Add git config block to bundle config (#356)

## Changes
This config block contains commit, branch and remote_url which will be
automatically loaded if specified in the repo, and can also be specified
by the user

## Tests
Unit and black-box tests

bundle/config/bundle.go

@@ -24,4 +24,8 @@ type Bundle struct {
 
 	// Lock configures locking behavior on deployment.
 	Lock Lock `json:"lock" bundle:"readonly"`
+
+	// Contains Git information like current commit, current branch and
+	// origin url. Automatically loaded by reading .git directory if not specified
+	Git Git `json:"git"`
 }

bundle/config/git.go

@@ -0,0 +1,7 @@
+package config
+
+type Git struct {
+	Branch    string `json:"branch,omitempty"`
+	OriginURL string `json:"origin_url,omitempty"`
+	Commit    string `json:"commit,omitempty" bundle:"readonly"`
+}

bundle/config/mutator/load_git_details.go

@@ -0,0 +1,51 @@
+package mutator
+
+import (
+	"context"
+
+	"github.com/databricks/bricks/bundle"
+	"github.com/databricks/bricks/libs/git"
+	"github.com/databricks/bricks/libs/log"
+)
+
+type loadGitDetails struct{}
+
+func LoadGitDetails() *loadGitDetails {
+	return &loadGitDetails{}
+}
+
+func (m *loadGitDetails) Name() string {
+	return "LoadGitDetails"
+}
+
+func (m *loadGitDetails) Apply(ctx context.Context, b *bundle.Bundle) ([]bundle.Mutator, error) {
+	// Load relevant git repository
+	repo, err := git.NewRepository(b.Config.Path)
+	if err != nil {
+		return nil, err
+	}
+	// load branch name if undefined
+	if b.Config.Bundle.Git.Branch == "" {
+		branch, err := repo.CurrentBranch()
+		if err != nil {
+			log.Warnf(ctx, "failed to load current branch: %s", err)
+		} else {
+			b.Config.Bundle.Git.Branch = branch
+		}
+	}
+	// load commit hash if undefined
+	if b.Config.Bundle.Git.Commit == "" {
+		commit, err := repo.LatestCommit()
+		if err != nil {
+			log.Warnf(ctx, "failed to load latest commit: %s", err)
+		} else {
+			b.Config.Bundle.Git.Commit = commit
+		}
+	}
+	// load origin url if undefined
+	if b.Config.Bundle.Git.OriginURL == "" {
+		remoteUrl := repo.OriginUrl()
+		b.Config.Bundle.Git.OriginURL = remoteUrl
+	}
+	return nil, nil
+}

bundle/config/mutator/mutator.go

@@ -9,6 +9,7 @@ func DefaultMutators() []bundle.Mutator {
 		DefineDefaultInclude(),
 		ProcessRootIncludes(),
 		DefineDefaultEnvironment(),
+		LoadGitDetails(),
 	}
 }
 

bundle/tests/autoload_git/bundle.yml

@@ -0,0 +1,4 @@
+bundle:
+  name: autoload git config test
+  git:
+    branch: foo

bundle/tests/autoload_git_test.go

@@ -0,0 +1,15 @@
+package config_tests
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGitConfig(t *testing.T) {
+	b := load(t, "./autoload_git")
+	assert.Equal(t, "foo", b.Config.Bundle.Git.Branch)
+	sshUrl := "git@github.com:databricks/bricks.git"
+	httpsUrl := "https://github.com/databricks/bricks"
+	assert.Contains(t, []string{sshUrl, httpsUrl}, b.Config.Bundle.Git.OriginURL)
+}

