1 //===- VirtualFileSystem.cpp - Virtual File System Layer --------*- 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 // This file implements the VirtualFileSystem interface.
10 //===----------------------------------------------------------------------===//
11
12 #include "clang/Basic/VirtualFileSystem.h"
13 #include "llvm/ADT/DenseMap.h"
14 #include "llvm/ADT/iterator_range.h"
15 #include "llvm/ADT/STLExtras.h"
16 #include "llvm/ADT/StringExtras.h"
17 #include "llvm/ADT/StringSet.h"
18 #include "llvm/Support/Errc.h"
19 #include "llvm/Support/MemoryBuffer.h"
20 #include "llvm/Support/Path.h"
21 #include "llvm/Support/YAMLParser.h"
22 #include <atomic>
23 #include <memory>
24
25 using namespace clang;
26 using namespace clang::vfs;
27 using namespace llvm;
28 using llvm::sys::fs::file_status;
29 using llvm::sys::fs::file_type;
30 using llvm::sys::fs::perms;
31 using llvm::sys::fs::UniqueID;
32
Status(const file_status & Status)33 Status::Status(const file_status &Status)
34 : UID(Status.getUniqueID()), MTime(Status.getLastModificationTime()),
35 User(Status.getUser()), Group(Status.getGroup()), Size(Status.getSize()),
36 Type(Status.type()), Perms(Status.permissions()), IsVFSMapped(false) {}
37
Status(StringRef Name,StringRef ExternalName,UniqueID UID,sys::TimeValue MTime,uint32_t User,uint32_t Group,uint64_t Size,file_type Type,perms Perms)38 Status::Status(StringRef Name, StringRef ExternalName, UniqueID UID,
39 sys::TimeValue MTime, uint32_t User, uint32_t Group,
40 uint64_t Size, file_type Type, perms Perms)
41 : Name(Name), UID(UID), MTime(MTime), User(User), Group(Group), Size(Size),
42 Type(Type), Perms(Perms), IsVFSMapped(false) {}
43
equivalent(const Status & Other) const44 bool Status::equivalent(const Status &Other) const {
45 return getUniqueID() == Other.getUniqueID();
46 }
isDirectory() const47 bool Status::isDirectory() const {
48 return Type == file_type::directory_file;
49 }
isRegularFile() const50 bool Status::isRegularFile() const {
51 return Type == file_type::regular_file;
52 }
isOther() const53 bool Status::isOther() const {
54 return exists() && !isRegularFile() && !isDirectory() && !isSymlink();
55 }
isSymlink() const56 bool Status::isSymlink() const {
57 return Type == file_type::symlink_file;
58 }
isStatusKnown() const59 bool Status::isStatusKnown() const {
60 return Type != file_type::status_error;
61 }
exists() const62 bool Status::exists() const {
63 return isStatusKnown() && Type != file_type::file_not_found;
64 }
65
~File()66 File::~File() {}
67
~FileSystem()68 FileSystem::~FileSystem() {}
69
getBufferForFile(const llvm::Twine & Name,std::unique_ptr<MemoryBuffer> & Result,int64_t FileSize,bool RequiresNullTerminator,bool IsVolatile)70 std::error_code FileSystem::getBufferForFile(
71 const llvm::Twine &Name, std::unique_ptr<MemoryBuffer> &Result,
72 int64_t FileSize, bool RequiresNullTerminator, bool IsVolatile) {
73 std::unique_ptr<File> F;
74 if (std::error_code EC = openFileForRead(Name, F))
75 return EC;
76
77 std::error_code EC =
78 F->getBuffer(Name, Result, FileSize, RequiresNullTerminator, IsVolatile);
79 return EC;
80 }
81
82 //===-----------------------------------------------------------------------===/
83 // RealFileSystem implementation
84 //===-----------------------------------------------------------------------===/
85
86 namespace {
87 /// \brief Wrapper around a raw file descriptor.
88 class RealFile : public File {
89 int FD;
90 Status S;
91 friend class RealFileSystem;
RealFile(int FD)92 RealFile(int FD) : FD(FD) {
93 assert(FD >= 0 && "Invalid or inactive file descriptor");
94 }
95
96 public:
97 ~RealFile();
98 ErrorOr<Status> status() override;
99 std::error_code getBuffer(const Twine &Name,
100 std::unique_ptr<MemoryBuffer> &Result,
101 int64_t FileSize = -1,
102 bool RequiresNullTerminator = true,
103 bool IsVolatile = false) override;
104 std::error_code close() override;
105 void setName(StringRef Name) override;
106 };
107 } // end anonymous namespace
~RealFile()108 RealFile::~RealFile() { close(); }
109
status()110 ErrorOr<Status> RealFile::status() {
111 assert(FD != -1 && "cannot stat closed file");
112 if (!S.isStatusKnown()) {
113 file_status RealStatus;
114 if (std::error_code EC = sys::fs::status(FD, RealStatus))
115 return EC;
116 Status NewS(RealStatus);
117 NewS.setName(S.getName());
118 S = std::move(NewS);
119 }
120 return S;
121 }
122
getBuffer(const Twine & Name,std::unique_ptr<MemoryBuffer> & Result,int64_t FileSize,bool RequiresNullTerminator,bool IsVolatile)123 std::error_code RealFile::getBuffer(const Twine &Name,
124 std::unique_ptr<MemoryBuffer> &Result,
125 int64_t FileSize,
126 bool RequiresNullTerminator,
127 bool IsVolatile) {
128 assert(FD != -1 && "cannot get buffer for closed file");
129 ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
130 MemoryBuffer::getOpenFile(FD, Name.str().c_str(), FileSize,
131 RequiresNullTerminator, IsVolatile);
132 if (std::error_code EC = BufferOrErr.getError())
133 return EC;
134 Result = std::move(BufferOrErr.get());
135 return std::error_code();
136 }
137
138 // FIXME: This is terrible, we need this for ::close.
139 #if !defined(_MSC_VER) && !defined(__MINGW32__)
140 #include <unistd.h>
141 #include <sys/uio.h>
142 #else
143 #include <io.h>
144 #ifndef S_ISFIFO
145 #define S_ISFIFO(x) (0)
146 #endif
147 #endif
close()148 std::error_code RealFile::close() {
149 if (::close(FD))
150 return std::error_code(errno, std::generic_category());
151 FD = -1;
152 return std::error_code();
153 }
154
setName(StringRef Name)155 void RealFile::setName(StringRef Name) {
156 S.setName(Name);
157 }
158
159 namespace {
160 /// \brief The file system according to your operating system.
