xref: /freebsd-src/contrib/llvm-project/llvm/tools/llvm-strings/llvm-strings.cpp (revision fe6060f10f634930ff71b7c50291ddc610da2475)
162cfcf62SDimitry Andric //===-- llvm-strings.cpp - Printable String dumping utility ---------------===//
262cfcf62SDimitry Andric //
362cfcf62SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
462cfcf62SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
562cfcf62SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
662cfcf62SDimitry Andric //
762cfcf62SDimitry Andric //===----------------------------------------------------------------------===//
862cfcf62SDimitry Andric //
962cfcf62SDimitry Andric // This program is a utility that works like binutils "strings", that is, it
1062cfcf62SDimitry Andric // prints out printable strings in a binary, objdump, or archive file.
1162cfcf62SDimitry Andric //
1262cfcf62SDimitry Andric //===----------------------------------------------------------------------===//
1362cfcf62SDimitry Andric 
14*fe6060f1SDimitry Andric #include "Opts.inc"
1562cfcf62SDimitry Andric #include "llvm/Object/Binary.h"
16*fe6060f1SDimitry Andric #include "llvm/Option/Arg.h"
17*fe6060f1SDimitry Andric #include "llvm/Option/ArgList.h"
18*fe6060f1SDimitry Andric #include "llvm/Option/Option.h"
1962cfcf62SDimitry Andric #include "llvm/Support/CommandLine.h"
2062cfcf62SDimitry Andric #include "llvm/Support/Error.h"
2162cfcf62SDimitry Andric #include "llvm/Support/Format.h"
2262cfcf62SDimitry Andric #include "llvm/Support/InitLLVM.h"
2362cfcf62SDimitry Andric #include "llvm/Support/MemoryBuffer.h"
2462cfcf62SDimitry Andric #include "llvm/Support/Program.h"
25*fe6060f1SDimitry Andric #include "llvm/Support/WithColor.h"
2662cfcf62SDimitry Andric #include <cctype>
2762cfcf62SDimitry Andric #include <string>
2862cfcf62SDimitry Andric 
2962cfcf62SDimitry Andric using namespace llvm;
3062cfcf62SDimitry Andric using namespace llvm::object;
3162cfcf62SDimitry Andric 
32*fe6060f1SDimitry Andric namespace {
33*fe6060f1SDimitry Andric enum ID {
34*fe6060f1SDimitry Andric   OPT_INVALID = 0, // This is not an option ID.
35*fe6060f1SDimitry Andric #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM,  \
36*fe6060f1SDimitry Andric                HELPTEXT, METAVAR, VALUES)                                      \
37*fe6060f1SDimitry Andric   OPT_##ID,
38*fe6060f1SDimitry Andric #include "Opts.inc"
39*fe6060f1SDimitry Andric #undef OPTION
40*fe6060f1SDimitry Andric };
41*fe6060f1SDimitry Andric 
42*fe6060f1SDimitry Andric #define PREFIX(NAME, VALUE) const char *const NAME[] = VALUE;
43*fe6060f1SDimitry Andric #include "Opts.inc"
44*fe6060f1SDimitry Andric #undef PREFIX
45*fe6060f1SDimitry Andric 
46*fe6060f1SDimitry Andric static const opt::OptTable::Info InfoTable[] = {
47*fe6060f1SDimitry Andric #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM,  \
48*fe6060f1SDimitry Andric                HELPTEXT, METAVAR, VALUES)                                      \
49*fe6060f1SDimitry Andric   {                                                                            \
50*fe6060f1SDimitry Andric       PREFIX,      NAME,      HELPTEXT,                                        \
51*fe6060f1SDimitry Andric       METAVAR,     OPT_##ID,  opt::Option::KIND##Class,                        \
52*fe6060f1SDimitry Andric       PARAM,       FLAGS,     OPT_##GROUP,                                     \
53*fe6060f1SDimitry Andric       OPT_##ALIAS, ALIASARGS, VALUES},
54*fe6060f1SDimitry Andric #include "Opts.inc"
55*fe6060f1SDimitry Andric #undef OPTION
56*fe6060f1SDimitry Andric };
57*fe6060f1SDimitry Andric 
58*fe6060f1SDimitry Andric class StringsOptTable : public opt::OptTable {
59*fe6060f1SDimitry Andric public:
60*fe6060f1SDimitry Andric   StringsOptTable() : OptTable(InfoTable) { setGroupedShortOptions(true); }
61*fe6060f1SDimitry Andric };
62*fe6060f1SDimitry Andric } // namespace
63*fe6060f1SDimitry Andric 
64*fe6060f1SDimitry Andric const char ToolName[] = "llvm-strings";
65*fe6060f1SDimitry Andric 
6662cfcf62SDimitry Andric static cl::list<std::string> InputFileNames(cl::Positional,