libs/git/reference.go

@@ -0,0 +1,96 @@
+package git
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+)
+
+type ReferenceType string
+
+var ErrNotAReferencePointer = fmt.Errorf("HEAD does not point to another reference")
+var ErrNotABranch = fmt.Errorf("HEAD is not a reference to a git branch")
+
+const (
+	// pointer to a secondary reference file path containing sha-1 object ID.
+	// eg: `ref: refs/heads/my-branch-name`
+	ReferenceTypePointer = ReferenceType("pointer")
+	// A hexadecimal encoded SHA1 hash
+	ReferenceTypeSHA1 = ReferenceType("sha-1")
+)
+
+// relevant documentation about git references:
+// https://git-scm.com/book/en/v2/Git-Internals-Git-References
+type Reference struct {
+	Type    ReferenceType
+	Content string
+}
+
+const ReferencePrefix = "ref: "
+const HeadPathPrefix = "refs/heads/"
+
+// asserts if a string is a 40 character hexadecimal encoded string
+func isSHA1(s string) bool {
+	re := regexp.MustCompile("^[0-9a-f]{40}$")
+	return re.MatchString(s)
+}
+
+func LoadReferenceFile(path string) (*Reference, error) {
+	// read referebce file content
+	b, err := os.ReadFile(path)
+	if os.IsNotExist(err) {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, err
+	}
+
+	// trim new line characters
+	content := strings.TrimRight(string(b), "\r\n")
+
+	// determine HEAD type
+	var refType ReferenceType
+	switch {
+	case strings.HasPrefix(content, ReferencePrefix):
+		refType = ReferenceTypePointer
+	case isSHA1(content):
+		refType = ReferenceTypeSHA1
+	default:
+		return nil, fmt.Errorf("unknown format for git HEAD: %s", content)
+	}
+
+	return &Reference{
+		Type:    refType,
+		Content: content,
+	}, nil
+}
+
+// resolves the path to the secondary reference file pointd to. eg: if the file
+// contents are `ref: a/b/c`, then this function returns `a/b/c`
+func (ref *Reference) ResolvePath() (string, error) {
+	if ref.Type != ReferenceTypePointer {
+		return "", ErrNotAReferencePointer
+	}
+	refPath := strings.TrimPrefix(ref.Content, ReferencePrefix)
+	return filepath.FromSlash(refPath), nil
+}
+
+// resolves the name of the current branch from the reference file content. For example
+// `ref: refs/heads/my-branch` returns `my-branch`
+func (ref *Reference) CurrentBranch() (string, error) {
+	branchRefPath, err := ref.ResolvePath()
+	if err == ErrNotAReferencePointer {
+		return "", ErrNotABranch
+	}
+	if err != nil {
+		return "", err
+	}
+	// normalize branch ref path to work accross different operating systems
+	branchRefPath = filepath.ToSlash(branchRefPath)
+	if !strings.HasPrefix(branchRefPath, HeadPathPrefix) {
+		return "", fmt.Errorf("reference path %s does not have expected prefix %s", branchRefPath, HeadPathPrefix)
+	}
+	return strings.TrimPrefix(branchRefPath, HeadPathPrefix), nil
+}

libs/git/reference_test.go

@@ -0,0 +1,100 @@
+package git
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestReferenceReferencePathForObjectID(t *testing.T) {
+	ref := &Reference{
+		Type:    ReferenceTypeSHA1,
+		Content: strings.Repeat("a", 40),
+	}
+	_, err := ref.ResolvePath()
+	assert.ErrorIs(t, err, ErrNotAReferencePointer)
+}
+
+func TestReferenceCurrentBranchForObjectID(t *testing.T) {
+	ref := &Reference{
+		Type:    ReferenceTypeSHA1,
+		Content: strings.Repeat("a", 40),
+	}
+	_, err := ref.CurrentBranch()
+	assert.ErrorIs(t, err, ErrNotABranch)
+}
+
+func TestReferenceCurrentBranchForReference(t *testing.T) {
+	ref := &Reference{
+		Type:    ReferenceTypePointer,
+		Content: `ref: refs/heads/my-branch`,
+	}
+	branch, err := ref.CurrentBranch()
+	assert.NoError(t, err)
+	assert.Equal(t, "my-branch", branch)
+}
+
+func TestReferenceReferencePathForReference(t *testing.T) {
+	ref := &Reference{
+		Type:    ReferenceTypePointer,
+		Content: `ref: refs/heads/my-branch`,
+	}
+	path, err := ref.ResolvePath()
+	assert.NoError(t, err)
+	assert.Equal(t, filepath.FromSlash("refs/heads/my-branch"), path)
+}
+
+func TestReferenceLoadingForObjectID(t *testing.T) {
+	tmp := t.TempDir()
+	f, err := os.Create(filepath.Join(tmp, "HEAD"))
+	require.NoError(t, err)
+	defer f.Close()
+	f.WriteString(strings.Repeat("e", 40) + "\r\n")
+
+	ref, err := LoadReferenceFile(filepath.Join(tmp, "HEAD"))
+	assert.NoError(t, err)
+	assert.Equal(t, ReferenceTypeSHA1, ref.Type)
+	assert.Equal(t, strings.Repeat("e", 40), ref.Content)
+}
+
+func TestReferenceLoadingForReference(t *testing.T) {
+	tmp := t.TempDir()
+	f, err := os.OpenFile(filepath.Join(tmp, "HEAD"), os.O_CREATE|os.O_WRONLY, os.ModePerm)
+	require.NoError(t, err)
+	defer f.Close()
+	f.WriteString("ref: refs/heads/foo\n")
+
+	ref, err := LoadReferenceFile(filepath.Join(tmp, "HEAD"))
+	assert.NoError(t, err)
+	assert.Equal(t, ReferenceTypePointer, ref.Type)
+	assert.Equal(t, "ref: refs/heads/foo", ref.Content)
+}
+
+func TestReferenceLoadingFailsForInvalidContent(t *testing.T) {
+	tmp := t.TempDir()
+	f, err := os.OpenFile(filepath.Join(tmp, "HEAD"), os.O_CREATE|os.O_WRONLY, os.ModePerm)
+	require.NoError(t, err)
+	defer f.Close()
+	f.WriteString("abc")
+
+	_, err = LoadReferenceFile(filepath.Join(tmp, "HEAD"))
+	assert.ErrorContains(t, err, "unknown format for git HEAD")
+}
+
+func TestReferenceIsSha1(t *testing.T) {
+	a := strings.Repeat("0", 40)
+	b := strings.Repeat("f", 40)
+	c := strings.Repeat("0", 39)
+	d := strings.Repeat("F", 40)
+	e := strings.Repeat("0", 41)
+
+	assert.True(t, isSHA1(a))
+	assert.True(t, isSHA1(b))
+	assert.False(t, isSHA1(c))
+	assert.False(t, isSHA1(d))
+	assert.False(t, isSHA1(e))
+}