161 class RealFileSystem : public FileSystem {
162 public:
163 ErrorOr<Status> status(const Twine &Path) override;
164 std::error_code openFileForRead(const Twine &Path,
165 std::unique_ptr<File> &Result) override;
166 directory_iterator dir_begin(const Twine &Dir, std::error_code &EC) override;
167 };
168 } // end anonymous namespace
169
status(const Twine & Path)170 ErrorOr<Status> RealFileSystem::status(const Twine &Path) {
171 sys::fs::file_status RealStatus;
172 if (std::error_code EC = sys::fs::status(Path, RealStatus))
173 return EC;
174 Status Result(RealStatus);
175 Result.setName(Path.str());
176 return Result;
177 }
178
openFileForRead(const Twine & Name,std::unique_ptr<File> & Result)179 std::error_code RealFileSystem::openFileForRead(const Twine &Name,
180 std::unique_ptr<File> &Result) {
181 int FD;
182 if (std::error_code EC = sys::fs::openFileForRead(Name, FD))
183 return EC;
184 Result.reset(new RealFile(FD));
185 Result->setName(Name.str());
186 return std::error_code();
187 }
188
getRealFileSystem()189 IntrusiveRefCntPtr<FileSystem> vfs::getRealFileSystem() {
190 static IntrusiveRefCntPtr<FileSystem> FS = new RealFileSystem();
191 return FS;
192 }
193
194 namespace {
195 class RealFSDirIter : public clang::vfs::detail::DirIterImpl {
196 std::string Path;
197 llvm::sys::fs::directory_iterator Iter;
198 public:
RealFSDirIter(const Twine & _Path,std::error_code & EC)199 RealFSDirIter(const Twine &_Path, std::error_code &EC)
200 : Path(_Path.str()), Iter(Path, EC) {
201 if (!EC && Iter != llvm::sys::fs::directory_iterator()) {
202 llvm::sys::fs::file_status S;
203 EC = Iter->status(S);
204 if (!EC) {
205 CurrentEntry = Status(S);
206 CurrentEntry.setName(Iter->path());
207 }
208 }
209 }
210
increment()211 std::error_code increment() override {
212 std::error_code EC;
213 Iter.increment(EC);
214 if (EC) {
215 return EC;
216 } else if (Iter == llvm::sys::fs::directory_iterator()) {
217 CurrentEntry = Status();
218 } else {
219 llvm::sys::fs::file_status S;
220 EC = Iter->status(S);
221 CurrentEntry = Status(S);
222 CurrentEntry.setName(Iter->path());
223 }
224 return EC;
225 }
226 };
227 }
228
dir_begin(const Twine & Dir,std::error_code & EC)229 directory_iterator RealFileSystem::dir_begin(const Twine &Dir,
230 std::error_code &EC) {
231 return directory_iterator(std::make_shared<RealFSDirIter>(Dir, EC));
232 }
233
234 //===-----------------------------------------------------------------------===/
235 // OverlayFileSystem implementation
236 //===-----------------------------------------------------------------------===/
OverlayFileSystem(IntrusiveRefCntPtr<FileSystem> BaseFS)237 OverlayFileSystem::OverlayFileSystem(IntrusiveRefCntPtr<FileSystem> BaseFS) {
238 pushOverlay(BaseFS);
239 }
240
pushOverlay(IntrusiveRefCntPtr<FileSystem> FS)241 void OverlayFileSystem::pushOverlay(IntrusiveRefCntPtr<FileSystem> FS) {
242 FSList.push_back(FS);
243 }
244
status(const Twine & Path)245 ErrorOr<Status> OverlayFileSystem::status(const Twine &Path) {
246 // FIXME: handle symlinks that cross file systems
247 for (iterator I = overlays_begin(), E = overlays_end(); I != E; ++I) {
248 ErrorOr<Status> Status = (*I)->status(Path);
249 if (Status || Status.getError() != llvm::errc::no_such_file_or_directory)
250 return Status;
251 }
252 return make_error_code(llvm::errc::no_such_file_or_directory);
253 }
254
255 std::error_code
openFileForRead(const llvm::Twine & Path,std::unique_ptr<File> & Result)256 OverlayFileSystem::openFileForRead(const llvm::Twine &Path,
257 std::unique_ptr<File> &Result) {
258 // FIXME: handle symlinks that cross file systems
259 for (iterator I = overlays_begin(), E = overlays_end(); I != E; ++I) {
260 std::error_code EC = (*I)->openFileForRead(Path, Result);
261 if (!EC || EC != llvm::errc::no_such_file_or_directory)
262 return EC;
263 }
264 return make_error_code(llvm::errc::no_such_file_or_directory);
265 }
266
~DirIterImpl()267 clang::vfs::detail::DirIterImpl::~DirIterImpl() { }
268
269 namespace {
270 class OverlayFSDirIterImpl : public clang::vfs::detail::DirIterImpl {
271 OverlayFileSystem &Overlays;
272 std::string Path;
273 OverlayFileSystem::iterator CurrentFS;
274 directory_iterator CurrentDirIter;
275 llvm::StringSet<> SeenNames;
276
incrementFS()277 std::error_code incrementFS() {
278 assert(CurrentFS != Overlays.overlays_end() && "incrementing past end");
279 ++CurrentFS;
280 for (auto E = Overlays.overlays_end(); CurrentFS != E; ++CurrentFS) {
281 std::error_code EC;
282 CurrentDirIter = (*CurrentFS)->dir_begin(Path, EC);
283 if (EC && EC != errc::no_such_file_or_directory)
284 return EC;
285 if (CurrentDirIter != directory_iterator())
286 break; // found
287 }
288 return std::error_code();
289 }
290
incrementDirIter(bool IsFirstTime)291 std::error_code incrementDirIter(bool IsFirstTime) {
292 assert((IsFirstTime || CurrentDirIter != directory_iterator()) &&
293 "incrementing past end");
294 std::error_code EC;
295 if (!IsFirstTime)
296 CurrentDirIter.increment(EC);
297 if (!EC && CurrentDirIter == directory_iterator())
298 EC = incrementFS();
299 return EC;
300 }
301
incrementImpl(bool IsFirstTime)302 std::error_code incrementImpl(bool IsFirstTime) {
