1 //===--- SortJavaScriptImports.h - Sort ES6 Imports -------------*- C++ -*-===//
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 /// \file
11 /// \brief This file implements a sort operation for JavaScript ES6 imports.
12 ///
13 //===----------------------------------------------------------------------===//
14 
15 #include "SortJavaScriptImports.h"
16 #include "SortJavaScriptImports.h"
17 #include "TokenAnalyzer.h"
18 #include "TokenAnnotator.h"
19 #include "clang/Basic/Diagnostic.h"
20 #include "clang/Basic/DiagnosticOptions.h"
21 #include "clang/Basic/LLVM.h"
22 #include "clang/Basic/SourceLocation.h"
23 #include "clang/Basic/SourceManager.h"
24 #include "clang/Format/Format.h"
25 #include "llvm/ADT/STLExtras.h"
26 #include "llvm/ADT/SmallVector.h"
27 #include "llvm/Support/Debug.h"
28 #include <algorithm>
29 #include <string>
30 
31 #define DEBUG_TYPE "format-formatter"
32 
33 namespace clang {
34 namespace format {
35 
36 class FormatTokenLexer;
37 
38 using clang::format::FormatStyle;
39 
40 // An imported symbol in a JavaScript ES6 import/export, possibly aliased.
41 struct JsImportedSymbol {
42   StringRef Symbol;
43   StringRef Alias;
44   SourceRange Range;
45 
46   bool operator==(const JsImportedSymbol &RHS) const {
47     // Ignore Range for comparison, it is only used to stitch code together,
48     // but imports at different code locations are still conceptually the same.
49     return Symbol == RHS.Symbol && Alias == RHS.Alias;
50   }
51 };
52 
53 // An ES6 module reference.
54 //
55 // ES6 implements a module system, where individual modules (~= source files)
56 // can reference other modules, either importing symbols from them, or exporting
57 // symbols from them:
58 //   import {foo} from 'foo';
59 //   export {foo};
60 //   export {bar} from 'bar';
61 //
62 // `export`s with URLs are syntactic sugar for an import of the symbol from the
63 // URL, followed by an export of the symbol, allowing this code to treat both
64 // statements more or less identically, with the exception being that `export`s
65 // are sorted last.
66 //
67 // imports and exports support individual symbols, but also a wildcard syntax:
68 //   import * as prefix from 'foo';
69 //   export * from 'bar';
70 //
71 // This struct represents both exports and imports to build up the information
72 // required for sorting module references.
73 struct JsModuleReference {
74   bool IsExport = false;
75   // Module references are sorted into these categories, in order.
76   enum ReferenceCategory {
77     SIDE_EFFECT,     // "import 'something';"
78     ABSOLUTE,        // from 'something'
79     RELATIVE_PARENT, // from '../*'
80     RELATIVE,        // from './*'
81   };
82   ReferenceCategory Category = ReferenceCategory::SIDE_EFFECT;
83   // The URL imported, e.g. `import .. from 'url';`. Empty for `export {a, b};`.
84   StringRef URL;
85   // Prefix from "import * as prefix". Empty for symbol imports and `export *`.
86   // Implies an empty names list.
87   StringRef Prefix;
88   // Symbols from `import {SymbolA, SymbolB, ...} from ...;`.
89   SmallVector<JsImportedSymbol, 1> Symbols;
90   // Textual position of the import/export, including preceding and trailing
91   // comments.
92   SourceRange Range;
93 };
94 
95 bool operator<(const JsModuleReference &LHS, const JsModuleReference &RHS) {
96   if (LHS.IsExport != RHS.IsExport)
97     return LHS.IsExport < RHS.IsExport;
98   if (LHS.Category != RHS.Category)
99     return LHS.Category < RHS.Category;
100   if (LHS.Category == JsModuleReference::ReferenceCategory::SIDE_EFFECT)
101     // Side effect imports might be ordering sensitive. Consider them equal so
102     // that they maintain their relative order in the stable sort below.
103     // This retains transitivity because LHS.Category == RHS.Category here.
104     return false;
105   // Empty URLs sort *last* (for export {...};).
106   if (LHS.URL.empty() != RHS.URL.empty())
107     return LHS.URL.empty() < RHS.URL.empty();
108   if (LHS.URL != RHS.URL)
109     return LHS.URL < RHS.URL;
110   // '*' imports (with prefix) sort before {a, b, ...} imports.
