Separate out a counts package

Author: mhagger

sizes/count.go counts/counts.gosizes/count.go renamed to counts/counts.go

@@ -1,4 +1,4 @@
-package sizes
+package counts
 
 import (
 	"math"

counts/human.go

@@ -0,0 +1,89 @@
+package counts
+
+import (
+	"fmt"
+	"math"
+)
+
+type Prefix struct {
+	Name       string
+	Multiplier uint64
+}
+
+type Humaner interface {
+	Human([]Prefix, string) (string, string)
+	ToUint64() uint64
+}
+
+var MetricPrefixes []Prefix
+
+func init() {
+	MetricPrefixes = []Prefix{
+		{"", 1},
+		{"k", 1e3},
+		{"M", 1e6},
+		{"G", 1e9},
+		{"T", 1e12},
+		{"P", 1e15},
+	}
+}
+
+var BinaryPrefixes []Prefix
+
+func init() {
+	BinaryPrefixes = []Prefix{
+		{"", 1 << (10 * 0)},
+		{"Ki", 1 << (10 * 1)},
+		{"Mi", 1 << (10 * 2)},
+		{"Gi", 1 << (10 * 3)},
+		{"Ti", 1 << (10 * 4)},
+		{"Pi", 1 << (10 * 5)},
+	}
+}
+
+// Format values, aligned, in `len(unit) + 10` or fewer characters
+// (except for extremely large numbers).
+func Human(n uint64, prefixes []Prefix, unit string) (string, string) {
+	prefix := prefixes[0]
+	wholePart := n
+	for _, p := range prefixes {
+		w := n / p.Multiplier
+		if w >= 1 {
+			wholePart = w
+			prefix = p
+		}
+	}
+
+	if prefix.Multiplier == 1 {
+		return fmt.Sprintf("%d", n), unit
+	} else {
+		mantissa := float64(n) / float64(prefix.Multiplier)
+		var format string
+
+		if wholePart >= 100 {
+			// `mantissa` can actually be up to 1023.999.
+			format = "%.0f"
+		} else if wholePart >= 10 {
+			format = "%.1f"
+		} else {
+			format = "%.2f"
+		}
+		return fmt.Sprintf(format, mantissa), prefix.Name + unit
+	}
+}
+
+func (n Count32) Human(prefixes []Prefix, unit string) (string, string) {
+	if n == math.MaxUint32 {
+		return "∞", ""
+	} else {
+		return Human(uint64(n), prefixes, unit)
+	}
+}
+
+func (n Count64) Human(prefixes []Prefix, unit string) (string, string) {
+	if n == math.MaxUint64 {
+		return "∞", unit
+	} else {
+		return Human(uint64(n), prefixes, unit)
+	}
+}

sizes/git.go

@@ -12,6 +12,8 @@ import (
 	"path/filepath"
 	"strconv"
 	"strings"
+
+	"github.com/github/git-sizer/counts"
 )
 
 // The type of an object ("blob", "tree", "commit", "tag", "missing").
@@ -97,7 +99,7 @@ func (repo *Repository) Close() error {
 type Reference struct {
 	Refname    string
 	ObjectType ObjectType
-	ObjectSize Count32
+	ObjectSize counts.Count32
 	Oid        Oid
 }
 
@@ -162,7 +164,7 @@ func (iter *ReferenceIter) Next() (Reference, bool, error) {
 	return Reference{
 		Refname:    refname,
 		ObjectType: objectType,
-		ObjectSize: Count32(objectSize),
+		ObjectSize: counts.Count32(objectSize),
 		Oid:        oid,
 	}, true, nil
 }
@@ -215,7 +217,7 @@ func (repo *Repository) NewBatchObjectIter() (*BatchObjectIter, io.WriteCloser,
 	}, in, nil
 }
 
-func (iter *BatchObjectIter) Next() (Oid, ObjectType, Count32, []byte, error) {
+func (iter *BatchObjectIter) Next() (Oid, ObjectType, counts.Count32, []byte, error) {
 	header, err := iter.f.ReadString('\n')
 	if err != nil {
 		return Oid{}, "", 0, nil, err
@@ -321,7 +323,7 @@ func NotFilter(filter ReferenceFilter) ReferenceFilter {
 
 // Parse a `cat-file --batch[-check]` output header line (including
 // the trailing LF). `spec`, if not "", is used in error messages.
-func parseBatchHeader(spec string, header string) (Oid, ObjectType, Count32, error) {
+func parseBatchHeader(spec string, header string) (Oid, ObjectType, counts.Count32, error) {
 	header = header[:len(header)-1]
 	words := strings.Split(header, " ")
 	if words[len(words)-1] == "missing" {
@@ -340,7 +342,7 @@ func parseBatchHeader(spec string, header string) (Oid, ObjectType, Count32, err
 	if err != nil {
 		return Oid{}, "missing", 0, err
 	}
-	return oid, ObjectType(words[1]), NewCount32(size), nil
+	return oid, ObjectType(words[1]), counts.NewCount32(size), nil
 }
 
 type ObjectIter struct {
@@ -443,7 +445,7 @@ func (repo *Repository) NewObjectIter(args ...string) (
 }
 
