Imported Upstream version 2011.01.12upstream/2011.01.12

6 files changed, 770 insertions, 376 deletions

diff --git a/src/pkg/bytes/Makefile b/src/pkg/bytes/Makefile
index d50e624d6..03395c7a4 100644
--- a/src/pkg/bytes/Makefile
+++ b/src/pkg/bytes/Makefile
@@ -2,7 +2,7 @@
 # Use of this source code is governed by a BSD-style
 # license that can be found in the LICENSE file.
 
-include ../../Make.$(GOARCH)
+include ../../Make.inc
 
 TARG=bytes
 GOFILES=\
diff --git a/src/pkg/bytes/asm_amd64.s b/src/pkg/bytes/asm_amd64.s
index 7e78700ec..c6793cbdc 100644
--- a/src/pkg/bytes/asm_amd64.s
+++ b/src/pkg/bytes/asm_amd64.s
@@ -3,15 +3,90 @@
 // license that can be found in the LICENSE file.
 
 TEXT ·IndexByte(SB),7,$0
-	MOVQ	p+0(FP), SI
-	MOVL	len+8(FP), CX
-	MOVB	b+16(FP), AL
-	MOVQ	SI, DI
+	MOVQ p+0(FP), SI
+	MOVL len+8(FP), BX
+	MOVB b+16(FP), AL
+	MOVQ SI, DI
+
+	CMPL BX, $16
+	JLT small
+
+	// round up to first 16-byte boundary
+	TESTQ $15, SI
+	JZ aligned
+	MOVQ SI, CX
+	ANDQ $~15, CX
+	ADDQ $16, CX
+
+	// search the beginning
+	SUBQ SI, CX
+	REPN; SCASB
+	JZ success
+
+// DI is 16-byte aligned; get ready to search using SSE instructions
+aligned:
+	// round down to last 16-byte boundary
+	MOVQ BX, R11
+	ADDQ SI, R11
+	ANDQ $~15, R11
+
+	// shuffle X0 around so that each byte contains c
+	MOVD AX, X0
+	PUNPCKLBW X0, X0
+	PUNPCKLBW X0, X0
+	PSHUFL $0, X0, X0
+	JMP condition
+
+sse:
+	// move the next 16-byte chunk of the buffer into X1
+	MOVO (DI), X1
+	// compare bytes in X0 to X1
+	PCMPEQB X0, X1
+	// take the top bit of each byte in X1 and put the result in DX
+	PMOVMSKB X1, DX
+	TESTL DX, DX
+	JNZ ssesuccess
+	ADDQ $16, DI
+
+condition:
+	CMPQ DI, R11
+	JLT sse
+
+	// search the end
+	MOVQ SI, CX
+	ADDQ BX, CX
+	SUBQ R11, CX
+	// if CX == 0, the zero flag will be set and we'll end up
+	// returning a false success
+	JZ failure
 	REPN; SCASB
-	JZ 3(PC)
-	MOVL	$-1, ret+24(FP)
+	JZ success
+
+failure:
+	MOVL $-1, ret+24(FP)
+	RET
+
+// handle for lengths < 16
+small:
+	MOVL BX, CX
+	REPN; SCASB
+	JZ success
+	MOVL $-1, ret+24(FP)
 	RET
-	SUBQ	SI, DI
-	SUBL	$1, DI
-	MOVL	DI, ret+24(FP)
+
+// we've found the chunk containing the byte
+// now just figure out which specific byte it is
+ssesuccess:
+	// get the index of the least significant set bit
+	BSFW DX, DX
+	SUBQ SI, DI
+	ADDQ DI, DX
+	MOVL DX, ret+24(FP)
+	RET
+
+success:
+	SUBQ SI, DI
+	SUBL $1, DI
+	MOVL DI, ret+24(FP)
 	RET
+
diff --git a/src/pkg/bytes/buffer.go b/src/pkg/bytes/buffer.go
index 01e6aef67..2574b4f43 100644
--- a/src/pkg/bytes/buffer.go
+++ b/src/pkg/bytes/buffer.go
@@ -12,14 +12,6 @@ import (
 	"utf8"
 )
 
-// Copy from string to byte array at offset doff.  Assume there's room.
-func copyString(dst []byte, doff int, str string) {
-	for soff := 0; soff < len(str); soff++ {
-		dst[doff] = str[soff]
-		doff++
-	}
-}
-
 // A Buffer is a variable-sized buffer of bytes with Read and Write methods.
 // The zero value for Buffer is an empty buffer ready to use.
 type Buffer struct {
@@ -27,8 +19,20 @@ type Buffer struct {
 	off       int               // read at &buf[off], write at &buf[len(buf)]
 	runeBytes [utf8.UTFMax]byte // avoid allocation of slice on each WriteByte or Rune
 	bootstrap [64]byte          // memory to hold first slice; helps small buffers (Printf) avoid allocation.
+	lastRead  readOp            // last read operation, so that Unread* can work correctly.
 }
 
+// The readOp constants describe the last action performed on
+// the buffer, so that UnreadRune and UnreadByte can
+// check for invalid usage.
+type readOp int
+
+const (
+	opInvalid  readOp = iota // Non-read operation.
+	opReadRune               // Read rune.
+	opRead                   // Any other read operation.
+)
+
 // Bytes returns a slice of the contents of the unread portion of the buffer;
 // len(b.Bytes()) == b.Len().  If the caller changes the contents of the
 // returned slice, the contents of the buffer will change provided there
@@ -52,6 +56,7 @@ func (b *Buffer) Len() int { return len(b.buf) - b.off }
 // Truncate discards all but the first n unread bytes from the buffer.
 // It is an error to call b.Truncate(n) with n > b.Len().
 func (b *Buffer) Truncate(n int) {
+	b.lastRead = opInvalid
 	if n == 0 {
 		// Reuse buffer space.
 		b.off = 0
@@ -90,6 +95,7 @@ func (b *Buffer) grow(n int) int {
 // Write appends the contents of p to the buffer.  The return
 // value n is the length of p; err is always nil.
 func (b *Buffer) Write(p []byte) (n int, err os.Error) {
+	b.lastRead = opInvalid
 	m := b.grow(len(p))
 	copy(b.buf[m:], p)
 	return len(p), nil
@@ -98,9 +104,9 @@ func (b *Buffer) Write(p []byte) (n int, err os.Error) {
 // WriteString appends the contents of s to the buffer.  The return
 // value n is the length of s; err is always nil.
 func (b *Buffer) WriteString(s string) (n int, err os.Error) {
+	b.lastRead = opInvalid
 	m := b.grow(len(s))
-	copyString(b.buf, m, s)
-	return len(s), nil
+	return copy(b.buf[m:], s), nil
 }
 
 // MinRead is the minimum slice size passed to a Read call by
@@ -114,6 +120,7 @@ const MinRead = 512
 // Any error except os.EOF encountered during the read
 // is also returned.
 func (b *Buffer) ReadFrom(r io.Reader) (n int64, err os.Error) {
+	b.lastRead = opInvalid
 	// If buffer is empty, reset to recover space.
 	if b.off >= len(b.buf) {
 		b.Truncate(0)
@@ -150,6 +157,7 @@ func (b *Buffer) ReadFrom(r io.Reader) (n int64, err os.Error) {
 // occurs. The return value n is the number of bytes written.
 // Any error encountered during the write is also returned.
 func (b *Buffer) WriteTo(w io.Writer) (n int64, err os.Error) {
+	b.lastRead = opInvalid
 	for b.off < len(b.buf) {
 		m, e := w.Write(b.buf[b.off:])
 		n += int64(m)
@@ -167,6 +175,7 @@ func (b *Buffer) WriteTo(w io.Writer) (n int64, err os.Error) {
 // The returned error is always nil, but is included
 // to match bufio.Writer's WriteByte.
 func (b *Buffer) WriteByte(c byte) os.Error {
+	b.lastRead = opInvalid
 	m := b.grow(1)
 	b.buf[m] = c
 	return nil
@@ -181,7 +190,7 @@ func (b *Buffer) WriteRune(r int) (n int, err os.Error) {
 		b.WriteByte(byte(r))
 		return 1, nil
 	}
-	n = utf8.EncodeRune(r, b.runeBytes[0:])
+	n = utf8.EncodeRune(b.runeBytes[0:], r)
 	b.Write(b.runeBytes[0:n])
 	return n, nil
 }
@@ -191,6 +200,7 @@ func (b *Buffer) WriteRune(r int) (n int, err os.Error) {
 // buffer has no data to return, err is os.EOF even if len(p) is zero;
 // otherwise it is nil.
 func (b *Buffer) Read(p []byte) (n int, err os.Error) {
+	b.lastRead = opInvalid
 	if b.off >= len(b.buf) {
 		// Buffer is empty, reset to recover space.
 		b.Truncate(0)
@@ -198,6 +208,9 @@ func (b *Buffer) Read(p []byte) (n int, err os.Error) {
 	}
 	n = copy(p, b.buf[b.off:])
 	b.off += n
+	if n > 0 {
+		b.lastRead = opRead
+	}
 	return
 }
 
@@ -206,18 +219,23 @@ func (b *Buffer) Read(p []byte) (n int, err os.Error) {
 // If there are fewer than n bytes in the buffer, Next returns the entire buffer.
 // The slice is only valid until the next call to a read or write method.
 func (b *Buffer) Next(n int) []byte {
+	b.lastRead = opInvalid
 	m := b.Len()
 	if n > m {
 		n = m
 	}
 	data := b.buf[b.off : b.off+n]
 	b.off += n
+	if n > 0 {
+		b.lastRead = opRead
+	}
 	return data
 }
 
