Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 1 | //===-- llvm-strings.cpp - Printable String dumping utility ---------------===// |
| 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 program is a utility that works like binutils "strings", that is, it |
| 11 | // prints out printable strings in a binary, objdump, or archive file. |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 15 | #include "llvm/Object/Binary.h" |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 16 | #include "llvm/Support/CommandLine.h" |
| 17 | #include "llvm/Support/Error.h" |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 18 | #include "llvm/Support/Format.h" |
Rui Ueyama | 0b9d56a | 2018-04-13 18:26:06 +0000 | [diff] [blame] | 19 | #include "llvm/Support/InitLLVM.h" |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 20 | #include "llvm/Support/MemoryBuffer.h" |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 21 | #include "llvm/Support/Program.h" |
Saleem Abdulrasool | ca29721 | 2016-11-11 04:00:59 +0000 | [diff] [blame] | 22 | #include <cctype> |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 23 | #include <string> |
| 24 | |
| 25 | using namespace llvm; |
| 26 | using namespace llvm::object; |
| 27 | |
| 28 | static cl::list<std::string> InputFileNames(cl::Positional, |
| 29 | cl::desc("<input object files>"), |
| 30 | cl::ZeroOrMore); |
| 31 | |
Saleem Abdulrasool | 4da3d72 | 2016-11-14 21:10:41 +0000 | [diff] [blame] | 32 | static cl::opt<bool> |
| 33 | PrintFileName("print-file-name", |
| 34 | cl::desc("Print the name of the file before each string")); |
| 35 | static cl::alias PrintFileNameShort("f", cl::desc(""), |
| 36 | cl::aliasopt(PrintFileName)); |
| 37 | |
Saleem Abdulrasool | fe23a5c | 2016-11-15 00:43:52 +0000 | [diff] [blame] | 38 | static cl::opt<int> |
| 39 | MinLength("bytes", cl::desc("Print sequences of the specified length"), |
| 40 | cl::init(4)); |
| 41 | static cl::alias MinLengthShort("n", cl::desc(""), cl::aliasopt(MinLength)); |
| 42 | |
Martin Storsjo | ba2da13 | 2017-11-14 19:58:36 +0000 | [diff] [blame] | 43 | static cl::opt<bool> |
| 44 | AllSections("all", |
| 45 | cl::desc("Check all sections, not just the data section")); |
| 46 | static cl::alias AllSectionsShort("a", cl::desc(""), |
| 47 | cl::aliasopt(AllSections)); |
| 48 | |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 49 | enum radix { none, octal, hexadecimal, decimal }; |
| 50 | static cl::opt<radix> |
| 51 | Radix("radix", cl::desc("print the offset within the file"), |
| 52 | cl::values(clEnumValN(octal, "o", "octal"), |
| 53 | clEnumValN(hexadecimal, "x", "hexadecimal"), |
| 54 | clEnumValN(decimal, "d", "decimal")), |
| 55 | cl::init(none)); |
| 56 | static cl::alias RadixShort("t", cl::desc(""), cl::aliasopt(Radix)); |
| 57 | |
Saleem Abdulrasool | 4da3d72 | 2016-11-14 21:10:41 +0000 | [diff] [blame] | 58 | static void strings(raw_ostream &OS, StringRef FileName, StringRef Contents) { |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 59 | auto print = [&OS, FileName](unsigned Offset, StringRef L) { |
Saleem Abdulrasool | fe23a5c | 2016-11-15 00:43:52 +0000 | [diff] [blame] | 60 | if (L.size() < static_cast<size_t>(MinLength)) |
| 61 | return; |
Saleem Abdulrasool | 4da3d72 | 2016-11-14 21:10:41 +0000 | [diff] [blame] | 62 | if (PrintFileName) |
