mime: encode CTL and non-US-ASCII characters in FormatMediaType

Encodes non-WSP CTL and non-US-ASCII UTF-8 characters using syntax specified in RFC 2231.

Fixes #7668
Fixes #9624

Change-Id: I433f167c5bdd84a7f811ac0410b08b10790e0d9f
GitHub-Last-Rev: 9c77146760341fdb3af35c1b94d4ee00ffb0daae
GitHub-Pull-Request: golang/go#29328
Reviewed-on: https://go-review.googlesource.com/c/go/+/154760
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

diff --git a/src/mime/mediatype.go b/src/mime/mediatype.go
index 05390773a8afa98272c7e2ca55d582248deb08ce..56ceb488533fc9b94063b68c6250c2e4e0f4fea6 100644 (file)
--- a/src/mime/mediatype.go
+++ b/src/mime/mediatype.go
@@ -19,7 +19,7 @@ import (
 // FormatMediaType returns the empty string.
 func FormatMediaType(t string, param map[string]string) string {
        var b strings.Builder
-       if slash := strings.Index(t, "/"); slash == -1 {
+       if slash := strings.IndexByte(t, '/'); slash == -1 {
                if !isToken(t) {
                        return ""
                }
@@ -48,7 +48,38 @@ func FormatMediaType(t string, param map[string]string) string {
                        return ""
                }
                b.WriteString(strings.ToLower(attribute))
+
+               needEnc := needsEncoding(value)
+               if needEnc {
+                       // RFC 2231 section 4
+                       b.WriteByte('*')
+               }
                b.WriteByte('=')
+
+               if needEnc {
+                       b.WriteString("utf-8''")
+
+                       offset := 0
+                       for index := 0; index < len(value); index++ {
+                               ch := value[index]
+                               // {RFC 2231 section 7}
+                               // attribute-char := <any (US-ASCII) CHAR except SPACE, CTLs, "*", "'", "%", or tspecials>
+                               if ch <= ' ' || ch >= 0x7F ||
+                                       ch == '*' || ch == '\'' || ch == '%' ||
+                                       isTSpecial(rune(ch)) {
+
+                                       b.WriteString(value[offset:index])
+                                       offset = index + 1
+
+                                       b.WriteByte('%')
+                                       b.WriteByte(upperhex[ch>>4])
+                                       b.WriteByte(upperhex[ch&0x0F])
+                               }
+                       }
+                       b.WriteString(value[offset:])
+                       continue
+               }
+
                if isToken(value) {
                        b.WriteString(value)
                        continue
@@ -63,9 +94,6 @@ func FormatMediaType(t string, param map[string]string) string {
                                offset = index
                                b.WriteByte('\\')
                        }
-                       if character&0x80 != 0 {
-                               return ""
-                       }
                }
                b.WriteString(value[offset:])
                b.WriteByte('"')

diff --git a/src/mime/mediatype_test.go b/src/mime/mediatype_test.go
index 945a8189e171c944ed58997368d396adc0bf98bb..e91ff38d68b4892d8ff1b8f1836db51ac0920c6e 100644 (file)
--- a/src/mime/mediatype_test.go
+++ b/src/mime/mediatype_test.go
@@ -6,6 +6,7 @@ package mime
 
 import (
        "reflect"
+       "strings"
        "testing"
 )
 
@@ -481,8 +482,9 @@ var formatTests = []formatTest{
        {"noslash", map[string]string{"X": "Y"}, "noslash; x=Y"}, // e.g. Content-Disposition values (RFC 2183); issue 11289
        {"foo bar/baz", nil, ""},
        {"foo/bar baz", nil, ""},
-       {"attachment", map[string]string{"filename": "ĄĄŽŽČČŠŠ"}, ""},
-       {"attachment", map[string]string{"filename": "ÁÁÊÊÇÇÎÎ"}, ""},
+       {"attachment", map[string]string{"filename": "ĄĄŽŽČČŠŠ"}, "attachment; filename*=utf-8''%C4%84%C4%84%C5%BD%C5%BD%C4%8C%C4%8C%C5%A0%C5%A0"},
+       {"attachment", map[string]string{"filename": "ÁÁÊÊÇÇÎÎ"}, "attachment; filename*=utf-8''%C3%81%C3%81%C3%8A%C3%8A%C3%87%C3%87%C3%8E%C3%8E"},
+       {"attachment", map[string]string{"filename": "数据统计.png"}, "attachment; filename*=utf-8''%E6%95%B0%E6%8D%AE%E7%BB%9F%E8%AE%A1.png"},
        {"foo/BAR", nil, "foo/bar"},
        {"foo/BAR", map[string]string{"X": "Y"}, "foo/bar; x=Y"},
        {"foo/BAR", map[string]string{"space": "With space"}, `foo/bar; space="With space"`},
@@ -491,7 +493,8 @@ var formatTests = []formatTest{
        {"foo/BAR", map[string]string{"both": `With \backslash and "quote`}, `foo/bar; both="With \\backslash and \"quote"`},
        {"foo/BAR", map[string]string{"": "empty attribute"}, ""},
        {"foo/BAR", map[string]string{"bad attribute": "baz"}, ""},
-       {"foo/BAR", map[string]string{"nonascii": "not an ascii character: ä"}, ""},
+       {"foo/BAR", map[string]string{"nonascii": "not an ascii character: ä"}, "foo/bar; nonascii*=utf-8''not%20an%20ascii%20character%3A%20%C3%A4"},
+       {"foo/BAR", map[string]string{"ctl": "newline: \n nil: \000"}, "foo/bar; ctl*=utf-8''newline%3A%20%0A%20nil%3A%20%00"},
        {"foo/bar", map[string]string{"a": "av", "b": "bv", "c": "cv"}, "foo/bar; a=av; b=bv; c=cv"},
        {"foo/bar", map[string]string{"0": "'", "9": "'"}, "foo/bar; 0='; 9='"},
        {"foo", map[string]string{"bar": ""}, `foo; bar=""`},
@@ -503,5 +506,21 @@ func TestFormatMediaType(t *testing.T) {
                if got != tt.want {
                        t.Errorf("%d. FormatMediaType(%q, %v) = %q; want %q", i, tt.typ, tt.params, got, tt.want)
                }
+               if got == "" {
+                       continue
+               }
+               typ, params, err := ParseMediaType(got)
+               if err != nil {
+                       t.Errorf("%d. ParseMediaType(%q) err: %v", i, got, err)
+               }
+               if typ != strings.ToLower(tt.typ) {
+                       t.Errorf("%d. ParseMediaType(%q) typ = %q; want %q", i, got, typ, tt.typ)
+               }
+               for k, v := range tt.params {
+                       k = strings.ToLower(k)
+                       if params[k] != v {
+                               t.Errorf("%d. ParseMediaType(%q) params[%s] = %q; want %q", i, got, k, params[k], v)
+                       }
+               }
        }
 }