Use commit graph files for listing pages (#7314)

* Experimental support for git commit graph files and bloom filter index

Signed-off-by: Filip Navara <filip.navara@gmail.com>

* Force vendor of commitgraph

Signed-off-by: Filip Navara <filip.navara@gmail.com>

* Remove bloom filter experiment and debug prints

* Remove old code for building commit graphs

* Remove unused function

* Remove mmap usage

* gofmt

* sort vendor/modules.txt

* Add copyright header and log commit-graph error

vendor/gopkg.in/src-d/go-git.v4/plumbing/object/commitgraph/commitnode.go

@@ -0,0 +1,98 @@
+package commitgraph
+
+import (
+	"io"
+	"time"
+
+	"gopkg.in/src-d/go-git.v4/plumbing"
+	"gopkg.in/src-d/go-git.v4/plumbing/object"
+	"gopkg.in/src-d/go-git.v4/plumbing/storer"
+)
+
+// CommitNode is generic interface encapsulating a lightweight commit object retrieved
+// from CommitNodeIndex
+type CommitNode interface {
+	// ID returns the Commit object id referenced by the commit graph node.
+	ID() plumbing.Hash
+	// Tree returns the Tree referenced by the commit graph node.
+	Tree() (*object.Tree, error)
+	// CommitTime returns the Commiter.When time of the Commit referenced by the commit graph node.
+	CommitTime() time.Time
+	// NumParents returns the number of parents in a commit.
+	NumParents() int
+	// ParentNodes return a CommitNodeIter for parents of specified node.
+	ParentNodes() CommitNodeIter
+	// ParentNode returns the ith parent of a commit.
+	ParentNode(i int) (CommitNode, error)
+	// ParentHashes returns hashes of the parent commits for a specified node
+	ParentHashes() []plumbing.Hash
+	// Generation returns the generation of the commit for reachability analysis.
+	// Objects with newer generation are not reachable from objects of older generation.
+	Generation() uint64
+	// Commit returns the full commit object from the node
+	Commit() (*object.Commit, error)
+}
+
+// CommitNodeIndex is generic interface encapsulating an index of CommitNode objects
+type CommitNodeIndex interface {
+	// Get returns a commit node from a commit hash
+	Get(hash plumbing.Hash) (CommitNode, error)
+}
+
+// CommitNodeIter is a generic closable interface for iterating over commit nodes.
+type CommitNodeIter interface {
+	Next() (CommitNode, error)
+	ForEach(func(CommitNode) error) error
+	Close()
+}
+
+// parentCommitNodeIter provides an iterator for parent commits from associated CommitNodeIndex.
+type parentCommitNodeIter struct {
+	node CommitNode
+	i    int
+}
+
+func newParentgraphCommitNodeIter(node CommitNode) CommitNodeIter {
+	return &parentCommitNodeIter{node, 0}
+}
+
+// Next moves the iterator to the next commit and returns a pointer to it. If
+// there are no more commits, it returns io.EOF.
+func (iter *parentCommitNodeIter) Next() (CommitNode, error) {
+	obj, err := iter.node.ParentNode(iter.i)
+	if err == object.ErrParentNotFound {
+		return nil, io.EOF
+	}
+	if err == nil {
+		iter.i++
+	}
+
+	return obj, err
+}
+
+// ForEach call the cb function for each commit contained on this iter until
+// an error appends or the end of the iter is reached. If ErrStop is sent
+// the iteration is stopped but no error is returned. The iterator is closed.
+func (iter *parentCommitNodeIter) ForEach(cb func(CommitNode) error) error {
+	for {
+		obj, err := iter.Next()
+		if err != nil {
+			if err == io.EOF {
+				return nil
+			}
+
+			return err
+		}
+
+		if err := cb(obj); err != nil {
+			if err == storer.ErrStop {
+				return nil
+			}
+
+			return err
+		}
+	}
+}
+
+func (iter *parentCommitNodeIter) Close() {
+}

