enhance sourcemgr to detect various UTF BOM's and emit a fatal error
about it instead of producing tons of garbage from the lexer.
It would be even better for sourcemgr to dynamically transcode (e.g.
from UTF16 -> UTF8).
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@101924 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Basic/SourceManager.cpp b/lib/Basic/SourceManager.cpp
index 053cfe3..c766241 100644
--- a/lib/Basic/SourceManager.cpp
+++ b/lib/Basic/SourceManager.cpp
@@ -119,6 +119,41 @@
Buffer.setInt(true);
#endif
}
+
+ // If the buffer is valid, check to see if it has a UTF Byte Order Mark
+ // (BOM). We only support UTF-8 without a BOM right now. See
+ // http://en.wikipedia.org/wiki/Byte_order_mark for more information.
+ if (!Buffer.getInt()) {
+ llvm::StringRef BufStr = Buffer.getPointer()->getBuffer();
+ const char *BOM = 0;
+ if (BufStr.startswith("\xFE\xBB\xBF"))
+ BOM = "UTF-8";
+ else if (BufStr.startswith("\xFE\xFF"))
+ BOM = "UTF-16 (BE)";
+ else if (BufStr.startswith("\xFF\xFE"))
+ BOM = "UTF-16 (LE)";
+ else if (BufStr.startswith(llvm::StringRef("\x00\x00\xFE\xFF", 4)))
+ BOM = "UTF-32 (BE)";
+ else if (BufStr.startswith(llvm::StringRef("\xFF\xFE\x00\x00", 4)))
+ BOM = "UTF-32 (LE)";
+ else if (BufStr.startswith("\x2B\x2F\x76"))
+ BOM = "UTF-7";
+ else if (BufStr.startswith("\xF7\x64\x4C"))
+ BOM = "UTF-1";
+ else if (BufStr.startswith("\xDD\x73\x66\x73"))
+ BOM = "UTF-EBCDIC";
+ else if (BufStr.startswith("\x0E\xFE\xFF"))
+ BOM = "SDSU";
+ else if (BufStr.startswith("\xFB\xEE\x28"))
+ BOM = "BOCU-1";
+ else if (BufStr.startswith("\x84\x31\x95\x33"))
+ BOM = "BOCU-1";
+
+ if (BOM) {
+ Diag.Report(diag::err_unsupported_bom) << BOM << Entry->getName();
+ Buffer.setInt(1);
+ }
+ }
}
if (Invalid)