mime/multipart and HTTP multipart/form-data support

Somewhat of a work-in-progress (in that MIME is a large spec), but this is
functional and enough for discussion and/or code review.

In addition to the unit tests, I've tested with curl and Chrome with
a variety of test files, making sure the digests of files are unaltered
when read via a multipart Part.

R=rsc, adg, dsymonds1, agl1
CC=golang-dev
https://golang.org/cl/1681049

src/pkg/Makefile

@@ -94,6 +94,7 @@ DIRS=\
 	log\
 	math\
 	mime\
+	mime/multipart\
 	net\
 	netchan\
 	nntp\

src/pkg/http/request.go

@@ -16,6 +16,8 @@ import (
 	"fmt"
 	"io"
 	"io/ioutil"
+	"mime"
+	"mime/multipart"
 	"os"
 	"strconv"
 	"strings"
@@ -40,6 +42,8 @@ var (
 	ErrNotSupported         = &ProtocolError{"feature not supported"}
 	ErrUnexpectedTrailer    = &ProtocolError{"trailer header without chunked transfer encoding"}
 	ErrMissingContentLength = &ProtocolError{"missing ContentLength in HEAD response"}
+	ErrNotMultipart         = &ProtocolError{"request Content-Type isn't multipart/form-data"}
+	ErrMissingBoundary      = &ProtocolError{"no multipart boundary param Content-Type"}
 )
 
 type badStringError struct {
@@ -139,6 +143,24 @@ func (r *Request) ProtoAtLeast(major, minor int) bool {
 		r.ProtoMajor == major && r.ProtoMinor >= minor
 }
 
+// MultipartReader returns a MIME multipart reader if this is a
+// multipart/form-data POST request, else returns nil and an error.
+func (r *Request) MultipartReader() (multipart.Reader, os.Error) {
+	v, ok := r.Header["Content-Type"]
+	if !ok {
+		return nil, ErrNotMultipart
+	}
+	d, params := mime.ParseMediaType(v)
+	if d != "multipart/form-data" {
+		return nil, ErrNotMultipart
+	}
+	boundary, ok := params["boundary"]
+	if !ok {
+		return nil, ErrMissingBoundary
+	}
+	return multipart.NewReader(r.Body, boundary), nil
+}
+
 // Return value if nonempty, def otherwise.
 func valueOrDefault(value, def string) string {
 	if value != "" {

src/pkg/http/request_test.go

@@ -101,6 +101,24 @@ func TestPostContentTypeParsing(t *testing.T) {
 	}
 }
 
+func TestMultipartReader(t *testing.T) {
+	req := &Request{
+		Method: "POST",
+		Header: stringMap{"Content-Type": `multipart/form-data; boundary="foo123"`},
+		Body:   nopCloser{new(bytes.Buffer)},
+	}
+	multipart, err := req.MultipartReader()
+	if multipart == nil {
+		t.Errorf("expected multipart; error: %v", err)
+	}
+
+	req.Header = stringMap{"Content-Type": "text/plain"}
+	multipart, err = req.MultipartReader()
+	if multipart != nil {
+		t.Errorf("unexpected multipart for text/plain")
+	}
+}
+
 func TestRedirect(t *testing.T) {
 	const (
 		start = "http://codesearch.google.com/"

src/pkg/mime/Makefile

@@ -6,6 +6,8 @@ include ../../Make.$(GOARCH)
 
 TARG=mime
 GOFILES=\
+	grammar.go\
+	mediatype.go\
 	type.go\
 
 include ../../Make.pkg

src/pkg/mime/grammar.go

@@ -0,0 +1,36 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package mime
+
+import (
+	"strings"
+)
+
+// isTSpecial returns true if rune is in 'tspecials' as defined by RFC
+// 1531 and RFC 2045.
+func isTSpecial(rune int) bool {
+	return strings.IndexRune(`()<>@,;:\"/[]?=`, rune) != -1
+}
+
+// IsTokenChar returns true if rune is in 'token' as defined by RFC
+// 1531 and RFC 2045.
+func IsTokenChar(rune int) bool {
+	// token := 1*<any (US-ASCII) CHAR except SPACE, CTLs,
+	//             or tspecials>
+	return rune > 0x20 && rune < 0x7f && !isTSpecial(rune)
+}
+
+// IsQText returns true if rune is in 'qtext' as defined by RFC 822.
+func IsQText(rune int) bool {
+	// CHAR        =  <any ASCII character>        ; (  0-177,  0.-127.)
+	// qtext       =  <any CHAR excepting <">,     ; => may be folded
+	//                "\" & CR, and including
+	//                linear-white-space>
+	switch rune {
+	case int('"'), int('\\'), int('\r'):
+		return false
+	}
+	return rune < 0x80
+}

