Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 1 | //===- Archive.h - ar archive file format -----------------------*- C++ -*-===// |
| 2 | // |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 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 |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | // |
| 9 | // This file declares the ar archive file format class. |
| 10 | // |
| 11 | //===----------------------------------------------------------------------===// |
| 12 | |
| 13 | #ifndef LLVM_OBJECT_ARCHIVE_H |
| 14 | #define LLVM_OBJECT_ARCHIVE_H |
| 15 | |
| 16 | #include "llvm/ADT/Optional.h" |
| 17 | #include "llvm/ADT/StringRef.h" |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 18 | #include "llvm/ADT/fallible_iterator.h" |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 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; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame] | 94 | /// Includes header but not padding byte. |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 95 | StringRef Data; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame] | 96 | /// Offset from Data to the start of the file. |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 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 | |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 146 | class ChildFallibleIterator { |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 147 | Child C; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 148 | |
| 149 | public: |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 150 | ChildFallibleIterator() : C(Child(nullptr, nullptr, nullptr)) {} |
| 151 | ChildFallibleIterator(const Child &C) : C(C) {} |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 152 | |
| 153 | const Child *operator->() const { return &C; } |
| 154 | const Child &operator*() const { return C; } |
| 155 | |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 156 | bool operator==(const ChildFallibleIterator &other) const { |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 157 | // Ignore errors here: If an error occurred during increment then getNext |
| 158 | // will have been set to child_end(), and the following comparison should |
| 159 | // do the right thing. |
| 160 | return C == other.C; |
| 161 | } |
| 162 | |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 163 | bool operator!=(const ChildFallibleIterator &other) const { |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 164 | return !(*this == other); |
| 165 | } |
| 166 | |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 167 | Error inc() { |
| 168 | auto NextChild = C.getNext(); |
| 169 | if (!NextChild) |
| 170 | return NextChild.takeError(); |
| 171 | C = std::move(*NextChild); |
| 172 | return Error::success(); |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 173 | } |
| 174 | }; |
| 175 | |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame^] | 176 | using child_iterator = fallible_iterator<ChildFallibleIterator>; |
| 177 | |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 178 | class Symbol { |
| 179 | const Archive *Parent; |
| 180 | uint32_t SymbolIndex; |
| 181 | uint32_t StringIndex; // Extra index to the string. |
| 182 | |
| 183 | public: |
| 184 | Symbol(const Archive *p, uint32_t symi, uint32_t stri) |
| 185 | : Parent(p) |
| 186 | , SymbolIndex(symi) |
| 187 | , StringIndex(stri) {} |
| 188 | |
| 189 | bool operator ==(const Symbol &other) const { |
| 190 | return (Parent == other.Parent) && (SymbolIndex == other.SymbolIndex); |
| 191 | } |
| 192 | |
| 193 | StringRef getName() const; |
| 194 | Expected<Child> getMember() const; |
| 195 | Symbol getNext() const; |
| 196 | }; |
| 197 | |
| 198 | class symbol_iterator { |
| 199 | Symbol symbol; |
| 200 | |
| 201 | public: |
| 202 | symbol_iterator(const Symbol &s) : symbol(s) {} |
| 203 | |
| 204 | const Symbol *operator->() const { return &symbol; } |
| 205 | const Symbol &operator*() const { return symbol; } |
| 206 | |
| 207 | bool operator==(const symbol_iterator &other) const { |
| 208 | return symbol == other.symbol; |
| 209 | } |
| 210 | |
| 211 | bool operator!=(const symbol_iterator &other) const { |
| 212 | return !(*this == other); |
| 213 | } |
| 214 | |
| 215 | symbol_iterator& operator++() { // Preincrement |
| 216 | symbol = symbol.getNext(); |
| 217 | return *this; |
| 218 | } |
| 219 | }; |
| 220 | |
| 221 | Archive(MemoryBufferRef Source, Error &Err); |
| 222 | static Expected<std::unique_ptr<Archive>> create(MemoryBufferRef Source); |
| 223 | |
| 224 | enum Kind { |
| 225 | K_GNU, |
| 226 | K_GNU64, |
| 227 | K_BSD, |
| 228 | K_DARWIN, |
| 229 | K_DARWIN64, |
| 230 | K_COFF |
| 231 | }; |
| 232 | |
| 233 | Kind kind() const { return (Kind)Format; } |
| 234 | bool isThin() const { return IsThin; } |
| 235 | |
| 236 | child_iterator child_begin(Error &Err, bool SkipInternal = true) const; |
| 237 | child_iterator child_end() const; |
| 238 | iterator_range<child_iterator> children(Error &Err, |
| 239 | bool SkipInternal = true) const { |
| 240 | return make_range(child_begin(Err, SkipInternal), child_end()); |
| 241 | } |
| 242 | |
| 243 | symbol_iterator symbol_begin() const; |
| 244 | symbol_iterator symbol_end() const; |
| 245 | iterator_range<symbol_iterator> symbols() const { |
| 246 | return make_range(symbol_begin(), symbol_end()); |
| 247 | } |
| 248 | |
| 249 | // Cast methods. |
| 250 | static bool classof(Binary const *v) { |
| 251 | return v->isArchive(); |
| 252 | } |
| 253 | |
| 254 | // check if a symbol is in the archive |
| 255 | Expected<Optional<Child>> findSym(StringRef name) const; |
| 256 | |
| 257 | bool isEmpty() const; |
| 258 | bool hasSymbolTable() const; |
| 259 | StringRef getSymbolTable() const { return SymbolTable; } |
| 260 | StringRef getStringTable() const { return StringTable; } |
| 261 | uint32_t getNumberOfSymbols() const; |
| 262 | |
| 263 | std::vector<std::unique_ptr<MemoryBuffer>> takeThinBuffers() { |
| 264 | return std::move(ThinBuffers); |
| 265 | } |
| 266 | |
| 267 | private: |
| 268 | StringRef SymbolTable; |
| 269 | StringRef StringTable; |
| 270 | |
| 271 | StringRef FirstRegularData; |
| 272 | uint16_t FirstRegularStartOfFile = -1; |
| 273 | void setFirstRegular(const Child &C); |
| 274 | |
| 275 | unsigned Format : 3; |
| 276 | unsigned IsThin : 1; |
| 277 | mutable std::vector<std::unique_ptr<MemoryBuffer>> ThinBuffers; |
| 278 | }; |
| 279 | |
| 280 | } // end namespace object |
| 281 | } // end namespace llvm |
| 282 | |
| 283 | #endif // LLVM_OBJECT_ARCHIVE_H |