 // ReadByte reads and returns the next byte from the buffer.
 // If no byte is available, it returns error os.EOF.
 func (b *Buffer) ReadByte() (c byte, err os.Error) {
+	b.lastRead = opInvalid
 	if b.off >= len(b.buf) {
 		// Buffer is empty, reset to recover space.
 		b.Truncate(0)
@@ -225,6 +243,7 @@ func (b *Buffer) ReadByte() (c byte, err os.Error) {
 	}
 	c = b.buf[b.off]
 	b.off++
+	b.lastRead = opRead
 	return c, nil
 }
 
@@ -234,11 +253,13 @@ func (b *Buffer) ReadByte() (c byte, err os.Error) {
 // If the bytes are an erroneous UTF-8 encoding, it
 // consumes one byte and returns U+FFFD, 1.
 func (b *Buffer) ReadRune() (r int, size int, err os.Error) {
+	b.lastRead = opInvalid
 	if b.off >= len(b.buf) {
 		// Buffer is empty, reset to recover space.
 		b.Truncate(0)
 		return 0, 0, os.EOF
 	}
+	b.lastRead = opReadRune
 	c := b.buf[b.off]
 	if c < utf8.RuneSelf {
 		b.off++
@@ -249,6 +270,37 @@ func (b *Buffer) ReadRune() (r int, size int, err os.Error) {
 	return r, n, nil
 }
 
+// UnreadRune unreads the last rune returned by ReadRune.
+// If the most recent read or write operation on the buffer was
+// not a ReadRune, UnreadRune returns an error.  (In this regard
+// it is stricter than UnreadByte, which will unread the last byte
+// from any read operation.)
+func (b *Buffer) UnreadRune() os.Error {
+	if b.lastRead != opReadRune {
+		return os.ErrorString("bytes.Buffer: UnreadRune: previous operation was not ReadRune")
+	}
+	b.lastRead = opInvalid
+	if b.off > 0 {
+		_, n := utf8.DecodeLastRune(b.buf[0:b.off])
+		b.off -= n
+	}
+	return nil
+}
+
+// UnreadByte unreads the last byte returned by the most recent
+// read operation.  If write has happened since the last read, UnreadByte
+// returns an error.
+func (b *Buffer) UnreadByte() os.Error {
+	if b.lastRead != opReadRune && b.lastRead != opRead {
+		return os.ErrorString("bytes.Buffer: UnreadByte: previous operation was not a read")
+	}
+	b.lastRead = opInvalid
+	if b.off > 0 {
+		b.off--
+	}
+	return nil
+}
+
 // NewBuffer creates and initializes a new Buffer using buf as its initial
 // contents.  It is intended to prepare a Buffer to read existing data.  It
 // can also be used to to size the internal buffer for writing.  To do that,
@@ -259,7 +311,5 @@ func NewBuffer(buf []byte) *Buffer { return &Buffer{buf: buf} }
 // initial contents.  It is intended to prepare a buffer to read an existing
 // string.
 func NewBufferString(s string) *Buffer {
-	buf := make([]byte, len(s))
-	copyString(buf, 0, s)
-	return &Buffer{buf: buf}
+	return &Buffer{buf: []byte(s)}
 }
diff --git a/src/pkg/bytes/buffer_test.go b/src/pkg/bytes/buffer_test.go
index bc696f4b5..509793d24 100644
--- a/src/pkg/bytes/buffer_test.go
+++ b/src/pkg/bytes/buffer_test.go
@@ -30,19 +30,19 @@ func check(t *testing.T, testname string, buf *Buffer, s string) {
 	bytes := buf.Bytes()
 	str := buf.String()
 	if buf.Len() != len(bytes) {
-		t.Errorf("%s: buf.Len() == %d, len(buf.Bytes()) == %d\n", testname, buf.Len(), len(bytes))
+		t.Errorf("%s: buf.Len() == %d, len(buf.Bytes()) == %d", testname, buf.Len(), len(bytes))
 	}
 
 	if buf.Len() != len(str) {
-		t.Errorf("%s: buf.Len() == %d, len(buf.String()) == %d\n", testname, buf.Len(), len(str))
+		t.Errorf("%s: buf.Len() == %d, len(buf.String()) == %d", testname, buf.Len(), len(str))
 	}
 
 	if buf.Len() != len(s) {
-		t.Errorf("%s: buf.Len() == %d, len(s) == %d\n", testname, buf.Len(), len(s))
+		t.Errorf("%s: buf.Len() == %d, len(s) == %d", testname, buf.Len(), len(s))
 	}
 
 	if string(bytes) != s {
-		t.Errorf("%s: string(buf.Bytes()) == %q, s == %q\n", testname, string(bytes), s)
+		t.Errorf("%s: string(buf.Bytes()) == %q, s == %q", testname, string(bytes), s)
 	}
 }
 
@@ -55,10 +55,10 @@ func fillString(t *testing.T, testname string, buf *Buffer, s string, n int, fus
 	for ; n > 0; n-- {
 		m, err := buf.WriteString(fus)
 		if m != len(fus) {
-			t.Errorf(testname+" (fill 2): m == %d, expected %d\n", m, len(fus))
+			t.Errorf(testname+" (fill 2): m == %d, expected %d", m, len(fus))
 		}
 		if err != nil {
-			t.Errorf(testname+" (fill 3): err should always be nil, found err == %s\n", err)
+			t.Errorf(testname+" (fill 3): err should always be nil, found err == %s", err)
 		}
 		s += fus
 		check(t, testname+" (fill 4)", buf, s)
@@ -75,10 +75,10 @@ func fillBytes(t *testing.T, testname string, buf *Buffer, s string, n int, fub
 	for ; n > 0; n-- {
 		m, err := buf.Write(fub)
 		if m != len(fub) {
-			t.Errorf(testname+" (fill 2): m == %d, expected %d\n", m, len(fub))
+			t.Errorf(testname+" (fill 2): m == %d, expected %d", m, len(fub))
 		}
 		if err != nil {
-			t.Errorf(testname+" (fill 3): err should always be nil, found err == %s\n", err)
+			t.Errorf(testname+" (fill 3): err should always be nil, found err == %s", err)
 		}
 		s += string(fub)
 		check(t, testname+" (fill 4)", buf, s)
@@ -110,7 +110,7 @@ func empty(t *testing.T, testname string, buf *Buffer, s string, fub []byte) {
 			break
 		}
 		if err != nil {
-			t.Errorf(testname+" (empty 2): err should always be nil, found err == %s\n", err)
+			t.Errorf(testname+" (empty 2): err should always be nil, found err == %s", err)
 		}
 		s = s[n:]
 		check(t, testname+" (empty 3)", buf, s)
@@ -132,21 +132,21 @@ func TestBasicOperations(t *testing.T) {
 		buf.Truncate(0)
 		check(t, "TestBasicOperations (3)", &buf, "")
 
-		n, err := buf.Write(Bytes(data[0:1]))
+		n, err := buf.Write([]byte(data[0:1]))
 		if n != 1 {
-			t.Errorf("wrote 1 byte, but n == %d\n", n)
+			t.Errorf("wrote 1 byte, but n == %d", n)
 		}
 		if err != nil {
-			t.Errorf("err should always be nil, but err == %s\n", err)
+			t.Errorf("err should always be nil, but err == %s", err)
 		}
 		check(t, "TestBasicOperations (4)", &buf, "a")
 
 		buf.WriteByte(data[1])
 		check(t, "TestBasicOperations (5)", &buf, "ab")
 
-		n, err = buf.Write(Bytes(data[2:26]))
+		n, err = buf.Write([]byte(data[2:26]))
 		if n != 24 {
-			t.Errorf("wrote 25 bytes, but n == %d\n", n)
+			t.Errorf("wrote 25 bytes, but n == %d", n)
 		}
 		check(t, "TestBasicOperations (6)", &buf, string(data[0:26]))
 
@@ -162,14 +162,14 @@ func TestBasicOperations(t *testing.T) {
 		buf.WriteByte(data[1])
 		c, err := buf.ReadByte()
 		if err != nil {
-			t.Errorf("ReadByte unexpected eof\n")
+			t.Error("ReadByte unexpected eof")
 		}
 		if c != data[1] {
-			t.Errorf("ReadByte wrong value c=%v\n", c)
+			t.Errorf("ReadByte wrong value c=%v", c)
 		}
 		c, err = buf.ReadByte()
 		if err == nil {
-			t.Errorf("ReadByte unexpected not eof\n")
+			t.Error("ReadByte unexpected not eof")
 		}
 	}
 }
@@ -238,7 +238,7 @@ func TestMixedReadsAndWrites(t *testing.T) {
 func TestNil(t *testing.T) {
 	var b *Buffer
 	if b.String() != "<nil>" {
-		t.Error("expcted <nil>; got %q", b.String())
+		t.Errorf("expcted <nil>; got %q", b.String())
 	}
 }
 
