Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame^] | 1 | //===- Archive.h - ar archive file format -----------------------*- 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 | // This file declares the ar archive file format class. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #ifndef LLVM_OBJECT_ARCHIVE_H |
| 15 | #define LLVM_OBJECT_ARCHIVE_H |
| 16 | |
| 17 | #include "llvm/ADT/Optional.h" |
| 18 | #include "llvm/ADT/StringRef.h" |
| 19 | #include "llvm/ADT/iterator_range.h" |
| 20 | #include "llvm/Object/Binary.h" |
| 21 | #include "llvm/Support/Chrono.h" |
| 22 | #include "llvm/Support/Error.h" |
| 23 | #include "llvm/Support/FileSystem.h" |
| 24 | #include "llvm/Support/MemoryBuffer.h" |
| 25 | #include <algorithm> |
| 26 | #include <cassert> |
| 27 | #include <cstdint> |
| 28 | #include <memory> |
| 29 | #include <string> |
| 30 | #include <vector> |
| 31 | |
| 32 | namespace llvm { |
| 33 | namespace object { |
| 34 | |
| 35 | class Archive; |
| 36 | |
| 37 | class ArchiveMemberHeader { |
| 38 | public: |
| 39 | friend class Archive; |
| 40 | |
| 41 | ArchiveMemberHeader(Archive const *Parent, const char *RawHeaderPtr, |
| 42 | uint64_t Size, Error *Err); |
| 43 | // ArchiveMemberHeader() = default; |
| 44 | |
| 45 | /// Get the name without looking up long names. |
| 46 | Expected<StringRef> getRawName() const; |
| 47 | |
| 48 | /// Get the name looking up long names. |
| 49 | Expected<StringRef> getName(uint64_t Size) const; |
| 50 | |
| 51 | /// Members are not larger than 4GB. |
| 52 | Expected<uint32_t> getSize() const; |
| 53 | |
| 54 | Expected<sys::fs::perms> getAccessMode() const; |
| 55 | Expected<sys::TimePoint<std::chrono::seconds>> getLastModified() const; |
| 56 | |
| 57 | StringRef getRawLastModified() const { |
| 58 | return StringRef(ArMemHdr->LastModified, |
| 59 | sizeof(ArMemHdr->LastModified)).rtrim(' '); |
| 60 | } |
| 61 | |
| 62 | Expected<unsigned> getUID() const; |
| 63 | Expected<unsigned> getGID() const; |
| 64 | |
| 65 | // This returns the size of the private struct ArMemHdrType |
| 66 | uint64_t getSizeOf() const { |
| 67 | return sizeof(ArMemHdrType); |
| 68 | } |
| 69 | |
| 70 | private: |
| 71 | struct ArMemHdrType { |
| 72 | char Name[16]; |
| 73 | char LastModified[12]; |
| 74 | char UID[6]; |
| 75 | char GID[6]; |
| 76 | char AccessMode[8]; |
| 77 | char Size[10]; ///< Size of data, not including header or padding. |
| 78 | char Terminator[2]; |
| 79 | }; |
| 80 | Archive const *Parent; |
| 81 | ArMemHdrType const *ArMemHdr; |
| 82 | }; |
| 83 | |
| 84 | class Archive : public Binary { |
| 85 | virtual void anchor(); |
| 86 | |
| 87 | public: |
| 88 | class Child { |
| 89 | friend Archive; |
| 90 | friend ArchiveMemberHeader; |
| 91 | |
| 92 | const Archive *Parent; |
| 93 | ArchiveMemberHeader Header; |
| 94 | /// \brief Includes header but not padding byte. |
| 95 | StringRef Data; |
| 96 | /// \brief Offset from Data to the start of the file. |
| 97 | uint16_t StartOfFile; |
| 98 | |
| 99 | Expected<bool> isThinMember() const; |
| 100 | |
| 101 | public: |
| 102 | Child(const Archive *Parent, const char *Start, Error *Err); |
| 103 | Child(const Archive *Parent, StringRef Data, uint16_t StartOfFile); |
| 104 | |
