dehub/commit.go

package dehub

import (
	"bytes"
	"dehub/accessctl"
	"dehub/fs"
	"dehub/sigcred"
	"dehub/yamlutil"
	"encoding"
	"encoding/base64"
	"errors"
	"fmt"
	"strings"
	"time"

	"gopkg.in/src-d/go-git.v4"
	"gopkg.in/src-d/go-git.v4/plumbing"
	"gopkg.in/src-d/go-git.v4/plumbing/object"
	yaml "gopkg.in/yaml.v2"
)

// ChangeCommit describes the structure of a change commit message.
type ChangeCommit struct {
	Message     string               `yaml:"message"`
	ChangeHash  yamlutil.Blob        `yaml:"change_hash"`
	Credentials []sigcred.Credential `yaml:"credentials"`
}

type ccYAML struct {
	Val ChangeCommit `yaml:",inline"`
}

func msgHead(msg string) string {
	i := strings.Index(msg, "\n")
	if i > 0 {
		return msg[:i]
	}
	return msg
}

// MarshalText implements the encoding.TextMarshaler interface by returning the
// form the ChangeCommit object takes in the git commit message.
func (cc ChangeCommit) MarshalText() ([]byte, error) {
	changeCommitEncoded, err := yaml.Marshal(ccYAML{cc})
	if err != nil {
		return nil, fmt.Errorf("failed to encode ChangeCommit message: %w", err)
	}

	fullMsg := msgHead(cc.Message) + "\n\n" + string(changeCommitEncoded)
	return []byte(fullMsg), nil
}

// UnmarshalText implements the encoding.TextUnmarshaler interface by decoding a
// ChangeCommit object which has been encoded into a git commit message.
func (cc *ChangeCommit) UnmarshalText(msg []byte) error {
	i := bytes.Index(msg, []byte("\n"))
	if i < 0 {
		return fmt.Errorf("commit message %q is malformed", msg)
	}
	msgHead, msg := msg[:i], msg[i:]

	var ccy ccYAML
	if err := yaml.Unmarshal(msg, &ccy); err != nil {
		return fmt.Errorf("could not unmarshal ChangeCommit message: %w", err)
	}

	*cc = ccy.Val
	if !strings.HasPrefix(cc.Message, string(msgHead)) {
		return errors.New("encoded ChangeCommit is malformed, it might not be an encoded ChangeCommit")
	}

	return nil
}

// Commit uses the given TextMarshaler to create a git commit object (with the
// specified accountID as the author) and commits it to the current HEAD,
// returning the hash of the commit.
func (r *Repo) Commit(m encoding.TextMarshaler, accountID string) (plumbing.Hash, error) {
	msgB, err := m.MarshalText()
	if err != nil {
		return plumbing.ZeroHash, fmt.Errorf("error marshaling %T to string: %v", m, err)
	}

	w, err := r.GitRepo.Worktree()
	if err != nil {
		return plumbing.ZeroHash, fmt.Errorf("could not get git worktree: %w", err)
	}
	return w.Commit(string(msgB), &git.CommitOptions{
		Author: &object.Signature{
			Name: accountID,
			When: time.Now(),
		},
	})
}

// HasStagedChanges returns true if there are file changes which have been
// staged (e.g. via "git add").
func (r *Repo) HasStagedChanges() (bool, error) {
	w, err := r.GitRepo.Worktree()
	if err != nil {
		return false, fmt.Errorf("error retrieving worktree: %w", err)
	}

	status, err := w.Status()
	if err != nil {
		return false, fmt.Errorf("error retrieving worktree status: %w", err)
	}

	var any bool
	for _, fileStatus := range status {
		if fileStatus.Staging != git.Unmodified {
			any = true
			break
		}
	}
	return any, nil
}

// NewChangeCommit constructs a ChangeCommit. If sig is given then it is used to
// create a Credential for the ChangeCommit.
func (r *Repo) NewChangeCommit(msg, accountID string, sig sigcred.SignifierInterface) (ChangeCommit, error) {
	_, headTree, err := r.head()
	if errors.Is(err, plumbing.ErrReferenceNotFound) {
		headTree = &object.Tree{}
	} else if err != nil {
		return ChangeCommit{}, err
	}