@@ -272,13 +272,13 @@ func TestRuneIO(t *testing.T) {
 	var buf Buffer
 	n := 0
 	for r := 0; r < NRune; r++ {
-		size := utf8.EncodeRune(r, b[n:])
+		size := utf8.EncodeRune(b[n:], r)
 		nbytes, err := buf.WriteRune(r)
 		if err != nil {
-			t.Fatalf("WriteRune(0x%x) error: %s", r, err)
+			t.Fatalf("WriteRune(%U) error: %s", r, err)
 		}
 		if nbytes != size {
-			t.Fatalf("WriteRune(0x%x) expected %d, got %d", r, size, nbytes)
+			t.Fatalf("WriteRune(%U) expected %d, got %d", r, size, nbytes)
 		}
 		n += size
 	}
@@ -289,12 +289,27 @@ func TestRuneIO(t *testing.T) {
 		t.Fatalf("incorrect result from WriteRune: %q not %q", buf.Bytes(), b)
 	}
 
+	p := make([]byte, utf8.UTFMax)
 	// Read it back with ReadRune
 	for r := 0; r < NRune; r++ {
-		size := utf8.EncodeRune(r, b)
+		size := utf8.EncodeRune(p, r)
 		nr, nbytes, err := buf.ReadRune()
 		if nr != r || nbytes != size || err != nil {
-			t.Fatalf("ReadRune(0x%x) got 0x%x,%d not 0x%x,%d (err=%s)", r, nr, nbytes, r, size, err)
+			t.Fatalf("ReadRune(%U) got %U,%d not %U,%d (err=%s)", r, nr, nbytes, r, size, err)
+		}
+	}
+
+	// Check that UnreadRune works
+	buf.Reset()
+	buf.Write(b)
+	for r := 0; r < NRune; r++ {
+		r1, size, _ := buf.ReadRune()
+		if err := buf.UnreadRune(); err != nil {
+			t.Fatalf("UnreadRune(%U) got error %q", r, err)
+		}
+		r2, nbytes, err := buf.ReadRune()
+		if r1 != r2 || r1 != r || nbytes != size || err != nil {
+			t.Fatalf("ReadRune(%U) after UnreadRune got %U,%d not %U,%d (err=%s)", r, r2, nbytes, r, size, err)
 		}
 	}
 }
diff --git a/src/pkg/bytes/bytes.go b/src/pkg/bytes/bytes.go
index bcf7b8609..bfe2ef39d 100644
--- a/src/pkg/bytes/bytes.go
+++ b/src/pkg/bytes/bytes.go
@@ -3,7 +3,7 @@
 // license that can be found in the LICENSE file.
 
 // The bytes package implements functions for the manipulation of byte slices.
-// Analagous to the facilities of the strings package.
+// Analogous to the facilities of the strings package.
 package bytes
 
 import (
@@ -127,7 +127,21 @@ func LastIndex(s, sep []byte) int {
 	return -1
 }
 
-// IndexAny interprets s as a sequence of UTF-8 encoded Unicode code points.
+// IndexRune interprets s as a sequence of UTF-8-encoded Unicode code points.
+// It returns the byte index of the first occurrence in s of the given rune.
+// It returns -1 if rune is not present in s.
+func IndexRune(s []byte, rune int) int {
+	for i := 0; i < len(s); {
+		r, size := utf8.DecodeRune(s[i:])
+		if r == rune {
+			return i
+		}
+		i += size
+	}
+	return -1
+}
+
+// IndexAny interprets s as a sequence of UTF-8-encoded Unicode code points.
 // It returns the byte index of the first occurrence in s of any of the Unicode
 // code points in chars.  It returns -1 if chars is empty or if there is no code
 // point in common.
@@ -151,6 +165,25 @@ func IndexAny(s []byte, chars string) int {
 	return -1
 }
 
+// LastIndexAny interprets s as a sequence of UTF-8-encoded Unicode code
+// points.  It returns the byte index of the last occurrence in s of any of
+// the Unicode code points in chars.  It returns -1 if chars is empty or if
+// there is no code point in common.
+func LastIndexAny(s []byte, chars string) int {
+	if len(chars) > 0 {
+		for i := len(s); i > 0; {
+			rune, size := utf8.DecodeLastRune(s[0:i])
+			i -= size
+			for _, m := range chars {
+				if rune == m {
+					return i
+				}
+			}
+		}
+	}
+	return -1
+}
+
 // Generic split: splits after each instance of sep,
 // including sepSave bytes of sep in the subarrays.
 func genSplit(s, sep []byte, sepSave, n int) [][]byte {
@@ -179,17 +212,22 @@ func genSplit(s, sep []byte, sepSave, n int) [][]byte {
 	return a[0 : na+1]
 }
 
-// Split splits the array s around each instance of sep, returning an array of subarrays of s.
-// If sep is empty, Split splits s after each UTF-8 sequence.
-// If n >= 0, Split splits s into at most n subarrays; the last subarray will contain an unsplit remainder.
-// Thus if n == 0, the result will ne nil.
+// Split slices s into subslices separated by sep and returns a slice of
+// the subslices between those separators.
+// If sep is empty, Split splits after each UTF-8 sequence.
+// The count determines the number of subslices to return:
+//   n > 0: at most n subslices; the last subslice will be the unsplit remainder.
+//   n == 0: the result is nil (zero subslices)
+//   n < 0: all subslices
 func Split(s, sep []byte, n int) [][]byte { return genSplit(s, sep, 0, n) }
 
-// SplitAfter splits the array s after each instance of sep, returning an array of subarrays of s.
-// If sep is empty, SplitAfter splits s after each UTF-8 sequence.
-// If n >= 0, SplitAfter splits s into at most n subarrays; the last subarray will contain an
-// unsplit remainder.
-// Thus if n == 0, the result will ne nil.
+// SplitAfter slices s into subslices after each instance of sep and
+// returns a slice of those subslices.
+// If sep is empty, Split splits after each UTF-8 sequence.
+// The count determines the number of subslices to return:
+//   n > 0: at most n subslices; the last subslice will be the unsplit remainder.
+//   n == 0: the result is nil (zero subslices)
+//   n < 0: all subslices
 func SplitAfter(s, sep []byte, n int) [][]byte {
 	return genSplit(s, sep, len(sep), n)
 }
@@ -197,12 +235,20 @@ func SplitAfter(s, sep []byte, n int) [][]byte {
 // Fields splits the array s around each instance of one or more consecutive white space
 // characters, returning a slice of subarrays of s or an empty list if s contains only white space.
 func Fields(s []byte) [][]byte {
+	return FieldsFunc(s, unicode.IsSpace)
+}
+
+// FieldsFunc interprets s as a sequence of UTF-8-encoded Unicode code points.
+// It splits the array s at each run of code points c satisfying f(c) and
+// returns a slice of subarrays of s.  If no code points in s satisfy f(c), an
+// empty slice is returned.
+func FieldsFunc(s []byte, f func(int) bool) [][]byte {
 	n := 0
 	inField := false
 	for i := 0; i < len(s); {
 		rune, size := utf8.DecodeRune(s[i:])
 		wasInField := inField
-		inField = !unicode.IsSpace(rune)
+		inField = !f(rune)
 		if inField && !wasInField {
 			n++
 		}
@@ -214,12 +260,12 @@ func Fields(s []byte) [][]byte {
 	fieldStart := -1
 	for i := 0; i <= len(s) && na < n; {
 		rune, size := utf8.DecodeRune(s[i:])
-		if fieldStart < 0 && size > 0 && !unicode.IsSpace(rune) {
+		if fieldStart < 0 && size > 0 && !f(rune) {
 			fieldStart = i
 			i += size
 			continue
 		}
-		if fieldStart >= 0 && (size == 0 || unicode.IsSpace(rune)) {
+		if fieldStart >= 0 && (size == 0 || f(rune)) {
 			a[na] = s[fieldStart:i]
 			na++
 			fieldStart = -1
@@ -278,7 +324,7 @@ func HasSuffix(s, suffix []byte) bool {
 // Map returns a copy of the byte array s with all its characters modified
 // according to the mapping function. If mapping returns a negative value, the character is
 // dropped from the string with no replacement.  The characters in s and the
-// output are interpreted as UTF-8 encoded Unicode code points.
+// output are interpreted as UTF-8-encoded Unicode code points.
 func Map(mapping func(rune int) int, s []byte) []byte {
 	// In the worst case, the array can grow when mapped, making
 	// things unpleasant.  But it's so rare we barge in assuming it's
@@ -298,12 +344,10 @@ func Map(mapping func(rune int) int, s []byte) []byte {
 				// Grow the buffer.
 				maxbytes = maxbytes*2 + utf8.UTFMax
 				nb := make([]byte, maxbytes)
-				for i, c := range b[0:nbytes] {
-					nb[i] = c
-				}
+				copy(nb, b[0:nbytes])
 				b = nb
 			}
-			nbytes += utf8.EncodeRune(rune, b[nbytes:maxbytes])
+			nbytes += utf8.EncodeRune(b[nbytes:maxbytes], rune)
 		}
 		i += wid
 	}
@@ -332,52 +376,147 @@ func ToLower(s []byte) []byte { return Map(unicode.ToLower, s) }
 // ToTitle returns a copy of the byte array s with all Unicode letters mapped to their title case.
 func ToTitle(s []byte) []byte { return Map(unicode.ToTitle, s) }
 
