Export: enhance trimming using build statement metadata

src/BUILD.plz

@@ -10,8 +10,8 @@ go_binary(
     deps = [
         "///third_party/go/github.com_thought-machine_go-flags//:go-flags",
         "///third_party/go/go.uber.org_automaxprocs//maxprocs",
-        "//src/audit",
         "//src/assets",
+        "//src/audit",
         "//src/build",
         "//src/cache",
         "//src/clean",
@@ -29,7 +29,6 @@ go_binary(
         "//src/help",
         "//src/metrics",
         "//src/output",
-        "//src/parse",
         "//src/plz",
         "//src/plzinit",
         "//src/process",

src/core/package.go

@@ -34,6 +34,8 @@ type Package struct {
 	targets map[string]*BuildTarget
 	// Set of output files from rules.
 	Outputs map[string]*BuildTarget
+	// Includes metadata from parsing the package BUILD file.
+	Metadata PackageMetadata
 	// Protects access to above
 	mutex sync.RWMutex
 }
@@ -43,13 +45,24 @@ func NewPackage(name string) *Package {
 	return NewPackageSubrepo(name, "")
 }
 
+// NewPackageWithOpts constructs a new package with the given name, and enables additional features
+// given the flags enabled in the build state.
+func NewPackageWithOpts(state *BuildState, name string) *Package {
+	pkg := NewPackage(name)
+	if state.ParseMetadata {
+		pkg.Metadata = newPackageMetadata()
+	}
+	return pkg
+}
+
 // NewPackageSubrepo constructs a new package with the given name and subrepo.
 func NewPackageSubrepo(name, subrepo string) *Package {
 	return &Package{
 		Name:        name,
 		SubrepoName: subrepo,
 		targets:     map[string]*BuildTarget{},
 		Outputs:     map[string]*BuildTarget{},
+		Metadata:    newNoopPackageMetadata(),
 	}
 }
 