| 105 | bool operator ==(const Child &other) const { |
| 106 | assert(!Parent || !other.Parent || Parent == other.Parent); |
| 107 | return Data.begin() == other.Data.begin(); |
| 108 | } |
| 109 | |
| 110 | const Archive *getParent() const { return Parent; } |
| 111 | Expected<Child> getNext() const; |
| 112 | |
| 113 | Expected<StringRef> getName() const; |
| 114 | Expected<std::string> getFullName() const; |
| 115 | Expected<StringRef> getRawName() const { return Header.getRawName(); } |
| 116 | |
| 117 | Expected<sys::TimePoint<std::chrono::seconds>> getLastModified() const { |
| 118 | return Header.getLastModified(); |
| 119 | } |
| 120 | |
| 121 | StringRef getRawLastModified() const { |
| 122 | return Header.getRawLastModified(); |
| 123 | } |
| 124 | |
| 125 | Expected<unsigned> getUID() const { return Header.getUID(); } |
| 126 | Expected<unsigned> getGID() const { return Header.getGID(); } |
| 127 | |
| 128 | Expected<sys::fs::perms> getAccessMode() const { |
| 129 | return Header.getAccessMode(); |
| 130 | } |
| 131 | |
| 132 | /// \return the size of the archive member without the header or padding. |
| 133 | Expected<uint64_t> getSize() const; |
| 134 | /// \return the size in the archive header for this member. |
| 135 | Expected<uint64_t> getRawSize() const; |
| 136 | |
| 137 | Expected<StringRef> getBuffer() const; |
| 138 | uint64_t getChildOffset() const; |
| 139 | |
| 140 | Expected<MemoryBufferRef> getMemoryBufferRef() const; |
| 141 | |
| 142 | Expected<std::unique_ptr<Binary>> |
| 143 | getAsBinary(LLVMContext *Context = nullptr) const; |
| 144 | }; |
| 145 | |
| 146 | class child_iterator { |
| 147 | Child C; |
| 148 | Error *E = nullptr; |
| 149 | |
| 150 | public: |
| 151 | child_iterator() : C(Child(nullptr, nullptr, nullptr)) {} |
| 152 | child_iterator(const Child &C, Error *E) : C(C), E(E) {} |
| 153 | |
| 154 | const Child *operator->() const { return &C; } |
| 155 | const Child &operator*() const { return C; } |
| 156 | |
| 157 | bool operator==(const child_iterator &other) const { |
| 158 | // Ignore errors here: If an error occurred during increment then getNext |
| 159 | // will have been set to child_end(), and the following comparison should |
| 160 | // do the right thing. |
| 161 | return C == other.C; |
| 162 | } |
| 163 | |
| 164 | bool operator!=(const child_iterator &other) const { |
| 165 | return !(*this == other); |
| 166 | } |
| 167 | |
| 168 | // Code in loops with child_iterators must check for errors on each loop |
| 169 | // iteration. And if there is an error break out of the loop. |
| 170 | child_iterator &operator++() { // Preincrement |
| 171 | assert(E && "Can't increment iterator with no Error attached"); |
| 172 | ErrorAsOutParameter ErrAsOutParam(E); |
| 173 | if (auto ChildOrErr = C.getNext()) |
| 174 | C = *ChildOrErr; |
| 175 | else { |
| 176 | C = C.getParent()->child_end().C; |
| 177 | *E = ChildOrErr.takeError(); |
| 178 | E = nullptr; |
| 179 | } |
| 180 | return *this; |
| 181 | } |
| 182 | }; |
| 183 | |
| 184 | class Symbol { |
| 185 | const Archive *Parent; |
| 186 | uint32_t SymbolIndex; |
| 187 | uint32_t StringIndex; // Extra index to the string. |
| 188 | |
| 189 | public: |
| 190 | Symbol(const Archive *p, uint32_t symi, uint32_t stri) |
