1 //===--- SortJavaScriptImports.cpp - Sort ES6 Imports -----------*- 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 a sort operation for JavaScript ES6 imports. 11 /// 12 //===----------------------------------------------------------------------===// 13 14 #include "SortJavaScriptImports.h" 15 #include "TokenAnalyzer.h" 16 #include "TokenAnnotator.h" 17 #include "clang/Basic/Diagnostic.h" 18 #include "clang/Basic/DiagnosticOptions.h" 19 #include "clang/Basic/LLVM.h" 20 #include "clang/Basic/SourceLocation.h" 21 #include "clang/Basic/SourceManager.h" 22 #include "clang/Basic/TokenKinds.h" 23 #include "clang/Format/Format.h" 24 #include "llvm/ADT/STLExtras.h" 25 #include "llvm/ADT/SmallVector.h" 26 #include "llvm/Support/Debug.h" 27 #include <algorithm> 28 #include <string> 29 30 #define DEBUG_TYPE "format-formatter" 31 32 namespace clang { 33 namespace format { 34 35 class FormatTokenLexer; 36 37 using clang::format::FormatStyle; 38 39 // An imported symbol in a JavaScript ES6 import/export, possibly aliased. 40 struct JsImportedSymbol { 41 StringRef Symbol; 42 StringRef Alias; 43 SourceRange Range; 44 45 bool operator==(const JsImportedSymbol &RHS) const { 46 // Ignore Range for comparison, it is only used to stitch code together, 47 // but imports at different code locations are still conceptually the same. 48 return Symbol == RHS.Symbol && Alias == RHS.Alias; 49 } 50 }; 51 52 // An ES6 module reference. 53 // 54 // ES6 implements a module system, where individual modules (~= source files) 55 // can reference other modules, either importing symbols from them, or exporting 56 // symbols from them: 57 // import {foo} from 'foo'; 58 // export {foo}; 59 // export {bar} from 'bar'; 60 // 61 // `export`s with URLs are syntactic sugar for an import of the symbol from the 62 // URL, followed by an export of the symbol, allowing this code to treat both 63 // statements more or less identically, with the exception being that `export`s 64 // are sorted last. 65 // 66 // imports and exports support individual symbols, but also a wildcard syntax: 67 // import * as prefix from 'foo'; 68 // export * from 'bar'; 69 // 70 // This struct represents both exports and imports to build up the information 71 // required for sorting module references. 72 struct JsModuleReference { 73 bool FormattingOff = false; 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 ALIAS, // import X = A.B; 82 }; 83 ReferenceCategory Category = ReferenceCategory::SIDE_EFFECT; 84 // The URL imported, e.g. `import .. from 'url';`. Empty for `export {a, b};`. 85 StringRef URL; 86 // Prefix from "import * as prefix". Empty for symbol imports and `export *`. 87 // Implies an empty names list. 88 StringRef Prefix; 89 // Default import from "import DefaultName from '...';". 90 StringRef DefaultImport; 91 // Symbols from `import {SymbolA, SymbolB, ...} from ...;`. 92 SmallVector<JsImportedSymbol, 1> Symbols; 93 // Whether some symbols were merged into this one. Controls if the module 94 // reference needs re-formatting. 95 bool SymbolsMerged = false; 96 // The source location just after { and just before } in the import. 97 // Extracted eagerly to allow modification of Symbols later on. 98 SourceLocation SymbolsStart, SymbolsEnd; 99 // Textual position of the import/export, including preceding and trailing 100 // comments. 