111   if (LHS.Prefix.empty() != RHS.Prefix.empty())
112     return LHS.Prefix.empty() < RHS.Prefix.empty();
113   if (LHS.Prefix != RHS.Prefix)
114     return LHS.Prefix > RHS.Prefix;
115   return false;
116 }
117 
118 // JavaScriptImportSorter sorts JavaScript ES6 imports and exports. It is
119 // implemented as a TokenAnalyzer because ES6 imports have substantial syntactic
120 // structure, making it messy to sort them using regular expressions.
121 class JavaScriptImportSorter : public TokenAnalyzer {
122 public:
123   JavaScriptImportSorter(const Environment &Env, const FormatStyle &Style)
124       : TokenAnalyzer(Env, Style),
125         FileContents(Env.getSourceManager().getBufferData(Env.getFileID())) {}
126 
127   tooling::Replacements
128   analyze(TokenAnnotator &Annotator,
129           SmallVectorImpl<AnnotatedLine *> &AnnotatedLines,
130           FormatTokenLexer &Tokens, tooling::Replacements &Result) override {
131     AffectedRangeMgr.computeAffectedLines(AnnotatedLines.begin(),
132                                           AnnotatedLines.end());
133 
134     const AdditionalKeywords &Keywords = Tokens.getKeywords();
135     SmallVector<JsModuleReference, 16> References;
136     AnnotatedLine *FirstNonImportLine;
137     std::tie(References, FirstNonImportLine) =
138         parseModuleReferences(Keywords, AnnotatedLines);
139 
140     if (References.empty())
141       return Result;
142 
143     SmallVector<unsigned, 16> Indices;
144     for (unsigned i = 0, e = References.size(); i != e; ++i)
145       Indices.push_back(i);
146     std::stable_sort(Indices.begin(), Indices.end(),
147                      [&](unsigned LHSI, unsigned RHSI) {
148                        return References[LHSI] < References[RHSI];
149                      });
150     bool ReferencesInOrder = std::is_sorted(Indices.begin(), Indices.end());
151 
152     std::string ReferencesText;
153     bool SymbolsInOrder = true;
154     for (unsigned i = 0, e = Indices.size(); i != e; ++i) {
155       JsModuleReference Reference = References[Indices[i]];
156       if (appendReference(ReferencesText, Reference))
157         SymbolsInOrder = false;
158       if (i + 1 < e) {
159         // Insert breaks between imports and exports.
160         ReferencesText += "\n";
161         // Separate imports groups with two line breaks, but keep all exports
162         // in a single group.
163         if (!Reference.IsExport &&
164             (Reference.IsExport != References[Indices[i + 1]].IsExport ||
165              Reference.Category != References[Indices[i + 1]].Category))
166           ReferencesText += "\n";
167       }
168     }
169 
170     if (ReferencesInOrder && SymbolsInOrder)
171       return Result;
172 
173     // Separate references from the main code body of the file.
174     if (FirstNonImportLine && FirstNonImportLine->First->NewlinesBefore < 2)
175       ReferencesText += "\n";
176 
177     SourceRange InsertionPoint = References[0].Range;
178     InsertionPoint.setEnd(References[References.size() - 1].Range.getEnd());
179     DEBUG(llvm::dbgs() << "Replacing imports:\n"
180                        << getSourceText(InsertionPoint) << "\nwith:\n"
181                        << ReferencesText << "\n");
182     Result.insert(tooling::Replacement(
183         Env.getSourceManager(), CharSourceRange::getCharRange(InsertionPoint),
184         ReferencesText));
185 
186     return Result;
187   }
188 
189 private:
190   FormatToken *Current;
191   FormatToken *LineEnd;
192 
193   FormatToken invalidToken;
194 
195   StringRef FileContents;
196 
197   void skipComments() { Current = skipComments(Current); }
198 
199   FormatToken *skipComments(FormatToken *Tok) {
200     while (Tok && Tok->is(tok::comment))
201       Tok = Tok->Next;
202     return Tok;
203   }
204 
205   void nextToken() {
206     Current = Current->Next;
207     skipComments();
208     if (!Current || Current == LineEnd->Next) {
209       // Set the current token to an invalid token, so that further parsing on
210       // this line fails.
