Dmitri Gribenko | 5bd1e5b | 2013-01-30 14:29:28 +0000 | [diff] [blame] | 1 | //===--- ClangCommentHTMLNamedCharacterReferenceEmitter.cpp -----------------=// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This tablegen backend emits an fficient function to translate HTML named |
| 11 | // character references to UTF-8 sequences. |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #include "llvm/ADT/SmallString.h" |
| 16 | #include "llvm/Support/ConvertUTF.h" |
| 17 | #include "llvm/TableGen/Error.h" |
| 18 | #include "llvm/TableGen/Record.h" |
| 19 | #include "llvm/TableGen/StringMatcher.h" |
| 20 | #include <vector> |
| 21 | |
| 22 | using namespace llvm; |
| 23 | |
| 24 | /// \brief Convert a code point to the corresponding UTF-8 sequence represented |
| 25 | /// as a C string literal. |
| 26 | /// |
| 27 | /// \returns true on success. |
| 28 | static bool translateCodePointToUTF8(unsigned CodePoint, |
| 29 | SmallVectorImpl<char> &CLiteral) { |
| 30 | char Translated[UNI_MAX_UTF8_BYTES_PER_CODE_POINT]; |
| 31 | char *TranslatedPtr = Translated; |
| 32 | if (!ConvertCodePointToUTF8(CodePoint, TranslatedPtr)) |
| 33 | return false; |
| 34 | |
| 35 | StringRef UTF8(Translated, TranslatedPtr - Translated); |
| 36 | |
| 37 | raw_svector_ostream OS(CLiteral); |
| 38 | OS << "\""; |
| 39 | for (size_t i = 0, e = UTF8.size(); i != e; ++i) { |
| 40 | OS << "\\x"; |
| 41 | OS.write_hex(static_cast<unsigned char>(UTF8[i])); |
| 42 | } |
| 43 | OS << "\""; |
| 44 | |
| 45 | return true; |
| 46 | } |
| 47 | |
| 48 | namespace clang { |
| 49 | void EmitClangCommentHTMLNamedCharacterReferences(RecordKeeper &Records, |
| 50 | raw_ostream &OS) { |
| 51 | std::vector<Record *> Tags = Records.getAllDerivedDefinitions("NCR"); |
| 52 | std::vector<StringMatcher::StringPair> NameToUTF8; |
| 53 | SmallString<32> CLiteral; |
| 54 | for (std::vector<Record *>::iterator I = Tags.begin(), E = Tags.end(); |
| 55 | I != E; ++I) { |
| 56 | Record &Tag = **I; |
| 57 | std::string Spelling = Tag.getValueAsString("Spelling"); |
| 58 | uint64_t CodePoint = Tag.getValueAsInt("CodePoint"); |
| 59 | CLiteral.clear(); |
| 60 | CLiteral.append("return "); |
| 61 | if (!translateCodePointToUTF8(CodePoint, CLiteral)) { |
| 62 | SrcMgr.PrintMessage(Tag.getLoc().front(), |
| 63 | SourceMgr::DK_Error, |
| 64 | Twine("invalid code point")); |
| 65 | continue; |
| 66 | } |
| 67 | CLiteral.append(";"); |
| 68 | |
| 69 | StringMatcher::StringPair Match(Spelling, CLiteral.str()); |
| 70 | NameToUTF8.push_back(Match); |
| 71 | } |
| 72 | |
| 73 | OS << "// This file is generated by TableGen. Do not edit.\n\n"; |
| 74 | |
| 75 | OS << "StringRef translateHTMLNamedCharacterReferenceToUTF8(\n" |
| 76 | " StringRef Name) {\n"; |
| 77 | StringMatcher("Name", NameToUTF8, OS).Emit(); |
| 78 | OS << " return StringRef();\n" |
| 79 | << "}\n\n"; |
| 80 | } |
| 81 | |
| 82 | } // end namespace clang |
| 83 | |