101 SourceRange Range; 102 }; 103 104 bool operator<(const JsModuleReference &LHS, const JsModuleReference &RHS) { 105 if (LHS.IsExport != RHS.IsExport) 106 return LHS.IsExport < RHS.IsExport; 107 if (LHS.Category != RHS.Category) 108 return LHS.Category < RHS.Category; 109 if (LHS.Category == JsModuleReference::ReferenceCategory::SIDE_EFFECT || 110 LHS.Category == JsModuleReference::ReferenceCategory::ALIAS) { 111 // Side effect imports and aliases might be ordering sensitive. Consider 112 // them equal so that they maintain their relative order in the stable sort 113 // below. This retains transitivity because LHS.Category == RHS.Category 114 // here. 115 return false; 116 } 117 // Empty URLs sort *last* (for export {...};). 118 if (LHS.URL.empty() != RHS.URL.empty()) 119 return LHS.URL.empty() < RHS.URL.empty(); 120 if (int Res = LHS.URL.compare_insensitive(RHS.URL)) 121 return Res < 0; 122 // '*' imports (with prefix) sort before {a, b, ...} imports. 123 if (LHS.Prefix.empty() != RHS.Prefix.empty()) 124 return LHS.Prefix.empty() < RHS.Prefix.empty(); 125 if (LHS.Prefix != RHS.Prefix) 126 return LHS.Prefix > RHS.Prefix; 127 return false; 128 } 129 130 // JavaScriptImportSorter sorts JavaScript ES6 imports and exports. It is 131 // implemented as a TokenAnalyzer because ES6 imports have substantial syntactic 132 // structure, making it messy to sort them using regular expressions. 133 class JavaScriptImportSorter : public TokenAnalyzer { 134 public: 135 JavaScriptImportSorter(const Environment &Env, const FormatStyle &Style) 136 : TokenAnalyzer(Env, Style), 137 FileContents(Env.getSourceManager().getBufferData(Env.getFileID())) { 138 // FormatToken.Tok starts out in an uninitialized state. 139 invalidToken.Tok.startToken(); 140 } 141 142 std::pair<tooling::Replacements, unsigned> 143 analyze(TokenAnnotator &Annotator, 144 SmallVectorImpl<AnnotatedLine *> &AnnotatedLines, 145 FormatTokenLexer &Tokens) override { 146 tooling::Replacements Result; 147 AffectedRangeMgr.computeAffectedLines(AnnotatedLines); 148 149 const AdditionalKeywords &Keywords = Tokens.getKeywords(); 150 SmallVector<JsModuleReference, 16> References; 151 AnnotatedLine *FirstNonImportLine; 152 std::tie(References, FirstNonImportLine) = 153 parseModuleReferences(Keywords, AnnotatedLines); 154 155 if (References.empty()) 156 return {Result, 0}; 157 158 // The text range of all parsed imports, to be replaced later. 159 SourceRange InsertionPoint = References[0].Range; 160 InsertionPoint.setEnd(References[References.size() - 1].Range.getEnd()); 161 162 References = sortModuleReferences(References); 163 164 std::string ReferencesText; 165 for (unsigned I = 0, E = References.size(); I != E; ++I) { 166 JsModuleReference Reference = References[I]; 167 appendReference(ReferencesText, Reference); 168 if (I + 1 < E) { 169 // Insert breaks between imports and exports. 170 ReferencesText += "\n"; 171 // Separate imports groups with two line breaks, but keep all exports 172 // in a single group. 173 if (!Reference.IsExport && 174 (Reference.IsExport != References[I + 1].IsExport || 175 Reference.Category != References[I + 1].Category)) { 176 ReferencesText += "\n"; 177 } 178 } 179 } 180 llvm::StringRef PreviousText = getSourceText(InsertionPoint); 181 if (ReferencesText == PreviousText) 182 return {Result, 0}; 183 184 // The loop above might collapse previously existing line breaks between 185 // import blocks, and thus shrink the file. SortIncludes must not shrink 186 // overall source length as there is currently no re-calculation of ranges 187 // after applying source sorting. 188 // This loop just backfills trailing spaces after the imports, which are 189 // harmless and will be stripped by the subsequent formatting pass. 190 // FIXME: A better long term fix is to re-calculate Ranges after sorting. 