src/pkg/mime/mediatype.go

@@ -0,0 +1,120 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package mime
+
+import (
+	"bytes"
+	"strings"
+	"unicode"
+)
+
+// ParseMediaType parses a media type value and any optional
+// parameters, per RFC 1531.  Media types are the values in
+// Content-Type and Content-Disposition headers (RFC 2183).  On
+// success, ParseMediaType returns the media type converted to
+// lowercase and trimmed of white space and a non-nil params.  On
+// error, it returns an empty string and a nil params.
+func ParseMediaType(v string) (mediatype string, params map[string]string) {
+	i := strings.Index(v, ";")
+	if i == -1 {
+		i = len(v)
+	}
+	mediatype = strings.TrimSpace(strings.ToLower(v[0:i]))
+	params = make(map[string]string)
+
+	v = v[i:]
+	for len(v) > 0 {
+		v = strings.TrimLeftFunc(v, unicode.IsSpace)
+		if len(v) == 0 {
+			return
+		}
+		key, value, rest := consumeMediaParam(v)
+		if key == "" {
+			// Parse error.
+			return "", nil
+		}
+		params[key] = value
+		v = rest
+	}
+	return
+}
+
+func isNotTokenChar(rune int) bool {
+	return !IsTokenChar(rune)
+}
+
+// consumeToken consumes a token from the beginning of provided
+// string, per RFC 2045 section 5.1 (referenced from 2183), and return
+// the token consumed and the rest of the string.  Returns ("", v) on
+// failure to consume at least one character.
+func consumeToken(v string) (token, rest string) {
+	notPos := strings.IndexFunc(v, isNotTokenChar)
+	if notPos == -1 {
+		return v, ""
+	}
+	if notPos == 0 {
+		return "", v
+	}
+	return v[0:notPos], v[notPos:]
+}
+
+// consumeValue consumes a "value" per RFC 2045, where a value is
+// either a 'token' or a 'quoted-string'.  On success, consumeValue
+// returns the value consumed (and de-quoted/escaped, if a
+// quoted-string) and the rest of the string.  On failure, returns
+// ("", v).
+func consumeValue(v string) (value, rest string) {
+	if !strings.HasPrefix(v, `"`) {
+		return consumeToken(v)
+	}
+
+	// parse a quoted-string
+	rest = v[1:] // consume the leading quote
+	buffer := new(bytes.Buffer)
+	var idx, rune int
+	var nextIsLiteral bool
+	for idx, rune = range rest {
+		switch {
+		case nextIsLiteral:
+			if rune >= 0x80 {
+				return "", v
+			}
+			buffer.WriteRune(rune)
+			nextIsLiteral = false
+		case rune == '"':
+			return buffer.String(), rest[idx+1:]
+		case IsQText(rune):
+			buffer.WriteRune(rune)
+		case rune == '\\':
+			nextIsLiteral = true
+		default:
+			return "", v
+		}
+	}
+	return "", v
+}
+
+func consumeMediaParam(v string) (param, value, rest string) {
+	rest = strings.TrimLeftFunc(v, unicode.IsSpace)
+	if !strings.HasPrefix(rest, ";") {
+		return "", "", v
+	}
+
+	rest = rest[1:] // consume semicolon
+	rest = strings.TrimLeftFunc(rest, unicode.IsSpace)
+	param, rest = consumeToken(rest)
+	if param == "" {
+		return "", "", v
+	}
+	if !strings.HasPrefix(rest, "=") {
+		return "", "", v
+	}
+	rest = rest[1:] // consume equals sign
+	value, rest = consumeValue(rest)
+	if value == "" {
+		return "", "", v
+	}
+	return param, value, rest
+}