303 while (true) {
304 std::error_code EC = incrementDirIter(IsFirstTime);
305 if (EC || CurrentDirIter == directory_iterator()) {
306 CurrentEntry = Status();
307 return EC;
308 }
309 CurrentEntry = *CurrentDirIter;
310 StringRef Name = llvm::sys::path::filename(CurrentEntry.getName());
311 if (SeenNames.insert(Name))
312 return EC; // name not seen before
313 }
314 llvm_unreachable("returned above");
315 }
316
317 public:
OverlayFSDirIterImpl(const Twine & Path,OverlayFileSystem & FS,std::error_code & EC)318 OverlayFSDirIterImpl(const Twine &Path, OverlayFileSystem &FS,
319 std::error_code &EC)
320 : Overlays(FS), Path(Path.str()), CurrentFS(Overlays.overlays_begin()) {
321 CurrentDirIter = (*CurrentFS)->dir_begin(Path, EC);
322 EC = incrementImpl(true);
323 }
324
increment()325 std::error_code increment() override { return incrementImpl(false); }
326 };
327 } // end anonymous namespace
328
dir_begin(const Twine & Dir,std::error_code & EC)329 directory_iterator OverlayFileSystem::dir_begin(const Twine &Dir,
330 std::error_code &EC) {
331 return directory_iterator(
332 std::make_shared<OverlayFSDirIterImpl>(Dir, *this, EC));
333 }
334
335 //===-----------------------------------------------------------------------===/
336 // VFSFromYAML implementation
337 //===-----------------------------------------------------------------------===/
338
339 // Allow DenseMap<StringRef, ...>. This is useful below because we know all the
340 // strings are literals and will outlive the map, and there is no reason to
341 // store them.
342 namespace llvm {
343 template<>
344 struct DenseMapInfo<StringRef> {
345 // This assumes that "" will never be a valid key.
getEmptyKeyllvm::DenseMapInfo346 static inline StringRef getEmptyKey() { return StringRef(""); }
getTombstoneKeyllvm::DenseMapInfo347 static inline StringRef getTombstoneKey() { return StringRef(); }
getHashValuellvm::DenseMapInfo348 static unsigned getHashValue(StringRef Val) { return HashString(Val); }
isEqualllvm::DenseMapInfo349 static bool isEqual(StringRef LHS, StringRef RHS) { return LHS == RHS; }
350 };
351 }
352
353 namespace {
354
355 enum EntryKind {
356 EK_Directory,
357 EK_File
358 };
359
360 /// \brief A single file or directory in the VFS.
361 class Entry {
362 EntryKind Kind;
363 std::string Name;
364
365 public:
366 virtual ~Entry();
Entry(EntryKind K,StringRef Name)367 Entry(EntryKind K, StringRef Name) : Kind(K), Name(Name) {}
getName() const368 StringRef getName() const { return Name; }
getKind() const369 EntryKind getKind() const { return Kind; }
370 };
371
372 class DirectoryEntry : public Entry {
373 std::vector<Entry *> Contents;
374 Status S;
375
376 public:
377 virtual ~DirectoryEntry();
DirectoryEntry(StringRef Name,std::vector<Entry * > Contents,Status S)378 DirectoryEntry(StringRef Name, std::vector<Entry *> Contents, Status S)
379 : Entry(EK_Directory, Name), Contents(std::move(Contents)),
380 S(std::move(S)) {}
getStatus()381 Status getStatus() { return S; }
382 typedef std::vector<Entry *>::iterator iterator;
contents_begin()383 iterator contents_begin() { return Contents.begin(); }
contents_end()384 iterator contents_end() { return Contents.end(); }
classof(const Entry * E)385 static bool classof(const Entry *E) { return E->getKind() == EK_Directory; }
386 };
387
388 class FileEntry : public Entry {
389 public:
390 enum NameKind {
391 NK_NotSet,
392 NK_External,
393 NK_Virtual
394 };
395 private:
396 std::string ExternalContentsPath;
397 NameKind UseName;
398 public:
FileEntry(StringRef Name,StringRef ExternalContentsPath,NameKind UseName)399 FileEntry(StringRef Name, StringRef ExternalContentsPath, NameKind UseName)
400 : Entry(EK_File, Name), ExternalContentsPath(ExternalContentsPath),
401 UseName(UseName) {}
getExternalContentsPath() const402 StringRef getExternalContentsPath() const { return ExternalContentsPath; }
403 /// \brief whether to use the external path as the name for this file.
useExternalName(bool GlobalUseExternalName) const404 bool useExternalName(bool GlobalUseExternalName) const {
405 return UseName == NK_NotSet ? GlobalUseExternalName
406 : (UseName == NK_External);
407 }
classof(const Entry * E)408 static bool classof(const Entry *E) { return E->getKind() == EK_File; }
409 };
410
411 class VFSFromYAML;
412
413 class VFSFromYamlDirIterImpl : public clang::vfs::detail::DirIterImpl {
414 std::string Dir;
415 VFSFromYAML &FS;
416 DirectoryEntry::iterator Current, End;
417 public:
418 VFSFromYamlDirIterImpl(const Twine &Path, VFSFromYAML &FS,
419 DirectoryEntry::iterator Begin,
420 DirectoryEntry::iterator End, std::error_code &EC);
421 std::error_code increment() override;
422 };
423
424 /// \brief A virtual file system parsed from a YAML file.
425 ///
426 /// Currently, this class allows creating virtual directories and mapping
427 /// virtual file paths to existing external files, available in \c ExternalFS.
428 ///
429 /// The basic structure of the parsed file is:
430 /// \verbatim
431 /// {
432 /// 'version': <version number>,
433 /// <optional configuration>
434 /// 'roots': [
435 /// <directory entries>
436 /// ]
437 /// }
438 /// \endverbatim
439 ///
440 /// All configuration options are optional.
