ReleaseKit/internal/validation/tree.go

package validation

import (
	"context"
	"fmt"
	"path"
	"path/filepath"
	"sort"
	"strings"

	"gitea.maximumdirect.net/eric/ReleaseKit/internal/git"
)

// CandidateTree is an inventory of one captured commit with blob contents
// loaded lazily by object ID.
type CandidateTree struct {
	context        context.Context
	repositoryDir  string
	entries        []git.TreeEntry
	inputEntries   []git.TreeEntry
	byPath         map[string]int
	directories    map[string]struct{}
	blobReader     *git.BlobReader
	blobReads      uint64
	symlinkTargets map[string]string
	closed         bool
	closeErr       error
}

// TreeScope selects the automatic input files for one physical module and
// optional repository-shared files. ModuleDir is an absolute path. A nil
// SharedFiles slice selects the optional default files; a non-nil slice
// replaces those defaults.
type TreeScope struct {
	ModuleDir   string
	SharedFiles []string
}

// ReadCandidateTree enumerates the captured commit without reading from the
// worktree. The supplied context also governs the scoped reader through closure.
func ReadCandidateTree(ctx context.Context, repositoryDir, commit string) (*CandidateTree, error) {
	repositoryAbs, err := filepath.Abs(repositoryDir)
	if err != nil {
		return nil, fmt.Errorf("resolve candidate repository directory: %w", err)
	}
	return ReadCandidateTreeForScope(ctx, repositoryDir, commit, TreeScope{ModuleDir: repositoryAbs})
}

// ReadCandidateTreeForScope retains the complete committed tree for reference
// lookup and separately constructs the selected automatic input paths.
func ReadCandidateTreeForScope(ctx context.Context, repositoryDir, commit string, scope TreeScope) (*CandidateTree, error) {
	entries, err := git.ReadTree(ctx, repositoryDir, commit)
	if err != nil {
		return nil, err
	}
	tree := &CandidateTree{
		context:        ctx,
		repositoryDir:  repositoryDir,
		entries:        entries,
		byPath:         make(map[string]int, len(entries)),
		directories:    make(map[string]struct{}),
		symlinkTargets: make(map[string]string),
	}
	tree.directories["."] = struct{}{}
	for index, entry := range entries {
		tree.byPath[entry.Path] = index
		for directory := path.Dir(entry.Path); directory != "."; directory = path.Dir(directory) {
			tree.directories[directory] = struct{}{}
		}
	}
	tree.inputEntries, err = selectInputEntries(repositoryDir, entries, tree.byPath, scope)
	if err != nil {
		_ = tree.Close()
		return nil, err
	}
	return tree, nil
}

// Entries returns the committed tree metadata in Git's tree order.
func (tree *CandidateTree) Entries() []git.TreeEntry {
	return append([]git.TreeEntry(nil), tree.entries...)
}

// InputEntries returns the ordered committed files selected for automatic
// checks. It is separate from Entries so reference resolution retains the full
// repository tree.
func (tree *CandidateTree) InputEntries() []git.TreeEntry {
	return append([]git.TreeEntry(nil), tree.inputEntries...)
}

func selectInputEntries(repositoryDir string, entries []git.TreeEntry, byPath map[string]int, scope TreeScope) ([]git.TreeEntry, error) {
	repositoryAbs, err := filepath.Abs(repositoryDir)
	if err != nil {
		return nil, fmt.Errorf("resolve candidate repository directory: %w", err)
	}
	moduleDir := scope.ModuleDir
	if moduleDir == "" {
		moduleDir = repositoryAbs
	}
	if !filepath.IsAbs(moduleDir) {
		return nil, fmt.Errorf("selected module directory must be absolute: %q", moduleDir)
	}
	moduleAbs, err := filepath.Abs(moduleDir)
	if err != nil {
		return nil, fmt.Errorf("resolve selected module directory: %w", err)
	}
	relative, err := filepath.Rel(repositoryAbs, moduleAbs)
	if err != nil {
		return nil, fmt.Errorf("make selected module relative to repository: %w", err)
	}
	if relative == ".." || strings.HasPrefix(relative, ".."+string(filepath.Separator)) || filepath.IsAbs(relative) {
		return nil, fmt.Errorf("selected module directory %q is outside repository %q", moduleDir, repositoryDir)
	}
	modulePath := filepath.ToSlash(relative)
	if relative == "." {
		modulePath = "."
	}

	boundaries := make([]string, 0)
	for _, entry := range entries {
		moduleRelative, ok := relativeToModule(entry.Path, modulePath)
		if !ok || moduleRelative == "go.mod" || path.Base(moduleRelative) != "go.mod" {
			continue
		}
		if insideVendor(moduleRelative) {
			continue
		}
		boundaries = append(boundaries, path.Dir(entry.Path))
	}
	selected := make(map[string]git.TreeEntry)
	for _, entry := range entries {
		moduleRelative, ok := relativeToModule(entry.Path, modulePath)
		if !ok || moduleRelative == "" {
			continue
		}
		excluded := false
		for _, boundary := range boundaries {
			if entry.Path == boundary || strings.HasPrefix(entry.Path, boundary+"/") {
				excluded = true
				break
			}
		}
		if !excluded {
			selected[entry.Path] = entry
		}
	}