vendor/gopkg.in/src-d/go-git.v4/plumbing/object/commitgraph/commitnode_graph.go

@@ -0,0 +1,131 @@
+package commitgraph
+
+import (
+	"fmt"
+	"time"
+
+	"gopkg.in/src-d/go-git.v4/plumbing"
+	"gopkg.in/src-d/go-git.v4/plumbing/format/commitgraph"
+	"gopkg.in/src-d/go-git.v4/plumbing/object"
+	"gopkg.in/src-d/go-git.v4/plumbing/storer"
+)
+
+// graphCommitNode is a reduced representation of Commit as presented in the commit
+// graph file (commitgraph.Node). It is merely useful as an optimization for walking
+// the commit graphs.
+//
+// graphCommitNode implements the CommitNode interface.
+type graphCommitNode struct {
+	// Hash for the Commit object
+	hash plumbing.Hash
+	// Index of the node in the commit graph file
+	index int
+
+	commitData *commitgraph.CommitData
+	gci        *graphCommitNodeIndex
+}
+
+// graphCommitNodeIndex is an index that can load CommitNode objects from both the commit
+// graph files and the object store.
+//
+// graphCommitNodeIndex implements the CommitNodeIndex interface
+type graphCommitNodeIndex struct {
+	commitGraph commitgraph.Index
+	s           storer.EncodedObjectStorer
+}
+
+// NewGraphCommitNodeIndex returns CommitNodeIndex implementation that uses commit-graph
+// files as backing storage and falls back to object storage when necessary
+func NewGraphCommitNodeIndex(commitGraph commitgraph.Index, s storer.EncodedObjectStorer) CommitNodeIndex {
+	return &graphCommitNodeIndex{commitGraph, s}
+}
+
+func (gci *graphCommitNodeIndex) Get(hash plumbing.Hash) (CommitNode, error) {
+	// Check the commit graph first
+	parentIndex, err := gci.commitGraph.GetIndexByHash(hash)
+	if err == nil {
+		parent, err := gci.commitGraph.GetCommitDataByIndex(parentIndex)
+		if err != nil {
+			return nil, err
+		}
+
+		return &graphCommitNode{
+			hash:       hash,
+			index:      parentIndex,
+			commitData: parent,
+			gci:        gci,
+		}, nil
+	}
+
+	// Fallback to loading full commit object
+	commit, err := object.GetCommit(gci.s, hash)
+	if err != nil {
+		return nil, err
+	}
+
+	return &objectCommitNode{
+		nodeIndex: gci,
+		commit:    commit,
+	}, nil
+}
+
+func (c *graphCommitNode) ID() plumbing.Hash {
+	return c.hash
+}
+
+func (c *graphCommitNode) Tree() (*object.Tree, error) {
+	return object.GetTree(c.gci.s, c.commitData.TreeHash)
+}
+
+func (c *graphCommitNode) CommitTime() time.Time {
+	return c.commitData.When
+}
+
+func (c *graphCommitNode) NumParents() int {
+	return len(c.commitData.ParentIndexes)
+}
+
+func (c *graphCommitNode) ParentNodes() CommitNodeIter {
+	return newParentgraphCommitNodeIter(c)
+}
+
+func (c *graphCommitNode) ParentNode(i int) (CommitNode, error) {
+	if i < 0 || i >= len(c.commitData.ParentIndexes) {
+		return nil, object.ErrParentNotFound
+	}
+
+	parent, err := c.gci.commitGraph.GetCommitDataByIndex(c.commitData.ParentIndexes[i])
+	if err != nil {
+		return nil, err
+	}
+
+	return &graphCommitNode{
+		hash:       c.commitData.ParentHashes[i],
+		index:      c.commitData.ParentIndexes[i],
+		commitData: parent,
+		gci:        c.gci,
+	}, nil
+}
+
+func (c *graphCommitNode) ParentHashes() []plumbing.Hash {
+	return c.commitData.ParentHashes
+}
+
+func (c *graphCommitNode) Generation() uint64 {
+	// If the commit-graph file was generated with older Git version that
+	// set the generation to zero for every commit the generation assumption
+	// is still valid. It is just less useful.
+	return uint64(c.commitData.Generation)
+}
+
+func (c *graphCommitNode) Commit() (*object.Commit, error) {
+	return object.GetCommit(c.gci.s, c.hash)
+}
+
+func (c *graphCommitNode) String() string {
+	return fmt.Sprintf(
+		"%s %s\nDate:   %s",
+		plumbing.CommitObject, c.ID(),
+		c.CommitTime().Format(object.DateFormat),
+	)
+}

