| 1 | //===-- llvm-strings.cpp - Printable String dumping utility ---------------===// |
| 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 | // This program is a utility that works like binutils "strings", that is, it |
| 10 | // prints out printable strings in a binary, objdump, or archive file. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #include "Opts.inc" |
| 15 | #include "llvm/ADT/StringExtras.h" |
| 16 | #include "llvm/Object/Binary.h" |
| 17 | #include "llvm/Option/Arg.h" |
| 18 | #include "llvm/Option/ArgList.h" |
| 19 | #include "llvm/Option/Option.h" |
| 20 | #include "llvm/Support/CommandLine.h" |
| 21 | #include "llvm/Support/Error.h" |
| 22 | #include "llvm/Support/FileSystem.h" |
| 23 | #include "llvm/Support/Format.h" |
| 24 | #include "llvm/Support/InitLLVM.h" |
| 25 | #include "llvm/Support/MemoryBuffer.h" |
| 26 | #include "llvm/Support/Program.h" |
| 27 | #include "llvm/Support/WithColor.h" |
| 28 | #include <cctype> |
| 29 | #include <string> |
| 30 | |
| 31 | using namespace llvm; |
| 32 | using namespace llvm::object; |
| 33 | |
| 34 | namespace { |
| 35 | enum ID { |
| 36 | OPT_INVALID = 0, // This is not an option ID. |
| 37 | #define OPTION(...) LLVM_MAKE_OPT_ID(__VA_ARGS__), |
| 38 | #include "Opts.inc" |
| 39 | #undef OPTION |
| 40 | }; |
| 41 | |
| 42 | using namespace llvm::opt; |
| 43 | #define OPTTABLE_CODE |
| 44 | #include "Opts.inc" |
| 45 | |
| 46 | class StringsOptTable : public opt::OptTable { |
| 47 | public: |
| 48 | StringsOptTable() : OptTable(optionTables()) { |
| 49 | setGroupedShortOptions(true); |
| 50 | setDashDashParsing(true); |
| 51 | } |
| 52 | }; |
| 53 | |
| 54 | static StringRef ToolName; |
| 55 | |
| 56 | static cl::list<std::string> InputFileNames(cl::Positional, |
| 57 | cl::desc("<input object files>" )); |
| 58 | |
| 59 | static constexpr int DefaultMinLength = 4; |
| 60 | static int MinLength = DefaultMinLength; |
| 61 | static bool PrintFileName; |
| 62 | |
| 63 | enum class Radix { None, Octal, Hexadecimal, Decimal }; |
| 64 | static Radix Radix; |
| 65 | } // namespace |
| 66 | |
| 67 | [[noreturn]] static void reportCmdLineError(const Twine &Message) { |
| 68 | WithColor::error(OS&: errs(), Prefix: ToolName) << Message << "\n" ; |
| 69 | exit(status: 1); |
| 70 | } |
| 71 | |
| 72 | template <typename T> |
| 73 | static void parseIntArg(const opt::InputArgList &Args, int ID, T &Value) { |
| 74 | if (const opt::Arg *A = Args.getLastArg(Ids: ID)) { |
| 75 | StringRef V(A->getValue()); |
| 76 | if (!llvm::to_integer(V, Value, 0) || Value <= 0) |
| 77 | reportCmdLineError(Message: "expected a positive integer, but got '" + V + "'" ); |
| 78 | } |
| 79 | } |
| 80 | |
| 81 | static bool isStringChar(char C) { return isPrint(C) || C == '\t'; } |
| 82 | |
| 83 | static void strings(raw_ostream &OS, StringRef FileName, |
| 84 | sys::fs::file_t Handle) { |
| 85 | SmallString<sys::fs::DefaultReadChunkSize> Buffer; |
| 86 | auto = [&OS, FileName](size_t StringStart) { |
| 87 | if (PrintFileName) |
| 88 | OS << FileName << ": " ; |
| 89 | switch (Radix) { |
| 90 | case Radix::None: |
| 91 | break; |
| 92 | case Radix::Octal: |
| 93 | OS << format(Fmt: "%7o " , Vals: StringStart); |
| 94 | break; |
| 95 | case Radix::Hexadecimal: |
| 96 | OS << format(Fmt: "%7x " , Vals: StringStart); |
| 97 | break; |
| 98 | case Radix::Decimal: |
| 99 | OS << format(Fmt: "%7u " , Vals: StringStart); |
| 100 | break; |
| 101 | } |
| 102 | }; |
| 103 | |
| 104 | // To handle very large files without consuming excessive memory, we read the |
| 105 | // file in a little at a time and process it then rather than reading the |
| 106 | // entire file at once. |
| 107 | // |
| 108 | // A string is only buffered until it is known to be long enough to print; |
| 109 | // from then on it is streamed out directly, so an arbitrarily long string |
| 110 | // never needs an arbitrarily large buffer. Candidate therefore only ever |