191 unsigned PreviousSize = PreviousText.size(); 192 while (ReferencesText.size() < PreviousSize) 193 ReferencesText += " "; 194 195 // Separate references from the main code body of the file. 196 if (FirstNonImportLine && FirstNonImportLine->First->NewlinesBefore < 2 && 197 !(FirstNonImportLine->First->is(tok::comment) && 198 FirstNonImportLine->First->TokenText.trim() == 199 "// clang-format on")) { 200 ReferencesText += "\n"; 201 } 202 203 LLVM_DEBUG(llvm::dbgs() << "Replacing imports:\n" 204 << PreviousText << "\nwith:\n" 205 << ReferencesText << "\n"); 206 auto Err = Result.add(tooling::Replacement( 207 Env.getSourceManager(), CharSourceRange::getCharRange(InsertionPoint), 208 ReferencesText)); 209 // FIXME: better error handling. For now, just print error message and skip 210 // the replacement for the release version. 211 if (Err) { 212 llvm::errs() << llvm::toString(std::move(Err)) << "\n"; 213 assert(false); 214 } 215 216 return {Result, 0}; 217 } 218 219 private: 220 FormatToken *Current; 221 FormatToken *LineEnd; 222 223 FormatToken invalidToken; 224 225 StringRef FileContents; 226 227 void skipComments() { Current = skipComments(Current); } 228 229 FormatToken *skipComments(FormatToken *Tok) { 230 while (Tok && Tok->is(tok::comment)) 231 Tok = Tok->Next; 232 return Tok; 233 } 234 235 void nextToken() { 236 Current = Current->Next; 237 skipComments(); 238 if (!Current || Current == LineEnd->Next) { 239 // Set the current token to an invalid token, so that further parsing on 240 // this line fails. 241 Current = &invalidToken; 242 } 243 } 244 245 StringRef getSourceText(SourceRange Range) { 246 return getSourceText(Range.getBegin(), Range.getEnd()); 247 } 248 249 StringRef getSourceText(SourceLocation Begin, SourceLocation End) { 250 const SourceManager &SM = Env.getSourceManager(); 251 return FileContents.substr(SM.getFileOffset(Begin), 252 SM.getFileOffset(End) - SM.getFileOffset(Begin)); 253 } 254 255 // Sorts the given module references. 256 // Imports can have formatting disabled (FormattingOff), so the code below 257 // skips runs of "no-formatting" module references, and sorts/merges the 258 // references that have formatting enabled in individual chunks. 259 SmallVector<JsModuleReference, 16> 260 sortModuleReferences(const SmallVector<JsModuleReference, 16> &References) { 261 // Sort module references. 262 // Imports can have formatting disabled (FormattingOff), so the code below 263 // skips runs of "no-formatting" module references, and sorts other 264 // references per group. 265 const auto *Start = References.begin(); 266 SmallVector<JsModuleReference, 16> ReferencesSorted; 267 while (Start != References.end()) { 268 while (Start != References.end() && Start->FormattingOff) { 269 // Skip over all imports w/ disabled formatting. 270 ReferencesSorted.push_back(*Start); 271 ++Start; 272 } 273 SmallVector<JsModuleReference, 16> SortChunk; 274 while (Start != References.end() && !Start->FormattingOff) { 275 // Skip over all imports w/ disabled formatting. 276 SortChunk.push_back(*Start); 277 ++Start; 278 } 279 llvm::stable_sort(SortChunk); 280 mergeModuleReferences(SortChunk); 281 ReferencesSorted.insert(ReferencesSorted.end(), SortChunk.begin(), 282 SortChunk.end()); 283 } 284 return ReferencesSorted; 285 } 286 287 // Merge module references. 288 // After sorting, find all references that import named symbols from the 289 // same URL and merge their names. E.g. 290 // import {X} from 'a'; 291 // import {Y} from 'a'; 292 // should be rewritten to: 293 // import {X, Y} from 'a'; 294 // Note: this modifies the passed in ``References`` vector (by removing no 295 // longer needed references). 