EARTHLY_GIT_AUTHOR and CO_AUTHOR args (#2260)

Adds `EARTHLY_GIT_AUTHOR` and `EARTHLY_GIT_CO_AUTHORS` args, when the
`--earthly-git-author-args` feature flag is set.

Signed-off-by: Alex Couture-Beil <alex@earthly.dev>

buildcontext/git.go

@@ -50,7 +50,9 @@ type resolvedGitProject struct {
 	// tags is the git tags.
 	tags []string
 	// ts is the git commit timestamp.
-	ts string
+	ts        string
+	author    string
+	coAuthors []string
 	// state is the state holding the git files.
 	state pllb.State
 }
@@ -149,6 +151,8 @@ func (gr *gitResolver) resolveEarthProject(ctx context.Context, gwClient gwclien
 			Branch:    rgp.branches,
 			Tags:      rgp.tags,
 			Timestamp: rgp.ts,
+			Author:    rgp.author,
+			CoAuthors: rgp.coAuthors,
 		},
 		Features: localBuildFile.ftrs,
 	}, nil
@@ -194,7 +198,10 @@ func (gr *gitResolver) resolveGitProject(ctx context.Context, gwClient gwclient.
 					"git rev-parse --short=8 HEAD >/dest/git-short-hash ; " +
 					"git rev-parse --abbrev-ref HEAD >/dest/git-branch  || touch /dest/git-branch ; " +
 					"git describe --exact-match --tags >/dest/git-tags || touch /dest/git-tags ; " +
-					"git log -1 --format=%ct >/dest/git-ts || touch /dest/git-ts",
+					"git log -1 --format=%ct >/dest/git-ts || touch /dest/git-ts ; " +
+					"git log -1 --format=%ae >/dest/git-author || touch /dest/git-author ; " +
+					"git log -1 --format=%b >/dest/git-body || touch /dest/git-body ; " +
+					"",
 			}),
 			llb.Dir("/git-src"),
 			llb.ReadonlyRootFS(),
@@ -241,10 +248,24 @@ func (gr *gitResolver) resolveGitProject(ctx context.Context, gwClient gwclient.
 		if err != nil {
 			return nil, errors.Wrap(err, "read git-ts")
 		}
+		gitAuthorBytes, err := gitMetaRef.ReadFile(ctx, gwclient.ReadRequest{
+			Filename: "git-author",
+		})
+		if err != nil {
+			return nil, errors.Wrap(err, "read git-author")
+		}
+		gitBodyBytes, err := gitMetaRef.ReadFile(ctx, gwclient.ReadRequest{
+			Filename: "git-body",
+		})
+		if err != nil {
+			return nil, errors.Wrap(err, "read git-body")
+		}
 
 		gitHash := strings.SplitN(string(gitHashBytes), "\n", 2)[0]
 		gitShortHash := strings.SplitN(string(gitShortHashBytes), "\n", 2)[0]
 		gitBranches := strings.SplitN(string(gitBranchBytes), "\n", 2)
+		gitAuthor := strings.SplitN(string(gitAuthorBytes), "\n", 2)[0]
+		gitCoAuthors := gitutil.ParseCoAuthorsFromBody(string(gitBodyBytes))
 		var gitBranches2 []string
 		for _, gitBranch := range gitBranches {
 			if gitBranch != "" {
@@ -274,6 +295,8 @@ func (gr *gitResolver) resolveGitProject(ctx context.Context, gwClient gwclient.
 			branches:  gitBranches2,
 			tags:      gitTags2,
 			ts:        gitTs,
+			author:    gitAuthor,
+			coAuthors: gitCoAuthors,
 			state: pllb.Git(
 				gitURL,
 				gitHash,

features/features.go

@@ -27,6 +27,7 @@ type Features struct {
 	CheckDuplicateImages       bool `long:"check-duplicate-images" description:"check for duplicate images during output"`
 	EarthlyVersionArg          bool `long:"earthly-version-arg" description:"includes EARTHLY_VERSION and EARTHLY_BUILD_SHA ARGs"`
 	EarthlyLocallyArg          bool `long:"earthly-locally-arg" description:"includes EARTHLY_LOCALLY ARG"`
+	EarthlyGitAuthorArgs       bool `long:"earthly-git-author-args" description:"includes EARTHLY_GIT_AUTHOR and EARTHLY_GIT_CO_AUTHORS ARGs"`
 	ExplicitGlobal             bool `long:"explicit-global" description:"require base target args to have explicit settings to be considered global args"`
 	UseCacheCommand            bool `long:"use-cache-command" description:"allow use of CACHE command in Earthfiles"`
 	UseHostCommand             bool `long:"use-host-command" description:"allow use of HOST command in Earthfiles"`