-// TrimLeftFunc returns a subslice of s by slicing off all leading UTF-8 encoded
+// ToUpperSpecial returns a copy of the byte array s with all Unicode letters mapped to their
+// upper case, giving priority to the special casing rules.
+func ToUpperSpecial(_case unicode.SpecialCase, s []byte) []byte {
+	return Map(func(r int) int { return _case.ToUpper(r) }, s)
+}
+
+// ToLowerSpecial returns a copy of the byte array s with all Unicode letters mapped to their
+// lower case, giving priority to the special casing rules.
+func ToLowerSpecial(_case unicode.SpecialCase, s []byte) []byte {
+	return Map(func(r int) int { return _case.ToLower(r) }, s)
+}
+
+// ToTitleSpecial returns a copy of the byte array s with all Unicode letters mapped to their
+// title case, giving priority to the special casing rules.
+func ToTitleSpecial(_case unicode.SpecialCase, s []byte) []byte {
+	return Map(func(r int) int { return _case.ToTitle(r) }, s)
+}
+
+
+// isSeparator reports whether the rune could mark a word boundary.
+// TODO: update when package unicode captures more of the properties.
+func isSeparator(rune int) bool {
+	// ASCII alphanumerics and underscore are not separators
+	if rune <= 0x7F {
+		switch {
+		case '0' <= rune && rune <= '9':
+			return false
+		case 'a' <= rune && rune <= 'z':
+			return false
+		case 'A' <= rune && rune <= 'Z':
+			return false
+		case rune == '_':
+			return false
+		}
+		return true
+	}
+	// Letters and digits are not separators
+	if unicode.IsLetter(rune) || unicode.IsDigit(rune) {
+		return false
+	}
+	// Otherwise, all we can do for now is treat spaces as separators.
+	return unicode.IsSpace(rune)
+}
+
+// BUG(r): The rule Title uses for word boundaries does not handle Unicode punctuation properly.
+
+// Title returns a copy of s with all Unicode letters that begin words
+// mapped to their title case.
+func Title(s []byte) []byte {
+	// Use a closure here to remember state.
+	// Hackish but effective. Depends on Map scanning in order and calling
+	// the closure once per rune.
+	prev := ' '
+	return Map(
+		func(r int) int {
+			if isSeparator(prev) {
+				prev = r
+				return unicode.ToTitle(r)
+			}
+			prev = r
+			return r
+		},
+		s)
+}
+
+// TrimLeftFunc returns a subslice of s by slicing off all leading UTF-8-encoded
 // Unicode code points c that satisfy f(c).
 func TrimLeftFunc(s []byte, f func(r int) bool) []byte {
-	var start, wid int
-	for start = 0; start < len(s); start += wid {
-		wid = 1
-		rune := int(s[start])
-		if rune >= utf8.RuneSelf {
-			rune, wid = utf8.DecodeRune(s[start:])
-		}
-		if !f(rune) {
-			break
-		}
+	i := indexFunc(s, f, false)
+	if i == -1 {
+		return nil
 	}
-	return s[start:]
+	return s[i:]
 }
 
 // TrimRightFunc returns a subslice of s by slicing off all trailing UTF-8
 // encoded Unicode code points c that satisfy f(c).
 func TrimRightFunc(s []byte, f func(r int) bool) []byte {
-	var end, wid int
-	for end = len(s); end > 0; end -= wid {
-		wid = 1
-		rune := int(s[end-wid])
-		if rune >= utf8.RuneSelf {
-			// Back up & look for beginning of rune. Mustn't pass start.
-			for wid = 2; end-wid >= 0 && !utf8.RuneStart(s[end-wid]); wid++ {
-			}
-			if end-wid < 0 { // invalid UTF-8 sequence; stop processing
-				break
-			}
-			rune, wid = utf8.DecodeRune(s[end-wid : end])
-		}
-		if !f(rune) {
-			break
-		}
+	i := lastIndexFunc(s, f, false)
+	if i >= 0 && s[i] >= utf8.RuneSelf {
+		_, wid := utf8.DecodeRune(s[i:])
+		i += wid
+	} else {
+		i++
 	}
-	return s[0:end]
+	return s[0:i]
 }
 
 // TrimFunc returns a subslice of s by slicing off all leading and trailing
-// UTF-8 encoded Unicode code points c that satisfy f(c).
+// UTF-8-encoded Unicode code points c that satisfy f(c).
 func TrimFunc(s []byte, f func(r int) bool) []byte {
 	return TrimRightFunc(TrimLeftFunc(s, f), f)
 }
 
+// IndexFunc interprets s as a sequence of UTF-8-encoded Unicode code points.
+// It returns the byte index in s of the first Unicode
+// code point satisfying f(c), or -1 if none do.
+func IndexFunc(s []byte, f func(r int) bool) int {
+	return indexFunc(s, f, true)
+}
+
+// LastIndexFunc interprets s as a sequence of UTF-8-encoded Unicode code points.
+// It returns the byte index in s of the last Unicode
+// code point satisfying f(c), or -1 if none do.
+func LastIndexFunc(s []byte, f func(r int) bool) int {
+	return lastIndexFunc(s, f, true)
+}
+
+// indexFunc is the same as IndexFunc except that if
+// truth==false, the sense of the predicate function is
+// inverted.
+func indexFunc(s []byte, f func(r int) bool, truth bool) int {
+	start := 0
+	for start < len(s) {
+		wid := 1
+		rune := int(s[start])
+		if rune >= utf8.RuneSelf {
+			rune, wid = utf8.DecodeRune(s[start:])
+		}
+		if f(rune) == truth {
+			return start
+		}
+		start += wid
+	}
+	return -1
+}
+
+// lastIndexFunc is the same as LastIndexFunc except that if
+// truth==false, the sense of the predicate function is
+// inverted.
+func lastIndexFunc(s []byte, f func(r int) bool, truth bool) int {
+	for i := len(s); i > 0; {
+		rune, size := utf8.DecodeLastRune(s[0:i])
+		i -= size
+		if f(rune) == truth {
+			return i
+		}
+	}
+	return -1
+}
+
 func makeCutsetFunc(cutset string) func(rune int) bool {
 	return func(rune int) bool {
 		for _, c := range cutset {
@@ -390,71 +529,29 @@ func makeCutsetFunc(cutset string) func(rune int) bool {
 }
 
 // Trim returns a subslice of s by slicing off all leading and
-// trailing UTF-8 encoded Unicode code points contained in cutset.
+// trailing UTF-8-encoded Unicode code points contained in cutset.
 func Trim(s []byte, cutset string) []byte {
 	return TrimFunc(s, makeCutsetFunc(cutset))
 }
 
 // TrimLeft returns a subslice of s by slicing off all leading
-// UTF-8 encoded Unicode code points contained in cutset.
+// UTF-8-encoded Unicode code points contained in cutset.
 func TrimLeft(s []byte, cutset string) []byte {
 	return TrimLeftFunc(s, makeCutsetFunc(cutset))
 }
 
 // TrimRight returns a subslice of s by slicing off all trailing
-// UTF-8 encoded Unicode code points that are contained in cutset.
+// UTF-8-encoded Unicode code points that are contained in cutset.
 func TrimRight(s []byte, cutset string) []byte {
 	return TrimRightFunc(s, makeCutsetFunc(cutset))
 }
 
 // TrimSpace returns a subslice of s by slicing off all leading and
-// trailing white space, as as defined by Unicode.
+// trailing white space, as defined by Unicode.
 func TrimSpace(s []byte) []byte {
 	return TrimFunc(s, unicode.IsSpace)
 }
 
-// How big to make a byte array when growing.
-// Heuristic: Scale by 50% to give n log n time.
-func resize(n int) int {
-	if n < 16 {
-		n = 16
-	}
-	return n + n/2
-}
-
-// Add appends the contents of t to the end of s and returns the result.
-// If s has enough capacity, it is extended in place; otherwise a
-// new array is allocated and returned.
-func Add(s, t []byte) []byte {
-	lens := len(s)
-	lent := len(t)
-	if lens+lent <= cap(s) {
-		s = s[0 : lens+lent]
-	} else {
-		news := make([]byte, lens+lent, resize(lens+lent))
-		copy(news, s)
-		s = news
-	}
-	copy(s[lens:lens+lent], t)
-	return s
-}
-
-// AddByte appends byte b to the end of s and returns the result.
-// If s has enough capacity, it is extended in place; otherwise a
-// new array is allocated and returned.
-func AddByte(s []byte, t byte) []byte {
-	lens := len(s)
-	if lens+1 <= cap(s) {
-		s = s[0 : lens+1]
-	} else {
-		news := make([]byte, lens+1, resize(lens+1))
-		copy(news, s)
-		s = news
-	}
-	s[lens] = t
-	return s
-}
-
 // Runes returns a slice of runes (Unicode code points) equivalent to s.
 func Runes(s []byte) []int {
 	t := make([]int, utf8.RuneCount(s))
diff --git a/src/pkg/bytes/bytes_test.go b/src/pkg/bytes/bytes_test.go
index 8197543dc..063686ec5 100644
--- a/src/pkg/bytes/bytes_test.go
+++ b/src/pkg/bytes/bytes_test.go
@@ -8,6 +8,7 @@ import (
 	. "bytes"
 	"testing"
 	"unicode"
+	"utf8"
 )
 
