ReleaseKit/internal/validation/tree.go

package validation

import (
	"context"
	"fmt"
	"path"
	"path/filepath"
	"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 {
	repositoryDir string
	entries       []git.TreeEntry
	byPath        map[string]int
	directories   map[string]struct{}
	blobs         map[string][]byte
	blobReader    *git.BlobReader
	closed        bool
	closeErr      error
}

// ReadCandidateTree enumerates the captured commit without reading from the
// worktree.
func ReadCandidateTree(ctx context.Context, repositoryDir, commit string) (*CandidateTree, error) {
	entries, err := git.ReadTree(ctx, repositoryDir, commit)
	if err != nil {
		return nil, err
	}
	tree := &CandidateTree{
		repositoryDir: repositoryDir,
		entries:       entries,
		byPath:        make(map[string]int, len(entries)),
		directories:   make(map[string]struct{}),
		blobs:         make(map[string][]byte),
	}
	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{}{}
		}
	}
	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...)
}

// ReadFile returns the committed blob bytes for path. Contents are cached by
// object ID for later reads from this tree.
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 contents, ok := tree.blobs[entry.ObjectID]; ok {
		return append([]byte(nil), contents...), nil
	}
	if tree.blobReader == nil {
		// The reader belongs to the tree and can outlive an individual read
		// context. Each Read call still applies its own cancellation to the
		// in-flight exchange.
		tree.blobReader = git.NewBlobReader(context.Background(), tree.repositoryDir)
	}
	contents, err := tree.blobReader.Read(ctx, entry.ObjectID)
	if err != nil {
		return nil, fmt.Errorf("read committed path %q: %w", path, err)
	}
	if tree.blobs == nil {
		tree.blobs = make(map[string][]byte)
	}
	tree.blobs[entry.ObjectID] = append([]byte(nil), contents...)
	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
			}
			linkText, err := tree.ReadFile(ctx, partial)
			if err != nil {
				return Destination{}, fmt.Errorf("read committed symlink %q: %w", partial, err)
			}
			target := string(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, "/")
}