tests/git-metadata/Earthfile

@@ -41,7 +41,10 @@ git config --global user.name \"test name\"
 git clone git@git.example.com:testuser/repo.git
 mv Earthfile repo/
 git -C repo add Earthfile
-git -C repo commit -m test
+git -C repo commit -m \"test
+
+Co-authored-by: Testy McTest <testy@earthly.dev>
+Co-authored-by: Coverage McCover <cover@earthly.dev>\"
 git -C repo push
 export expectedsha=\$(git -C repo rev-parse HEAD)
 

tests/git-metadata/test.earth

@@ -1,10 +1,19 @@
-VERSION 0.6
+VERSION --earthly-git-author-args 0.6
+
+# Note: this is tests both locally and remotely
+# the population of git metadata occurs in two **different** functions respectively:
+#   - gitutil/detectgit.go: Metadata            (locally-reference)
+#   - buildcontext/git.go: resolveGitProject    (remotely-referenced)
 
 test-git-metadata:
     FROM alpine
     ARG --required expectedsha
     ARG EARTHLY_GIT_SHORT_HASH
     ARG EARTHLY_GIT_HASH
+    ARG EARTHLY_GIT_AUTHOR
+    ARG EARTHLY_GIT_CO_AUTHORS
     RUN test "$EARTHLY_GIT_HASH" = "$expectedsha"
     RUN test -n "$EARTHLY_GIT_SHORT_HASH"
     RUN echo "$EARTHLY_GIT_HASH" | grep "$EARTHLY_GIT_SHORT_HASH"
+    RUN test "$EARTHLY_GIT_AUTHOR" = "onlyspammersemailthis@earthly.dev"
+    RUN test "$EARTHLY_GIT_CO_AUTHORS" = "testy@earthly.dev cover@earthly.dev"

util/gitutil/detectgit.go

@@ -38,6 +38,8 @@ type GitMetadata struct {
 	Branch    []string
 	Tags      []string
 	Timestamp string
+	Author    string
+	CoAuthors []string
 }
 
 // Metadata performs git metadata detection on the provided directory.
@@ -92,6 +94,16 @@ func Metadata(ctx context.Context, dir string) (*GitMetadata, error) {
 		retErr = err
 		// Keep going.
 	}
+	author, err := detectGitAuthor(ctx, dir)
+	if err != nil {
+		retErr = err
+		// Keep going.
+	}
+	coAuthors, err := detectGitCoAuthors(ctx, dir)
+	if err != nil {
+		retErr = err
+		// Keep going.
+	}
 
 	relDir, isRel, err := gitRelDir(baseDir, dir)
 	if err != nil {
@@ -111,6 +123,8 @@ func Metadata(ctx context.Context, dir string) (*GitMetadata, error) {
 		Branch:    branch,
 		Tags:      tags,
 		Timestamp: timestamp,
+		Author:    author,
+		CoAuthors: coAuthors,
 	}, retErr
 }
 
@@ -126,6 +140,8 @@ func (gm *GitMetadata) Clone() *GitMetadata {
 		Branch:    gm.Branch,
 		Tags:      gm.Tags,
 		Timestamp: gm.Timestamp,
+		Author:    gm.Author,
+		CoAuthors: gm.CoAuthors,
 	}
 }
 
@@ -251,17 +267,87 @@ func detectGitTags(ctx context.Context, dir string) ([]string, error) {
 func detectGitTimestamp(ctx context.Context, dir string) (string, error) {
 	cmd := exec.CommandContext(ctx, "git", "log", "-1", "--format=%ct")
 	cmd.Dir = dir
+	cmd.Stderr = nil // force capture of stderr on errors
 	out, err := cmd.Output()
 	if err != nil {
+		exitError, ok := err.(*exec.ExitError)
+		if ok && strings.Contains(string(exitError.Stderr), "does not have any commits yet") {
+			return "", nil
+		}
+		return "", errors.Wrap(err, "detect git timestamp")
+	}
+	outStr := string(out)
+	if outStr == "" {
 		return "", nil
 	}
+	return strings.SplitN(outStr, "\n", 2)[0], nil
+}
+
+func detectGitAuthor(ctx context.Context, dir string) (string, error) {
+	cmd := exec.CommandContext(ctx, "git", "log", "-1", "--format=%ae")
+	cmd.Dir = dir
+	cmd.Stderr = nil // force capture of stderr on errors
+	out, err := cmd.Output()
+	if err != nil {
+		exitError, ok := err.(*exec.ExitError)
+		if ok && strings.Contains(string(exitError.Stderr), "does not have any commits yet") {
+			return "", nil
+		}
+		return "", errors.Wrap(err, "detect git author")
+	}
 	outStr := string(out)
 	if outStr == "" {
 		return "", nil
 	}
 	return strings.SplitN(outStr, "\n", 2)[0], nil
 }
 