211       invalidToken.Tok.setKind(tok::unknown);
212       Current = &invalidToken;
213     }
214   }
215 
216   StringRef getSourceText(SourceRange Range) {
217     return getSourceText(Range.getBegin(), Range.getEnd());
218   }
219 
220   StringRef getSourceText(SourceLocation Begin, SourceLocation End) {
221     const SourceManager &SM = Env.getSourceManager();
222     return FileContents.substr(SM.getFileOffset(Begin),
223                                SM.getFileOffset(End) - SM.getFileOffset(Begin));
224   }
225 
226   // Appends ``Reference`` to ``Buffer``, returning true if text within the
227   // ``Reference`` changed (e.g. symbol order).
228   bool appendReference(std::string &Buffer, JsModuleReference &Reference) {
229     // Sort the individual symbols within the import.
230     // E.g. `import {b, a} from 'x';` -> `import {a, b} from 'x';`
231     SmallVector<JsImportedSymbol, 1> Symbols = Reference.Symbols;
232     std::stable_sort(
233         Symbols.begin(), Symbols.end(),
234         [&](const JsImportedSymbol &LHS, const JsImportedSymbol &RHS) {
235           return LHS.Symbol < RHS.Symbol;
236         });
237     if (Symbols == Reference.Symbols) {
238       // No change in symbol order.
239       StringRef ReferenceStmt = getSourceText(Reference.Range);
240       Buffer += ReferenceStmt;
241       return false;
242     }
243     // Stitch together the module reference start...
244     SourceLocation SymbolsStart = Reference.Symbols.front().Range.getBegin();
245     SourceLocation SymbolsEnd = Reference.Symbols.back().Range.getEnd();
246     Buffer += getSourceText(Reference.Range.getBegin(), SymbolsStart);
247     // ... then the references in order ...
248     for (auto I = Symbols.begin(), E = Symbols.end(); I != E; ++I) {
249       if (I != Symbols.begin())
250         Buffer += ",";
251       Buffer += getSourceText(I->Range);
252     }
253     // ... followed by the module reference end.
254     Buffer += getSourceText(SymbolsEnd, Reference.Range.getEnd());
255     return true;
256   }
257 
258   // Parses module references in the given lines. Returns the module references,
259   // and a pointer to the first "main code" line if that is adjacent to the
260   // affected lines of module references, nullptr otherwise.
261   std::pair<SmallVector<JsModuleReference, 16>, AnnotatedLine*>
262   parseModuleReferences(const AdditionalKeywords &Keywords,
263                         SmallVectorImpl<AnnotatedLine *> &AnnotatedLines) {
264     SmallVector<JsModuleReference, 16> References;
265     SourceLocation Start;
266     bool FoundLines = false;
267     AnnotatedLine *FirstNonImportLine = nullptr;
268     for (auto Line : AnnotatedLines) {
269       if (!Line->Affected) {
270         // Only sort the first contiguous block of affected lines.
271         if (FoundLines)
272           break;
273         else
274           continue;
275       }
276       Current = Line->First;
277       LineEnd = Line->Last;
278       skipComments();
279       if (Start.isInvalid() || References.empty())
280         // After the first file level comment, consider line comments to be part
281         // of the import that immediately follows them by using the previously
282         // set Start.
283         Start = Line->First->Tok.getLocation();
284       if (!Current)
285         continue; // Only comments on this line.
286       FoundLines = true;
287       JsModuleReference Reference;
288       Reference.Range.setBegin(Start);
289       if (!parseModuleReference(Keywords, Reference)) {
290         FirstNonImportLine = Line;
291         break;
292       }
293       Reference.Range.setEnd(LineEnd->Tok.getEndLoc());
294       DEBUG({
295         llvm::dbgs() << "JsModuleReference: {"
296                      << "is_export: " << Reference.IsExport
297                      << ", cat: " << Reference.Category
298                      << ", url: " << Reference.URL
299                      << ", prefix: " << Reference.Prefix;
300         for (size_t i = 0; i < Reference.Symbols.size(); ++i)
301           llvm::dbgs() << ", " << Reference.Symbols[i].Symbol << " as "
302                        << Reference.Symbols[i].Alias;
303         llvm::dbgs() << ", text: " << getSourceText(Reference.Range);
304         llvm::dbgs() << "}\n";
305       });
306       References.push_back(Reference);
307       Start = SourceLocation();
308     }
309     return std::make_pair(References, FirstNonImportLine);
310   }
311 
312   // Parses a JavaScript/ECMAScript 6 module reference.