| 191 | : Parent(p) |
| 192 | , SymbolIndex(symi) |
| 193 | , StringIndex(stri) {} |
| 194 | |
| 195 | bool operator ==(const Symbol &other) const { |
| 196 | return (Parent == other.Parent) && (SymbolIndex == other.SymbolIndex); |
| 197 | } |
| 198 | |
| 199 | StringRef getName() const; |
| 200 | Expected<Child> getMember() const; |
| 201 | Symbol getNext() const; |
| 202 | }; |
| 203 | |
| 204 | class symbol_iterator { |
| 205 | Symbol symbol; |
| 206 | |
| 207 | public: |
| 208 | symbol_iterator(const Symbol &s) : symbol(s) {} |
| 209 | |
| 210 | const Symbol *operator->() const { return &symbol; } |
| 211 | const Symbol &operator*() const { return symbol; } |
| 212 | |
| 213 | bool operator==(const symbol_iterator &other) const { |
| 214 | return symbol == other.symbol; |
| 215 | } |
| 216 | |
| 217 | bool operator!=(const symbol_iterator &other) const { |
| 218 | return !(*this == other); |
| 219 | } |
| 220 | |
| 221 | symbol_iterator& operator++() { // Preincrement |
| 222 | symbol = symbol.getNext(); |
| 223 | return *this; |
| 224 | } |
| 225 | }; |
| 226 | |
| 227 | Archive(MemoryBufferRef Source, Error &Err); |
| 228 | static Expected<std::unique_ptr<Archive>> create(MemoryBufferRef Source); |
| 229 | |
| 230 | enum Kind { |
| 231 | K_GNU, |
| 232 | K_GNU64, |
| 233 | K_BSD, |
| 234 | K_DARWIN, |
| 235 | K_DARWIN64, |
| 236 | K_COFF |
| 237 | }; |
| 238 | |
| 239 | Kind kind() const { return (Kind)Format; } |
| 240 | bool isThin() const { return IsThin; } |
| 241 | |
| 242 | child_iterator child_begin(Error &Err, bool SkipInternal = true) const; |
| 243 | child_iterator child_end() const; |
| 244 | iterator_range<child_iterator> children(Error &Err, |
| 245 | bool SkipInternal = true) const { |
| 246 | return make_range(child_begin(Err, SkipInternal), child_end()); |
| 247 | } |
| 248 | |
| 249 | symbol_iterator symbol_begin() const; |
| 250 | symbol_iterator symbol_end() const; |
| 251 | iterator_range<symbol_iterator> symbols() const { |
| 252 | return make_range(symbol_begin(), symbol_end()); |
| 253 | } |
| 254 | |
| 255 | // Cast methods. |
| 256 | static bool classof(Binary const *v) { |
| 257 | return v->isArchive(); |
| 258 | } |
| 259 | |
| 260 | // check if a symbol is in the archive |
| 261 | Expected<Optional<Child>> findSym(StringRef name) const; |
| 262 | |
| 263 | bool isEmpty() const; |
| 264 | bool hasSymbolTable() const; |
| 265 | StringRef getSymbolTable() const { return SymbolTable; } |
| 266 | StringRef getStringTable() const { return StringTable; } |
| 267 | uint32_t getNumberOfSymbols() const; |
| 268 | |
| 269 | std::vector<std::unique_ptr<MemoryBuffer>> takeThinBuffers() { |
| 270 | return std::move(ThinBuffers); |
| 271 | } |
| 272 | |
| 273 | private: |
| 274 | StringRef SymbolTable; |
| 275 | StringRef StringTable; |
| 276 | |
| 277 | StringRef FirstRegularData; |
| 278 | uint16_t FirstRegularStartOfFile = -1; |
| 279 | void setFirstRegular(const Child &C); |
| 280 | |
| 281 | unsigned Format : 3; |
| 282 | unsigned IsThin : 1; |
| 283 | mutable std::vector<std::unique_ptr<MemoryBuffer>> ThinBuffers; |
| 284 | }; |
| 285 | |
| 286 | } // end namespace object |
| 287 | } // end namespace llvm |
| 288 | |
| 289 | #endif // LLVM_OBJECT_ARCHIVE_H |