+func detectGitCoAuthors(ctx context.Context, dir string) ([]string, error) {
+	cmd := exec.CommandContext(ctx, "git", "log", "-1", "--format=%b")
+	cmd.Dir = dir
+	cmd.Stderr = nil // force capture of stderr on errors
+	out, err := cmd.Output()
+	if err != nil {
+		exitError, ok := err.(*exec.ExitError)
+		if ok && strings.Contains(string(exitError.Stderr), "does not have any commits yet") {
+			return nil, nil
+		}
+		if out != nil && strings.Contains(string(out), "does not have any commits yet") {
+			return nil, nil
+		}
+		return nil, errors.Wrap(err, "detect git co-authors")
+	}
+	return ParseCoAuthorsFromBody(string(out)), nil
+}
+
+// ParseCoAuthorsFromBody returns a list of coauthor emails from a git body
+func ParseCoAuthorsFromBody(body string) []string {
+	coAuthors := []string{}
+	coAuthorsSeen := map[string]struct{}{}
+	for _, s := range strings.Split(body, "\n") {
+		s = strings.TrimSpace(s)
+		splits := strings.Split(s, " ")
+		n := len(splits)
+		if n > 2 {
+			if splits[0] == "Co-authored-by:" {
+				email := splits[n-1]
+				n = len(email)
+				if n > 2 {
+					if email[0] == '<' && email[n-1] == '>' {
+						email = email[1:(n - 1)]
+						_, seen := coAuthorsSeen[email]
+						if !seen {
+							coAuthors = append(coAuthors, email)
+							coAuthorsSeen[email] = struct{}{}
+						}
+					}
+				}
+			}
+		}
+	}
+	return coAuthors
+}
+
 // gitRelDir returns the relative path from git root (where .git directory locates in the project)
 // This function validates the input data (basePath, path) as well.
 func gitRelDir(basePath string, path string) (string, bool, error) {

variables/builtin.go

@@ -74,6 +74,10 @@ func BuiltinArgs(target domain.Target, platr *platutil.Resolver, gitMeta *gituti
 		} else {
 			ret.AddInactive(arg.EarthlySourceDateEpoch, gitMeta.Timestamp)
 		}
+		if ftrs.EarthlyGitAuthorArgs {
+			ret.AddInactive(arg.EarthlyGitAuthor, gitMeta.Author)
+			ret.AddInactive(arg.EarthlyGitCoAuthors, strings.Join(gitMeta.CoAuthors, " "))
+		}
 	} else {
 		// Ensure SOURCE_DATE_EPOCH is always available
 		ret.AddInactive(arg.EarthlySourceDateEpoch, "0")

variables/reserved/names.go

@@ -8,6 +8,8 @@ const (
 	EarthlyGitOriginURL         = "EARTHLY_GIT_ORIGIN_URL"
 	EarthlyGitOriginURLScrubbed = "EARTHLY_GIT_ORIGIN_URL_SCRUBBED"
 	EarthlyGitProjectName       = "EARTHLY_GIT_PROJECT_NAME"
+	EarthlyGitAuthor            = "EARTHLY_GIT_AUTHOR"
+	EarthlyGitCoAuthors         = "EARTHLY_GIT_CO_AUTHORS"
 	EarthlyGitShortHash         = "EARTHLY_GIT_SHORT_HASH"
 	EarthlyGitTag               = "EARTHLY_GIT_TAG"
 	EarthlyLocally              = "EARTHLY_LOCALLY"
@@ -41,6 +43,8 @@ func init() {
 		EarthlyBuildSha:             struct{}{},
 		EarthlyGitBranch:            struct{}{},
 		EarthlyGitCommitTimestamp:   struct{}{},
+		EarthlyGitAuthor:            struct{}{},
+		EarthlyGitCoAuthors:         struct{}{},
 		EarthlyGitHash:              struct{}{},
 		EarthlyGitOriginURL:         struct{}{},
 		EarthlyGitOriginURLScrubbed: struct{}{},