441 /// 'case-sensitive': <boolean, default=true>
442 /// 'use-external-names': <boolean, default=true>
443 ///
444 /// Virtual directories are represented as
445 /// \verbatim
446 /// {
447 /// 'type': 'directory',
448 /// 'name': <string>,
449 /// 'contents': [ <file or directory entries> ]
450 /// }
451 /// \endverbatim
452 ///
453 /// The default attributes for virtual directories are:
454 /// \verbatim
455 /// MTime = now() when created
456 /// Perms = 0777
457 /// User = Group = 0
458 /// Size = 0
459 /// UniqueID = unspecified unique value
460 /// \endverbatim
461 ///
462 /// Re-mapped files are represented as
463 /// \verbatim
464 /// {
465 /// 'type': 'file',
466 /// 'name': <string>,
467 /// 'use-external-name': <boolean> # Optional
468 /// 'external-contents': <path to external file>)
469 /// }
470 /// \endverbatim
471 ///
472 /// and inherit their attributes from the external contents.
473 ///
474 /// In both cases, the 'name' field may contain multiple path components (e.g.
475 /// /path/to/file). However, any directory that contains more than one child
476 /// must be uniquely represented by a directory entry.
477 class VFSFromYAML : public vfs::FileSystem {
478 std::vector<Entry *> Roots; ///< The root(s) of the virtual file system.
479 /// \brief The file system to use for external references.
480 IntrusiveRefCntPtr<FileSystem> ExternalFS;
481
482 /// @name Configuration
483 /// @{
484
485 /// \brief Whether to perform case-sensitive comparisons.
486 ///
487 /// Currently, case-insensitive matching only works correctly with ASCII.
488 bool CaseSensitive;
489
490 /// \brief Whether to use to use the value of 'external-contents' for the
491 /// names of files. This global value is overridable on a per-file basis.
492 bool UseExternalNames;
493 /// @}
494
495 friend class VFSFromYAMLParser;
496
497 private:
VFSFromYAML(IntrusiveRefCntPtr<FileSystem> ExternalFS)498 VFSFromYAML(IntrusiveRefCntPtr<FileSystem> ExternalFS)
499 : ExternalFS(ExternalFS), CaseSensitive(true), UseExternalNames(true) {}
500
501 /// \brief Looks up \p Path in \c Roots.
502 ErrorOr<Entry *> lookupPath(const Twine &Path);
503
504 /// \brief Looks up the path <tt>[Start, End)</tt> in \p From, possibly
505 /// recursing into the contents of \p From if it is a directory.
506 ErrorOr<Entry *> lookupPath(sys::path::const_iterator Start,
507 sys::path::const_iterator End, Entry *From);
508
509 /// \brief Get the status of a given an \c Entry.
510 ErrorOr<Status> status(const Twine &Path, Entry *E);
511
512 public:
513 ~VFSFromYAML();
514
515 /// \brief Parses \p Buffer, which is expected to be in YAML format and
516 /// returns a virtual file system representing its contents.
517 ///
518 /// Takes ownership of \p Buffer.
519 static VFSFromYAML *create(MemoryBuffer *Buffer,
520 SourceMgr::DiagHandlerTy DiagHandler,
521 void *DiagContext,
522 IntrusiveRefCntPtr<FileSystem> ExternalFS);
523
524 ErrorOr<Status> status(const Twine &Path) override;
525 std::error_code openFileForRead(const Twine &Path,
526 std::unique_ptr<File> &Result) override;
527
dir_begin(const Twine & Dir,std::error_code & EC)528 directory_iterator dir_begin(const Twine &Dir, std::error_code &EC) override{
529 ErrorOr<Entry *> E = lookupPath(Dir);
530 if (!E) {
531 EC = E.getError();
532 return directory_iterator();
533 }
534 ErrorOr<Status> S = status(Dir, *E);
535 if (!S) {
536 EC = S.getError();
537 return directory_iterator();
538 }
539 if (!S->isDirectory()) {
540 EC = std::error_code(static_cast<int>(errc::not_a_directory),
541 std::system_category());
542 return directory_iterator();
543 }
544
545 DirectoryEntry *D = cast<DirectoryEntry>(*E);
546 return directory_iterator(std::make_shared<VFSFromYamlDirIterImpl>(Dir,
547 *this, D->contents_begin(), D->contents_end(), EC));
548 }
549 };
550
551 /// \brief A helper class to hold the common YAML parsing state.
552 class VFSFromYAMLParser {
553 yaml::Stream &Stream;
554
error(yaml::Node * N,const Twine & Msg)555 void error(yaml::Node *N, const Twine &Msg) {
556 Stream.printError(N, Msg);
557 }
558
559 // false on error
parseScalarString(yaml::Node * N,StringRef & Result,SmallVectorImpl<char> & Storage)560 bool parseScalarString(yaml::Node *N, StringRef &Result,
561 SmallVectorImpl<char> &Storage) {
562 yaml::ScalarNode *S = dyn_cast<yaml::ScalarNode>(N);
563 if (!S) {
564 error(N, "expected string");
565 return false;
566 }
567 Result = S->getValue(Storage);
568 return true;
569 }
570
571 // false on error
parseScalarBool(yaml::Node * N,bool & Result)572 bool parseScalarBool(yaml::Node *N, bool &Result) {
573 SmallString<5> Storage;
574 StringRef Value;
575 if (!parseScalarString(N, Value, Storage))
576 return false;
577
578 if (Value.equals_lower("true") || Value.equals_lower("on") ||
579 Value.equals_lower("yes") || Value == "1") {
580 Result = true;
581 return true;
582 } else if (Value.equals_lower("false") || Value.equals_lower("off") ||
583 Value.equals_lower("no") || Value == "0") {
584 Result = false;
585 return true;
586 }
587
588 error(N, "expected boolean value");
589 return false;
590 }
591
592 struct KeyStatus {
KeyStatus__anon63d25d9d0511::VFSFromYAMLParser::KeyStatus593 KeyStatus(bool Required=false) : Required(Required), Seen(false) {}
594 bool Required;
595 bool Seen;
