Merge pull request #80 from splitsh/excludes

Support excluding directories

README.md

@@ -128,10 +128,18 @@ splitsh-lite --prefix=lib/ --origin=origin/1.0 --path=/path/to/repo
 
 Available options:
 
- * `--prefix` is the prefix of the directory to split; you can put the split
+ * `--prefix` is the prefix of the directory to split; the value can be one of
-   contents in a sub-directory of the target repository by using the
+   the following:
-   `--prefix=from:to` syntax; split several directories by passing multiple
+
-   `--prefix` flags;
+     * `from`: the origin directory to split;
+
+     * `from:to`: move the split content to a sub-directory on the target;
+
+     * `from:to:exclude`: exclude a directory from the origin `from` directory
+       (use `from:to:exclude1:exclude2:...` to exclude more than one
+       directory).
+
+   Split several directories by passing multiple `--prefix` flags;
 
  * `--path` is the path of the repository to split (current directory by default);
 

main.go

@@ -24,19 +24,24 @@ func (p *prefixesFlag) Set(value string) error {
 	parts := strings.Split(value, ":")
 	from := parts[0]
 	to := ""
-	if len(parts) > 1 {
+	excludes := make([]string, 0)
-		to = parts[1]
+	if len(parts) >= 2 {
+		to = strings.TrimRight(parts[1], "/")
+		if len(parts) > 2 {
+			for _, exclude := range parts[2:] {
+				excludes = append(excludes, exclude)
+			}
+		}
 	}
 
 	// value must be unique
-	for _, prefix := range []*splitter.Prefix(*p) {
+	for _, prefix := range *p {
-		// FIXME: to should be normalized (xxx vs xxx/ for instance)
 		if prefix.To == to {
 			return fmt.Errorf("cannot have two prefix splits under the same directory: %s -> %s vs %s -> %s", prefix.From, prefix.To, from, to)
 		}
 	}
 
-	*p = append(*p, &splitter.Prefix{From: from, To: to})
+	*p = append(*p, splitter.NewPrefix(from, to, excludes))
 	return nil
 }
 
@@ -73,7 +78,7 @@ func main() {
 	config := &splitter.Config{
 		Path:       path,
 		Origin:     origin,
-		Prefixes:   []*splitter.Prefix(prefixes),
+		Prefixes:   prefixes,
 		Target:     target,
 		Commit:     commit,
 		Debug:      debug,

splitter/cache.go

@@ -77,6 +77,9 @@ func key(config *Config) []byte {
 	for _, prefix := range config.Prefixes {
 		io.WriteString(h, prefix.From)
 		io.WriteString(h, prefix.To)
+		for _, exclude := range prefix.Excludes {
+			io.WriteString(h, exclude)
+		}
 	}
 
 	return h.Sum(nil)

splitter/config.go

@@ -3,6 +3,7 @@ package splitter
 import (
 	"fmt"
 	"log"
+	"strings"
 	"sync"
 
 	git "github.com/libgit2/git2go/v34"
@@ -11,8 +12,27 @@ import (
 
 // Prefix represents which paths to split
 type Prefix struct {
-	From string
+	From     string
-	To   string
+	To       string
+	Excludes []string
+}
+
+// NewPrefix returns a new prefix, sanitizing the input
+func NewPrefix(from, to string, excludes []string) *Prefix {
+	// remove the trailing slash (to avoid duplicating cache)
+	from = strings.TrimRight(from, "/")
+	to = strings.TrimRight(to, "/")
+
+	// remove trailing slashes from excludes (as it does not mean anything)
+	for i, exclude := range excludes {
+		excludes[i] = strings.TrimRight(exclude, "/")
+	}
+
+	return &Prefix{
+		From:     from,
+		To:       to,
+		Excludes: excludes,
+	}
 }
 
 // Config represents a split configuration

splitter/state.go

@@ -61,13 +61,17 @@ func newState(config *Config, result *Result) (*state, error) {
 	}
 
 	if config.Debug {
-		state.logger.Printf("Splitting %s\n", state.originBranch)
+		state.logger.Printf("Splitting %s", state.originBranch)
 		for _, v := range config.Prefixes {
 			to := v.To
 			if to == "" {
 				to = "ROOT"
 			}
-			state.logger.Printf("  From \"%s\" to \"%s\"\n", v.From, to)
+			state.logger.Printf(`  From "%s" to "%s"`, v.From, to)
+			if (len(v.Excludes)) == 0 {
+			} else {
+				state.logger.Printf(`  Excluding "%s"`, strings.Join(v.Excludes, `", "`))
+			}
 		}
 	}
 
