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
35namespace clang {
36namespace 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.
40class FatalDiagnosticConsumer : public DiagnosticConsumer {
41public:
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(OutStr&: Message);
48 llvm::errs() << Message << "\n";
49 }
50 }
51
52 bool fatalError() const { return Fatal; }
53
54private:
55 bool Fatal = false;
56};
57
58std::unique_ptr<Environment>
59Environment::make(StringRef Code, StringRef FileName,
60 ArrayRef<tooling::Range> Ranges, unsigned FirstStartColumn,
61 unsigned NextStartColumn, unsigned LastStartColumn) {
62 auto Env = std::make_unique<Environment>(args&: Code, args&: FileName, args&: FirstStartColumn,
63 args&: NextStartColumn, args&: LastStartColumn);
64 FatalDiagnosticConsumer Diags;
65 Env->SM.getDiagnostics().setClient(client: &Diags, /*ShouldOwnClient=*/false);
66 SourceLocation StartOfFile = Env->SM.getLocForStartOfFile(FID: Env->ID);
67 for (const tooling::Range &Range : Ranges) {
68 SourceLocation Start = StartOfFile.getLocWithOffset(Offset: Range.getOffset());
69 SourceLocation End = Start.getLocWithOffset(Offset: Range.getLength());
70 Env->CharRanges.push_back(Elt: CharSourceRange::getCharRange(B: Start, E: End));
71 }
72 // Validate that we can get the buffer data without a fatal error.
73 Env->SM.getBufferData(FID: Env->ID);
74 if (Diags.fatalError())
75 return nullptr;
76 return Env;
77}
78
79Environment::Environment(StringRef Code, StringRef FileName,
80 unsigned FirstStartColumn, unsigned NextStartColumn,
81 unsigned LastStartColumn)
82 : VirtualSM(new SourceManagerForFile(FileName, Code)), SM(VirtualSM->get()),
83 ID(VirtualSM->get().getMainFileID()), FirstStartColumn(FirstStartColumn),
84 NextStartColumn(NextStartColumn), LastStartColumn(LastStartColumn) {}
85
86TokenAnalyzer::TokenAnalyzer(const Environment &Env, const FormatStyle &Style)
87 : Style(Style), LangOpts(getFormattingLangOpts(Style)), Env(Env),
88 AffectedRangeMgr(Env.getSourceManager(), Env.getCharRanges()),
89 UnwrappedLines(1),
90 Encoding(encoding::detectEncoding(
91 Text: Env.getSourceManager().getBufferData(FID: 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
100std::pair<tooling::Replacements, unsigned>
101TokenAnalyzer::process(bool SkipAnnotation) {
102 tooling::Replacements Result;
103 llvm::SpecificBumpPtrAllocator<FormatToken> Allocator;
104 IdentifierTable IdentTable(LangOpts);
105 FormatTokenLexer Lex(Env.getSourceManager(), Env.getFileID(),
106 Env.getFirstStartColumn(), Style, Encoding, Allocator,
107 IdentTable);
108 ArrayRef<FormatToken *> Toks(Lex.lex());
109 SmallVector<FormatToken *, 10> Tokens(Toks.begin(), Toks.end());
110 UnwrappedLineParser Parser(Env.getSourceManager(), Style, Lex.getKeywords(),
111 Env.getFirstStartColumn(), Tokens, *this,
112 Allocator, IdentTable);
113 Parser.parse();
114 assert(UnwrappedLines.back().empty());
115 unsigned Penalty = 0;
116 for (unsigned Run = 0, RunE = UnwrappedLines.size(); Run + 1 != RunE; ++Run) {
117 const auto &Lines = UnwrappedLines[Run];
118 LLVM_DEBUG(llvm::dbgs() << "Run " << Run << "...\n");
119 SmallVector<AnnotatedLine *, 16> AnnotatedLines;
120 AnnotatedLines.reserve(N: Lines.size());
121
122 TokenAnnotator Annotator(Style, Lex.getKeywords());
123 for (const UnwrappedLine &Line : Lines) {
124 AnnotatedLines.push_back(Elt: new AnnotatedLine(Line));
125 if (!SkipAnnotation)
126 Annotator.annotate(Line&: *AnnotatedLines.back());
127 }
128
129 std::pair<tooling::Replacements, unsigned> RunResult =
130 analyze(Annotator, AnnotatedLines, Tokens&: Lex);
131
132 LLVM_DEBUG({
133 llvm::dbgs() << "Replacements for run " << Run << ":\n";
134 for (const tooling::Replacement &Fix : RunResult.first)
135 llvm::dbgs() << Fix.toString() << "\n";
136 });
137 for (AnnotatedLine *Line : AnnotatedLines)
138 delete Line;
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(E: std::move(Err)) << "\n";
147 return {tooling::Replacements(), 0};
148 }
149 }
150 }
151 return {Result, Penalty};
152}
153
154void TokenAnalyzer::consumeUnwrappedLine(const UnwrappedLine &TheLine) {
155 assert(!UnwrappedLines.empty());
156 UnwrappedLines.back().push_back(Elt: TheLine);
157}
158
159void TokenAnalyzer::finishRun() {
160 UnwrappedLines.push_back(Elt: SmallVector<UnwrappedLine, 16>());
161}
162
163} // end namespace format
164} // end namespace clang
165