 func eq(a, b []string) bool {
@@ -45,16 +46,16 @@ type BinOpTest struct {
 }
 
 var comparetests = []BinOpTest{
-	BinOpTest{"", "", 0},
-	BinOpTest{"a", "", 1},
-	BinOpTest{"", "a", -1},
-	BinOpTest{"abc", "abc", 0},
-	BinOpTest{"ab", "abc", -1},
-	BinOpTest{"abc", "ab", 1},
-	BinOpTest{"x", "ab", 1},
-	BinOpTest{"ab", "x", -1},
-	BinOpTest{"x", "a", 1},
-	BinOpTest{"b", "x", -1},
+	{"", "", 0},
+	{"a", "", 1},
+	{"", "a", -1},
+	{"abc", "abc", 0},
+	{"ab", "abc", -1},
+	{"abc", "ab", 1},
+	{"x", "ab", 1},
+	{"ab", "x", -1},
+	{"x", "a", 1},
+	{"b", "x", -1},
 }
 
 func TestCompare(t *testing.T) {
@@ -73,55 +74,81 @@ func TestCompare(t *testing.T) {
 }
 
 var indexTests = []BinOpTest{
-	BinOpTest{"", "", 0},
-	BinOpTest{"", "a", -1},
-	BinOpTest{"", "foo", -1},
-	BinOpTest{"fo", "foo", -1},
-	BinOpTest{"foo", "foo", 0},
-	BinOpTest{"oofofoofooo", "f", 2},
-	BinOpTest{"oofofoofooo", "foo", 4},
-	BinOpTest{"barfoobarfoo", "foo", 3},
-	BinOpTest{"foo", "", 0},
-	BinOpTest{"foo", "o", 1},
-	BinOpTest{"abcABCabc", "A", 3},
+	{"", "", 0},
+	{"", "a", -1},
+	{"", "foo", -1},
+	{"fo", "foo", -1},
+	{"foo", "foo", 0},
+	{"oofofoofooo", "f", 2},
+	{"oofofoofooo", "foo", 4},
+	{"barfoobarfoo", "foo", 3},
+	{"foo", "", 0},
+	{"foo", "o", 1},
+	{"abcABCabc", "A", 3},
 	// cases with one byte strings - test IndexByte and special case in Index()
-	BinOpTest{"", "a", -1},
-	BinOpTest{"x", "a", -1},
-	BinOpTest{"x", "x", 0},
-	BinOpTest{"abc", "a", 0},
-	BinOpTest{"abc", "b", 1},
-	BinOpTest{"abc", "c", 2},
-	BinOpTest{"abc", "x", -1},
+	{"", "a", -1},
+	{"x", "a", -1},
+	{"x", "x", 0},
+	{"abc", "a", 0},
+	{"abc", "b", 1},
+	{"abc", "c", 2},
+	{"abc", "x", -1},
+	{"barfoobarfooyyyzzzyyyzzzyyyzzzyyyxxxzzzyyy", "x", 33},
+	{"foofyfoobarfoobar", "y", 4},
+	{"oooooooooooooooooooooo", "r", -1},
 }
 
 var lastIndexTests = []BinOpTest{
-	BinOpTest{"", "", 0},
-	BinOpTest{"", "a", -1},
-	BinOpTest{"", "foo", -1},
-	BinOpTest{"fo", "foo", -1},
-	BinOpTest{"foo", "foo", 0},
-	BinOpTest{"foo", "f", 0},
-	BinOpTest{"oofofoofooo", "f", 7},
-	BinOpTest{"oofofoofooo", "foo", 7},
-	BinOpTest{"barfoobarfoo", "foo", 9},
-	BinOpTest{"foo", "", 3},
-	BinOpTest{"foo", "o", 2},
-	BinOpTest{"abcABCabc", "A", 3},
-	BinOpTest{"abcABCabc", "a", 6},
+	{"", "", 0},
+	{"", "a", -1},
+	{"", "foo", -1},
+	{"fo", "foo", -1},
+	{"foo", "foo", 0},
+	{"foo", "f", 0},
+	{"oofofoofooo", "f", 7},
+	{"oofofoofooo", "foo", 7},
+	{"barfoobarfoo", "foo", 9},
+	{"foo", "", 3},
+	{"foo", "o", 2},
+	{"abcABCabc", "A", 3},
+	{"abcABCabc", "a", 6},
 }
 
 var indexAnyTests = []BinOpTest{
-	BinOpTest{"", "", -1},
-	BinOpTest{"", "a", -1},
-	BinOpTest{"", "abc", -1},
-	BinOpTest{"a", "", -1},
-	BinOpTest{"a", "a", 0},
-	BinOpTest{"aaa", "a", 0},
-	BinOpTest{"abc", "xyz", -1},
-	BinOpTest{"abc", "xcz", 2},
-	BinOpTest{"ab☺c", "x☺yz", 2},
-	BinOpTest{"aRegExp*", ".(|)*+?^$[]", 7},
-	BinOpTest{dots + dots + dots, " ", -1},
+	{"", "", -1},
+	{"", "a", -1},
+	{"", "abc", -1},
+	{"a", "", -1},
+	{"a", "a", 0},
+	{"aaa", "a", 0},
+	{"abc", "xyz", -1},
+	{"abc", "xcz", 2},
+	{"ab☺c", "x☺yz", 2},
+	{"aRegExp*", ".(|)*+?^$[]", 7},
+	{dots + dots + dots, " ", -1},
+}
+
+var lastIndexAnyTests = []BinOpTest{
+	{"", "", -1},
+	{"", "a", -1},
+	{"", "abc", -1},
+	{"a", "", -1},
+	{"a", "a", 0},
+	{"aaa", "a", 2},
+	{"abc", "xyz", -1},
+	{"abc", "ab", 1},
+	{"a☺b☻c☹d", "uvw☻xyz", 2 + len("☺")},
+	{"a.RegExp*", ".(|)*+?^$[]", 8},
+	{dots + dots + dots, " ", -1},
+}
+
+var indexRuneTests = []BinOpTest{
+	{"", "a", -1},
+	{"", "☺", -1},
+	{"foo", "☹", -1},
+	{"foo", "o", 1},
+	{"foo☺bar", "☺", 3},
+	{"foo☺☻☹bar", "☹", 9},
 }
 
 // Execute f on each test case.  funcName should be the name of f; it's used
@@ -137,18 +164,23 @@ func runIndexTests(t *testing.T, f func(s, sep []byte) int, funcName string, tes
 	}
 }
 
-func TestIndex(t *testing.T)     { runIndexTests(t, Index, "Index", indexTests) }
-func TestLastIndex(t *testing.T) { runIndexTests(t, LastIndex, "LastIndex", lastIndexTests) }
-func TestIndexAny(t *testing.T) {
-	for _, test := range indexAnyTests {
+func runIndexAnyTests(t *testing.T, f func(s []byte, chars string) int, funcName string, testCases []BinOpTest) {
+	for _, test := range testCases {
 		a := []byte(test.a)
-		actual := IndexAny(a, test.b)
+		actual := f(a, test.b)
 		if actual != test.i {
-			t.Errorf("IndexAny(%q,%q) = %v; want %v", a, test.b, actual, test.i)
+			t.Errorf("%s(%q,%q) = %v; want %v", funcName, a, test.b, actual, test.i)
 		}
 	}
 }
 
+func TestIndex(t *testing.T)     { runIndexTests(t, Index, "Index", indexTests) }
+func TestLastIndex(t *testing.T) { runIndexTests(t, LastIndex, "LastIndex", lastIndexTests) }
+func TestIndexAny(t *testing.T)  { runIndexAnyTests(t, IndexAny, "IndexAny", indexAnyTests) }
+func TestLastIndexAny(t *testing.T) {
+	runIndexAnyTests(t, LastIndexAny, "LastIndexAny", lastIndexAnyTests)
+}
+
 func TestIndexByte(t *testing.T) {
 	for _, tt := range indexTests {
 		if len(tt.b) != 1 {
@@ -167,6 +199,67 @@ func TestIndexByte(t *testing.T) {
 	}
 }
 
+// test a larger buffer with different sizes and alignments
+func TestIndexByteBig(t *testing.T) {
+	const n = 1024
+	b := make([]byte, n)
+	for i := 0; i < n; i++ {
+		// different start alignments
+		b1 := b[i:]
+		for j := 0; j < len(b1); j++ {
+			b1[j] = 'x'
+			pos := IndexByte(b1, 'x')
+			if pos != j {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+			b1[j] = 0
+			pos = IndexByte(b1, 'x')
+			if pos != -1 {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+		}
+		// different end alignments
+		b1 = b[:i]
+		for j := 0; j < len(b1); j++ {
+			b1[j] = 'x'
+			pos := IndexByte(b1, 'x')
+			if pos != j {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+			b1[j] = 0
+			pos = IndexByte(b1, 'x')
+			if pos != -1 {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+		}
+		// different start and end alignments
+		b1 = b[i/2 : n-(i+1)/2]
+		for j := 0; j < len(b1); j++ {
+			b1[j] = 'x'
+			pos := IndexByte(b1, 'x')
+			if pos != j {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+			b1[j] = 0
+			pos = IndexByte(b1, 'x')
+			if pos != -1 {
+				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
+			}
+		}
+	}
+}
+
+func TestIndexRune(t *testing.T) {
+	for _, tt := range indexRuneTests {
+		a := []byte(tt.a)
+		r, _ := utf8.DecodeRuneInString(tt.b)
+		pos := IndexRune(a, r)
+		if pos != tt.i {
+			t.Errorf(`IndexRune(%q, '%c') = %v`, tt.a, r, pos)
+		}
+	}
+}
+
 func BenchmarkIndexByte4K(b *testing.B) { bmIndex(b, IndexByte, 4<<10) }
 