@@ -79,7 +83,7 @@ func newState(config *Config, result *Result) (*state, error) {
 
 	// simplePrefix contains the prefix when there is only one
 	// with an empty value (target)
-	if len(config.Prefixes) == 1 && config.Prefixes[0].To == "" {
+	if len(config.Prefixes) == 1 && config.Prefixes[0].To == "" && len(config.Prefixes[0].Excludes) == 0 {
 		state.simplePrefix = config.Prefixes[0].From
 	}
 
@@ -284,6 +288,14 @@ func (s *state) treeByPaths(tree *git.Tree) (*git.Tree, error) {
 			continue
 		}
 
+		if len(prefix.Excludes) > 0 {
+			prunedTree, err := s.pruneTree(splitTree, prefix.Excludes)
+			if err != nil {
+				return nil, err
+			}
+			splitTree = prunedTree
+		}
+
 		// adding the prefix
 		if prefix.To != "" {
 			prefixedTree, err = s.addPrefixToTree(splitTree, prefix.To)
@@ -360,6 +372,53 @@ func (s *state) addPrefixToTree(tree *git.Tree, prefix string) (*git.Tree, error
 	return prefixedTree, nil
 }
 
+func (s *state) pruneTree(tree *git.Tree, excludes []string) (*git.Tree, error) {
+	var err error
+	treeBuilder, err := s.repo.TreeBuilder()
+	if err != nil {
+		return nil, err
+	}
+	defer treeBuilder.Free()
+
+	err = tree.Walk(func(path string, entry *git.TreeEntry) error {
+		// always add files at the root directory
+		if entry.Type == git.ObjectBlob {
+			if err := treeBuilder.Insert(entry.Name, entry.Id, git.FilemodeBlob); err != nil {
+				return err
+			}
+			return nil
+		}
+
+		if entry.Type != git.ObjectTree {
+			// should never happen
+			return fmt.Errorf("Unexpected entry %s/%s (type %s)", path, entry.Name, entry.Type)
+		}
+
+		// exclude directory in excludes
+		for _, exclude := range excludes {
+			if entry.Name == exclude {
+				return git.TreeWalkSkip
+			}
+		}
+
+		if err := treeBuilder.Insert(entry.Name, entry.Id, git.FilemodeTree); err != nil {
+			return err
+		}
+		return git.TreeWalkSkip
+	})
+
+	if err != nil {
+		return nil, err
+	}
+
+	treeOid, err := treeBuilder.Write()
+	if err != nil {
+		return nil, err
+	}
+
+	return s.repo.LookupTree(treeOid)
+}
+
 func (s *state) copyOrSkip(rev *git.Commit, tree *git.Tree, newParents []*git.Oid) (*git.Oid, bool, error) {
 	var identical, nonIdentical *git.Oid
 	var gotParents []*git.Oid
@@ -522,12 +581,18 @@ func (s *state) legacyMessage(rev *git.Commit) string {
 
 // pushRevs sets the range to split
 func (s *state) pushRevs(revWalk *git.RevWalk) error {
-	// this is needed as origin might be in the process of being updated by git.FetchOrigin()
 	s.repoMu.Lock()
 	defer s.repoMu.Unlock()
 
-	// find the latest split sha1 if any on origin
 	var start *git.Oid
+	start = s.cache.getHead()
+	if start != nil {
+		s.result.moveHead(s.cache.get(start))
+		// FIXME: CHECK that this is an ancestor of the branch?
+		return revWalk.PushRange(fmt.Sprintf("%s..%s", start, s.originBranch))
+	}
+
+	// find the latest split sha1 if any on origin
 	var err error
 	if s.config.Commit != "" {
 		start, err = git.NewOid(s.config.Commit)
@@ -538,13 +603,6 @@ func (s *state) pushRevs(revWalk *git.RevWalk) error {
 		return revWalk.PushRange(fmt.Sprintf("%s^..%s", start, s.originBranch))
 	}
 
-	start = s.cache.getHead()
-	if start != nil {
-		s.result.moveHead(s.cache.get(start))
-		// FIXME: CHECK that this is an ancestor of the branch?
-		return revWalk.PushRange(fmt.Sprintf("%s..%s", start, s.originBranch))
-	}
-
 	branch, err := s.repo.RevparseSingle(s.originBranch)
 	if err != nil {
 		return err