1 //===--- TokenAnalyzer.cpp - Analyze Token Streams --------------*- C++ -*-===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 /// 9 /// \file 10 /// This file implements an abstract TokenAnalyzer and associated helper 11 /// classes. TokenAnalyzer can be extended to generate replacements based on 12 /// an annotated and pre-processed token stream. 13 /// 14 //===----------------------------------------------------------------------===// 15 16 #include "TokenAnalyzer.h" 17 #include "AffectedRangeManager.h" 18 #include "Encoding.h" 19 #include "FormatToken.h" 20 #include "FormatTokenLexer.h" 21 #include "TokenAnnotator.h" 22 #include "UnwrappedLineParser.h" 23 #include "clang/Basic/Diagnostic.h" 24 #include "clang/Basic/DiagnosticOptions.h" 25 #include "clang/Basic/FileManager.h" 26 #include "clang/Basic/SourceManager.h" 27 #include "clang/Format/Format.h" 28 #include "llvm/ADT/STLExtras.h" 29 #include "llvm/ADT/SmallVector.h" 30 #include "llvm/Support/Debug.h" 31 #include <type_traits> 32 33 #define DEBUG_TYPE "format-formatter" 34 35 namespace clang { 36 namespace format { 37 38 // FIXME: Instead of printing the diagnostic we should store it and have a 39 // better way to return errors through the format APIs. 40 class FatalDiagnosticConsumer: public DiagnosticConsumer { 41 public: 42 void HandleDiagnostic(DiagnosticsEngine::Level DiagLevel, 43 const Diagnostic &Info) override { 44 if (DiagLevel == DiagnosticsEngine::Fatal) { 45 Fatal = true; 46 llvm::SmallVector<char, 128> Message; 47 Info.FormatDiagnostic(Message); 48 llvm::errs() << Message << "\n"; 49 } 50 } 51 52 bool fatalError() const { return Fatal; } 53 54 private: 55 bool Fatal = false; 56 }; 57 58 std::unique_ptr<Environment> 59 Environment::make(StringRef Code, StringRef FileName, 60 ArrayRef<tooling::Range> Ranges, unsigned FirstStartColumn, 61 unsigned NextStartColumn, unsigned LastStartColumn) { 62 auto Env = std::make_unique<Environment>(Code, FileName, FirstStartColumn, 63 NextStartColumn, LastStartColumn); 64 FatalDiagnosticConsumer Diags; 65 Env->SM.getDiagnostics().setClient(&Diags, /*ShouldOwnClient=*/false); 66 SourceLocation StartOfFile = Env->SM.getLocForStartOfFile(Env->ID); 67 for (const tooling::Range &Range : Ranges) { 68 SourceLocation Start = StartOfFile.getLocWithOffset(Range.getOffset()); 69 SourceLocation End = Start.getLocWithOffset(Range.getLength()); 70 Env->CharRanges.push_back(CharSourceRange::getCharRange(Start, End)); 71 } 72 // Validate that we can get the buffer data without a fatal error. 73 Env->SM.getBufferData(Env->ID); 74 if (Diags.fatalError()) return nullptr; 75 return Env; 76 } 77 78 Environment::Environment(StringRef Code, StringRef FileName, 79 unsigned FirstStartColumn, unsigned NextStartColumn, 80 unsigned LastStartColumn) 81 : VirtualSM(new SourceManagerForFile(FileName, Code)), SM(VirtualSM->get()), 82 ID(VirtualSM->get().getMainFileID()), FirstStartColumn(FirstStartColumn), 83 NextStartColumn(NextStartColumn), LastStartColumn(LastStartColumn) { 84 } 85 86 TokenAnalyzer::TokenAnalyzer(const Environment &Env, const FormatStyle &Style) 87 : Style(Style), Env(Env), 88 AffectedRangeMgr(Env.getSourceManager(), Env.getCharRanges()), 89 UnwrappedLines(1), 90 Encoding(encoding::detectEncoding( 91 Env.getSourceManager().getBufferData(Env.getFileID()))) { 92 LLVM_DEBUG( 93 llvm::dbgs() << "File encoding: " 94 << (Encoding == encoding::Encoding_UTF8 ? "UTF8" : "unknown") 95 << "\n"); 96 LLVM_DEBUG(llvm::dbgs() << "Language: " << getLanguageName(Style.Language) 97 << "\n"); 98 } 99 100 std::pair<tooling::Replacements, unsigned> TokenAnalyzer::process() { 101 tooling::Replacements Result; 102 llvm::SpecificBumpPtrAllocator<FormatToken> Allocator; 103 IdentifierTable IdentTable(getFormattingLangOpts(Style)); 104 FormatTokenLexer Lex(Env.getSourceManager(), Env.getFileID(), 105 Env.getFirstStartColumn(), Style, Encoding, Allocator, 106 107 IdentTable); 108 ArrayRef<FormatToken *> Toks(Lex.lex()); 109 SmallVector<FormatToken *, 10> Tokens(Toks.begin(), Toks.end()); 110 UnwrappedLineParser Parser(Style, Lex.getKeywords(), 111 Env.getFirstStartColumn(), Tokens, *this); 112 Parser.parse(); 113 assert(UnwrappedLines.rbegin()->empty()); 114 unsigned Penalty = 0; 115 for (unsigned Run = 0, RunE = UnwrappedLines.size(); Run + 1 != RunE; ++Run) { 116 LLVM_DEBUG(llvm::dbgs() << "Run " << Run << "...\n"); 117 SmallVector<AnnotatedLine *, 16> AnnotatedLines; 118 119 TokenAnnotator Annotator(Style, Lex.getKeywords()); 120 for (unsigned i = 0, e = UnwrappedLines[Run].size(); i != e; ++i) { 121 AnnotatedLines.push_back(new AnnotatedLine(UnwrappedLines[Run][i])); 122 Annotator.annotate(*AnnotatedLines.back()); 123 } 124 125 std::pair<tooling::Replacements, unsigned> RunResult = 126 analyze(Annotator, AnnotatedLines, Lex); 127 128 LLVM_DEBUG({ 129 llvm::dbgs() << "Replacements for run " << Run << ":\n"; 130 for (tooling::Replacements::const_iterator I = RunResult.first.begin(), 131 E = RunResult.first.end(); 132 I != E; ++I) { 133 llvm::dbgs() << I->toString() << "\n"; 134 } 135 }); 136 for (unsigned i = 0, e = AnnotatedLines.size(); i != e; ++i) { 137 delete AnnotatedLines[i]; 138 } 139 140 Penalty += RunResult.second; 141 for (const auto &R : RunResult.first) { 142 auto Err = Result.add(R); 143 // FIXME: better error handling here. For now, simply return an empty 144 // Replacements to indicate failure. 145 if (Err) { 146 llvm::errs() << llvm::toString(std::move(Err)) << "\n"; 147 return {tooling::Replacements(), 0}; 148 } 149 } 150 } 151 return {Result, Penalty}; 152 } 153 154 void TokenAnalyzer::consumeUnwrappedLine(const UnwrappedLine &TheLine) { 155 assert(!UnwrappedLines.empty()); 156 UnwrappedLines.back().push_back(TheLine); 157 } 158 159 void TokenAnalyzer::finishRun() { 160 UnwrappedLines.push_back(SmallVector<UnwrappedLine, 16>()); 161 } 162 163 } // end namespace format 164 } // end namespace clang 165