313   // See http://www.ecma-international.org/ecma-262/6.0/#sec-scripts-and-modules
314   // for grammar EBNF (production ModuleItem).
315   bool parseModuleReference(const AdditionalKeywords &Keywords,
316                             JsModuleReference &Reference) {
317     if (!Current || !Current->isOneOf(Keywords.kw_import, tok::kw_export))
318       return false;
319     Reference.IsExport = Current->is(tok::kw_export);
320 
321     nextToken();
322     if (Current->isStringLiteral() && !Reference.IsExport) {
323       // "import 'side-effect';"
324       Reference.Category = JsModuleReference::ReferenceCategory::SIDE_EFFECT;
325       Reference.URL =
326           Current->TokenText.substr(1, Current->TokenText.size() - 2);
327       return true;
328     }
329 
330     if (!parseModuleBindings(Keywords, Reference))
331       return false;
332     nextToken();
333 
334     if (Current->is(Keywords.kw_from)) {
335       // imports have a 'from' clause, exports might not.
336       nextToken();
337       if (!Current->isStringLiteral())
338         return false;
339       // URL = TokenText without the quotes.
340       Reference.URL =
341           Current->TokenText.substr(1, Current->TokenText.size() - 2);
342       if (Reference.URL.startswith(".."))
343         Reference.Category =
344             JsModuleReference::ReferenceCategory::RELATIVE_PARENT;
345       else if (Reference.URL.startswith("."))
346         Reference.Category = JsModuleReference::ReferenceCategory::RELATIVE;
347       else
348         Reference.Category = JsModuleReference::ReferenceCategory::ABSOLUTE;
349     } else {
350       // w/o URL groups with "empty".
351       Reference.Category = JsModuleReference::ReferenceCategory::RELATIVE;
352     }
353     return true;
354   }
355 
356   bool parseModuleBindings(const AdditionalKeywords &Keywords,
357                            JsModuleReference &Reference) {
358     if (parseStarBinding(Keywords, Reference))
359       return true;
360     return parseNamedBindings(Keywords, Reference);
361   }
362 
363   bool parseStarBinding(const AdditionalKeywords &Keywords,
364                         JsModuleReference &Reference) {
365     // * as prefix from '...';
366     if (Current->isNot(tok::star))
367       return false;
368     nextToken();
369     if (Current->isNot(Keywords.kw_as))
370       return false;
371     nextToken();
372     if (Current->isNot(tok::identifier))
373       return false;
374     Reference.Prefix = Current->TokenText;
375     return true;
376   }
377 
378   bool parseNamedBindings(const AdditionalKeywords &Keywords,
379                           JsModuleReference &Reference) {
380     if (Current->isNot(tok::l_brace))
381       return false;
382 
383     // {sym as alias, sym2 as ...} from '...';
384     nextToken();
385     while (true) {
386       if (Current->isNot(tok::identifier))
387         return false;
388 
389       JsImportedSymbol Symbol;
390       Symbol.Symbol = Current->TokenText;
391       // Make sure to include any preceding comments.
392       Symbol.Range.setBegin(
393           Current->getPreviousNonComment()->Next->WhitespaceRange.getBegin());
394       nextToken();
395 
396       if (Current->is(Keywords.kw_as)) {
397         nextToken();
398         if (Current->isNot(tok::identifier))
399           return false;
400         Symbol.Alias = Current->TokenText;
401         nextToken();
402       }
403       Symbol.Range.setEnd(Current->Tok.getLocation());
404       Reference.Symbols.push_back(Symbol);
405 
406       if (Current->is(tok::r_brace))
407         return true;
408       if (Current->isNot(tok::comma))
409         return false;
410       nextToken();
411     }
412   }
413 };
414 
415 tooling::Replacements sortJavaScriptImports(const FormatStyle &Style,
416                                             StringRef Code,
417                                             ArrayRef<tooling::Range> Ranges,
418                                             StringRef FileName) {
419   // FIXME: Cursor support.
420   std::unique_ptr<Environment> Env =
421       Environment::CreateVirtualEnvironment(Code, FileName, Ranges);
422   JavaScriptImportSorter Sorter(*Env, Style);
423   return Sorter.process();
424 }
425 
426 } // end namespace format
427 } // end namespace clang
428