 func BenchmarkIndexByte4M(b *testing.B) { bmIndex(b, IndexByte, 4<<20) }
@@ -211,9 +304,10 @@ type ExplodeTest struct {
 }
 
 var explodetests = []ExplodeTest{
-	ExplodeTest{abcd, -1, []string{"a", "b", "c", "d"}},
-	ExplodeTest{faces, -1, []string{"☺", "☻", "☹"}},
-	ExplodeTest{abcd, 2, []string{"a", "bcd"}},
+	{"", -1, []string{}},
+	{abcd, -1, []string{"a", "b", "c", "d"}},
+	{faces, -1, []string{"☺", "☻", "☹"}},
+	{abcd, 2, []string{"a", "bcd"}},
 }
 
 func TestExplode(t *testing.T) {
@@ -240,19 +334,19 @@ type SplitTest struct {
 }
 
 var splittests = []SplitTest{
-	SplitTest{abcd, "a", 0, nil},
-	SplitTest{abcd, "a", -1, []string{"", "bcd"}},
-	SplitTest{abcd, "z", -1, []string{"abcd"}},
-	SplitTest{abcd, "", -1, []string{"a", "b", "c", "d"}},
-	SplitTest{commas, ",", -1, []string{"1", "2", "3", "4"}},
-	SplitTest{dots, "...", -1, []string{"1", ".2", ".3", ".4"}},
-	SplitTest{faces, "☹", -1, []string{"☺☻", ""}},
-	SplitTest{faces, "~", -1, []string{faces}},
-	SplitTest{faces, "", -1, []string{"☺", "☻", "☹"}},
-	SplitTest{"1 2 3 4", " ", 3, []string{"1", "2", "3 4"}},
-	SplitTest{"1 2", " ", 3, []string{"1", "2"}},
-	SplitTest{"123", "", 2, []string{"1", "23"}},
-	SplitTest{"123", "", 17, []string{"1", "2", "3"}},
+	{abcd, "a", 0, nil},
+	{abcd, "a", -1, []string{"", "bcd"}},
+	{abcd, "z", -1, []string{"abcd"}},
+	{abcd, "", -1, []string{"a", "b", "c", "d"}},
+	{commas, ",", -1, []string{"1", "2", "3", "4"}},
+	{dots, "...", -1, []string{"1", ".2", ".3", ".4"}},
+	{faces, "☹", -1, []string{"☺☻", ""}},
+	{faces, "~", -1, []string{faces}},
+	{faces, "", -1, []string{"☺", "☻", "☹"}},
+	{"1 2 3 4", " ", 3, []string{"1", "2", "3 4"}},
+	{"1 2", " ", 3, []string{"1", "2"}},
+	{"123", "", 2, []string{"1", "23"}},
+	{"123", "", 17, []string{"1", "2", "3"}},
 }
 
 func TestSplit(t *testing.T) {
@@ -274,19 +368,19 @@ func TestSplit(t *testing.T) {
 }
 
 var splitaftertests = []SplitTest{
-	SplitTest{abcd, "a", -1, []string{"a", "bcd"}},
-	SplitTest{abcd, "z", -1, []string{"abcd"}},
-	SplitTest{abcd, "", -1, []string{"a", "b", "c", "d"}},
-	SplitTest{commas, ",", -1, []string{"1,", "2,", "3,", "4"}},
-	SplitTest{dots, "...", -1, []string{"1...", ".2...", ".3...", ".4"}},
-	SplitTest{faces, "☹", -1, []string{"☺☻☹", ""}},
-	SplitTest{faces, "~", -1, []string{faces}},
-	SplitTest{faces, "", -1, []string{"☺", "☻", "☹"}},
-	SplitTest{"1 2 3 4", " ", 3, []string{"1 ", "2 ", "3 4"}},
-	SplitTest{"1 2 3", " ", 3, []string{"1 ", "2 ", "3"}},
-	SplitTest{"1 2", " ", 3, []string{"1 ", "2"}},
-	SplitTest{"123", "", 2, []string{"1", "23"}},
-	SplitTest{"123", "", 17, []string{"1", "2", "3"}},
+	{abcd, "a", -1, []string{"a", "bcd"}},
+	{abcd, "z", -1, []string{"abcd"}},
+	{abcd, "", -1, []string{"a", "b", "c", "d"}},
+	{commas, ",", -1, []string{"1,", "2,", "3,", "4"}},
+	{dots, "...", -1, []string{"1...", ".2...", ".3...", ".4"}},
+	{faces, "☹", -1, []string{"☺☻☹", ""}},
+	{faces, "~", -1, []string{faces}},
+	{faces, "", -1, []string{"☺", "☻", "☹"}},
+	{"1 2 3 4", " ", 3, []string{"1 ", "2 ", "3 4"}},
+	{"1 2 3", " ", 3, []string{"1 ", "2 ", "3"}},
+	{"1 2", " ", 3, []string{"1 ", "2"}},
+	{"123", "", 2, []string{"1", "23"}},
+	{"123", "", 17, []string{"1", "2", "3"}},
 }
 
 func TestSplitAfter(t *testing.T) {
@@ -310,17 +404,17 @@ type FieldsTest struct {
 }
 
 var fieldstests = []FieldsTest{
-	FieldsTest{"", []string{}},
-	FieldsTest{" ", []string{}},
-	FieldsTest{" \t ", []string{}},
-	FieldsTest{"  abc  ", []string{"abc"}},
-	FieldsTest{"1 2 3 4", []string{"1", "2", "3", "4"}},
-	FieldsTest{"1  2  3  4", []string{"1", "2", "3", "4"}},
-	FieldsTest{"1\t\t2\t\t3\t4", []string{"1", "2", "3", "4"}},
-	FieldsTest{"1\u20002\u20013\u20024", []string{"1", "2", "3", "4"}},
-	FieldsTest{"\u2000\u2001\u2002", []string{}},
-	FieldsTest{"\n™\t™\n", []string{"™", "™"}},
-	FieldsTest{faces, []string{faces}},
+	{"", []string{}},
+	{" ", []string{}},
+	{" \t ", []string{}},
+	{"  abc  ", []string{"abc"}},
+	{"1 2 3 4", []string{"1", "2", "3", "4"}},
+	{"1  2  3  4", []string{"1", "2", "3", "4"}},
+	{"1\t\t2\t\t3\t4", []string{"1", "2", "3", "4"}},
+	{"1\u20002\u20013\u20024", []string{"1", "2", "3", "4"}},
+	{"\u2000\u2001\u2002", []string{}},
+	{"\n™\t™\n", []string{"™", "™"}},
+	{faces, []string{faces}},
 }
 
 func TestFields(t *testing.T) {
@@ -334,6 +428,23 @@ func TestFields(t *testing.T) {
 	}
 }
 
+func TestFieldsFunc(t *testing.T) {
+	pred := func(c int) bool { return c == 'X' }
+	var fieldsFuncTests = []FieldsTest{
+		{"", []string{}},
+		{"XX", []string{}},
+		{"XXhiXXX", []string{"hi"}},
+		{"aXXbXXXcX", []string{"a", "b", "c"}},
+	}
+	for _, tt := range fieldsFuncTests {
+		a := FieldsFunc([]byte(tt.s), pred)
+		result := arrayOfString(a)
+		if !eq(result, tt.a) {
+			t.Errorf("FieldsFunc(%q) = %v, want %v", tt.s, a, tt.a)
+		}
+	}
+}
+
 // Test case for any function which accepts and returns a byte array.
 // For ease of creation, we write the byte arrays as strings.
 type StringTest struct {
@@ -341,51 +452,47 @@ type StringTest struct {
 }
 
 var upperTests = []StringTest{
-	StringTest{"", ""},
-	StringTest{"abc", "ABC"},
-	StringTest{"AbC123", "ABC123"},
-	StringTest{"azAZ09_", "AZAZ09_"},
-	StringTest{"\u0250\u0250\u0250\u0250\u0250", "\u2C6F\u2C6F\u2C6F\u2C6F\u2C6F"}, // grows one byte per char
+	{"", ""},
+	{"abc", "ABC"},
+	{"AbC123", "ABC123"},
+	{"azAZ09_", "AZAZ09_"},
+	{"\u0250\u0250\u0250\u0250\u0250", "\u2C6F\u2C6F\u2C6F\u2C6F\u2C6F"}, // grows one byte per char
 }
 
 var lowerTests = []StringTest{
-	StringTest{"", ""},
-	StringTest{"abc", "abc"},
-	StringTest{"AbC123", "abc123"},
-	StringTest{"azAZ09_", "azaz09_"},
-	StringTest{"\u2C6D\u2C6D\u2C6D\u2C6D\u2C6D", "\u0251\u0251\u0251\u0251\u0251"}, // shrinks one byte per char
+	{"", ""},
+	{"abc", "abc"},
+	{"AbC123", "abc123"},
+	{"azAZ09_", "azaz09_"},
+	{"\u2C6D\u2C6D\u2C6D\u2C6D\u2C6D", "\u0251\u0251\u0251\u0251\u0251"}, // shrinks one byte per char
 }
 
 const space = "\t\v\r\f\n\u0085\u00a0\u2000\u3000"
 