6762cfcf62SDimitry Andric                                             cl::desc("<input object files>"),
6862cfcf62SDimitry Andric                                             cl::ZeroOrMore);
6962cfcf62SDimitry Andric 
70*fe6060f1SDimitry Andric static int MinLength = 4;
71*fe6060f1SDimitry Andric static bool PrintFileName;
7262cfcf62SDimitry Andric 
7362cfcf62SDimitry Andric enum radix { none, octal, hexadecimal, decimal };
74*fe6060f1SDimitry Andric static radix Radix;
7562cfcf62SDimitry Andric 
76*fe6060f1SDimitry Andric LLVM_ATTRIBUTE_NORETURN static void reportCmdLineError(const Twine &Message) {
77*fe6060f1SDimitry Andric   WithColor::error(errs(), ToolName) << Message << "\n";
78*fe6060f1SDimitry Andric   exit(1);
79*fe6060f1SDimitry Andric }
80*fe6060f1SDimitry Andric 
81*fe6060f1SDimitry Andric template <typename T>
82*fe6060f1SDimitry Andric static void parseIntArg(const opt::InputArgList &Args, int ID, T &Value) {
83*fe6060f1SDimitry Andric   if (const opt::Arg *A = Args.getLastArg(ID)) {
84*fe6060f1SDimitry Andric     StringRef V(A->getValue());
85*fe6060f1SDimitry Andric     if (!llvm::to_integer(V, Value, 0) || Value <= 0)
86*fe6060f1SDimitry Andric       reportCmdLineError("expected a positive integer, but got '" + V + "'");
87*fe6060f1SDimitry Andric   }
88*fe6060f1SDimitry Andric }
8962cfcf62SDimitry Andric 
9062cfcf62SDimitry Andric static void strings(raw_ostream &OS, StringRef FileName, StringRef Contents) {
9162cfcf62SDimitry Andric   auto print = [&OS, FileName](unsigned Offset, StringRef L) {
9262cfcf62SDimitry Andric     if (L.size() < static_cast<size_t>(MinLength))
9362cfcf62SDimitry Andric       return;
9462cfcf62SDimitry Andric     if (PrintFileName)
9562cfcf62SDimitry Andric       OS << FileName << ": ";
9662cfcf62SDimitry Andric     switch (Radix) {
9762cfcf62SDimitry Andric     case none:
9862cfcf62SDimitry Andric       break;
9962cfcf62SDimitry Andric     case octal:
10062cfcf62SDimitry Andric       OS << format("%7o ", Offset);
10162cfcf62SDimitry Andric       break;
10262cfcf62SDimitry Andric     case hexadecimal:
10362cfcf62SDimitry Andric       OS << format("%7x ", Offset);
10462cfcf62SDimitry Andric       break;
10562cfcf62SDimitry Andric     case decimal:
10662cfcf62SDimitry Andric       OS << format("%7u ", Offset);
10762cfcf62SDimitry Andric       break;
10862cfcf62SDimitry Andric     }
10962cfcf62SDimitry Andric     OS << L << '\n';
11062cfcf62SDimitry Andric   };
11162cfcf62SDimitry Andric 
11262cfcf62SDimitry Andric   const char *B = Contents.begin();
11362cfcf62SDimitry Andric   const char *P = nullptr, *E = nullptr, *S = nullptr;
11462cfcf62SDimitry Andric   for (P = Contents.begin(), E = Contents.end(); P < E; ++P) {
11562cfcf62SDimitry Andric     if (isPrint(*P) || *P == '\t') {
11662cfcf62SDimitry Andric       if (S == nullptr)
11762cfcf62SDimitry Andric         S = P;
11862cfcf62SDimitry Andric     } else if (S) {
11962cfcf62SDimitry Andric       print(S - B, StringRef(S, P - S));
12062cfcf62SDimitry Andric       S = nullptr;
12162cfcf62SDimitry Andric     }
12262cfcf62SDimitry Andric   }
12362cfcf62SDimitry Andric   if (S)
12462cfcf62SDimitry Andric     print(S - B, StringRef(S, E - S));
12562cfcf62SDimitry Andric }
12662cfcf62SDimitry Andric 
12762cfcf62SDimitry Andric int main(int argc, char **argv) {
12862cfcf62SDimitry Andric   InitLLVM X(argc, argv);
129*fe6060f1SDimitry Andric   BumpPtrAllocator A;
130*fe6060f1SDimitry Andric   StringSaver Saver(A);
131*fe6060f1SDimitry Andric   StringsOptTable Tbl;
132*fe6060f1SDimitry Andric   opt::InputArgList Args =
133*fe6060f1SDimitry Andric       Tbl.parseArgs(argc, argv, OPT_UNKNOWN, Saver,
134*fe6060f1SDimitry Andric                     [&](StringRef Msg) { reportCmdLineError(Msg); });
135*fe6060f1SDimitry Andric   if (Args.hasArg(OPT_help)) {
136*fe6060f1SDimitry Andric     Tbl.printHelp(
137*fe6060f1SDimitry Andric         outs(),
138*fe6060f1SDimitry Andric         (Twine(ToolName) + " [options] <input object files>").str().c_str(),
139*fe6060f1SDimitry Andric         "llvm string dumper");
140*fe6060f1SDimitry Andric     // TODO Replace this with OptTable API once it adds extrahelp support.