 // Next returns the next object, or EOF when done.
-func (l *ObjectIter) Next() (Oid, ObjectType, Count32, error) {
+func (l *ObjectIter) Next() (Oid, ObjectType, counts.Count32, error) {
 	line, err := l.f.ReadString('\n')
 	if err != nil {
 		return Oid{}, "", 0, err
@@ -508,7 +510,7 @@ func (iter *ObjectHeaderIter) Next() (string, string, error) {
 }
 
 type Commit struct {
-	Size    Count32
+	Size    counts.Count32
 	Parents []Oid
 	Tree    Oid
 }
@@ -548,7 +550,7 @@ func ParseCommit(oid Oid, data []byte) (*Commit, error) {
 		return nil, fmt.Errorf("no tree found in commit %s", oid)
 	}
 	return &Commit{
-		Size:    NewCount32(uint64(len(data))),
+		Size:    counts.NewCount32(uint64(len(data))),
 		Parents: parents,
 		Tree:    tree,
 	}, nil
@@ -619,7 +621,7 @@ func (iter *TreeIter) NextEntry() (TreeEntry, bool, error) {
 }
 
 type Tag struct {
-	Size         Count32
+	Size         counts.Count32
 	Referent     Oid
 	ReferentType ObjectType
 }
@@ -663,7 +665,7 @@ func ParseTag(oid Oid, data []byte) (*Tag, error) {
 		return nil, fmt.Errorf("no type found in tag %s", oid)
 	}
 	return &Tag{
-		Size:         NewCount32(uint64(len(data))),
+		Size:         counts.NewCount32(uint64(len(data))),
 		Referent:     referent,
 		ReferentType: referentType,
 	}, nil

sizes/graph.go

@@ -8,6 +8,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/github/git-sizer/counts"
 	"github.com/github/git-sizer/meter"
 )
 
@@ -82,7 +83,7 @@ func ScanRepositoryUsingGraph(
 
 	type ObjectHeader struct {
 		oid        Oid
-		objectSize Count32
+		objectSize counts.Count32
 	}
 
 	type CommitHeader struct {
@@ -399,7 +400,7 @@ func (g *Graph) HistorySize() HistorySize {
 }
 
 // Record that the specified `oid` is a blob with the specified size.
-func (g *Graph) RegisterBlob(oid Oid, objectSize Count32) {
+func (g *Graph) RegisterBlob(oid Oid, objectSize counts.Count32) {
 	size := BlobSize{Size: objectSize}
 	// There are no listeners. Since this is a blob, we know all that
 	// we need to know about it. So skip the record and just fill in
@@ -485,7 +486,9 @@ func (g *Graph) RegisterTree(oid Oid, tree *Tree) error {
 	return record.initialize(g, oid, tree)
 }
 
-func (g *Graph) finalizeTreeSize(oid Oid, size TreeSize, objectSize Count32, treeEntries Count32) {
+func (g *Graph) finalizeTreeSize(
+	oid Oid, size TreeSize, objectSize counts.Count32, treeEntries counts.Count32,
+) {
 	g.treeLock.Lock()
 	g.treeSizes[oid] = size
 	delete(g.treeRecords, oid)
@@ -504,11 +507,11 @@ type treeRecord struct {
 
 	// The size of this object, in bytes. Initialized iff pending !=
 	// -1.
-	objectSize Count32
+	objectSize counts.Count32
 
 	// The number of entries directly in this tree. Initialized iff
 	// pending != -1.
-	entryCount Count32
+	entryCount counts.Count32
 
 	// The size of the items we know so far:
 	size TreeSize
@@ -537,7 +540,7 @@ func (r *treeRecord) initialize(g *Graph, oid Oid, tree *Tree) error {
 	r.lock.Lock()
 	defer r.lock.Unlock()
 
-	r.objectSize = NewCount32(uint64(len(tree.data)))
+	r.objectSize = counts.NewCount32(uint64(len(tree.data)))
 	r.pending = 0
 
 	iter := tree.Iter()
@@ -642,7 +645,7 @@ func (g *Graph) RegisterCommit(oid Oid, commit *Commit) {
 	g.commitLock.Unlock()
 
 	// The number of direct parents of this commit.
-	parentCount := NewCount32(uint64(len(commit.Parents)))
+	parentCount := counts.NewCount32(uint64(len(commit.Parents)))
 
 	// The size of the items we know so far:
 	size := CommitSize{}
@@ -711,7 +714,7 @@ func (g *Graph) RegisterTag(oid Oid, tag *Tag) {
 	record.initialize(g, oid, tag)
 }
 
-func (g *Graph) finalizeTagSize(oid Oid, size TagSize, objectSize Count32) {
+func (g *Graph) finalizeTagSize(oid Oid, size TagSize, objectSize counts.Count32) {
 	g.tagLock.Lock()
 	g.tagSizes[oid] = size
 	delete(g.tagRecords, oid)
@@ -729,7 +732,7 @@ type tagRecord struct {
 	lock sync.Mutex
 
 	// The size of this commit object in bytes.
-	objectSize Count32
+	objectSize counts.Count32
 
 	// The size of the items we know so far:
 	size TagSize