596 };
597 typedef std::pair<StringRef, KeyStatus> KeyStatusPair;
598
599 // false on error
checkDuplicateOrUnknownKey(yaml::Node * KeyNode,StringRef Key,DenseMap<StringRef,KeyStatus> & Keys)600 bool checkDuplicateOrUnknownKey(yaml::Node *KeyNode, StringRef Key,
601 DenseMap<StringRef, KeyStatus> &Keys) {
602 if (!Keys.count(Key)) {
603 error(KeyNode, "unknown key");
604 return false;
605 }
606 KeyStatus &S = Keys[Key];
607 if (S.Seen) {
608 error(KeyNode, Twine("duplicate key '") + Key + "'");
609 return false;
610 }
611 S.Seen = true;
612 return true;
613 }
614
615 // false on error
checkMissingKeys(yaml::Node * Obj,DenseMap<StringRef,KeyStatus> & Keys)616 bool checkMissingKeys(yaml::Node *Obj, DenseMap<StringRef, KeyStatus> &Keys) {
617 for (DenseMap<StringRef, KeyStatus>::iterator I = Keys.begin(),
618 E = Keys.end();
619 I != E; ++I) {
620 if (I->second.Required && !I->second.Seen) {
621 error(Obj, Twine("missing key '") + I->first + "'");
622 return false;
623 }
624 }
625 return true;
626 }
627
parseEntry(yaml::Node * N)628 Entry *parseEntry(yaml::Node *N) {
629 yaml::MappingNode *M = dyn_cast<yaml::MappingNode>(N);
630 if (!M) {
631 error(N, "expected mapping node for file or directory entry");
632 return nullptr;
633 }
634
635 KeyStatusPair Fields[] = {
636 KeyStatusPair("name", true),
637 KeyStatusPair("type", true),
638 KeyStatusPair("contents", false),
639 KeyStatusPair("external-contents", false),
640 KeyStatusPair("use-external-name", false),
641 };
642
643 DenseMap<StringRef, KeyStatus> Keys(
644 &Fields[0], Fields + sizeof(Fields)/sizeof(Fields[0]));
645
646 bool HasContents = false; // external or otherwise
647 std::vector<Entry *> EntryArrayContents;
648 std::string ExternalContentsPath;
649 std::string Name;
650 FileEntry::NameKind UseExternalName = FileEntry::NK_NotSet;
651 EntryKind Kind;
652
653 for (yaml::MappingNode::iterator I = M->begin(), E = M->end(); I != E;
654 ++I) {
655 StringRef Key;
656 // Reuse the buffer for key and value, since we don't look at key after
657 // parsing value.
658 SmallString<256> Buffer;
659 if (!parseScalarString(I->getKey(), Key, Buffer))
660 return nullptr;
661
662 if (!checkDuplicateOrUnknownKey(I->getKey(), Key, Keys))
663 return nullptr;
664
665 StringRef Value;
666 if (Key == "name") {
667 if (!parseScalarString(I->getValue(), Value, Buffer))
668 return nullptr;
669 Name = Value;
670 } else if (Key == "type") {
671 if (!parseScalarString(I->getValue(), Value, Buffer))
672 return nullptr;
673 if (Value == "file")
674 Kind = EK_File;
675 else if (Value == "directory")
676 Kind = EK_Directory;
677 else {
678 error(I->getValue(), "unknown value for 'type'");
679 return nullptr;
680 }
681 } else if (Key == "contents") {
682 if (HasContents) {
683 error(I->getKey(),
684 "entry already has 'contents' or 'external-contents'");
685 return nullptr;
686 }
687 HasContents = true;
688 yaml::SequenceNode *Contents =
689 dyn_cast<yaml::SequenceNode>(I->getValue());
690 if (!Contents) {
691 // FIXME: this is only for directories, what about files?
692 error(I->getValue(), "expected array");
693 return nullptr;
694 }
695
696 for (yaml::SequenceNode::iterator I = Contents->begin(),
697 E = Contents->end();
698 I != E; ++I) {
699 if (Entry *E = parseEntry(&*I))
700 EntryArrayContents.push_back(E);
701 else
702 return nullptr;
703 }
704 } else if (Key == "external-contents") {
705 if (HasContents) {
706 error(I->getKey(),
707 "entry already has 'contents' or 'external-contents'");
708 return nullptr;
709 }
710 HasContents = true;
711 if (!parseScalarString(I->getValue(), Value, Buffer))
712 return nullptr;
713 ExternalContentsPath = Value;
714 } else if (Key == "use-external-name") {
715 bool Val;
716 if (!parseScalarBool(I->getValue(), Val))
717 return nullptr;
718 UseExternalName = Val ? FileEntry::NK_External : FileEntry::NK_Virtual;
719 } else {
720 llvm_unreachable("key missing from Keys");
721 }
722 }
723
724 if (Stream.failed())
725 return nullptr;
726
727 // check for missing keys
728 if (!HasContents) {
729 error(N, "missing key 'contents' or 'external-contents'");
730 return nullptr;
731 }
732 if (!checkMissingKeys(N, Keys))
733 return nullptr;
734
735 // check invalid configuration
736 if (Kind == EK_Directory && UseExternalName != FileEntry::NK_NotSet) {
737 error(N, "'use-external-name' is not supported for directories");
738 return nullptr;
739 }
740
741 // Remove trailing slash(es), being careful not to remove the root path
742 StringRef Trimmed(Name);
743 size_t RootPathLen = sys::path::root_path(Trimmed).size();
744 while (Trimmed.size() > RootPathLen &&
745 sys::path::is_separator(Trimmed.back()))
746 Trimmed = Trimmed.slice(0, Trimmed.size()-1);
747 // Get the last component
748 StringRef LastComponent = sys::path::filename(Trimmed);
749
750 Entry *Result = nullptr;
751 switch (Kind) {
752 case EK_File:
753 Result = new FileEntry(LastComponent, std::move(ExternalContentsPath),
754 UseExternalName);
755 break;
756 case EK_Directory:
757 Result = new DirectoryEntry(LastComponent, std::move(EntryArrayContents),
758 Status("", "", getNextVirtualUniqueID(), sys::TimeValue::now(), 0, 0,
759 0, file_type::directory_file, sys::fs::all_all));
760 break;
761 }
762
763 StringRef Parent = sys::path::parent_path(Trimmed);
764 if (Parent.empty())