 var trimSpaceTests = []StringTest{
-	StringTest{"", ""},
-	StringTest{"abc", "abc"},
-	StringTest{space + "abc" + space, "abc"},
-	StringTest{" ", ""},
-	StringTest{" \t\r\n \t\t\r\r\n\n ", ""},
-	StringTest{" \t\r\n x\t\t\r\r\n\n ", "x"},
-	StringTest{" \u2000\t\r\n x\t\t\r\r\ny\n \u3000", "x\t\t\r\r\ny"},
-	StringTest{"1 \t\r\n2", "1 \t\r\n2"},
-	StringTest{" x\x80", "x\x80"}, // invalid UTF-8 on end
-	StringTest{" x\xc0", "x\xc0"}, // invalid UTF-8 on end
-}
-
-// Bytes returns a new slice containing the bytes in s.
-// Borrowed from strings to avoid dependency.
-func Bytes(s string) []byte {
-	b := make([]byte, len(s))
-	for i := 0; i < len(s); i++ {
-		b[i] = s[i]
-	}
-	return b
+	{"", ""},
+	{"abc", "abc"},
+	{space + "abc" + space, "abc"},
+	{" ", ""},
+	{" \t\r\n \t\t\r\r\n\n ", ""},
+	{" \t\r\n x\t\t\r\r\n\n ", "x"},
+	{" \u2000\t\r\n x\t\t\r\r\ny\n \u3000", "x\t\t\r\r\ny"},
+	{"1 \t\r\n2", "1 \t\r\n2"},
+	{" x\x80", "x\x80"},
+	{" x\xc0", "x\xc0"},
+	{"x \xc0\xc0 ", "x \xc0\xc0"},
+	{"x \xc0", "x \xc0"},
+	{"x \xc0 ", "x \xc0"},
+	{"x \xc0\xc0 ", "x \xc0\xc0"},
+	{"x ☺\xc0\xc0 ", "x ☺\xc0\xc0"},
+	{"x ☺ ", "x ☺"},
 }
 
 // Execute f on each test case.  funcName should be the name of f; it's used
 // in failure reports.
 func runStringTests(t *testing.T, f func([]byte) []byte, funcName string, testCases []StringTest) {
 	for _, tc := range testCases {
-		actual := string(f(Bytes(tc.in)))
+		actual := string(f([]byte(tc.in)))
 		if actual != tc.out {
 			t.Errorf("%s(%q) = %q; want %q", funcName, tc.in, actual, tc.out)
 		}
@@ -418,7 +525,7 @@ func TestMap(t *testing.T) {
 
 	// 1.  Grow.  This triggers two reallocations in Map.
 	maxRune := func(rune int) int { return unicode.MaxRune }
-	m := Map(maxRune, Bytes(a))
+	m := Map(maxRune, []byte(a))
 	expect := tenRunes(unicode.MaxRune)
 	if string(m) != expect {
 		t.Errorf("growing: expected %q got %q", expect, m)
@@ -426,21 +533,21 @@ func TestMap(t *testing.T) {
 
 	// 2. Shrink
 	minRune := func(rune int) int { return 'a' }
-	m = Map(minRune, Bytes(tenRunes(unicode.MaxRune)))
+	m = Map(minRune, []byte(tenRunes(unicode.MaxRune)))
 	expect = a
 	if string(m) != expect {
 		t.Errorf("shrinking: expected %q got %q", expect, m)
 	}
 
 	// 3. Rot13
-	m = Map(rot13, Bytes("a to zed"))
+	m = Map(rot13, []byte("a to zed"))
 	expect = "n gb mrq"
 	if string(m) != expect {
 		t.Errorf("rot13: expected %q got %q", expect, m)
 	}
 
 	// 4. Rot13^2
-	m = Map(rot13, Map(rot13, Bytes("a to zed")))
+	m = Map(rot13, Map(rot13, []byte("a to zed")))
 	expect = "a to zed"
 	if string(m) != expect {
 		t.Errorf("rot13: expected %q got %q", expect, m)
@@ -453,7 +560,7 @@ func TestMap(t *testing.T) {
 		}
 		return -1
 	}
-	m = Map(dropNotLatin, Bytes("Hello, 세계"))
+	m = Map(dropNotLatin, []byte("Hello, 세계"))
 	expect = "Hello"
 	if string(m) != expect {
 		t.Errorf("drop: expected %q got %q", expect, m)
@@ -466,60 +573,19 @@ func TestToLower(t *testing.T) { runStringTests(t, ToLower, "ToLower", lowerTest
 
 func TestTrimSpace(t *testing.T) { runStringTests(t, TrimSpace, "TrimSpace", trimSpaceTests) }
 
-type AddTest struct {
-	s, t string
-	cap  int
-}
-
-var addtests = []AddTest{
-	AddTest{"", "", 0},
-	AddTest{"a", "", 1},
-	AddTest{"a", "b", 1},
-	AddTest{"abc", "def", 100},
-}
-
-func TestAdd(t *testing.T) {
-	for _, test := range addtests {
-		b := make([]byte, len(test.s), test.cap)
-		for i := 0; i < len(test.s); i++ {
-			b[i] = test.s[i]
-		}
-		b = Add(b, []byte(test.t))
-		if string(b) != test.s+test.t {
-			t.Errorf("Add(%q,%q) = %q", test.s, test.t, string(b))
-		}
-	}
-}
-
-func TestAddByte(t *testing.T) {
-	const N = 2e5
-	b := make([]byte, 0)
-	for i := 0; i < N; i++ {
-		b = AddByte(b, byte(i))
-	}
-	if len(b) != N {
-		t.Errorf("AddByte: too small; expected %d got %d", N, len(b))
-	}
-	for i, c := range b {
-		if c != byte(i) {
-			t.Fatalf("AddByte: b[%d] should be %d is %d", i, c, byte(i))
-		}
-	}
-}
-
 type RepeatTest struct {
 	in, out string
 	count   int
 }
 
 var RepeatTests = []RepeatTest{
-	RepeatTest{"", "", 0},
-	RepeatTest{"", "", 1},
-	RepeatTest{"", "", 2},
-	RepeatTest{"-", "", 0},
-	RepeatTest{"-", "-", 1},
-	RepeatTest{"-", "----------", 10},
-	RepeatTest{"abc ", "abc abc abc ", 3},
+	{"", "", 0},
+	{"", "", 1},
+	{"", "", 2},
+	{"-", "", 0},
+	{"-", "-", 1},
+	{"-", "----------", 10},
+	{"abc ", "abc abc abc ", 3},
 }
 
 func TestRepeat(t *testing.T) {
@@ -553,13 +619,13 @@ type RunesTest struct {
 }
 
 var RunesTests = []RunesTest{
-	RunesTest{"", []int{}, false},
-	RunesTest{" ", []int{32}, false},
-	RunesTest{"ABC", []int{65, 66, 67}, false},
-	RunesTest{"abc", []int{97, 98, 99}, false},
-	RunesTest{"\u65e5\u672c\u8a9e", []int{26085, 26412, 35486}, false},
-	RunesTest{"ab\x80c", []int{97, 98, 0xFFFD, 99}, true},
-	RunesTest{"ab\xc0c", []int{97, 98, 0xFFFD, 99}, true},
+	{"", []int{}, false},
+	{" ", []int{32}, false},
+	{"ABC", []int{65, 66, 67}, false},
+	{"abc", []int{97, 98, 99}, false},
+	{"\u65e5\u672c\u8a9e", []int{26085, 26412, 35486}, false},
+	{"ab\x80c", []int{97, 98, 0xFFFD, 99}, true},
+	{"ab\xc0c", []int{97, 98, 0xFFFD, 99}, true},
 }
 
 func TestRunes(t *testing.T) {
@@ -587,26 +653,27 @@ type TrimTest struct {
 }
 
 var trimTests = []TrimTest{
-	TrimTest{Trim, "abba", "a", "bb"},
-	TrimTest{Trim, "abba", "ab", ""},
-	TrimTest{TrimLeft, "abba", "ab", ""},
-	TrimTest{TrimRight, "abba", "ab", ""},
-	TrimTest{TrimLeft, "abba", "a", "bba"},
-	TrimTest{TrimRight, "abba", "a", "abb"},
-	TrimTest{Trim, "<tag>", "<>", "tag"},
-	TrimTest{Trim, "* listitem", " *", "listitem"},
-	TrimTest{Trim, `"quote"`, `"`, "quote"},
-	TrimTest{Trim, "\u2C6F\u2C6F\u0250\u0250\u2C6F\u2C6F", "\u2C6F", "\u0250\u0250"},
+	{Trim, "abba", "a", "bb"},
+	{Trim, "abba", "ab", ""},
+	{TrimLeft, "abba", "ab", ""},
+	{TrimRight, "abba", "ab", ""},
+	{TrimLeft, "abba", "a", "bba"},
+	{TrimRight, "abba", "a", "abb"},
+	{Trim, "<tag>", "<>", "tag"},
+	{Trim, "* listitem", " *", "listitem"},
+	{Trim, `"quote"`, `"`, "quote"},
+	{Trim, "\u2C6F\u2C6F\u0250\u0250\u2C6F\u2C6F", "\u2C6F", "\u0250\u0250"},
 	//empty string tests
-	TrimTest{Trim, "abba", "", "abba"},
-	TrimTest{Trim, "", "123", ""},
-	TrimTest{Trim, "", "", ""},
-	TrimTest{TrimLeft, "abba", "", "abba"},
-	TrimTest{TrimLeft, "", "123", ""},
-	TrimTest{TrimLeft, "", "", ""},
-	TrimTest{TrimRight, "abba", "", "abba"},
-	TrimTest{TrimRight, "", "123", ""},
-	TrimTest{TrimRight, "", "", ""},
+	{Trim, "abba", "", "abba"},
+	{Trim, "", "123", ""},
+	{Trim, "", "", ""},
+	{TrimLeft, "abba", "", "abba"},
+	{TrimLeft, "", "123", ""},
+	{TrimLeft, "", "", ""},
+	{TrimRight, "abba", "", "abba"},
+	{TrimRight, "", "123", ""},
+	{TrimRight, "", "", ""},
+	{TrimRight, "☺\xc0", "☺", "☺\xc0"},
 }
 
 func TestTrim(t *testing.T) {
@@ -629,22 +696,90 @@ func TestTrim(t *testing.T) {
 	}
 }
 