	_, stagedTree, err := fs.FromStagedChangesTree(r.GitRepo)
	if err != nil {
		return ChangeCommit{}, err
	}

	changeHash := genChangeHash(nil, msg, headTree, stagedTree)

	var creds []sigcred.Credential
	if sig != nil {
		// this is necessarily different than headTree for the case of there
		// being no HEAD (ie it's the first commit). In that case we want
		// headTree to be empty (because it's being used to generate the change
		// hash), but we want the signifier to use the raw fs (because that's
		// where the signifier's data might be).
		sigFS, err := r.headOrRawFS()
		if err != nil {
			return ChangeCommit{}, err
		}

		cred, err := sig.Sign(sigFS, changeHash)
		if err != nil {
			return ChangeCommit{}, fmt.Errorf("failed to sign commit hash: %w", err)
		}
		cred.AccountID = accountID
		creds = append(creds, cred)
	}

	return ChangeCommit{
		Message:     msg,
		ChangeHash:  changeHash,
		Credentials: creds,
	}, nil
}

func (r *Repo) assertAccessControls(
	accessCtls []accessctl.BranchAccessControl, creds []sigcred.Credential,
	branch plumbing.ReferenceName, from, to *object.Tree,
) error {
	filesChanged, err := calcDiff(from, to)
	if err != nil {
		return err
	}

	pathsChanged := make([]string, len(filesChanged))
	for i := range filesChanged {
		pathsChanged[i] = filesChanged[i].path
	}

	matchRes, err := accessctl.Match(accessCtls, accessctl.MatchInteractions{
		Branch:           branch.Short(),
		FilePathsChanged: pathsChanged,
	})
	if err != nil {
		return fmt.Errorf("could not determine applicable access controls: %w", err)
	}

	for _, matchedAC := range matchRes.ChangeAccessControls {
		ac := matchedAC.ChangeAccessControl
		condInt, err := ac.Condition.Interface()
		if err != nil {
			return fmt.Errorf("could not cast Condition of file path pattern %q to interface: %w",
				ac.FilePathPattern, err)
		} else if err := condInt.Satisfied(creds); err != nil {
			return fmt.Errorf("access control of file path pattern %q not satisfied: %w\nFiles matched:\n%s",
				ac.FilePathPattern, err, strings.Join(matchedAC.FilePaths, "\n"))
		}
	}

	return nil
}

// VerifyChangeCommit verifies that the change commit at the given hash, which
// is presumably on the given branch, is gucci.
func (r *Repo) VerifyChangeCommit(branch plumbing.ReferenceName, h plumbing.Hash) error {
	commit, err := r.GitRepo.CommitObject(h)
	if err != nil {
		return fmt.Errorf("could not retrieve commit object: %w", err)
	}

	commitTree, err := r.GitRepo.TreeObject(commit.TreeHash)
	if err != nil {
		return fmt.Errorf("could not retrieve tree object: %w", err)
	}

	var changeCommit ChangeCommit
	if err := changeCommit.UnmarshalText([]byte(commit.Message)); err != nil {
		return err
	}

	sigTree := commitTree // only for root commit
	parentTree := &object.Tree{}
	if commit.NumParents() > 0 {
		parent, err := commit.Parent(0)
		if err != nil {
			return fmt.Errorf("could not retrieve parent of commit: %w", err)
		} else if parentTree, err = r.GitRepo.TreeObject(parent.TreeHash); err != nil {
			return fmt.Errorf("could not retrieve tree object of parent %q: %w", parent.Hash, err)
		}
		sigTree = parentTree
	}
	sigFS := fs.FromTree(sigTree)

	cfg, err := r.loadConfig(sigFS)
	if err != nil {
		return fmt.Errorf("error loading config: %w", err)
	}

	err = r.assertAccessControls(
		cfg.AccessControls, changeCommit.Credentials,
		branch, parentTree, commitTree,
	)
	if err != nil {
		return fmt.Errorf("failed to satisfy all access controls: %w", err)
	}

	expectedChangeHash := genChangeHash(nil, changeCommit.Message, parentTree, commitTree)
	if !bytes.Equal(changeCommit.ChangeHash, expectedChangeHash) {
		return fmt.Errorf("malformed change_hash in commit body, is %s but should be %s",
			base64.StdEncoding.EncodeToString(expectedChangeHash),
			base64.StdEncoding.EncodeToString(changeCommit.ChangeHash))
	}

	for _, cred := range changeCommit.Credentials {
		sig, err := r.signifierForCredential(sigFS, cred)
		if err != nil {
			return fmt.Errorf("error finding signifier for credential %+v: %w", cred, err)
		} else if err := sig.Verify(sigFS, expectedChangeHash, cred); err != nil {
			return fmt.Errorf("error verifying credential %+v: %w", cred, err)
		}
	}

	return nil
}