libs/git/repository.go

@@ -30,6 +30,12 @@ type Repository struct {
 	// Note: prefixes use the forward slash instead of the
 	// OS-specific path separator. This matches Git convention.
 	ignore map[string][]ignoreRules
+
+	// config contains a merged view of the user specific and the repository
+	// specific git configuration loaded from .git/config files.
+	//
+	// Also see: https://git-scm.com/docs/git-config.
+	config *config
 }
 
 // Root returns the repository root.
@@ -37,17 +43,75 @@ func (r *Repository) Root() string {
 	return r.rootPath
 }
 
+func (r *Repository) CurrentBranch() (string, error) {
+	// load .git/HEAD
+	ref, err := LoadReferenceFile(filepath.Join(r.rootPath, ".git", "HEAD"))
+	if err != nil {
+		return "", err
+	}
+	if ref == nil {
+		return "", nil
+	}
+
+	// case: when a git object like commit,tag or remote branch is checked out
+	if ref.Type == ReferenceTypeSHA1 {
+		return "", nil
+	}
+	return ref.CurrentBranch()
+}
+
+func (r *Repository) LatestCommit() (string, error) {
+	// load .git/HEAD
+	ref, err := LoadReferenceFile(filepath.Join(r.rootPath, ".git", "HEAD"))
+	if err != nil {
+		return "", err
+	}
+	if ref == nil {
+		// return empty string when head file does not exist
+		return "", nil
+	}
+
+	// case: when a git object like commit,tag or remote branch is checked out
+	if ref.Type == ReferenceTypeSHA1 {
+		return ref.Content, nil
+	}
+
+	// read reference from .git/HEAD
+	branchHeadPath, err := ref.ResolvePath()
+	if err != nil {
+		return "", err
+	}
+	branchHeadRef, err := LoadReferenceFile(filepath.Join(r.rootPath, ".git", branchHeadPath))
+	if err != nil {
+		return "", err
+	}
+	if branchHeadRef == nil {
+		// return empty string when head file does not exist
+		return "", nil
+	}
+	if branchHeadRef.Type != ReferenceTypeSHA1 {
+		return "", fmt.Errorf("git reference at %s was expected to be a SHA-1 commit id", branchHeadPath)
+	}
+	return branchHeadRef.Content, nil
+}
+
+// return origin url if it's defined, otherwise an empty string
+func (r *Repository) OriginUrl() string {
+	return r.config.variables["remote.origin.url"]
+}
+
 // loadConfig loads and combines user specific and repository specific configuration files.
-func (r *Repository) loadConfig() (*config, error) {
+func (r *Repository) loadConfig() error {
 	config, err := globalGitConfig()
 	if err != nil {
-		return nil, fmt.Errorf("unable to load user specific gitconfig: %w", err)
+		return fmt.Errorf("unable to load user specific gitconfig: %w", err)
 	}
 	err = config.loadFile(filepath.Join(r.rootPath, ".git/config"))
 	if err != nil {
-		return nil, fmt.Errorf("unable to load repository specific gitconfig: %w", err)
+		return fmt.Errorf("unable to load repository specific gitconfig: %w", err)
 	}
-	return config, nil
+	r.config = config
+	return nil
 }
 