296 void mergeModuleReferences(SmallVector<JsModuleReference, 16> &References) { 297 if (References.empty()) 298 return; 299 JsModuleReference *PreviousReference = References.begin(); 300 auto *Reference = std::next(References.begin()); 301 while (Reference != References.end()) { 302 // Skip: 303 // import 'foo'; 304 // import * as foo from 'foo'; on either previous or this. 305 // import Default from 'foo'; on either previous or this. 306 // mismatching 307 if (Reference->Category == JsModuleReference::SIDE_EFFECT || 308 PreviousReference->Category == JsModuleReference::SIDE_EFFECT || 309 Reference->IsExport != PreviousReference->IsExport || 310 !PreviousReference->Prefix.empty() || !Reference->Prefix.empty() || 311 !PreviousReference->DefaultImport.empty() || 312 !Reference->DefaultImport.empty() || Reference->Symbols.empty() || 313 PreviousReference->URL != Reference->URL) { 314 PreviousReference = Reference; 315 ++Reference; 316 continue; 317 } 318 // Merge symbols from identical imports. 319 PreviousReference->Symbols.append(Reference->Symbols); 320 PreviousReference->SymbolsMerged = true; 321 // Remove the merged import. 322 Reference = References.erase(Reference); 323 } 324 } 325 326 // Appends ``Reference`` to ``Buffer``. 327 void appendReference(std::string &Buffer, JsModuleReference &Reference) { 328 if (Reference.FormattingOff) { 329 Buffer += 330 getSourceText(Reference.Range.getBegin(), Reference.Range.getEnd()); 331 return; 332 } 333 // Sort the individual symbols within the import. 334 // E.g. `import {b, a} from 'x';` -> `import {a, b} from 'x';` 335 SmallVector<JsImportedSymbol, 1> Symbols = Reference.Symbols; 336 llvm::stable_sort( 337 Symbols, [&](const JsImportedSymbol &LHS, const JsImportedSymbol &RHS) { 338 return LHS.Symbol.compare_insensitive(RHS.Symbol) < 0; 339 }); 340 if (!Reference.SymbolsMerged && Symbols == Reference.Symbols) { 341 // Symbols didn't change, just emit the entire module reference. 342 StringRef ReferenceStmt = getSourceText(Reference.Range); 343 Buffer += ReferenceStmt; 344 return; 345 } 346 // Stitch together the module reference start... 347 Buffer += getSourceText(Reference.Range.getBegin(), Reference.SymbolsStart); 348 // ... then the references in order ... 349 if (!Symbols.empty()) { 350 Buffer += getSourceText(Symbols.front().Range); 351 for (const JsImportedSymbol &Symbol : llvm::drop_begin(Symbols)) { 352 Buffer += ","; 353 Buffer += getSourceText(Symbol.Range); 354 } 355 } 356 // ... followed by the module reference end. 357 Buffer += getSourceText(Reference.SymbolsEnd, Reference.Range.getEnd()); 358 } 359 360 // Parses module references in the given lines. Returns the module references, 361 // and a pointer to the first "main code" line if that is adjacent to the 362 // affected lines of module references, nullptr otherwise. 363 std::pair<SmallVector<JsModuleReference, 16>, AnnotatedLine *> 364 parseModuleReferences(const AdditionalKeywords &Keywords, 365 SmallVectorImpl<AnnotatedLine *> &AnnotatedLines) { 366 SmallVector<JsModuleReference, 16> References; 367 SourceLocation Start; 368 AnnotatedLine *FirstNonImportLine = nullptr; 369 bool AnyImportAffected = false; 370 bool FormattingOff = false; 371 for (auto *Line : AnnotatedLines) { 372 assert(Line->First); 373 Current = Line->First; 374 LineEnd = Line->Last; 375 // clang-format comments toggle formatting on/off. 376 // This is tracked in FormattingOff here and on JsModuleReference. 