src/pkg/mime/mediatype_test.go

@@ -0,0 +1,117 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package mime
+
+import (
+	"testing"
+)
+
+func TestConsumeToken(t *testing.T) {
+	tests := [...][3]string{
+		[3]string{"foo bar", "foo", " bar"},
+		[3]string{"bar", "bar", ""},
+		[3]string{"", "", ""},
+		[3]string{" foo", "", " foo"},
+	}
+	for _, test := range tests {
+		token, rest := consumeToken(test[0])
+		expectedToken := test[1]
+		expectedRest := test[2]
+		if token != expectedToken {
+			t.Errorf("expected to consume token '%s', not '%s' from '%s'",
+				expectedToken, token, test[0])
+		} else if rest != expectedRest {
+			t.Errorf("expected to have left '%s', not '%s' after reading token '%s' from '%s'",
+				expectedRest, rest, token, test[0])
+		}
+	}
+}
+
+func TestConsumeValue(t *testing.T) {
+	tests := [...][3]string{
+		[3]string{"foo bar", "foo", " bar"},
+		[3]string{"bar", "bar", ""},
+		[3]string{" bar ", "", " bar "},
+		[3]string{`"My value"end`, "My value", "end"},
+		[3]string{`"My value" end`, "My value", " end"},
+		[3]string{`"\\" rest`, "\\", " rest"},
+		[3]string{`"My \" value"end`, "My \" value", "end"},
+		[3]string{`"\" rest`, "", `"\" rest`},
+	}
+	for _, test := range tests {
+		value, rest := consumeValue(test[0])
+		expectedValue := test[1]
+		expectedRest := test[2]
+		if value != expectedValue {
+			t.Errorf("expected to consume value [%s], not [%s] from [%s]",
+				expectedValue, value, test[0])
+		} else if rest != expectedRest {
+			t.Errorf("expected to have left [%s], not [%s] after reading value [%s] from [%s]",
+				expectedRest, rest, value, test[0])
+		}
+	}
+}
+
+func TestConsumeMediaParam(t *testing.T) {
+	tests := [...][4]string{
+		[4]string{" ; foo=bar", "foo", "bar", ""},
+		[4]string{"; foo=bar", "foo", "bar", ""},
+		[4]string{";foo=bar", "foo", "bar", ""},
+		[4]string{`;foo="bar"`, "foo", "bar", ""},
+		[4]string{`;foo="bar"; `, "foo", "bar", "; "},
+		[4]string{`;foo="bar"; foo=baz`, "foo", "bar", "; foo=baz"},
+		[4]string{` ; boundary=----CUT;`, "boundary", "----CUT", ";"},
+		[4]string{` ; key=value;  blah="value";name="foo" `, "key", "value", `;  blah="value";name="foo" `},
+		[4]string{`;  blah="value";name="foo" `, "blah", "value", `;name="foo" `},
+		[4]string{`;name="foo" `, "name", "foo", ` `},
+	}
+	for _, test := range tests {
+		param, value, rest := consumeMediaParam(test[0])
+		expectedParam := test[1]
+		expectedValue := test[2]
+		expectedRest := test[3]
+		if param != expectedParam {
+			t.Errorf("expected to consume param [%s], not [%s] from [%s]",
+				expectedParam, param, test[0])
+		} else if value != expectedValue {
+			t.Errorf("expected to consume value [%s], not [%s] from [%s]",
+				expectedValue, value, test[0])
+		} else if rest != expectedRest {
+			t.Errorf("expected to have left [%s], not [%s] after reading [%s/%s] from [%s]",
+				expectedRest, rest, param, value, test[0])
+		}
+	}
+}
+
+func TestParseMediaType(t *testing.T) {
+	tests := [...]string{
+		`form-data; name="foo"`,
+		` form-data ; name=foo`,
+		`FORM-DATA;name="foo"`,
+		` FORM-DATA ; name="foo"`,
+		` FORM-DATA ; name="foo"`,
+		`form-data; key=value;  blah="value";name="foo" `,
+	}
+	for _, test := range tests {
+		mt, params := ParseMediaType(test)
+		if mt != "form-data" {
+			t.Errorf("expected type form-data for %s, got [%s]", test, mt)
+			continue
+		}
+		if params["name"] != "foo" {
+			t.Errorf("expected name=foo for %s", test)
+		}
+	}
+}
+
+func TestParseMediaTypeBogus(t *testing.T) {
+	mt, params := ParseMediaType("bogus ;=========")
+	if mt != "" {
+		t.Error("expected empty type")
+	}
+	if params != nil {
+		t.Error("expected nil params")
+	}
+}

