authorgravatar for johnnymarler@gmail.comJonathan Marler <johnnymarler@gmail.com> 2021-07-01 22:23:10-06:00
committergravatar for git@vexu.euVeikka Tuominen <git@vexu.eu> 2021-07-15 20:37:16+03:00
log08e5daa7d5ff0226ced462308d06b3b17ee4472f
tree1a4693bfd12b81d099b5769a1cd29c6120267d1a
parentee173d512721df521b5a29bc939770e098931e21

Add std.unicode.fmtUtf16le

1 files changed, 43 insertions(+), 2 deletions(-)

lib/std/unicode.zig+43-2
...@@ -317,9 +317,9 @@ pub const Utf16LeIterator = struct {...@@ -317,9 +317,9 @@ pub const Utf16LeIterator = struct {
317 assert(it.i <= it.bytes.len);317 assert(it.i <= it.bytes.len);
318 if (it.i == it.bytes.len) return null;318 if (it.i == it.bytes.len) return null;
319 const c0: u21 = mem.readIntLittle(u16, it.bytes[it.i..][0..2]);319 const c0: u21 = mem.readIntLittle(u16, it.bytes[it.i..][0..2]);
320 it.i += 2;
320 if (c0 & ~@as(u21, 0x03ff) == 0xd800) {321 if (c0 & ~@as(u21, 0x03ff) == 0xd800) {
321 // surrogate pair322 // surrogate pair
322 it.i += 2;
323 if (it.i >= it.bytes.len) return error.DanglingSurrogateHalf;323 if (it.i >= it.bytes.len) return error.DanglingSurrogateHalf;
324 const c1: u21 = mem.readIntLittle(u16, it.bytes[it.i..][0..2]);324 const c1: u21 = mem.readIntLittle(u16, it.bytes[it.i..][0..2]);
325 if (c1 & ~@as(u21, 0x03ff) != 0xdc00) return error.ExpectedSecondSurrogateHalf;325 if (c1 & ~@as(u21, 0x03ff) != 0xdc00) return error.ExpectedSecondSurrogateHalf;
...@@ -328,7 +328,6 @@ pub const Utf16LeIterator = struct {...@@ -328,7 +328,6 @@ pub const Utf16LeIterator = struct {
328 } else if (c0 & ~@as(u21, 0x03ff) == 0xdc00) {328 } else if (c0 & ~@as(u21, 0x03ff) == 0xdc00) {
329 return error.UnexpectedSecondSurrogateHalf;329 return error.UnexpectedSecondSurrogateHalf;
330 } else {330 } else {
331 it.i += 2;
332 return c0;331 return c0;
333 }332 }
334 }333 }
...@@ -769,6 +768,48 @@ fn calcUtf16LeLen(utf8: []const u8) usize {...@@ -769,6 +768,48 @@ fn calcUtf16LeLen(utf8: []const u8) usize {
769 return dest_len;768 return dest_len;
770}769}
771770
771/// Print the given `utf16le` string
772fn formatUtf16le(
773 utf16le: []const u16,
774 comptime fmt: []const u8,
775 options: std.fmt.FormatOptions,
776 writer: anytype,
777) !void {
778 const unknown_codepoint = 0xfffd;
779 _ = fmt;
780 _ = options;
781 var buf: [300]u8 = undefined; // just a random size I chose
782 var it = Utf16LeIterator.init(utf16le);
783 var u8len: usize = 0;
784 while (it.nextCodepoint() catch unknown_codepoint) |codepoint| {
785 u8len += utf8Encode(codepoint, buf[u8len..]) catch
786 utf8Encode(unknown_codepoint, buf[u8len..]) catch unreachable;
787 if (u8len + 3 >= buf.len) {
788 try writer.writeAll(buf[0..u8len]);
789 u8len = 0;
790 }
791 }
792 try writer.writeAll(buf[0..u8len]);
793}
794
795/// Return a Formatter for a Utf16le string
796pub fn fmtUtf16le(utf16le: []const u16) std.fmt.Formatter(formatUtf16le) {
797 return .{ .data = utf16le };
798}
799
800test "fmtUtf16le" {
801 const expectFmt = std.testing.expectFmt;
802 try expectFmt("", "{}", .{fmtUtf16le(utf8ToUtf16LeStringLiteral(""))});
803 try expectFmt("foo", "{}", .{fmtUtf16le(utf8ToUtf16LeStringLiteral("foo"))});
804 try expectFmt("𐐷", "{}", .{fmtUtf16le(utf8ToUtf16LeStringLiteral("𐐷"))});
805 try expectFmt("퟿", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\xff\xd7")})});
806 try expectFmt("�", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\x00\xd8")})});
807 try expectFmt("�", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\xff\xdb")})});
808 try expectFmt("�", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\x00\xdc")})});
809 try expectFmt("�", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\xff\xdf")})});
810 try expectFmt("", "{}", .{fmtUtf16le(&[_]u16{std.mem.readIntNative(u16, "\x00\xe0")})});
811}
812
772test "utf8ToUtf16LeStringLiteral" {813test "utf8ToUtf16LeStringLiteral" {
773 {814 {
774 const bytes = [_:0]u16{815 const bytes = [_:0]u16{