vendor/gopkg.in/src-d/go-git.v4/plumbing/object/commitgraph/commitnode_object.go

@@ -0,0 +1,90 @@
+package commitgraph
+
+import (
+	"math"
+	"time"
+
+	"gopkg.in/src-d/go-git.v4/plumbing"
+	"gopkg.in/src-d/go-git.v4/plumbing/object"
+	"gopkg.in/src-d/go-git.v4/plumbing/storer"
+)
+
+// objectCommitNode is a representation of Commit as presented in the GIT object format.
+//
+// objectCommitNode implements the CommitNode interface.
+type objectCommitNode struct {
+	nodeIndex CommitNodeIndex
+	commit    *object.Commit
+}
+
+// NewObjectCommitNodeIndex returns CommitNodeIndex implementation that uses
+// only object storage to load the nodes
+func NewObjectCommitNodeIndex(s storer.EncodedObjectStorer) CommitNodeIndex {
+	return &objectCommitNodeIndex{s}
+}
+
+func (oci *objectCommitNodeIndex) Get(hash plumbing.Hash) (CommitNode, error) {
+	commit, err := object.GetCommit(oci.s, hash)
+	if err != nil {
+		return nil, err
+	}
+
+	return &objectCommitNode{
+		nodeIndex: oci,
+		commit:    commit,
+	}, nil
+}
+
+// objectCommitNodeIndex is an index that can load CommitNode objects only from the
+// object store.
+//
+// objectCommitNodeIndex implements the CommitNodeIndex interface
+type objectCommitNodeIndex struct {
+	s storer.EncodedObjectStorer
+}
+
+func (c *objectCommitNode) CommitTime() time.Time {
+	return c.commit.Committer.When
+}
+
+func (c *objectCommitNode) ID() plumbing.Hash {
+	return c.commit.ID()
+}
+
+func (c *objectCommitNode) Tree() (*object.Tree, error) {
+	return c.commit.Tree()
+}
+
+func (c *objectCommitNode) NumParents() int {
+	return c.commit.NumParents()
+}
+
+func (c *objectCommitNode) ParentNodes() CommitNodeIter {
+	return newParentgraphCommitNodeIter(c)
+}
+
+func (c *objectCommitNode) ParentNode(i int) (CommitNode, error) {
+	if i < 0 || i >= len(c.commit.ParentHashes) {
+		return nil, object.ErrParentNotFound
+	}
+
+	// Note: It's necessary to go through CommitNodeIndex here to ensure
+	// that if the commit-graph file covers only part of the history we
+	// start using it when that part is reached.
+	return c.nodeIndex.Get(c.commit.ParentHashes[i])
+}
+
+func (c *objectCommitNode) ParentHashes() []plumbing.Hash {
+	return c.commit.ParentHashes
+}
+
+func (c *objectCommitNode) Generation() uint64 {
+	// Commit nodes representing objects outside of the commit graph can never
+	// be reached by objects from the commit-graph thus we return the highest
+	// possible value.
+	return math.MaxUint64
+}
+
+func (c *objectCommitNode) Commit() (*object.Commit, error) {
+	return c.commit, nil
+}

vendor/gopkg.in/src-d/go-git.v4/plumbing/object/commitgraph/commitnode_walker_ctime.go

@@ -0,0 +1,105 @@
+package commitgraph
+
+import (
+	"io"
+
+	"github.com/emirpasic/gods/trees/binaryheap"
+
+	"gopkg.in/src-d/go-git.v4/plumbing"
+	"gopkg.in/src-d/go-git.v4/plumbing/storer"
+)
+
+type commitNodeIteratorByCTime struct {
+	heap         *binaryheap.Heap
+	seenExternal map[plumbing.Hash]bool
+	seen         map[plumbing.Hash]bool
+}
+
+// NewCommitNodeIterCTime returns a CommitNodeIter that walks the commit history,
+// starting at the given commit and visiting its parents while preserving Committer Time order.
+// this appears to be the closest order to `git log`
+// The given callback will be called for each visited commit. Each commit will
+// be visited only once. If the callback returns an error, walking will stop
+// and will return the error. Other errors might be returned if the history
+// cannot be traversed (e.g. missing objects). Ignore allows to skip some
+// commits from being iterated.
+func NewCommitNodeIterCTime(
+	c CommitNode,
+	seenExternal map[plumbing.Hash]bool,
+	ignore []plumbing.Hash,
+) CommitNodeIter {
+	seen := make(map[plumbing.Hash]bool)
+	for _, h := range ignore {
+		seen[h] = true
+	}
+
+	heap := binaryheap.NewWith(func(a, b interface{}) int {
+		if a.(CommitNode).CommitTime().Before(b.(CommitNode).CommitTime()) {
+			return 1
+		}
+		return -1
+	})
+
+	heap.Push(c)
+
+	return &commitNodeIteratorByCTime{
+		heap:         heap,
+		seenExternal: seenExternal,
+		seen:         seen,
+	}
+}
+
+func (w *commitNodeIteratorByCTime) Next() (CommitNode, error) {
+	var c CommitNode
+	for {
+		cIn, ok := w.heap.Pop()
+		if !ok {
+			return nil, io.EOF
+		}
+		c = cIn.(CommitNode)
+		cID := c.ID()
+
+		if w.seen[cID] || w.seenExternal[cID] {
+			continue
+		}
+
+		w.seen[cID] = true
+
+		for i, h := range c.ParentHashes() {
+			if w.seen[h] || w.seenExternal[h] {
+				continue
+			}
+			pc, err := c.ParentNode(i)
+			if err != nil {
+				return nil, err
+			}
+			w.heap.Push(pc)
+		}
+
+		return c, nil
+	}
+}
+
+func (w *commitNodeIteratorByCTime) ForEach(cb func(CommitNode) error) error {
+	for {
+		c, err := w.Next()
+		if err == io.EOF {
+			break
+		}
+		if err != nil {
+			return err
+		}
+
+		err = cb(c)
+		if err == storer.ErrStop {
+			break
+		}
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func (w *commitNodeIteratorByCTime) Close() {}

vendor/gopkg.in/src-d/go-git.v4/plumbing/object/commitgraph/doc.go

@@ -0,0 +1,7 @@
+// Package commitgraph provides an interface for efficient traversal over Git
+// commit graph either through the regular object storage, or optionally with
+// the index stored in commit-graph file (Git 2.18+).
+//
+// The API and functionality of this package are considered EXPERIMENTAL and is
+// not considered stable nor production ready.
+package commitgraph