src/pkg/mime/multipart/Makefile

@@ -0,0 +1,11 @@
+# Copyright 2010 The Go Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style
+# license that can be found in the LICENSE file.
+
+include ../../../Make.$(GOARCH)
+
+TARG=mime/multipart
+GOFILES=\
+	multipart.go\
+
+include ../../../Make.pkg

src/pkg/mime/multipart/multipart.go

@@ -0,0 +1,280 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+//
+
+/*
+Package multipart implements MIME multipart parsing, as defined in RFC
+2046.
+
+The implementation is sufficient for HTTP (RFC 2388) and the multipart
+bodies generated by popular browsers.
+*/
+package multipart
+
+import (
+	"bufio"
+	"bytes"
+	"io"
+	"mime"
+	"os"
+	"regexp"
+	"strings"
+)
+
+var headerRegexp *regexp.Regexp = regexp.MustCompile("^([a-zA-Z0-9\\-]+): *([^\r\n]+)")
+
+// Reader is an iterator over parts in a MIME multipart body.
+// Reader's underlying parser consumes its input as needed.  Seeking
+// isn't supported.
+type Reader interface {
+	// NextPart returns the next part in the multipart, or (nil,
+	// nil) on EOF.  An error is returned if the underlying reader
+	// reports errors, or on truncated or otherwise malformed
+	// input.
+	NextPart() (*Part, os.Error)
+}
+
+// A Part represents a single part in a multipart body.
+type Part struct {
+	// The headers of the body, if any, with the keys canonicalized
+	// in the same fashion that the Go http.Request headers are.
+	// i.e. "foo-bar" changes case to "Foo-Bar"
+	Header map[string]string
+
+	buffer *bytes.Buffer
+	mr     *multiReader
+}
+
+// FormName returns the name parameter if p has a Content-Disposition
+// of type "form-data".  Otherwise it returns the empty string.
+func (p *Part) FormName() string {
+	// See http://tools.ietf.org/html/rfc2183 section 2 for EBNF
+	// of Content-Disposition value format.
+	v, ok := p.Header["Content-Disposition"]
+	if !ok {
+		return ""
+	}
+	d, params := mime.ParseMediaType(v)
+	if d != "form-data" {
+		return ""
+	}
+	return params["name"]
+}
+
+// NewReader creates a new multipart Reader reading from r using the
+// given MIME boundary.
+func NewReader(reader io.Reader, boundary string) Reader {
+	return &multiReader{
+		boundary:     boundary,
+		dashBoundary: "--" + boundary,
+		endLine:      "--" + boundary + "--",
+		bufReader:    bufio.NewReader(reader),
+	}
+}
+
+// Implementation ....
+
+type devNullWriter bool
+
+func (*devNullWriter) Write(p []byte) (n int, err os.Error) {
+	return len(p), nil
+}
+
+var devNull = devNullWriter(false)
+
+func newPart(mr *multiReader) (bp *Part, err os.Error) {
+	bp = new(Part)
+	bp.Header = make(map[string]string)
+	bp.mr = mr
+	bp.buffer = new(bytes.Buffer)
+	if err = bp.populateHeaders(); err != nil {
+		bp = nil
+	}
+	return
+}
+
+func (bp *Part) populateHeaders() os.Error {
+	for {
+		line, err := bp.mr.bufReader.ReadString('\n')
+		if err != nil {
+			return err
+		}
+		if line == "\n" || line == "\r\n" {
+			return nil
+		}
+		if matches := headerRegexp.MatchStrings(line); len(matches) == 3 {
+			key := matches[1]
+			value := matches[2]
+			// TODO: canonicalize headers ala http.Request.Header?
+			bp.Header[key] = value
+			continue
+		}
+		return os.NewError("Unexpected header line found parsing multipart body")
+	}
+	panic("unreachable")
+}
+
+// Read reads the body of a part, after its headers and before the
+// next part (if any) begins.
+func (bp *Part) Read(p []byte) (n int, err os.Error) {
+	for {
+		if bp.buffer.Len() >= len(p) {
+			// Internal buffer of unconsumed data is large enough for
+			// the read request.  No need to parse more at the moment.
+			break
+		}
+		if !bp.mr.ensureBufferedLine() {