 // newIgnoreFile constructs a new [ignoreRules] implementation backed by
@@ -139,13 +203,13 @@ func NewRepository(path string) (*Repository, error) {
 		ignore:   make(map[string][]ignoreRules),
 	}
 
-	config, err := repo.loadConfig()
+	err = repo.loadConfig()
 	if err != nil {
 		// Error doesn't need to be rewrapped.
 		return nil, err
 	}
 
-	coreExcludesPath, err := config.coreExcludesFile()
+	coreExcludesPath, err := repo.config.coreExcludesFile()
 	if err != nil {
 		return nil, fmt.Errorf("unable to access core excludes file: %w", err)
 	}

libs/git/repository_test.go

@@ -1,7 +1,10 @@
 package git
 
 import (
+	"fmt"
+	"os"
 	"path/filepath"
+	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -14,12 +17,115 @@ type testRepository struct {
 	r *Repository
 }
 
-func (r *testRepository) Ignore(relPath string) bool {
-	ign, err := r.r.Ignore(relPath)
-	require.NoError(r.t, err)
+func newTestRepository(t *testing.T) *testRepository {
+	tmp := t.TempDir()
+	err := os.Mkdir(filepath.Join(tmp, ".git"), os.ModePerm)
+	require.NoError(t, err)
+
+	f1, err := os.Create(filepath.Join(tmp, ".git", "config"))
+	require.NoError(t, err)
+	defer f1.Close()
+
+	f1.WriteString(
+		`[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = false
+	logallrefupdates = true
+	ignorecase = true
+	precomposeunicode = true
+`)
+
+	f2, err := os.Create(filepath.Join(tmp, ".git", "HEAD"))
+	require.NoError(t, err)
+	defer f2.Close()
+
+	_, err = f2.WriteString(`ref: refs/heads/main`)
+	require.NoError(t, err)
+
+	repo, err := NewRepository(tmp)
+	require.NoError(t, err)
+
+	return &testRepository{
+		t: t,
+		r: repo,
+	}
+}
+
+func (testRepo *testRepository) checkoutCommit(commitId string) {
+	f, err := os.OpenFile(filepath.Join(testRepo.r.rootPath, ".git", "HEAD"), os.O_WRONLY|os.O_TRUNC, os.ModePerm)
+	require.NoError(testRepo.t, err)
+	defer f.Close()
+
+	_, err = f.WriteString(commitId)
+	require.NoError(testRepo.t, err)
+}
+
+func (testRepo *testRepository) addBranch(name string, latestCommit string) {
+	// create dir for branch head reference
+	branchDir := filepath.Join(testRepo.r.rootPath, ".git", "refs", "heads")
+	err := os.MkdirAll(branchDir, os.ModePerm)
+	require.NoError(testRepo.t, err)
+
+	// create branch head reference file
+	f, err := os.OpenFile(filepath.Join(branchDir, name), os.O_CREATE|os.O_WRONLY, os.ModePerm)
+	require.NoError(testRepo.t, err)
+	defer f.Close()
+
+	// enter the latest commit in the branch reference file
+	_, err = f.WriteString(latestCommit)
+	require.NoError(testRepo.t, err)
+}
+
+func (testRepo *testRepository) checkoutBranch(name string) {
+	f, err := os.OpenFile(filepath.Join(testRepo.r.rootPath, ".git", "HEAD"), os.O_WRONLY|os.O_TRUNC, os.ModePerm)
+	require.NoError(testRepo.t, err)
+	defer f.Close()
+
+	_, err = f.WriteString("ref: refs/heads/" + name)
+	require.NoError(testRepo.t, err)
+}
+
+// add remote origin url to test repo
+func (testRepo *testRepository) addOriginUrl(url string) {
+	// open config in append mode
+	f, err := os.OpenFile(filepath.Join(testRepo.r.rootPath, ".git", "config"), os.O_WRONLY|os.O_APPEND, os.ModePerm)
+	require.NoError(testRepo.t, err)
+	defer f.Close()
+
+	_, err = f.WriteString(
+		fmt.Sprintf(`[remote "origin"]
+	url = %s`, url))
+	require.NoError(testRepo.t, err)
+
+	// reload config to reflect the remote url
+	err = testRepo.r.loadConfig()
+	require.NoError(testRepo.t, err)
+}
+
+func (testRepo *testRepository) Ignore(relPath string) bool {
+	ign, err := testRepo.r.Ignore(relPath)
+	require.NoError(testRepo.t, err)
 	return ign
 }
 