377 while (Current && Current->is(tok::comment)) { 378 StringRef CommentText = Current->TokenText.trim(); 379 if (CommentText == "// clang-format off") { 380 FormattingOff = true; 381 } else if (CommentText == "// clang-format on") { 382 FormattingOff = false; 383 // Special case: consider a trailing "clang-format on" line to be part 384 // of the module reference, so that it gets moved around together with 385 // it (as opposed to the next module reference, which might get sorted 386 // around). 387 if (!References.empty()) { 388 References.back().Range.setEnd(Current->Tok.getEndLoc()); 389 Start = Current->Tok.getEndLoc().getLocWithOffset(1); 390 } 391 } 392 // Handle all clang-format comments on a line, e.g. for an empty block. 393 Current = Current->Next; 394 } 395 skipComments(); 396 if (Start.isInvalid() || References.empty()) { 397 // After the first file level comment, consider line comments to be part 398 // of the import that immediately follows them by using the previously 399 // set Start. 400 Start = Line->First->Tok.getLocation(); 401 } 402 if (!Current) { 403 // Only comments on this line. Could be the first non-import line. 404 FirstNonImportLine = Line; 405 continue; 406 } 407 JsModuleReference Reference; 408 Reference.FormattingOff = FormattingOff; 409 Reference.Range.setBegin(Start); 410 // References w/o a URL, e.g. export {A}, groups with RELATIVE. 411 Reference.Category = JsModuleReference::ReferenceCategory::RELATIVE; 412 if (!parseModuleReference(Keywords, Reference)) { 413 if (!FirstNonImportLine) 414 FirstNonImportLine = Line; // if no comment before. 415 break; 416 } 417 FirstNonImportLine = nullptr; 418 AnyImportAffected = AnyImportAffected || Line->Affected; 419 Reference.Range.setEnd(LineEnd->Tok.getEndLoc()); 420 LLVM_DEBUG({ 421 llvm::dbgs() << "JsModuleReference: {" 422 << "formatting_off: " << Reference.FormattingOff 423 << ", is_export: " << Reference.IsExport 424 << ", cat: " << Reference.Category 425 << ", url: " << Reference.URL 426 << ", prefix: " << Reference.Prefix; 427 for (const JsImportedSymbol &Symbol : Reference.Symbols) 428 llvm::dbgs() << ", " << Symbol.Symbol << " as " << Symbol.Alias; 429 llvm::dbgs() << ", text: " << getSourceText(Reference.Range); 430 llvm::dbgs() << "}\n"; 431 }); 432 References.push_back(Reference); 433 Start = SourceLocation(); 434 } 435 // Sort imports if any import line was affected. 436 if (!AnyImportAffected) 437 References.clear(); 438 return std::make_pair(References, FirstNonImportLine); 439 } 440 441 // Parses a JavaScript/ECMAScript 6 module reference. 442 // See http://www.ecma-international.org/ecma-262/6.0/#sec-scripts-and-modules 443 // for grammar EBNF (production ModuleItem). 444 bool parseModuleReference(const AdditionalKeywords &Keywords, 445 JsModuleReference &Reference) { 446 if (!Current || !Current->isOneOf(Keywords.kw_import, tok::kw_export)) 447 return false; 448 Reference.IsExport = Current->is(tok::kw_export); 449 450 nextToken(); 451 if (Current->isStringLiteral() && !Reference.IsExport) { 452 // "import 'side-effect';" 453 Reference.Category = JsModuleReference::ReferenceCategory::SIDE_EFFECT; 454 Reference.URL = 455 Current->TokenText.substr(1, Current->TokenText.size() - 2); 456 return true; 457 } 458 459 if (!parseModuleBindings(Keywords, Reference)) 460 return false; 461 462 if (Current->is(Keywords.kw_from)) { 463 // imports have a 'from' clause, exports might not. 464 nextToken(); 465 if (!Current->isStringLiteral()) 466 return false; 467 // URL = TokenText without the quotes. 