765 return Result;
766
767 // if 'name' contains multiple components, create implicit directory entries
768 for (sys::path::reverse_iterator I = sys::path::rbegin(Parent),
769 E = sys::path::rend(Parent);
770 I != E; ++I) {
771 Result = new DirectoryEntry(*I, llvm::makeArrayRef(Result),
772 Status("", "", getNextVirtualUniqueID(), sys::TimeValue::now(), 0, 0,
773 0, file_type::directory_file, sys::fs::all_all));
774 }
775 return Result;
776 }
777
778 public:
VFSFromYAMLParser(yaml::Stream & S)779 VFSFromYAMLParser(yaml::Stream &S) : Stream(S) {}
780
781 // false on error
parse(yaml::Node * Root,VFSFromYAML * FS)782 bool parse(yaml::Node *Root, VFSFromYAML *FS) {
783 yaml::MappingNode *Top = dyn_cast<yaml::MappingNode>(Root);
784 if (!Top) {
785 error(Root, "expected mapping node");
786 return false;
787 }
788
789 KeyStatusPair Fields[] = {
790 KeyStatusPair("version", true),
791 KeyStatusPair("case-sensitive", false),
792 KeyStatusPair("use-external-names", false),
793 KeyStatusPair("roots", true),
794 };
795
796 DenseMap<StringRef, KeyStatus> Keys(
797 &Fields[0], Fields + sizeof(Fields)/sizeof(Fields[0]));
798
799 // Parse configuration and 'roots'
800 for (yaml::MappingNode::iterator I = Top->begin(), E = Top->end(); I != E;
801 ++I) {
802 SmallString<10> KeyBuffer;
803 StringRef Key;
804 if (!parseScalarString(I->getKey(), Key, KeyBuffer))
805 return false;
806
807 if (!checkDuplicateOrUnknownKey(I->getKey(), Key, Keys))
808 return false;
809
810 if (Key == "roots") {
811 yaml::SequenceNode *Roots = dyn_cast<yaml::SequenceNode>(I->getValue());
812 if (!Roots) {
813 error(I->getValue(), "expected array");
814 return false;
815 }
816
817 for (yaml::SequenceNode::iterator I = Roots->begin(), E = Roots->end();
818 I != E; ++I) {
819 if (Entry *E = parseEntry(&*I))
820 FS->Roots.push_back(E);
821 else
822 return false;
823 }
824 } else if (Key == "version") {
825 StringRef VersionString;
826 SmallString<4> Storage;
827 if (!parseScalarString(I->getValue(), VersionString, Storage))
828 return false;
829 int Version;
830 if (VersionString.getAsInteger<int>(10, Version)) {
831 error(I->getValue(), "expected integer");
832 return false;
833 }
834 if (Version < 0) {
835 error(I->getValue(), "invalid version number");
836 return false;
837 }
838 if (Version != 0) {
839 error(I->getValue(), "version mismatch, expected 0");
840 return false;
841 }
842 } else if (Key == "case-sensitive") {
843 if (!parseScalarBool(I->getValue(), FS->CaseSensitive))
844 return false;
845 } else if (Key == "use-external-names") {
846 if (!parseScalarBool(I->getValue(), FS->UseExternalNames))
847 return false;
848 } else {
849 llvm_unreachable("key missing from Keys");
850 }
851 }
852
853 if (Stream.failed())
854 return false;
855
856 if (!checkMissingKeys(Top, Keys))
857 return false;
858 return true;
859 }
860 };
861 } // end of anonymous namespace
862
~Entry()863 Entry::~Entry() {}
~DirectoryEntry()864 DirectoryEntry::~DirectoryEntry() { llvm::DeleteContainerPointers(Contents); }
865
~VFSFromYAML()866 VFSFromYAML::~VFSFromYAML() { llvm::DeleteContainerPointers(Roots); }
867
create(MemoryBuffer * Buffer,SourceMgr::DiagHandlerTy DiagHandler,void * DiagContext,IntrusiveRefCntPtr<FileSystem> ExternalFS)868 VFSFromYAML *VFSFromYAML::create(MemoryBuffer *Buffer,
869 SourceMgr::DiagHandlerTy DiagHandler,
870 void *DiagContext,
871 IntrusiveRefCntPtr<FileSystem> ExternalFS) {
872
873 SourceMgr SM;
874 yaml::Stream Stream(Buffer, SM);
875
876 SM.setDiagHandler(DiagHandler, DiagContext);
877 yaml::document_iterator DI = Stream.begin();
878 yaml::Node *Root = DI->getRoot();
879 if (DI == Stream.end() || !Root) {
880 SM.PrintMessage(SMLoc(), SourceMgr::DK_Error, "expected root node");
881 return nullptr;
882 }
883
884 VFSFromYAMLParser P(Stream);
885
886 std::unique_ptr<VFSFromYAML> FS(new VFSFromYAML(ExternalFS));
887 if (!P.parse(Root, FS.get()))
888 return nullptr;
889
890 return FS.release();
891 }
892
lookupPath(const Twine & Path_)893 ErrorOr<Entry *> VFSFromYAML::lookupPath(const Twine &Path_) {
894 SmallString<256> Path;
895 Path_.toVector(Path);
896
897 // Handle relative paths
898 if (std::error_code EC = sys::fs::make_absolute(Path))
899 return EC;
900
901 if (Path.empty())
902 return make_error_code(llvm::errc::invalid_argument);
903
904 sys::path::const_iterator Start = sys::path::begin(Path);
905 sys::path::const_iterator End = sys::path::end(Path);
906 for (std::vector<Entry *>::iterator I = Roots.begin(), E = Roots.end();
907 I != E; ++I) {
908 ErrorOr<Entry *> Result = lookupPath(Start, End, *I);
909 if (Result || Result.getError() != llvm::errc::no_such_file_or_directory)
910 return Result;
911 }
912 return make_error_code(llvm::errc::no_such_file_or_directory);
913 }
914
lookupPath(sys::path::const_iterator Start,sys::path::const_iterator End,Entry * From)915 ErrorOr<Entry *> VFSFromYAML::lookupPath(sys::path::const_iterator Start,
916 sys::path::const_iterator End,
917 Entry *From) {
918 if (Start->equals("."))
919 ++Start;
920
921 // FIXME: handle ..
922 if (CaseSensitive ? !Start->equals(From->getName())
923 : !Start->equals_lower(From->getName()))
924 // failure to match
925 return make_error_code(llvm::errc::no_such_file_or_directory);
926
927 ++Start;
928
929 if (Start == End) {
930 // Match!
