translate-c: Add Wide, UTF-16, and UTF-32 character literals

Add support for L'<wchar_t>', u'<char16_t>', and U'<char32_t>'. Currently
this just translates wide char literals to \u{NNNNNN} escape codes
(e.g. U'💯' -> '\u{1f4af}')

Another approach would be to emit UTF-8 encoded character literals
directly, but in my opinion this approaches Unicode-complete because it
would require knowledge of which Unicode codepoints have graphical
representations for the emitted source to be readable.

We could also just emit integer literals, but the current method makes
it clear that we have translated a wide character literal and not just
an integer constant.
This commit is contained in:
Evan Haas 2021-01-17 19:22:48 -08:00 committed by Andrew Kelley
parent 384ccaa27a
commit c3dadfa95b
2 changed files with 28 additions and 8 deletions

View File

@ -2942,9 +2942,9 @@ fn transCharLiteral(
suppress_as: SuppressCast,
) TransError!*ast.Node {
const kind = stmt.getKind();
const val = stmt.getValue();
const int_lit_node = switch (kind) {
.Ascii, .UTF8 => blk: {
const val = stmt.getValue();
if (kind == .Ascii) {
// C has a somewhat obscure feature called multi-character character
// constant
@ -2960,13 +2960,15 @@ fn transCharLiteral(
};
break :blk &node.base;
},
.UTF16, .UTF32, .Wide => return revertAndWarn(
rp,
error.UnsupportedTranslation,
@ptrCast(*const clang.Stmt, stmt).getBeginLoc(),
"TODO: support character literal kind {}",
.{kind},
),
.Wide, .UTF16, .UTF32 => blk: {
const token = try appendTokenFmt(rp.c, .CharLiteral, "'\\u{{{x}}}'", .{val});
const node = try rp.c.arena.create(ast.Node.OneToken);
node.* = .{
.base = .{ .tag = .CharLiteral },
.token = token,
};
break :blk &node.base;
},
};
if (suppress_as == .no_as) {
return maybeSuppressResult(rp, scope, result_used, int_lit_node);

View File

@ -719,4 +719,22 @@ pub fn addCases(cases: *tests.RunTranslatedCContext) void {
\\ return 0;
\\}
, "");
cases.add("Wide, UTF-16, and UTF-32 character literals",
\\#include <wchar.h>
\\#include <stdlib.h>
\\int main() {
\\ wchar_t wc = L'™';
\\ int utf16_char = u'™';
\\ int utf32_char = U'💯';
\\ if (wc != 8482) abort();
\\ if (utf16_char != 8482) abort();
\\ if (utf32_char != 128175) abort();
\\ unsigned char c = wc;
\\ if (c != 0x22) abort();
\\ c = utf32_char;
\\ if (c != 0xaf) abort();
\\ return 0;
\\}
, "");
}