+func (testRepo *testRepository) assertBranch(expected string) {
+	branch, err := testRepo.r.CurrentBranch()
+	assert.NoError(testRepo.t, err)
+	assert.Equal(testRepo.t, expected, branch)
+}
+
+func (testRepo *testRepository) assertCommit(expected string) {
+	commit, err := testRepo.r.LatestCommit()
+	assert.NoError(testRepo.t, err)
+	assert.Equal(testRepo.t, expected, commit)
+}
+
+func (testRepo *testRepository) assertOriginUrl(expected string) {
+	originUrl := testRepo.r.OriginUrl()
+	assert.Equal(testRepo.t, expected, originUrl)
+}
+
 func TestRepository(t *testing.T) {
 	// Load this repository as test.
 	repo, err := NewRepository("../..")
@@ -38,3 +144,65 @@ func TestRepository(t *testing.T) {
 	// Check that ignores under testdata work.
 	assert.True(t, tr.Ignore(filepath.Join("libs", "git", "testdata", "root.ignoreme")))
 }
+
+func TestRepositoryGitConfigForEmptyRepo(t *testing.T) {
+	repo := newTestRepository(t)
+	repo.assertBranch("main")
+	repo.assertCommit("")
+	repo.assertOriginUrl("")
+}
+
+func TestRepositoryGitConfig(t *testing.T) {
+	repo := newTestRepository(t)
+	repo.addBranch("foo", strings.Repeat("1", 40))
+	repo.addBranch("bar", strings.Repeat("2", 40))
+	repo.assertBranch("main")
+	repo.assertCommit("")
+	repo.assertOriginUrl("")
+
+	repo.checkoutBranch("foo")
+	repo.assertBranch("foo")
+	repo.assertCommit(strings.Repeat("1", 40))
+	repo.assertOriginUrl("")
+
+	repo.addOriginUrl("https://www.foo.com/bar")
+	repo.assertBranch("foo")
+	repo.assertCommit(strings.Repeat("1", 40))
+	repo.assertOriginUrl("https://www.foo.com/bar")
+
+	repo.checkoutBranch("bar")
+	repo.assertBranch("bar")
+	repo.assertCommit(strings.Repeat("2", 40))
+	repo.assertOriginUrl("https://www.foo.com/bar")
+
+	repo.checkoutCommit(strings.Repeat("3", 40))
+	repo.assertBranch("")
+	repo.assertCommit(strings.Repeat("3", 40))
+	repo.assertOriginUrl("https://www.foo.com/bar")
+}
+
+func TestRepositoryGitConfigForSshUrl(t *testing.T) {
+	repo := newTestRepository(t)
+	repo.addOriginUrl(`git@foo.com:databricks/bar.git`)
+
+	repo.assertBranch("main")
+	repo.assertCommit("")
+	repo.assertOriginUrl("git@foo.com:databricks/bar.git")
+}
+
+func TestRepositoryGitConfigWhenNotARepo(t *testing.T) {
+	tmp := t.TempDir()
+	repo, err := NewRepository(tmp)
+	require.NoError(t, err)
+
+	branch, err := repo.CurrentBranch()
+	assert.NoError(t, err)
+	assert.Equal(t, "", branch)
+
+	commit, err := repo.LatestCommit()
+	assert.NoError(t, err)
+	assert.Equal(t, "", commit)
+
+	originUrl := repo.OriginUrl()
+	assert.Equal(t, "", originUrl)
+}

libs/git/utils.go

@@ -0,0 +1,28 @@
+package git
+
+import (
+	"strings"
+
+	giturls "github.com/whilp/git-urls"
+)
+
+func ToHttpsUrl(url string) (string, error) {
+	originUrl, err := giturls.Parse(url)
+	if err != nil {
+		return "", err
+	}
+	if originUrl.Scheme == "https" {
+		return originUrl.String(), nil
+	}
+	// if current repo is checked out with a SSH key
+	if originUrl.Scheme != "https" {
+		originUrl.Scheme = "https"
+	}
+	// `git@` is not required for HTTPS
+	if originUrl.User != nil {
+		originUrl.User = nil
+	}
+	// Remove `.git` suffix, if present.
+	originUrl.Path = strings.TrimSuffix(originUrl.Path, ".git")
+	return originUrl.String(), nil
+}

libs/git/utils_test.go

@@ -0,0 +1,13 @@
+package git
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestToHttpsUrlForSsh(t *testing.T) {
+	url, err := ToHttpsUrl("user@foo.com:org/repo-name.git")
+	assert.NoError(t, err)
+	assert.Equal(t, "https://foo.com/org/repo-name", url)
+}