File and Blob objects

6 files changed, 389 insertions, 0 deletions

diff --git a/objects/object.go b/objects/object.go
index fbe1f98..ced78bd 100644
--- a/objects/object.go
+++ b/objects/object.go
@@ -112,3 +112,9 @@ func Unserialize(r io.Reader) (RawObject, error) {
 
 	return o, nil
 }
+
+type Object interface {
+	Type() ObjectType
+	Payload() []byte
+	FromPayload([]byte) error
+}
diff --git a/objects/object_blob.go b/objects/object_blob.go
new file mode 100644
index 0000000..c35cb2a
--- /dev/null
+++ b/objects/object_blob.go
@@ -0,0 +1,17 @@
+package objects
+
+type Blob []byte
+
+func (b Blob) Type() ObjectType {
+	return OTBlob
+}
+
+func (b Blob) Payload() []byte {
+	return []byte(b)
+}
+
+func (b *Blob) FromPayload(bytes []byte) error {
+	// TODO: perhaps it is better to copy the bytes?
+	*b = bytes
+	return nil
+}
diff --git a/objects/object_file.go b/objects/object_file.go
new file mode 100644
index 0000000..7551193
--- /dev/null
+++ b/objects/object_file.go
@@ -0,0 +1,103 @@
+package objects
+
+import (
+	"bufio"
+	"bytes"
+	"errors"
+	"strconv"
+)
+
+type FileFragment struct {
+	Blob ObjectId
+	Size uint64
+}
+
+func (ff FileFragment) toProperties() properties {
+	return properties{"blob": ff.Blob.String(), "size": strconv.FormatUint(ff.Size, 10)}
+}
+
+func (ff *FileFragment) fromProperties(p properties) error {
+	blob, ok := p["blob"]
+	if !ok {
+		return errors.New("Field `blob` is missing")
+	}
+
+	var err error
+	ff.Blob, err = ParseObjectId(blob)
+	if err != nil {
+		return err
+	}
+
+	size, ok := p["size"]
+	if !ok {
+		return errors.New("Field `size` is missing")
+	}
+
+	ff.Size, err = strconv.ParseUint(size, 10, 64)
+	return err
+}
+
+func (a FileFragment) Equals(b FileFragment) bool {
+	return a.Blob.Equals(b.Blob) && a.Size == b.Size
+}
+
+type File []FileFragment
+
+func (f File) Type() ObjectType {
+	return OTFile
+}
+
+func (f File) Payload() []byte {
+	out := []byte{}
+
+	for _, ff := range f {
+		b, err := ff.toProperties().MarshalText()
+		if err != nil {
+			panic(err)
+		}
+
+		out = append(out, b...)
+		out = append(out, '\n')
+	}
+
+	return out
+}
+
+func (f *File) FromPayload(payload []byte) error {
+	sc := bufio.NewScanner(bytes.NewReader(payload))
+
+	for sc.Scan() {
+		line := sc.Bytes()
+		if len(line) == 0 {
+			continue
+		}
+
+		props := make(properties)
+		if err := props.UnmarshalText(line); err != nil {
+			return nil
+		}
+
+		ff := FileFragment{}
+		if err := ff.fromProperties(props); err != nil {
+			return err
+		}
+
+		*f = append(*f, ff)
+	}
+
+	return sc.Err()
+}
+
+func (a File) Equals(b File) bool {
+	if len(a) != len(b) {
+		return false
+	}
+
+	for i := range a {
+		if !a[i].Equals(b[i]) {
+			return false
+		}
+	}
+
+	return true
+}
diff --git a/objects/object_file_test.go b/objects/object_file_test.go
new file mode 100644
index 0000000..f1cb584
--- /dev/null
+++ b/objects/object_file_test.go
@@ -0,0 +1,98 @@
+package objects
+
+import (
+	"bytes"
+	"testing"
+)
+
+var (
+	testFileObj = File{
+		FileFragment{Blob: genId(0x11), Size: 10},
+		FileFragment{Blob: genId(0x22), Size: 20},
+		FileFragment{Blob: genId(0x33), Size: 30},
+		FileFragment{Blob: genId(0x44), Size: 40},
+		FileFragment{Blob: genId(0x55), Size: 50},
+	}
+
+	testFileSerialization = []byte("" +
+		"blob=sha3-256:1111111111111111111111111111111111111111111111111111111111111111&size=10\n" +