| 111 | // holds a run that is shorter than MinLength and that was cut off by the end |
| 112 | // of a chunk. |
| 113 | const size_t Min = MinLength; |
| 114 | SmallString<DefaultMinLength> Candidate; |
| 115 | bool InString = false; |
| 116 | // Offset of the start of the current chunk within the file. |
| 117 | size_t ChunkOffset = 0; |
| 118 | |
| 119 | Buffer.resize_for_overwrite(N: sys::fs::DefaultReadChunkSize); |
| 120 | |
| 121 | while (true) { |
| 122 | Expected<size_t> ReadBytesOrErr = sys::fs::readNativeFile( |
| 123 | FileHandle: Handle, Buf: MutableArrayRef(Buffer.data(), Buffer.size())); |
| 124 | if (!ReadBytesOrErr) { |
| 125 | errs() << FileName << ": " |
| 126 | << errorToErrorCode(Err: ReadBytesOrErr.takeError()).message() << '\n'; |
| 127 | return; |
| 128 | } |
| 129 | size_t ChunkSize = *ReadBytesOrErr; |
| 130 | if (ChunkSize == 0) |
| 131 | break; |
| 132 | |
| 133 | // To prevent performance regression under O0, access the raw pointer |
| 134 | // instead of using methods provided by the standard library, which are not |
| 135 | // inlined under O0. |
| 136 | const char *const Begin = Buffer.data(); |
| 137 | const char *const End = Begin + ChunkSize; |
| 138 | const char *Cur = Begin; |
| 139 | |
| 140 | // Handle the remaining part from the previous chunk. |
| 141 | // The previous chunk can be either shorter than MinSize or part of the |
| 142 | // string. |
| 143 | // Keep the buffer size bounded. With a small Min, a long string spanning |
| 144 | // multiple chunks will have at most DefaultReadChunkSize bytes, since the |
| 145 | // buffer is printed immediately with the header (guarded by the second if). |
| 146 | // With a large Min, the buffer must hold at least Min bytes, since we need |
| 147 | // enough data to decide whether to print it. |
| 148 | if (InString || !Candidate.empty()) { |
| 149 | // Find the end of the current string. |
| 150 | while (Cur != End && isStringChar(C: *Cur)) |
| 151 | ++Cur; |
| 152 | size_t Len = Cur - Begin; |
| 153 | if (InString) { |
| 154 | // Print the remaining part if the previous chunk has already printed |
| 155 | // the header. E.g. header: aaaaa | bbbbb, where | is the chunk |
| 156 | // boundary. |
| 157 | // Output: Header: aaaaabbbbb, where bbbbb is printed in here. |
| 158 | OS << StringRef(Begin, Len); |
| 159 | } else if (Candidate.size() + Len >= Min) { |
| 160 | // If the header hasn't been printed yet (e.g. the previous candidate |
| 161 | // was smaller than Min), but we can print it now, print the header |
| 162 | // first, followed by the candidate from the previous chunk and the |
| 163 | // current string. E.g. aa | bbbbbb |
| 164 | // Output Header: aabbbbbb, where aabbbbbb is printed in here. |
| 165 | PrintHeader(ChunkOffset - Candidate.size()); |
| 166 | OS << Candidate << StringRef(Begin, Len); |
| 167 | Candidate.clear(); |
| 168 | InString = true; |
| 169 | } else if (Cur == End) { |
| 170 | // If the current chunk + previous candidate is still smaller than Min, |
| 171 | // append it to Candidate. |
| 172 | Candidate.append(in_start: Begin, in_end: End); |
| 173 | } else { |
| 174 | // If the string has terminated but is still smaller than Min, clear the |
| 175 | // buffer since it is too short to print. |
| 176 | Candidate.clear(); |
| 177 | } |
| 178 | |
| 179 | if (Cur == End) { |
| 180 | // Finish handling the current chunk and update ChunkOffset. |
| 181 | ChunkOffset += ChunkSize; |
| 182 | continue; |
| 183 | } |
| 184 | if (InString) { |
| 185 | // We haven't reached the end of the chunk, which means the string is |
| 186 | // terminated. Add a '\n' to start printing a new string. |
| 187 | OS << '\n'; |
| 188 | InString = false; |
| 189 | } |
| 190 | } |
| 191 | |
| 192 | // At this point, we are always at the start of a new string because the |
| 193 | // remaining part of the previous string has already been handled. |
| 194 | const char *StrHead = nullptr; |
| 195 | for (; Cur != End; ++Cur) { |