src/core/package_metadata.go

@@ -0,0 +1,154 @@
+package core
+
+import (
+	"fmt"
+	"slices"
+)
+
+// BuildStatement represents the start and end byte positions of a parsed statement in a BUILD file.
+type BuildStatement struct {
+	Start, End int
+}
+
+// Len returns the byte length of the build statement.
+func (bs *BuildStatement) Len() int64 {
+	return int64(bs.End - bs.Start)
+}
+
+// StartPos returns the starting byte position of the build statement.
+func (bs *BuildStatement) StartPos() int64 {
+	return int64(bs.Start)
+}
+
+// BuildStatements is a slice of BuildStatement that implements sort.Interface.
+type BuildStatements []BuildStatement
+
+func (s BuildStatements) Len() int           { return len(s) }
+func (s BuildStatements) Swap(i, j int)      { s[i], s[j] = s[j], s[i] }
+func (s BuildStatements) Less(i, j int) bool { return s[i].StartPos() < s[j].StartPos() }
+
+// BuildStatementProvider is a type for methods that generate new build statements.
+type BuildStatementProvider func() *BuildStatement
+
+// SubincludesLabelProvider is a type for methods that generate labels from a subincludes.
+type SubincludesLabelProvider func() BuildLabels
+
+// PackageMetadata stores metadata about parsed BUILD files, mapping statements and subincludes
+// to their respective targets.
+type PackageMetadata interface {
+	// RegisterStatementTarget maps a build statement to a target it generated.
+	RegisterStatementTarget(target *BuildTarget, stmtProvider BuildStatementProvider)
+	// RegisterRequiredSubinclude maps a target to the subincludes required to build it.
+	RegisterRequiredSubinclude(target *BuildTarget, labelProvider SubincludesLabelProvider)
+	// RegisterSubincludeStmt maps a subinclude statement to a label it includes.
+	RegisterSubincludeStmt(label BuildLabel, stmtProvider BuildStatementProvider)
+	// FindStatement returns the build statement that generated the given target.
+	FindStatement(target *BuildTarget) (*BuildStatement, error)
+	// FindTargets returns all targets generated by the given build statement.
+	FindTargets(stmt *BuildStatement) ([]*BuildTarget, error)
+	// FindRequiredSubincludes returns all subinclude labels required by the given target.
+	FindRequiredSubincludes(target *BuildTarget) (BuildLabels, error)
+	// GetSubincludedLabels returns the labels included by a given subinclude statement.
+	GetSubincludedLabels(stmt *BuildStatement) (BuildLabels, bool)
+}
+
+// packageMetadataImpl stores metadata about parsed BUILD files, mapping statements and subincludes
+// to their respective targets.
+type packageMetadataImpl struct {
+	// a list of targets generated from each built statement
+	StmtToTarget map[BuildStatement][]*BuildTarget
+	// the subincluded label dependencies per target
+	TargetToSubinclude map[*BuildTarget]BuildLabels
+	// all the labels included for each subincludes statement
+	LabelsPerSubincludeStmt map[BuildStatement]BuildLabels
+}
+
+func newPackageMetadata() PackageMetadata {
+	return &packageMetadataImpl{
+		StmtToTarget:            map[BuildStatement][]*BuildTarget{},
+		TargetToSubinclude:      map[*BuildTarget]BuildLabels{},
+		LabelsPerSubincludeStmt: map[BuildStatement]BuildLabels{},
+	}
+}
+
+// RegisterStatementTarget maps a build statement to a target it generated.
+func (bfm *packageMetadataImpl) RegisterStatementTarget(target *BuildTarget, stmtProvider BuildStatementProvider) {
+	stmt := stmtProvider()
+	bfm.StmtToTarget[*stmt] = append(bfm.StmtToTarget[*stmt], target)
+}
+
+// RegisterRequiredSubinclude maps a target to the subincludes required to build it.
+func (bfm *packageMetadataImpl) RegisterRequiredSubinclude(target *BuildTarget, labelProvider SubincludesLabelProvider) {
+	labels := labelProvider()
+	if len(labels) == 0 {
+		log.Infof("Attempted to register empty subinclude labels for target %s", target.String())
+		return
+	}
+
+	bfm.TargetToSubinclude[target] = append(bfm.TargetToSubinclude[target], labels...)
+}
+
+// RegisterSubincludeStmt maps a subinclude statement to a label it includes.
+func (bfm *packageMetadataImpl) RegisterSubincludeStmt(label BuildLabel, stmtProvider BuildStatementProvider) {
+	stmt := stmtProvider()
+	bfm.LabelsPerSubincludeStmt[*stmt] = append(bfm.LabelsPerSubincludeStmt[*stmt], label)
+}
+
+// FindStatement returns the build statement that generated the given target.
+func (bfm *packageMetadataImpl) FindStatement(target *BuildTarget) (*BuildStatement, error) {
+	for stmt, targets := range bfm.StmtToTarget {
+		if slices.Contains(targets, target) {
+			return &stmt, nil
+		}
+	}
+	return nil, fmt.Errorf("Target %s not found in statement metadata.", target.String())
+}
+
+// FindTargets returns all targets generated by the given build statement.
+func (bfm *packageMetadataImpl) FindTargets(stmt *BuildStatement) ([]*BuildTarget, error) {
+	targets, ok := bfm.StmtToTarget[*stmt]
+	if !ok {
+		return nil, fmt.Errorf("Targets not found for statement %v.", stmt)
+	}
+	return targets, nil
+}
+
+// FindRequiredSubincludes returns all subinclude labels required by the given target.
+func (bfm *packageMetadataImpl) FindRequiredSubincludes(target *BuildTarget) (BuildLabels, error) {
+	subincludes, ok := bfm.TargetToSubinclude[target]
+	if !ok {
+		return nil, fmt.Errorf("Subincludes not found for target %v.", target)
+	}
+	return subincludes, nil
+}
+
+// GetSubincludedLabels returns the labels included by a given subinclude statement.
+func (bfm *packageMetadataImpl) GetSubincludedLabels(stmt *BuildStatement) (BuildLabels, bool) {
+	v, ok := bfm.LabelsPerSubincludeStmt[*stmt]
+	return v, ok
+}
+
+type noopPackageMetadata struct{}
+
+func newNoopPackageMetadata() PackageMetadata {
+	return &noopPackageMetadata{}
+}
+
+func (n *noopPackageMetadata) RegisterStatementTarget(target *BuildTarget, stmtProvider BuildStatementProvider) {
+}
+func (n *noopPackageMetadata) RegisterRequiredSubinclude(target *BuildTarget, labelProvider SubincludesLabelProvider) {
+}
+func (n *noopPackageMetadata) RegisterSubincludeStmt(label BuildLabel, stmtProvider BuildStatementProvider) {
+}
+func (n *noopPackageMetadata) FindStatement(target *BuildTarget) (*BuildStatement, error) {
+	return nil, fmt.Errorf("metadata not tracked")
+}
+func (n *noopPackageMetadata) FindTargets(stmt *BuildStatement) ([]*BuildTarget, error) {
+	return nil, fmt.Errorf("metadata not tracked")
+}
+func (n *noopPackageMetadata) FindRequiredSubincludes(target *BuildTarget) (BuildLabels, error) {
+	return nil, fmt.Errorf("metadata not tracked")
+}
+func (n *noopPackageMetadata) GetSubincludedLabels(stmt *BuildStatement) (BuildLabels, bool) {
+	return nil, false
+}

src/core/state.go

@@ -240,6 +240,8 @@ type BuildState struct {
 	// NeedDebugDeps is true if we're doing a `plz debug` and we need to build the debug tools and
 	// data
 	NeedDebugDeps bool
+	// ParseMetadata is true if we want to store build file metadata
+	ParseMetadata bool
 
 	// initOnce is used to control loading the subrepo .plzconfig
 	initOnce *sync.Once

src/export/BUILD

@@ -4,10 +4,21 @@ go_library(
     pgo_file = "//:pgo",
     visibility = ["PUBLIC"],
     deps = [
+        "///third_party/go/github.com_please-build_buildtools//build",
         "//src/cli/logging",
         "//src/core",
         "//src/fs",
-        "//src/gc",
         "//src/parse",
+        "//src/parse/asp",
+    ],
+)
+
+go_test(
+    name = "export_test",
+    srcs = ["export_test.go"],
+    deps = [
+        ":export",
+        "///third_party/go/github.com_stretchr_testify//assert",
+        "//src/core",
     ],
 )