+		"blob=sha3-256:2222222222222222222222222222222222222222222222222222222222222222&size=20\n" +
+		"blob=sha3-256:3333333333333333333333333333333333333333333333333333333333333333&size=30\n" +
+		"blob=sha3-256:4444444444444444444444444444444444444444444444444444444444444444&size=40\n" +
+		"blob=sha3-256:5555555555555555555555555555555555555555555555555555555555555555&size=50\n")
+)
+
+func genId(b byte) (oid ObjectId) {
+	oid.Algo = OIdAlgoSHA3_256
+	oid.Sum = make([]byte, OIdAlgoSHA3_256.sumLength())
+	for i := 0; i < OIdAlgoSHA3_256.sumLength(); i++ {
+		oid.Sum[i] = b
+	}
+
+	return
+}
+
+func TestSerializeFile(t *testing.T) {
+	have := testFileObj.Payload()
+
+	if !bytes.Equal(have, testFileSerialization) {
+		t.Errorf("Unexpected serialization result: %s", have)
+	}
+}
+
+func TestSerializeEmptyFile(t *testing.T) {
+	f := File{}
+
+	have := f.Payload()
+	want := []byte{}
+
+	if !bytes.Equal(have, want) {
+		t.Errorf("Unexpected serialization result: %s", have)
+	}
+}
+
+func TestUnserializeFile(t *testing.T) {
+	have := File{}
+	err := have.FromPayload(testFileSerialization)
+
+	if err != nil {
+		t.Fatalf("Unexpected error: %s", err)
+	}
+
+	if !have.Equals(testFileObj) {
+		t.Errorf("Unexpeced unserialization result: %v", have)
+	}
+}
+
+func TestUnserializeEmptyFile(t *testing.T) {
+	have := File{}
+	err := have.FromPayload([]byte{})
+
+	if err != nil {
+		t.Fatalf("Unexpected error: %s", err)
+	}
+
+	if len(have) != 0 {
+		t.Errorf("Unexpeced unserialization result: %v", have)
+	}
+}
+
+func TestUnserializeFailure(t *testing.T) {
+	subtests := []struct{ name, payload string }{
+		{"missing blob", "size=100\n"},
+		{"empty blob", "blob=&size=100"},
+		{"invalid blob", "blob=foobar&size=100"}, // Variations of invalid IDs are tested elsewhere
+		{"missing size", "blob=sha3-256:0000000000000000000000000000000000000000000000000000000000000000\n"},
+		{"empty size", "blob=sha3-256:0000000000000000000000000000000000000000000000000000000000000000&size=\n"},
+		{"invalid size", "blob=sha3-256:0000000000000000000000000000000000000000000000000000000000000000&size=foobar\n"},
+		{"no props", "foobar\n"},
+	}
+
+	for _, subtest := range subtests {
+		have := File{}
+		err := have.FromPayload([]byte(subtest.payload))
+		if err == nil {
+			t.Errorf("Unexpected unserialization success: %v", have)
+		}
+	}
+}
diff --git a/objects/properties.go b/objects/properties.go
new file mode 100644
index 0000000..6b4acc1
--- /dev/null
+++ b/objects/properties.go
@@ -0,0 +1,93 @@
+package objects
+
+import (
+	"encoding/hex"
+	"fmt"
+	"net/url"
+	"sort"
+)
+
+// properties are mappings from strings to strings that are encoded as a restricted version of URL query strings
+// (only the characters [a-zA-Z0-9.:_-] are allowed, values are ordered by their key)
+type properties map[string]string
+
+// escapePropertyString escapes all bytes not in [a-zA-Z0-9.:_-] as %XX, where XX represents the hexadecimal value of the byte.
+// Compatible with URL query strings
+func escapePropertyString(s string) []byte {
+	out := []byte{}
+	esc := []byte("%XX")
+
+	for _, b := range []byte(s) {
+		if (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z') || (b >= '0' && b <= '9') || b == '.' || b == ':' || b == '_' || b == '-' {
+			out = append(out, b)
+		} else {
+			hex.Encode(esc[1:], []byte{b})
+			out = append(out, esc...)