141*fe6060f1SDimitry Andric     outs() << "\nPass @FILE as argument to read options from FILE.\n";
142*fe6060f1SDimitry Andric     return 0;
143*fe6060f1SDimitry Andric   }
144*fe6060f1SDimitry Andric   if (Args.hasArg(OPT_version)) {
145*fe6060f1SDimitry Andric     outs() << ToolName << '\n';
146*fe6060f1SDimitry Andric     cl::PrintVersionMessage();
147*fe6060f1SDimitry Andric     return 0;
148*fe6060f1SDimitry Andric   }
14962cfcf62SDimitry Andric 
150*fe6060f1SDimitry Andric   parseIntArg(Args, OPT_bytes_EQ, MinLength);
151*fe6060f1SDimitry Andric   PrintFileName = Args.hasArg(OPT_print_file_name);
152*fe6060f1SDimitry Andric   StringRef R = Args.getLastArgValue(OPT_radix_EQ);
153*fe6060f1SDimitry Andric   if (R.empty())
154*fe6060f1SDimitry Andric     Radix = none;
155*fe6060f1SDimitry Andric   else if (R == "o")
156*fe6060f1SDimitry Andric     Radix = octal;
157*fe6060f1SDimitry Andric   else if (R == "d")
158*fe6060f1SDimitry Andric     Radix = decimal;
159*fe6060f1SDimitry Andric   else if (R == "x")
160*fe6060f1SDimitry Andric     Radix = hexadecimal;
161*fe6060f1SDimitry Andric   else
162*fe6060f1SDimitry Andric     reportCmdLineError("--radix value should be one of: '' (no offset), 'o' "
163*fe6060f1SDimitry Andric                        "(octal), 'd' (decimal), 'x' (hexadecimal)");
164*fe6060f1SDimitry Andric 
16562cfcf62SDimitry Andric   if (MinLength == 0) {
16662cfcf62SDimitry Andric     errs() << "invalid minimum string length 0\n";
16762cfcf62SDimitry Andric     return EXIT_FAILURE;
16862cfcf62SDimitry Andric   }
16962cfcf62SDimitry Andric 
170*fe6060f1SDimitry Andric   std::vector<std::string> InputFileNames = Args.getAllArgValues(OPT_INPUT);
17162cfcf62SDimitry Andric   if (InputFileNames.empty())
17262cfcf62SDimitry Andric     InputFileNames.push_back("-");
17362cfcf62SDimitry Andric 
17462cfcf62SDimitry Andric   for (const auto &File : InputFileNames) {
17562cfcf62SDimitry Andric     ErrorOr<std::unique_ptr<MemoryBuffer>> Buffer =
17662cfcf62SDimitry Andric         MemoryBuffer::getFileOrSTDIN(File);
17762cfcf62SDimitry Andric     if (std::error_code EC = Buffer.getError())
17862cfcf62SDimitry Andric       errs() << File << ": " << EC.message() << '\n';
17962cfcf62SDimitry Andric     else
18062cfcf62SDimitry Andric       strings(llvm::outs(), File == "-" ? "{standard input}" : File,
18162cfcf62SDimitry Andric               Buffer.get()->getMemBufferRef().getBuffer());
18262cfcf62SDimitry Andric   }
18362cfcf62SDimitry Andric 
18462cfcf62SDimitry Andric   return EXIT_SUCCESS;
18562cfcf62SDimitry Andric }
186