| 196 | if (isStringChar(C: *Cur)) { |
| 197 | // Find the start of the next string. |
| 198 | if (!StrHead) |
| 199 | StrHead = Cur; |
| 200 | } else if (StrHead) { |
| 201 | // If it is not a printable character, we have reached the end of the |
| 202 | // current string. Print it if long enough. |
| 203 | if (static_cast<size_t>(Cur - StrHead) >= Min) { |
| 204 | PrintHeader(ChunkOffset + (StrHead - Begin)); |
| 205 | OS << StringRef(StrHead, Cur - StrHead) << '\n'; |
| 206 | } |
| 207 | StrHead = nullptr; |
| 208 | } |
| 209 | } |
| 210 | |
| 211 | // The last string could span multiple chunks. If it is larger than Min, |
| 212 | // print the header immediately and set the InString flag to avoid printing |
| 213 | // it again. |
| 214 | if (StrHead) { |
| 215 | size_t Len = End - StrHead; |
| 216 | // Print it, or append it to Candidate if it is too short. |
| 217 | if (Len >= Min) { |
| 218 | PrintHeader(ChunkOffset + (StrHead - Begin)); |
| 219 | OS << StringRef(StrHead, Len); |
| 220 | InString = true; |
| 221 | } else { |
| 222 | Candidate.append(in_start: StrHead, in_end: End); |
| 223 | } |
| 224 | } |
| 225 | ChunkOffset += ChunkSize; |
| 226 | } |
| 227 | |
| 228 | if (InString) |
| 229 | OS << '\n'; |
| 230 | } |
| 231 | |
| 232 | int main(int argc, char **argv) { |
| 233 | InitLLVM X(argc, argv); |
| 234 | BumpPtrAllocator A; |
| 235 | StringSaver Saver(A); |
| 236 | StringsOptTable Tbl; |
| 237 | ToolName = argv[0]; |
| 238 | opt::InputArgList Args = |
| 239 | Tbl.parseArgs(Argc: argc, Argv: argv, Unknown: OPT_UNKNOWN, Saver, |
| 240 | ErrorFn: [&](StringRef Msg) { reportCmdLineError(Message: Msg); }); |
| 241 | if (Args.hasArg(Ids: OPT_help)) { |
| 242 | Tbl.printHelp( |
| 243 | OS&: outs(), |
| 244 | Usage: (Twine(ToolName) + " [options] <input object files>" ).str().c_str(), |
| 245 | Title: "llvm string dumper" ); |
| 246 | // TODO Replace this with OptTable API once it adds extrahelp support. |
| 247 | outs() << "\nPass @FILE as argument to read options from FILE.\n" ; |
| 248 | return 0; |
| 249 | } |
| 250 | if (Args.hasArg(Ids: OPT_version)) { |
| 251 | outs() << ToolName << '\n'; |
| 252 | cl::PrintVersionMessage(); |
| 253 | return 0; |
| 254 | } |
| 255 | |
| 256 | parseIntArg(Args, ID: OPT_bytes_EQ, Value&: MinLength); |
| 257 | PrintFileName = Args.hasArg(Ids: OPT_print_file_name); |
| 258 | Arg *RadixArg = Args.getLastArg(Ids: OPT_radix_EQ); |
| 259 | if (!RadixArg) { |
| 260 | Radix = Radix::None; |
| 261 | } else { |
| 262 | Radix = llvm::StringSwitch<enum Radix>(RadixArg->getValue()) |
| 263 | .Case(S: "o" , Value: Radix::Octal) |
| 264 | .Case(S: "d" , Value: Radix::Decimal) |
| 265 | .Case(S: "x" , Value: Radix::Hexadecimal) |
| 266 | .Default(Value: Radix::None); |
| 267 | if (Radix == Radix::None) |
| 268 | reportCmdLineError(Message: "'" + StringRef(RadixArg->getValue()) + |
| 269 | "' is not a valid value for '" + |
| 270 | RadixArg->getSpelling() + "'" ); |
| 271 | } |
| 272 | |
| 273 | if (MinLength == 0) { |
| 274 | errs() << "invalid minimum string length 0\n" ; |
| 275 | return EXIT_FAILURE; |
| 276 | } |
| 277 | |
| 278 | std::vector<std::string> InputFileNames = Args.getAllArgValues(Id: OPT_INPUT); |
| 279 | if (InputFileNames.empty()) |
| 280 | InputFileNames.push_back(x: "-" ); |
| 281 | |
| 282 | for (const auto &File : InputFileNames) { |
| 283 | if (File == "-" ) { |
| 284 | strings(OS&: llvm::outs(), FileName: "{standard input}" , Handle: sys::fs::getStdinHandle()); |
| 285 | } else { |
| 286 | Expected<sys::fs::file_t> FDOrErr = |
| 287 | sys::fs::openNativeFileForRead(Name: File, Flags: sys::fs::OF_TextWithCRLF); |
| 288 | if (!FDOrErr) { |
| 289 | errs() << File |
| 290 | << ": cannot open file: " << toString(E: FDOrErr.takeError()) |
| 291 | << '\n'; |
| 292 | continue; |
| 293 | } |
| 294 | strings(OS&: llvm::outs(), FileName: File, Handle: *FDOrErr); |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | return EXIT_SUCCESS; |
| 299 | } |
| 300 | |