	sharedFiles := scope.SharedFiles
	usingDefaults := sharedFiles == nil
	if usingDefaults {
		sharedFiles = []string{"COPYRIGHT", "LICENSE", "docs/oss/attribution.md"}
	}
	for _, requested := range sharedFiles {
		name, err := normalizeSharedFile(requested)
		if err != nil {
			return nil, &Failure{Check: SharedFiles, Findings: []Finding{{Path: requested, Message: err.Error()}}}
		}
		index, exists := byPath[name]
		if !exists || entries[index].Type != "blob" {
			if usingDefaults {
				continue
			}
			return nil, &Failure{Check: SharedFiles, Findings: []Finding{{Path: name, Message: "selected shared file is not present in the committed tree"}}}
		}
		selected[name] = entries[index]
	}

	paths := make([]string, 0, len(selected))
	for name := range selected {
		paths = append(paths, name)
	}
	sort.Strings(paths)
	inputs := make([]git.TreeEntry, 0, len(paths))
	for _, name := range paths {
		inputs = append(inputs, selected[name])
	}
	return inputs, nil
}

func relativeToModule(repositoryPath, modulePath string) (string, bool) {
	if modulePath == "." {
		return repositoryPath, true
	}
	if repositoryPath == modulePath {
		return "", true
	}
	prefix := modulePath + "/"
	if !strings.HasPrefix(repositoryPath, prefix) {
		return "", false
	}
	return strings.TrimPrefix(repositoryPath, prefix), true
}

func insideVendor(moduleRelative string) bool {
	for _, component := range strings.Split(moduleRelative, "/") {
		if component == "vendor" {
			return true
		}
	}
	return false
}

func normalizeSharedFile(name string) (string, error) {
	if name == "" || path.IsAbs(name) || strings.Contains(name, "\\") {
		return "", fmt.Errorf("shared file path must be a repository-relative file path")
	}
	clean := path.Clean(name)
	if clean == "." || clean == ".." || strings.HasPrefix(clean, "../") {
		return "", fmt.Errorf("shared file path must stay within the repository")
	}
	return clean, nil
}

// ReadFile returns newly allocated, caller-owned bytes for a committed blob.
// Regular-file content is not cached, so repeated reads request the object
// again through the tree's scoped blob reader.
func (tree *CandidateTree) ReadFile(ctx context.Context, path string) ([]byte, error) {
	if tree.closed {
		return nil, fmt.Errorf("read committed path %q: candidate tree is closed", path)
	}
	index, ok := tree.byPath[path]
	if !ok {
		return nil, fmt.Errorf("committed path %q does not exist", path)
	}
	entry := tree.entries[index]
	if entry.Type != "blob" {
		return nil, fmt.Errorf("committed path %q is a %s entry, not a blob", path, entry.Type)
	}
	if entry.Mode == "120000" {
		target, err := tree.readSymlinkTarget(ctx, entry)
		if err != nil {
			return nil, err
		}
		return []byte(target), nil
	}
	return tree.readBlob(ctx, entry)
}

func (tree *CandidateTree) readSymlinkTarget(ctx context.Context, entry git.TreeEntry) (string, error) {
	if tree.closed {
		return "", fmt.Errorf("read committed path %q: candidate tree is closed", entry.Path)
	}
	if entry.Type != "blob" || entry.Mode != "120000" {
		return "", fmt.Errorf("committed path %q is not a symlink blob", entry.Path)
	}
	if target, ok := tree.symlinkTargets[entry.ObjectID]; ok {
		return target, nil
	}
	contents, err := tree.readBlob(ctx, entry)
	if err != nil {
		return "", err
	}
	target := string(contents)
	if tree.symlinkTargets == nil {
		tree.symlinkTargets = make(map[string]string)
	}
	tree.symlinkTargets[entry.ObjectID] = target
	return target, nil
}

func (tree *CandidateTree) readBlob(ctx context.Context, entry git.TreeEntry) ([]byte, error) {
	if tree.closed {
		return nil, fmt.Errorf("read committed path %q: candidate tree is closed", entry.Path)
	}
	if tree.blobReader == nil {
		// The operation context governs shutdown independently of each read.
		operationContext := tree.context
		if operationContext == nil {
			operationContext = context.Background()
		}
		tree.blobReader = git.NewBlobReader(operationContext, tree.repositoryDir)
	}
	tree.blobReads++
	contents, err := tree.blobReader.Read(ctx, entry.ObjectID)
	if err != nil {
		return nil, fmt.Errorf("read committed path %q: %w", entry.Path, err)
	}
	return contents, nil
}

// Close closes the scoped local Git blob reader, if content was requested.
// Metadata-only trees have no child process to close.
func (tree *CandidateTree) Close() error {
	if tree.closed {
		return tree.closeErr
	}
	tree.closed = true
	if tree.blobReader != nil {
		tree.closeErr = tree.blobReader.Close()
	}
	return tree.closeErr
}