+			return 0, io.ErrUnexpectedEOF
+		}
+		if bp.mr.bufferedLineIsBoundary() {
+			// Don't consume this line
+			break
+		}
+
+		// Write all of this line, except the final CRLF
+		s := *bp.mr.bufferedLine
+		if strings.HasSuffix(s, "\r\n") {
+			bp.mr.consumeLine()
+			if !bp.mr.ensureBufferedLine() {
+				return 0, io.ErrUnexpectedEOF
+			}
+			if bp.mr.bufferedLineIsBoundary() {
+				// The final \r\n isn't ours.  It logically belongs
+				// to the boundary line which follows.
+				bp.buffer.WriteString(s[0 : len(s)-2])
+			} else {
+				bp.buffer.WriteString(s)
+			}
+			break
+		}
+		if strings.HasSuffix(s, "\n") {
+			bp.buffer.WriteString(s)
+			bp.mr.consumeLine()
+			continue
+		}
+		return 0, os.NewError("multipart parse error during Read; unexpected line: " + s)
+	}
+	return bp.buffer.Read(p)
+}
+
+func (bp *Part) Close() os.Error {
+	io.Copy(&devNull, bp)
+	return nil
+}
+
+type multiReader struct {
+	boundary     string
+	dashBoundary string // --boundary
+	endLine      string // --boundary--
+
+	bufferedLine *string
+
+	bufReader   *bufio.Reader
+	currentPart *Part
+	partsRead   int
+}
+
+func (mr *multiReader) eof() bool {
+	return mr.bufferedLine == nil &&
+		!mr.readLine()
+}
+
+func (mr *multiReader) readLine() bool {
+	line, err := mr.bufReader.ReadString('\n')
+	if err != nil {
+		// TODO: care about err being EOF or not?
+		return false
+	}
+	mr.bufferedLine = &line
+	return true
+}
+
+func (mr *multiReader) bufferedLineIsBoundary() bool {
+	return strings.HasPrefix(*mr.bufferedLine, mr.dashBoundary)
+}
+
+func (mr *multiReader) ensureBufferedLine() bool {
+	if mr.bufferedLine == nil {
+		return mr.readLine()
+	}
+	return true
+}
+
+func (mr *multiReader) consumeLine() {
+	mr.bufferedLine = nil
+}
+
+func (mr *multiReader) NextPart() (*Part, os.Error) {
+	if mr.currentPart != nil {
+		mr.currentPart.Close()
+	}
+
+	for {
+		if mr.eof() {
+			return nil, io.ErrUnexpectedEOF
+		}
+
+		if isBoundaryDelimiterLine(*mr.bufferedLine, mr.dashBoundary) {
+			mr.consumeLine()
+			mr.partsRead++
+			bp, err := newPart(mr)
+			if err != nil {
+				return nil, err
+			}
+			mr.currentPart = bp
+			return bp, nil
+		}
+
+		if hasPrefixThenNewline(*mr.bufferedLine, mr.endLine) {
+			mr.consumeLine()
+			// Expected EOF (no error)
+			return nil, nil
+		}
+
+		if mr.partsRead == 0 {
+			// skip line
+			mr.consumeLine()
+			continue
+		}
+
+		return nil, os.NewError("Unexpected line in Next().")
+	}
+	panic("unreachable")
+}
+
+func isBoundaryDelimiterLine(line, dashPrefix string) bool {
+	// http://tools.ietf.org/html/rfc2046#section-5.1
+	//   The boundary delimiter line is then defined as a line
+	//   consisting entirely of two hyphen characters ("-",
+	//   decimal value 45) followed by the boundary parameter
+	//   value from the Content-Type header field, optional linear
+	//   whitespace, and a terminating CRLF.
+	if !strings.HasPrefix(line, dashPrefix) {
+		return false
+	}
+	if strings.HasSuffix(line, "\r\n") {
+		return onlyHorizontalWhitespace(line[len(dashPrefix) : len(line)-2])
+	}
+	// Violate the spec and also support newlines without the
+	// carriage return...
+	if strings.HasSuffix(line, "\n") {
+		return onlyHorizontalWhitespace(line[len(dashPrefix) : len(line)-1])
+	}
+	return false
+}
+
+func onlyHorizontalWhitespace(s string) bool {
+	for i := 0; i < len(s); i++ {
+		if s[i] != ' ' && s[i] != '\t' {
+			return false
+		}
+	}
+	return true
+}
+
+func hasPrefixThenNewline(s, prefix string) bool {
+	return strings.HasPrefix(s, prefix) &&
+		(len(s) == len(prefix)+1 && strings.HasSuffix(s, "\n") ||
+			len(s) == len(prefix)+2 && strings.HasSuffix(s, "\r\n"))
+}