931 return From;
932 }
933
934 DirectoryEntry *DE = dyn_cast<DirectoryEntry>(From);
935 if (!DE)
936 return make_error_code(llvm::errc::not_a_directory);
937
938 for (DirectoryEntry::iterator I = DE->contents_begin(),
939 E = DE->contents_end();
940 I != E; ++I) {
941 ErrorOr<Entry *> Result = lookupPath(Start, End, *I);
942 if (Result || Result.getError() != llvm::errc::no_such_file_or_directory)
943 return Result;
944 }
945 return make_error_code(llvm::errc::no_such_file_or_directory);
946 }
947
status(const Twine & Path,Entry * E)948 ErrorOr<Status> VFSFromYAML::status(const Twine &Path, Entry *E) {
949 assert(E != nullptr);
950 std::string PathStr(Path.str());
951 if (FileEntry *F = dyn_cast<FileEntry>(E)) {
952 ErrorOr<Status> S = ExternalFS->status(F->getExternalContentsPath());
953 assert(!S || S->getName() == F->getExternalContentsPath());
954 if (S && !F->useExternalName(UseExternalNames))
955 S->setName(PathStr);
956 if (S)
957 S->IsVFSMapped = true;
958 return S;
959 } else { // directory
960 DirectoryEntry *DE = cast<DirectoryEntry>(E);
961 Status S = DE->getStatus();
962 S.setName(PathStr);
963 return S;
964 }
965 }
966
status(const Twine & Path)967 ErrorOr<Status> VFSFromYAML::status(const Twine &Path) {
968 ErrorOr<Entry *> Result = lookupPath(Path);
969 if (!Result)
970 return Result.getError();
971 return status(Path, *Result);
972 }
973
974 std::error_code
openFileForRead(const Twine & Path,std::unique_ptr<vfs::File> & Result)975 VFSFromYAML::openFileForRead(const Twine &Path,
976 std::unique_ptr<vfs::File> &Result) {
977 ErrorOr<Entry *> E = lookupPath(Path);
978 if (!E)
979 return E.getError();
980
981 FileEntry *F = dyn_cast<FileEntry>(*E);
982 if (!F) // FIXME: errc::not_a_file?
983 return make_error_code(llvm::errc::invalid_argument);
984
985 if (std::error_code EC =
986 ExternalFS->openFileForRead(F->getExternalContentsPath(), Result))
987 return EC;
988
989 if (!F->useExternalName(UseExternalNames))
990 Result->setName(Path.str());
991
992 return std::error_code();
993 }
994
995 IntrusiveRefCntPtr<FileSystem>
getVFSFromYAML(MemoryBuffer * Buffer,SourceMgr::DiagHandlerTy DiagHandler,void * DiagContext,IntrusiveRefCntPtr<FileSystem> ExternalFS)996 vfs::getVFSFromYAML(MemoryBuffer *Buffer, SourceMgr::DiagHandlerTy DiagHandler,
997 void *DiagContext,
998 IntrusiveRefCntPtr<FileSystem> ExternalFS) {
999 return VFSFromYAML::create(Buffer, DiagHandler, DiagContext, ExternalFS);
1000 }
1001
getNextVirtualUniqueID()1002 UniqueID vfs::getNextVirtualUniqueID() {
1003 static std::atomic<unsigned> UID;
1004 unsigned ID = ++UID;
1005 // The following assumes that uint64_t max will never collide with a real
1006 // dev_t value from the OS.
1007 return UniqueID(std::numeric_limits<uint64_t>::max(), ID);
1008 }
1009
1010 #ifndef NDEBUG
pathHasTraversal(StringRef Path)1011 static bool pathHasTraversal(StringRef Path) {
1012 using namespace llvm::sys;
1013 for (StringRef Comp : llvm::make_range(path::begin(Path), path::end(Path)))
1014 if (Comp == "." || Comp == "..")
1015 return true;
1016 return false;
1017 }
1018 #endif
1019
addFileMapping(StringRef VirtualPath,StringRef RealPath)1020 void YAMLVFSWriter::addFileMapping(StringRef VirtualPath, StringRef RealPath) {
1021 assert(sys::path::is_absolute(VirtualPath) && "virtual path not absolute");
1022 assert(sys::path::is_absolute(RealPath) && "real path not absolute");
1023 assert(!pathHasTraversal(VirtualPath) && "path traversal is not supported");
1024 Mappings.emplace_back(VirtualPath, RealPath);
1025 }
1026
1027 namespace {
1028 class JSONWriter {
1029 llvm::raw_ostream &OS;
1030 SmallVector<StringRef, 16> DirStack;
getDirIndent()1031 inline unsigned getDirIndent() { return 4 * DirStack.size(); }
getFileIndent()1032 inline unsigned getFileIndent() { return 4 * (DirStack.size() + 1); }
1033 bool containedIn(StringRef Parent, StringRef Path);
1034 StringRef containedPart(StringRef Parent, StringRef Path);
1035 void startDirectory(StringRef Path);
1036 void endDirectory();
1037 void writeEntry(StringRef VPath, StringRef RPath);
1038
1039 public:
JSONWriter(llvm::raw_ostream & OS)1040 JSONWriter(llvm::raw_ostream &OS) : OS(OS) {}
1041 void write(ArrayRef<YAMLVFSEntry> Entries, Optional<bool> IsCaseSensitive);
1042 };
1043 }
1044
containedIn(StringRef Parent,StringRef Path)1045 bool JSONWriter::containedIn(StringRef Parent, StringRef Path) {
1046 using namespace llvm::sys;
1047 // Compare each path component.
1048 auto IParent = path::begin(Parent), EParent = path::end(Parent);
1049 for (auto IChild = path::begin(Path), EChild = path::end(Path);
1050 IParent != EParent && IChild != EChild; ++IParent, ++IChild) {
1051 if (*IParent != *IChild)
1052 return false;
1053 }
1054 // Have we exhausted the parent path?