468 Reference.URL = 469 Current->TokenText.substr(1, Current->TokenText.size() - 2); 470 if (Reference.URL.startswith("..")) { 471 Reference.Category = 472 JsModuleReference::ReferenceCategory::RELATIVE_PARENT; 473 } else if (Reference.URL.startswith(".")) { 474 Reference.Category = JsModuleReference::ReferenceCategory::RELATIVE; 475 } else { 476 Reference.Category = JsModuleReference::ReferenceCategory::ABSOLUTE; 477 } 478 } 479 return true; 480 } 481 482 bool parseModuleBindings(const AdditionalKeywords &Keywords, 483 JsModuleReference &Reference) { 484 if (parseStarBinding(Keywords, Reference)) 485 return true; 486 return parseNamedBindings(Keywords, Reference); 487 } 488 489 bool parseStarBinding(const AdditionalKeywords &Keywords, 490 JsModuleReference &Reference) { 491 // * as prefix from '...'; 492 if (Current->isNot(tok::star)) 493 return false; 494 nextToken(); 495 if (Current->isNot(Keywords.kw_as)) 496 return false; 497 nextToken(); 498 if (Current->isNot(tok::identifier)) 499 return false; 500 Reference.Prefix = Current->TokenText; 501 nextToken(); 502 return true; 503 } 504 505 bool parseNamedBindings(const AdditionalKeywords &Keywords, 506 JsModuleReference &Reference) { 507 // eat a potential "import X, " prefix. 508 if (Current->is(tok::identifier)) { 509 Reference.DefaultImport = Current->TokenText; 510 nextToken(); 511 if (Current->is(Keywords.kw_from)) 512 return true; 513 // import X = A.B.C; 514 if (Current->is(tok::equal)) { 515 Reference.Category = JsModuleReference::ReferenceCategory::ALIAS; 516 nextToken(); 517 while (Current->is(tok::identifier)) { 518 nextToken(); 519 if (Current->is(tok::semi)) 520 return true; 521 if (!Current->is(tok::period)) 522 return false; 523 nextToken(); 524 } 525 } 526 if (Current->isNot(tok::comma)) 527 return false; 528 nextToken(); // eat comma. 529 } 530 if (Current->isNot(tok::l_brace)) 531 return false; 532 533 // {sym as alias, sym2 as ...} from '...'; 534 Reference.SymbolsStart = Current->Tok.getEndLoc(); 535 while (Current->isNot(tok::r_brace)) { 536 nextToken(); 537 if (Current->is(tok::r_brace)) 538 break; 539 if (!Current->isOneOf(tok::identifier, tok::kw_default)) 540 return false; 541 542 JsImportedSymbol Symbol; 543 Symbol.Symbol = Current->TokenText; 544 // Make sure to include any preceding comments. 545 Symbol.Range.setBegin( 546 Current->getPreviousNonComment()->Next->WhitespaceRange.getBegin()); 547 nextToken(); 548 549 if (Current->is(Keywords.kw_as)) { 550 nextToken(); 551 if (!Current->isOneOf(tok::identifier, tok::kw_default)) 552 return false; 553 Symbol.Alias = Current->TokenText; 554 nextToken(); 555 } 556 Symbol.Range.setEnd(Current->Tok.getLocation()); 557 Reference.Symbols.push_back(Symbol); 558 559 if (!Current->isOneOf(tok::r_brace, tok::comma)) 560 return false; 561 } 562 Reference.SymbolsEnd = Current->Tok.getLocation(); 563 // For named imports with a trailing comma ("import {X,}"), consider the 564 // comma to be the end of the import list, so that it doesn't get removed. 565 if (Current->Previous->is(tok::comma)) 566 Reference.SymbolsEnd = Current->Previous->Tok.getLocation(); 567 nextToken(); // consume r_brace 568 return true; 569 } 570 }; 571 572 tooling::Replacements sortJavaScriptImports(const FormatStyle &Style, 573 StringRef Code, 574 ArrayRef<tooling::Range> Ranges, 575 StringRef FileName) { 576 // FIXME: Cursor support. 577 auto Env = Environment::make(Code, FileName, Ranges); 578 if (!Env) 579 return {}; 580 return JavaScriptImportSorter(*Env, Style).process().first; 581 } 582 583 } // end namespace format 584 } // end namespace clang 585