src/pkg/mime/multipart/multipart_test.go

@@ -0,0 +1,204 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package multipart
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"json"
+	"regexp"
+	"strings"
+	"testing"
+)
+
+func TestHorizontalWhitespace(t *testing.T) {
+	if !onlyHorizontalWhitespace(" \t") {
+		t.Error("expected pass")
+	}
+	if onlyHorizontalWhitespace("foo bar") {
+		t.Error("expected failure")
+	}
+}
+
+func TestBoundaryLine(t *testing.T) {
+	boundary := "myBoundary"
+	prefix := "--" + boundary
+	if !isBoundaryDelimiterLine("--myBoundary\r\n", prefix) {
+		t.Error("expected")
+	}
+	if !isBoundaryDelimiterLine("--myBoundary \r\n", prefix) {
+		t.Error("expected")
+	}
+	if !isBoundaryDelimiterLine("--myBoundary \n", prefix) {
+		t.Error("expected")
+	}
+	if isBoundaryDelimiterLine("--myBoundary bogus \n", prefix) {
+		t.Error("expected fail")
+	}
+	if isBoundaryDelimiterLine("--myBoundary bogus--", prefix) {
+		t.Error("expected fail")
+	}
+}
+
+func escapeString(v string) string {
+	bytes, _ := json.Marshal(v)
+	return string(bytes)
+}
+
+func expectEq(t *testing.T, expected, actual, what string) {
+	if expected == actual {
+		return
+	}
+	t.Errorf("Unexpected value for %s; got %s (len %d) but expected: %s (len %d)",
+		what, escapeString(actual), len(actual), escapeString(expected), len(expected))
+}
+
+func TestFormName(t *testing.T) {
+	p := new(Part)
+	p.Header = make(map[string]string)
+	tests := [...][2]string{
+		[2]string{`form-data; name="foo"`, "foo"},
+		[2]string{` form-data ; name=foo`, "foo"},
+		[2]string{`FORM-DATA;name="foo"`, "foo"},
+		[2]string{` FORM-DATA ; name="foo"`, "foo"},
+		[2]string{` FORM-DATA ; name="foo"`, "foo"},
+		[2]string{` FORM-DATA ; name=foo`, "foo"},
+		[2]string{` FORM-DATA ; filename="foo.txt"; name=foo; baz=quux`, "foo"},
+	}
+	for _, test := range tests {
+		p.Header["Content-Disposition"] = test[0]
+		expected := test[1]
+		actual := p.FormName()
+		if actual != expected {
+			t.Errorf("expected \"%s\"; got: \"%s\"", expected, actual)
+		}
+	}
+}
+
+func TestMultipart(t *testing.T) {
+	testBody := `
+This is a multi-part message.  This line is ignored.
+--MyBoundary
+Header1: value1
+HEADER2: value2
+foo-bar: baz
+
+My value
+The end.
+--MyBoundary
+Header1: value1b
+HEADER2: value2b
+foo-bar: bazb
+
+Line 1
+Line 2
+Line 3 ends in a newline, but just one.
+
+--MyBoundary
+
+never read data
+--MyBoundary--
+`
+	testBody = regexp.MustCompile("\n").ReplaceAllString(testBody, "\r\n")
+	bodyReader := strings.NewReader(testBody)
+
+	reader := NewReader(bodyReader, "MyBoundary")
+	buf := new(bytes.Buffer)
+
+	// Part1
+	part, err := reader.NextPart()
+	if part == nil || err != nil {
+		t.Error("Expected part1")
+		return
+	}
+	if part.Header["Header1"] != "value1" {
+		t.Error("Expected Header1: value")
+	}
+	if part.Header["foo-bar"] != "baz" {
+		t.Error("Expected foo-bar: baz")
+	}
+	buf.Reset()
+	io.Copy(buf, part)
+	expectEq(t, "My value\r\nThe end.",
