0
			} else if feature.Score > 100 {
				feature.Score = 100
			}
		}
		// TODO: also take into account how rare it is for a suspect commit to
		// have a high score for this feature. We might want to lower the
		// weighting for features where many suspects are determined to have the
		// same score.
		total += feature.Score * feature.Weight
		possible += 100 * feature.Weight
	}
	return (100 * total) / possible
}

// features returns all the individually computed feature scores from the
// various data sources used to assess the suspect commit.
func (c *suspectCommit) features() []culpritFeature {
	var affectedScore int
	if c.AffectedTest {
		affectedScore = 100
	}
	var tagMatchesScore int
	if c.TagMatchesTest {
		tagMatchesScore = 100
	}
	res := []culpritFeature{
		{
			Name:   "blamelist distances",
			Score:  scoreBlamelistDistances(maps.Values(c.BlamelistDistances)),
			Weight: 4,
		},
		{
			Name:   "test affected by change",
			Score:  affectedScore,
			Weight: 3,
		},
		{
			Name:   "commit tag matches test name",
			Score:  tagMatchesScore,
			Weight: 1,
		},
	}
	if c.ChangeInfo != nil {
		res = append(res, culpritFeature{
			Name:   "changed file proximity",
			Score:  scoreChangedFileProximity(c.changedFiles(), c.signature.TestGNLabel),
			Weight: 4,
		})
	}
	return res
}