+		}
+	}
+
+	return out
+}
+
+func (p properties) MarshalText() ([]byte, error) { // Guaranteed to not fail, error is only here to satisfy encoding.TextMarshaler
+	keys := make([]string, len(p))
+	i := 0
+	for k := range p {
+		keys[i] = k
+		i++
+	}
+
+	sort.Strings(keys)
+
+	out := []byte{}
+
+	first := true
+	for _, k := range keys {
+		if first {
+			first = false
+		} else {
+			out = append(out, '&')
+		}
+
+		out = append(out, escapePropertyString(k)...)
+		out = append(out, '=')
+		out = append(out, escapePropertyString(p[k])...)
+	}
+
+	return out, nil
+}
+
+func (p properties) UnmarshalText(text []byte) error {
+	vals, err := url.ParseQuery(string(text))
+	if err != nil {
+		return err
+	}
+
+	for k, v := range vals {
+		if len(v) != 1 {
+			return fmt.Errorf("Got %d values for key %s, expected 1", len(v), k)
+		}
+
+		p[k] = v[0]
+	}
+
+	return nil
+}
+
+func (a properties) Equals(b properties) bool {
+	for k, va := range a {
+		vb, ok := b[k]
+		if !ok || vb != va {
+			return false
+		}
+	}
+
+	for k := range b {
+		_, ok := a[k]
+		if !ok {
+			return false
+		}
+	}
+
+	return true
+}
diff --git a/objects/properties_test.go b/objects/properties_test.go
new file mode 100644
index 0000000..96df5c8
--- /dev/null
+++ b/objects/properties_test.go
@@ -0,0 +1,72 @@
+package objects
+
+import (
+	"testing"
+)
+
+func TestPropEscape(t *testing.T) {
+	tests := []struct{ name, in, want string }{
+		{"empty", "", ""},
+		{"no escape", "foo:bar_BAZ-123", "foo:bar_BAZ-123"},
+		{"reserved chars", "foo=bar%baz%%=", "foo%3dbar%25baz%25%25%3d"},
+	}
+
+	for _, subtest := range tests {
+		have := string(escapePropertyString(subtest.in))
+		if have != subtest.want {
+			t.Errorf("%s: want: '%s', have: '%s'", subtest.name, subtest.want, have)
+		}
+	}
+}
+
+func TestPropertyMarshalling(t *testing.T) {
+	tests := []struct {
+		name string
+		in   properties
+		want string
+	}{
+		{"empty", properties{}, ""},
+		{"single", properties{"foo": "bar"}, "foo=bar"},
+		{"simple", properties{"foo": "bar", "bar": "baz"}, "bar=baz&foo=bar"},
+		{"escapes", properties{"foo&bar": "%=baz", "?": "!"}, "%3f=%21&foo%26bar=%25%3dbaz"},
+	}
+
+	for _, subtest := range tests {
+		have, err := subtest.in.MarshalText()
+		if err != nil {
+			t.Errorf("%s: Got an error: %s", err)
+			continue
+		}
+
+		if string(have) != subtest.want {
+			t.Errorf("%s: want: '%s', have: '%s'", subtest.name, subtest.want, have)
+		}
+	}
+}
+
+func TestPropertyUnmarshalling(t *testing.T) {
+	tests := []struct {
+		name string
+		in   string
+		want properties
+	}{
+		{"empty", "", properties{}},
+		{"single", "foo=bar", properties{"foo": "bar"}},
+		{"simple", "bar=baz&foo=bar", properties{"foo": "bar", "bar": "baz"}},
+		{"escapes", "%3f=%21&foo%26bar=%25%3dbaz", properties{"foo&bar": "%=baz", "?": "!"}},
+	}
+
+	for _, subtest := range tests {
+		have := make(properties)
+		err := have.UnmarshalText([]byte(subtest.in))
+
+		if err != nil {
+			t.Errorf("%s: Got an error: %s", err)
+			continue
+		}
+
+		if !have.Equals(subtest.want) {
+			t.Errorf("%s: want: '%v', have: '%v'", subtest.name, subtest.want, have)
+		}
+	}
+}