// Entry returns the committed metadata for path when present.
func (tree *CandidateTree) Entry(path string) (git.TreeEntry, bool) {
	index, ok := tree.byPath[path]
	if !ok {
		return git.TreeEntry{}, false
	}
	return tree.entries[index], true
}

// DestinationKind classifies a committed local destination.
type DestinationKind string

const (
	DestinationEntry           DestinationKind = "entry"
	DestinationDirectory       DestinationKind = "directory"
	DestinationMissing         DestinationKind = "missing"
	DestinationOutsideCheckout DestinationKind = "outside-checkout"
	DestinationSubmodule       DestinationKind = "submodule"
	DestinationSymlinkCycle    DestinationKind = "symlink-cycle"
)

// Destination records the resolved identity and its classification.
type Destination struct {
	Kind DestinationKind
	Path string
}

// ResolveDestination resolves local destinations using the captured Git tree.
// Relative destinations start at the source directory; absolute destinations
// within the repository directory map back to committed paths.
func (tree *CandidateTree) ResolveDestination(ctx context.Context, sourcePath, destination string) (Destination, error) {
	if err := ctx.Err(); err != nil {
		return Destination{}, err
	}
	directory := path.Dir(sourcePath)
	if path.IsAbs(destination) {
		relative, inside, err := tree.relativeAbsoluteDestination(destination)
		if err != nil {
			return Destination{}, err
		}
		if !inside {
			return Destination{Kind: DestinationOutsideCheckout, Path: destination}, nil
		}
		directory, destination = ".", relative
	}
	return tree.resolveDestination(ctx, directory, strings.Split(destination, "/"), make(map[string]bool))
}

func (tree *CandidateTree) resolveDestination(ctx context.Context, directory string, components []string, active map[string]bool) (Destination, error) {
	for index, component := range components {
		if err := ctx.Err(); err != nil {
			return Destination{}, err
		}
		remaining := components[index+1:]
		switch component {
		case "", ".":
			continue
		case "..":
			if directory == "." {
				return Destination{Kind: DestinationOutsideCheckout, Path: unresolvedPath("..", remaining)}, nil
			}
			directory = path.Dir(directory)
			continue
		}
		partial := path.Join(directory, component)
		entry, exists := tree.Entry(partial)
		if !exists {
			if _, isDirectory := tree.directories[partial]; !isDirectory {
				return Destination{Kind: DestinationMissing, Path: unresolvedPath(partial, remaining)}, nil
			}
			directory = partial
			continue
		}
		if entry.Mode == "160000" {
			return Destination{Kind: DestinationSubmodule, Path: unresolvedPath(partial, remaining)}, nil
		}
		if entry.Mode == "120000" {
			if active[partial] {
				return Destination{Kind: DestinationSymlinkCycle, Path: partial}, nil
			}
			entry, _ := tree.Entry(partial)
			linkText, err := tree.readSymlinkTarget(ctx, entry)
			if err != nil {
				return Destination{}, fmt.Errorf("read committed symlink %q: %w", partial, err)
			}
			target := linkText
			targetDirectory := directory
			if path.IsAbs(target) {
				relative, inside, err := tree.relativeAbsoluteDestination(target)
				if err != nil {
					return Destination{}, err
				}
				if !inside {
					return Destination{Kind: DestinationOutsideCheckout, Path: unresolvedPath(target, remaining)}, nil
				}
				targetDirectory, target = ".", relative
			}
			active[partial] = true
			destination, err := tree.resolveDestination(ctx, targetDirectory, strings.Split(target, "/"), active)
			delete(active, partial)
			if err != nil || len(remaining) == 0 {
				return destination, err
			}
			switch destination.Kind {
			case DestinationDirectory:
				return tree.resolveDestination(ctx, destination.Path, remaining, active)
			case DestinationEntry:
				destination.Kind = DestinationMissing
			case DestinationSymlinkCycle:
				return destination, nil
			}
			destination.Path = unresolvedPath(destination.Path, remaining)
			return destination, nil
		}
		if len(remaining) != 0 {
			return Destination{Kind: DestinationMissing, Path: unresolvedPath(partial, remaining)}, nil
		}
		return Destination{Kind: DestinationEntry, Path: partial}, nil
	}
	return Destination{Kind: DestinationDirectory, Path: directory}, nil
}

func (tree *CandidateTree) relativeAbsoluteDestination(destination string) (string, bool, error) {
	root, err := filepath.Abs(tree.repositoryDir)
	if err != nil {
		return "", false, fmt.Errorf("resolve candidate repository directory: %w", err)
	}
	root = filepath.ToSlash(root)
	if destination == root {
		return ".", true, nil
	}
	// Preserve the suffix for component traversal, including symlinks before '..'.
	relative, inside := strings.CutPrefix(destination, strings.TrimSuffix(root, "/")+"/")
	return relative, inside, nil
}

// unresolvedPath preserves components that could not be traversed. Cleaning
// them would hide missing entries, non-directories, or checkout boundaries.
func unresolvedPath(base string, remaining []string) string {
	if len(remaining) == 0 {
		return base
	}
	return base + "/" + strings.Join(remaining, "/")
}