+type predicate struct {
+	f    func(r int) bool
+	name string
+}
+
+var isSpace = predicate{unicode.IsSpace, "IsSpace"}
+var isDigit = predicate{unicode.IsDigit, "IsDigit"}
+var isUpper = predicate{unicode.IsUpper, "IsUpper"}
+var isValidRune = predicate{
+	func(r int) bool {
+		return r != utf8.RuneError
+	},
+	"IsValidRune",
+}
+
 type TrimFuncTest struct {
-	f             func(r int) bool
-	name, in, out string
+	f       predicate
+	in, out string
+}
+
+func not(p predicate) predicate {
+	return predicate{
+		func(r int) bool {
+			return !p.f(r)
+		},
+		"not " + p.name,
+	}
 }
 
 var trimFuncTests = []TrimFuncTest{
-	TrimFuncTest{unicode.IsSpace, "IsSpace", space + " hello " + space, "hello"},
-	TrimFuncTest{unicode.IsDigit, "IsDigit", "\u0e50\u0e5212hello34\u0e50\u0e51", "hello"},
-	TrimFuncTest{unicode.IsUpper, "IsUpper", "\u2C6F\u2C6F\u2C6F\u2C6FABCDhelloEF\u2C6F\u2C6FGH\u2C6F\u2C6F", "hello"},
+	{isSpace, space + " hello " + space, "hello"},
+	{isDigit, "\u0e50\u0e5212hello34\u0e50\u0e51", "hello"},
+	{isUpper, "\u2C6F\u2C6F\u2C6F\u2C6FABCDhelloEF\u2C6F\u2C6FGH\u2C6F\u2C6F", "hello"},
+	{not(isSpace), "hello" + space + "hello", space},
+	{not(isDigit), "hello\u0e50\u0e521234\u0e50\u0e51helo", "\u0e50\u0e521234\u0e50\u0e51"},
+	{isValidRune, "ab\xc0a\xc0cd", "\xc0a\xc0"},
+	{not(isValidRune), "\xc0a\xc0", "a"},
 }
 
 func TestTrimFunc(t *testing.T) {
 	for _, tc := range trimFuncTests {
-		actual := string(TrimFunc([]byte(tc.in), tc.f))
+		actual := string(TrimFunc([]byte(tc.in), tc.f.f))
 		if actual != tc.out {
-			t.Errorf("TrimFunc(%q, %q) = %q; want %q", tc.in, tc.name, actual, tc.out)
+			t.Errorf("TrimFunc(%q, %q) = %q; want %q", tc.in, tc.f.name, actual, tc.out)
+		}
+	}
+}
+
+type IndexFuncTest struct {
+	in          string
+	f           predicate
+	first, last int
+}
+
+var indexFuncTests = []IndexFuncTest{
+	{"", isValidRune, -1, -1},
+	{"abc", isDigit, -1, -1},
+	{"0123", isDigit, 0, 3},
+	{"a1b", isDigit, 1, 1},
+	{space, isSpace, 0, len(space) - 3}, // last rune in space is 3 bytes
+	{"\u0e50\u0e5212hello34\u0e50\u0e51", isDigit, 0, 18},
+	{"\u2C6F\u2C6F\u2C6F\u2C6FABCDhelloEF\u2C6F\u2C6FGH\u2C6F\u2C6F", isUpper, 0, 34},
+	{"12\u0e50\u0e52hello34\u0e50\u0e51", not(isDigit), 8, 12},
+
+	// tests of invalid UTF-8
+	{"\x801", isDigit, 1, 1},
+	{"\x80abc", isDigit, -1, -1},
+	{"\xc0a\xc0", isValidRune, 1, 1},
+	{"\xc0a\xc0", not(isValidRune), 0, 2},
+	{"\xc0☺\xc0", not(isValidRune), 0, 4},
+	{"\xc0☺\xc0\xc0", not(isValidRune), 0, 5},
+	{"ab\xc0a\xc0cd", not(isValidRune), 2, 4},
+	{"a\xe0\x80cd", not(isValidRune), 1, 2},
+}
+
+func TestIndexFunc(t *testing.T) {
+	for _, tc := range indexFuncTests {
+		first := IndexFunc([]byte(tc.in), tc.f.f)
+		if first != tc.first {
+			t.Errorf("IndexFunc(%q, %s) = %d; want %d", tc.in, tc.f.name, first, tc.first)
+		}
+		last := LastIndexFunc([]byte(tc.in), tc.f.f)
+		if last != tc.last {
+			t.Errorf("LastIndexFunc(%q, %s) = %d; want %d", tc.in, tc.f.name, last, tc.last)
 		}
 	}
 }
@@ -657,25 +792,25 @@ type ReplaceTest struct {
 }
 
 var ReplaceTests = []ReplaceTest{
-	ReplaceTest{"hello", "l", "L", 0, "hello"},
-	ReplaceTest{"hello", "l", "L", -1, "heLLo"},
-	ReplaceTest{"hello", "x", "X", -1, "hello"},
-	ReplaceTest{"", "x", "X", -1, ""},
-	ReplaceTest{"radar", "r", "<r>", -1, "<r>ada<r>"},
-	ReplaceTest{"", "", "<>", -1, "<>"},
-	ReplaceTest{"banana", "a", "<>", -1, "b<>n<>n<>"},
-	ReplaceTest{"banana", "a", "<>", 1, "b<>nana"},
-	ReplaceTest{"banana", "a", "<>", 1000, "b<>n<>n<>"},
-	ReplaceTest{"banana", "an", "<>", -1, "b<><>a"},
-	ReplaceTest{"banana", "ana", "<>", -1, "b<>na"},
-	ReplaceTest{"banana", "", "<>", -1, "<>b<>a<>n<>a<>n<>a<>"},
-	ReplaceTest{"banana", "", "<>", 10, "<>b<>a<>n<>a<>n<>a<>"},
-	ReplaceTest{"banana", "", "<>", 6, "<>b<>a<>n<>a<>n<>a"},
-	ReplaceTest{"banana", "", "<>", 5, "<>b<>a<>n<>a<>na"},
-	ReplaceTest{"banana", "", "<>", 1, "<>banana"},
-	ReplaceTest{"banana", "a", "a", -1, "banana"},
-	ReplaceTest{"banana", "a", "a", 1, "banana"},
-	ReplaceTest{"☺☻☹", "", "<>", -1, "<>☺<>☻<>☹<>"},
+	{"hello", "l", "L", 0, "hello"},
+	{"hello", "l", "L", -1, "heLLo"},
+	{"hello", "x", "X", -1, "hello"},
+	{"", "x", "X", -1, ""},
+	{"radar", "r", "<r>", -1, "<r>ada<r>"},
+	{"", "", "<>", -1, "<>"},
+	{"banana", "a", "<>", -1, "b<>n<>n<>"},
+	{"banana", "a", "<>", 1, "b<>nana"},
+	{"banana", "a", "<>", 1000, "b<>n<>n<>"},
+	{"banana", "an", "<>", -1, "b<><>a"},
+	{"banana", "ana", "<>", -1, "b<>na"},
+	{"banana", "", "<>", -1, "<>b<>a<>n<>a<>n<>a<>"},
+	{"banana", "", "<>", 10, "<>b<>a<>n<>a<>n<>a<>"},
+	{"banana", "", "<>", 6, "<>b<>a<>n<>a<>n<>a"},
+	{"banana", "", "<>", 5, "<>b<>a<>n<>a<>na"},
+	{"banana", "", "<>", 1, "<>banana"},
+	{"banana", "a", "a", -1, "banana"},
+	{"banana", "a", "a", 1, "banana"},
+	{"☺☻☹", "", "<>", -1, "<>☺<>☻<>☹<>"},
 }
 
 func TestReplace(t *testing.T) {
@@ -685,3 +820,25 @@ func TestReplace(t *testing.T) {
 		}
 	}
 }
+
+type TitleTest struct {
+	in, out string
+}
+
+var TitleTests = []TitleTest{
+	{"", ""},
+	{"a", "A"},
+	{" aaa aaa aaa ", " Aaa Aaa Aaa "},
+	{" Aaa Aaa Aaa ", " Aaa Aaa Aaa "},
+	{"123a456", "123a456"},
+	{"double-blind", "Double-Blind"},
+	{"ÿøû", "Ÿøû"},
+}
+
+func TestTitle(t *testing.T) {
+	for _, tt := range TitleTests {
+		if s := string(Title([]byte(tt.in))); s != tt.out {
+			t.Errorf("Title(%q) = %q, want %q", tt.in, s, tt.out)
+		}
+	}
+}