+		buf.String(), "Value of first part")
+
+	// Part2
+	part, err = reader.NextPart()
+	if part == nil || err != nil {
+		t.Error("Expected part2")
+		return
+	}
+	if part.Header["foo-bar"] != "bazb" {
+		t.Error("Expected foo-bar: bazb")
+	}
+	buf.Reset()
+	io.Copy(buf, part)
+	expectEq(t, "Line 1\r\nLine 2\r\nLine 3 ends in a newline, but just one.\r\n",
+		buf.String(), "Value of second part")
+
+	// Part3
+	part, err = reader.NextPart()
+	if part == nil || err != nil {
+		t.Error("Expected part3 without errors")
+		return
+	}
+
+	// Non-existent part4
+	part, err = reader.NextPart()
+	if part != nil {
+		t.Error("Didn't expect a third part.")
+	}
+	if err != nil {
+		t.Errorf("Unexpected error getting third part: %v", err)
+	}
+}
+
+func TestVariousTextLineEndings(t *testing.T) {
+	tests := [...]string{
+		"Foo\nBar",
+		"Foo\nBar\n",
+		"Foo\r\nBar",
+		"Foo\r\nBar\r\n",
+		"Foo\rBar",
+		"Foo\rBar\r",
+		"\x00\x01\x02\x09\x0a\x0b\x0c\x0d\x0e\x0f\x10",
+	}
+
+	for testNum, expectedBody := range tests {
+		body := "--BOUNDARY\r\n" +
+			"Content-Disposition: form-data; name=\"value\"\r\n" +
+			"\r\n" +
+			expectedBody +
+			"\r\n--BOUNDARY--\r\n"
+		bodyReader := strings.NewReader(body)
+
+		reader := NewReader(bodyReader, "BOUNDARY")
+		buf := new(bytes.Buffer)
+		part, err := reader.NextPart()
+		if part == nil {
+			t.Errorf("Expected a body part on text %d", testNum)
+			continue
+		}
+		if err != nil {
+			t.Errorf("Unexpected error on text %d: %v", testNum, err)
+			continue
+		}
+		written, err := io.Copy(buf, part)
+		expectEq(t, expectedBody, buf.String(), fmt.Sprintf("test %d", testNum))
+		if err != nil {
+			t.Errorf("Error copying multipart; bytes=%v, error=%v", written, err)
+		}
+
+		part, err = reader.NextPart()
+		if part != nil {
+			t.Errorf("Unexpected part in test %d", testNum)
+		}
+		if err != nil {
+			t.Errorf("Unexpected error in test %d: %v", testNum, err)
+		}
+
+	}
+}

src/pkg/mime/type.go

@@ -2,14 +2,7 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-// The mime package translates file name extensions to MIME types.
+// The mime package implements parts of the MIME spec.
-// It consults the local system's mime.types file, which must be installed
-// under one of these names:
-//
-//   /etc/mime.types
-//   /etc/apache2/mime.types
-//   /etc/apache/mime.types
-//
 package mime
 
 import (
@@ -76,6 +69,14 @@ func initMime() {
 // TypeByExtension returns the MIME type associated with the file extension ext.
 // The extension ext should begin with a leading dot, as in ".html".
 // When ext has no associated type, TypeByExtension returns "".
+//
+// The built-in table is small but is is augmented by the local
+// system's mime.types file(s) if available under one or more of these
+// names:
+//
+//   /etc/mime.types
+//   /etc/apache2/mime.types
+//   /etc/apache/mime.types
 func TypeByExtension(ext string) string {
 	once.Do(initMime)
 	return mimeTypes[ext]