1055 return IParent == EParent;
1056 }
1057
containedPart(StringRef Parent,StringRef Path)1058 StringRef JSONWriter::containedPart(StringRef Parent, StringRef Path) {
1059 assert(!Parent.empty());
1060 assert(containedIn(Parent, Path));
1061 return Path.slice(Parent.size() + 1, StringRef::npos);
1062 }
1063
startDirectory(StringRef Path)1064 void JSONWriter::startDirectory(StringRef Path) {
1065 StringRef Name =
1066 DirStack.empty() ? Path : containedPart(DirStack.back(), Path);
1067 DirStack.push_back(Path);
1068 unsigned Indent = getDirIndent();
1069 OS.indent(Indent) << "{\n";
1070 OS.indent(Indent + 2) << "'type': 'directory',\n";
1071 OS.indent(Indent + 2) << "'name': \"" << llvm::yaml::escape(Name) << "\",\n";
1072 OS.indent(Indent + 2) << "'contents': [\n";
1073 }
1074
endDirectory()1075 void JSONWriter::endDirectory() {
1076 unsigned Indent = getDirIndent();
1077 OS.indent(Indent + 2) << "]\n";
1078 OS.indent(Indent) << "}";
1079
1080 DirStack.pop_back();
1081 }
1082
writeEntry(StringRef VPath,StringRef RPath)1083 void JSONWriter::writeEntry(StringRef VPath, StringRef RPath) {
1084 unsigned Indent = getFileIndent();
1085 OS.indent(Indent) << "{\n";
1086 OS.indent(Indent + 2) << "'type': 'file',\n";
1087 OS.indent(Indent + 2) << "'name': \"" << llvm::yaml::escape(VPath) << "\",\n";
1088 OS.indent(Indent + 2) << "'external-contents': \""
1089 << llvm::yaml::escape(RPath) << "\"\n";
1090 OS.indent(Indent) << "}";
1091 }
1092
write(ArrayRef<YAMLVFSEntry> Entries,Optional<bool> IsCaseSensitive)1093 void JSONWriter::write(ArrayRef<YAMLVFSEntry> Entries,
1094 Optional<bool> IsCaseSensitive) {
1095 using namespace llvm::sys;
1096
1097 OS << "{\n"
1098 " 'version': 0,\n";
1099 if (IsCaseSensitive.hasValue())
1100 OS << " 'case-sensitive': '"
1101 << (IsCaseSensitive.getValue() ? "true" : "false") << "',\n";
1102 OS << " 'roots': [\n";
1103
1104 if (Entries.empty())
1105 return;
1106
1107 const YAMLVFSEntry &Entry = Entries.front();
1108 startDirectory(path::parent_path(Entry.VPath));
1109 writeEntry(path::filename(Entry.VPath), Entry.RPath);
1110
1111 for (const auto &Entry : Entries.slice(1)) {
1112 StringRef Dir = path::parent_path(Entry.VPath);
1113 if (Dir == DirStack.back())
1114 OS << ",\n";
1115 else {
1116 while (!DirStack.empty() && !containedIn(DirStack.back(), Dir)) {
1117 OS << "\n";
1118 endDirectory();
1119 }
1120 OS << ",\n";
1121 startDirectory(Dir);
1122 }
1123 writeEntry(path::filename(Entry.VPath), Entry.RPath);
1124 }
1125
1126 while (!DirStack.empty()) {
1127 OS << "\n";
1128 endDirectory();
1129 }
1130
1131 OS << "\n"
1132 << " ]\n"
1133 << "}\n";
1134 }
1135
write(llvm::raw_ostream & OS)1136 void YAMLVFSWriter::write(llvm::raw_ostream &OS) {
1137 std::sort(Mappings.begin(), Mappings.end(),
1138 [](const YAMLVFSEntry &LHS, const YAMLVFSEntry &RHS) {
1139 return LHS.VPath < RHS.VPath;
1140 });
1141
1142 JSONWriter(OS).write(Mappings, IsCaseSensitive);
1143 }
1144
VFSFromYamlDirIterImpl(const Twine & _Path,VFSFromYAML & FS,DirectoryEntry::iterator Begin,DirectoryEntry::iterator End,std::error_code & EC)1145 VFSFromYamlDirIterImpl::VFSFromYamlDirIterImpl(const Twine &_Path,
1146 VFSFromYAML &FS,
1147 DirectoryEntry::iterator Begin,
1148 DirectoryEntry::iterator End,
1149 std::error_code &EC)
1150 : Dir(_Path.str()), FS(FS), Current(Begin), End(End) {
1151 if (Current != End) {
1152 SmallString<128> PathStr(Dir);
1153 llvm::sys::path::append(PathStr, (*Current)->getName());
1154 llvm::ErrorOr<vfs::Status> S = FS.status(PathStr.str());
1155 if (S)
1156 CurrentEntry = *S;
1157 else
1158 EC = S.getError();
1159 }
1160 }
1161
increment()1162 std::error_code VFSFromYamlDirIterImpl::increment() {
1163 assert(Current != End && "cannot iterate past end");
1164 if (++Current != End) {
1165 SmallString<128> PathStr(Dir);
1166 llvm::sys::path::append(PathStr, (*Current)->getName());
1167 llvm::ErrorOr<vfs::Status> S = FS.status(PathStr.str());
1168 if (!S)
1169 return S.getError();
1170 CurrentEntry = *S;
1171 } else {
1172 CurrentEntry = Status();
1173 }
1174 return std::error_code();
1175 }
1176
recursive_directory_iterator(FileSystem & FS_,const Twine & Path,std::error_code & EC)1177 vfs::recursive_directory_iterator::recursive_directory_iterator(FileSystem &FS_,
1178 const Twine &Path,
1179 std::error_code &EC)
1180 : FS(&FS_) {
1181 directory_iterator I = FS->dir_begin(Path, EC);
1182 if (!EC && I != directory_iterator()) {
1183 State = std::make_shared<IterState>();
1184 State->push(I);
1185 }
1186 }
1187
1188 vfs::recursive_directory_iterator &
increment(std::error_code & EC)1189 recursive_directory_iterator::increment(std::error_code &EC) {
1190 assert(FS && State && !State->empty() && "incrementing past end");
1191 assert(State->top()->isStatusKnown() && "non-canonical end iterator");
1192 vfs::directory_iterator End;
1193 if (State->top()->isDirectory()) {
1194 vfs::directory_iterator I = FS->dir_begin(State->top()->getName(), EC);
1195 if (EC)
1196 return *this;
1197 if (I != End) {
1198 State->push(I);
1199 return *this;
1200 }
1201 }
1202
1203 while (!State->empty() && State->top().increment(EC) == End)
1204 State->pop();
1205
1206 if (State->empty())
1207 State.reset(); // end iterator
1208
1209 return *this;
1210 }
1211