Jordan Rupprecht | 1070226 | 2018-11-09 18:03:21 +0000 | [diff] [blame] | 63 | OS << FileName << ": "; |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 64 | switch (Radix) { |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 65 | case none: |
| 66 | break; |
| 67 | case octal: |
Jordan Rupprecht | 1070226 | 2018-11-09 18:03:21 +0000 | [diff] [blame] | 68 | OS << format("%7o ", Offset); |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 69 | break; |
| 70 | case hexadecimal: |
Jordan Rupprecht | 1070226 | 2018-11-09 18:03:21 +0000 | [diff] [blame] | 71 | OS << format("%7x ", Offset); |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 72 | break; |
| 73 | case decimal: |
Jordan Rupprecht | 1070226 | 2018-11-09 18:03:21 +0000 | [diff] [blame] | 74 | OS << format("%7u ", Offset); |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 75 | break; |
| 76 | } |
Jordan Rupprecht | 1070226 | 2018-11-09 18:03:21 +0000 | [diff] [blame] | 77 | OS << L << '\n'; |
Saleem Abdulrasool | 4da3d72 | 2016-11-14 21:10:41 +0000 | [diff] [blame] | 78 | }; |
| 79 | |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 80 | const char *B = Contents.begin(); |
Saleem Abdulrasool | 421c744 | 2016-11-12 03:39:21 +0000 | [diff] [blame] | 81 | const char *P = nullptr, *E = nullptr, *S = nullptr; |
| 82 | for (P = Contents.begin(), E = Contents.end(); P < E; ++P) { |
James Henderson | 306868b | 2018-10-24 13:16:16 +0000 | [diff] [blame] | 83 | if (isPrint(*P) || *P == '\t') { |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 84 | if (S == nullptr) |
| 85 | S = P; |
| 86 | } else if (S) { |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 87 | print(S - B, StringRef(S, P - S)); |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 88 | S = nullptr; |
| 89 | } |
| 90 | } |
Saleem Abdulrasool | fe23a5c | 2016-11-15 00:43:52 +0000 | [diff] [blame] | 91 | if (S) |
Saleem Abdulrasool | 6d72920 | 2017-01-21 02:36:28 +0000 | [diff] [blame] | 92 | print(S - B, StringRef(S, E - S)); |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 93 | } |
| 94 | |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 95 | int main(int argc, char **argv) { |
Rui Ueyama | 0b9d56a | 2018-04-13 18:26:06 +0000 | [diff] [blame] | 96 | InitLLVM X(argc, argv); |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 97 | |
| 98 | cl::ParseCommandLineOptions(argc, argv, "llvm string dumper\n"); |
Saleem Abdulrasool | fe23a5c | 2016-11-15 00:43:52 +0000 | [diff] [blame] | 99 | if (MinLength == 0) { |
| 100 | errs() << "invalid minimum string length 0\n"; |
| 101 | return EXIT_FAILURE; |
| 102 | } |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 103 | |
| 104 | if (InputFileNames.empty()) |
| 105 | InputFileNames.push_back("-"); |
| 106 | |
Saleem Abdulrasool | 98825db | 2016-11-12 18:37:04 +0000 | [diff] [blame] | 107 | for (const auto &File : InputFileNames) { |
| 108 | ErrorOr<std::unique_ptr<MemoryBuffer>> Buffer = |
| 109 | MemoryBuffer::getFileOrSTDIN(File); |
| 110 | if (std::error_code EC = Buffer.getError()) |
| 111 | errs() << File << ": " << EC.message() << '\n'; |
| 112 | else |
Saleem Abdulrasool | 4da3d72 | 2016-11-14 21:10:41 +0000 | [diff] [blame] | 113 | strings(llvm::outs(), File == "-" ? "{standard input}" : File, |
| 114 | Buffer.get()->getMemBufferRef().getBuffer()); |
Saleem Abdulrasool | 98825db | 2016-11-12 18:37:04 +0000 | [diff] [blame] | 115 | } |
| 116 | |
Saleem Abdulrasool | 6a1b55e | 2016-11-11 03:44:12 +0000 | [diff] [blame] | 117 | return EXIT_SUCCESS; |
| 118 | } |