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
14fe6060f1SDimitry Andric #include "Opts.inc"
1506c3fb27SDimitry Andric #include "llvm/ADT/StringExtras.h"
1662cfcf62SDimitry Andric #include "llvm/Object/Binary.h"
17fe6060f1SDimitry Andric #include "llvm/Option/Arg.h"
18fe6060f1SDimitry Andric #include "llvm/Option/ArgList.h"
19fe6060f1SDimitry Andric #include "llvm/Option/Option.h"
2062cfcf62SDimitry Andric #include "llvm/Support/CommandLine.h"
2162cfcf62SDimitry Andric #include "llvm/Support/Error.h"
2262cfcf62SDimitry Andric #include "llvm/Support/Format.h"
2362cfcf62SDimitry Andric #include "llvm/Support/InitLLVM.h"
2462cfcf62SDimitry Andric #include "llvm/Support/MemoryBuffer.h"
2562cfcf62SDimitry Andric #include "llvm/Support/Program.h"
26fe6060f1SDimitry Andric #include "llvm/Support/WithColor.h"
2762cfcf62SDimitry Andric #include <cctype>
2862cfcf62SDimitry Andric #include <string>
2962cfcf62SDimitry Andric
3062cfcf62SDimitry Andric using namespace llvm;
3162cfcf62SDimitry Andric using namespace llvm::object;
3262cfcf62SDimitry Andric
33fe6060f1SDimitry Andric namespace {
34fe6060f1SDimitry Andric enum ID {
35fe6060f1SDimitry Andric OPT_INVALID = 0, // This is not an option ID.
36*5f757f3fSDimitry Andric #define OPTION(...) LLVM_MAKE_OPT_ID(__VA_ARGS__),
37fe6060f1SDimitry Andric #include "Opts.inc"
38fe6060f1SDimitry Andric #undef OPTION
39fe6060f1SDimitry Andric };
40fe6060f1SDimitry Andric
41bdd1243dSDimitry Andric #define PREFIX(NAME, VALUE) \
42bdd1243dSDimitry Andric static constexpr StringLiteral NAME##_init[] = VALUE; \
43bdd1243dSDimitry Andric static constexpr ArrayRef<StringLiteral> NAME(NAME##_init, \
44bdd1243dSDimitry Andric std::size(NAME##_init) - 1);
45fe6060f1SDimitry Andric #include "Opts.inc"
46fe6060f1SDimitry Andric #undef PREFIX
47fe6060f1SDimitry Andric
48*5f757f3fSDimitry Andric using namespace llvm::opt;
49bdd1243dSDimitry Andric static constexpr opt::OptTable::Info InfoTable[] = {
50*5f757f3fSDimitry Andric #define OPTION(...) LLVM_CONSTRUCT_OPT_INFO(__VA_ARGS__),
51fe6060f1SDimitry Andric #include "Opts.inc"
52fe6060f1SDimitry Andric #undef OPTION
53fe6060f1SDimitry Andric };
54fe6060f1SDimitry Andric
55bdd1243dSDimitry Andric class StringsOptTable : public opt::GenericOptTable {
56fe6060f1SDimitry Andric public:
StringsOptTable()57bdd1243dSDimitry Andric StringsOptTable() : GenericOptTable(InfoTable) {
58bdd1243dSDimitry Andric setGroupedShortOptions(true);
5906c3fb27SDimitry Andric setDashDashParsing(true);
60bdd1243dSDimitry Andric }
61fe6060f1SDimitry Andric };
62fe6060f1SDimitry Andric } // namespace
63fe6060f1SDimitry Andric
6404eeddc0SDimitry Andric static StringRef ToolName;
65fe6060f1SDimitry Andric
6662cfcf62SDimitry Andric static cl::list<std::string> InputFileNames(cl::Positional,
6781ad6265SDimitry Andric cl::desc("<input object files>"));
6862cfcf62SDimitry Andric
69fe6060f1SDimitry Andric static int MinLength = 4;
70fe6060f1SDimitry Andric static bool PrintFileName;
7162cfcf62SDimitry Andric
7262cfcf62SDimitry Andric enum radix { none, octal, hexadecimal, decimal };
73fe6060f1SDimitry Andric static radix Radix;
7462cfcf62SDimitry Andric
reportCmdLineError(const Twine & Message)75349cc55cSDimitry Andric [[noreturn]] static void reportCmdLineError(const Twine &Message) {
76fe6060f1SDimitry Andric WithColor::error(errs(), ToolName) << Message << "\n";
77fe6060f1SDimitry Andric exit(1);
78fe6060f1SDimitry Andric }
79fe6060f1SDimitry Andric
80fe6060f1SDimitry Andric template <typename T>
parseIntArg(const opt::InputArgList & Args,int ID,T & Value)81fe6060f1SDimitry Andric static void parseIntArg(const opt::InputArgList &Args, int ID, T &Value) {
82fe6060f1SDimitry Andric if (const opt::Arg *A = Args.getLastArg(ID)) {
83fe6060f1SDimitry Andric StringRef V(A->getValue());
84fe6060f1SDimitry Andric if (!llvm::to_integer(V, Value, 0) || Value <= 0)
85fe6060f1SDimitry Andric reportCmdLineError("expected a positive integer, but got '" + V + "'");
86fe6060f1SDimitry Andric }
87fe6060f1SDimitry Andric }
8862cfcf62SDimitry Andric
strings(raw_ostream & OS,StringRef FileName,StringRef Contents)8962cfcf62SDimitry Andric static void strings(raw_ostream &OS, StringRef FileName, StringRef Contents) {
9062cfcf62SDimitry Andric auto print = [&OS, FileName](unsigned Offset, StringRef L) {
9162cfcf62SDimitry Andric if (L.size() < static_cast<size_t>(MinLength))
9262cfcf62SDimitry Andric return;
9362cfcf62SDimitry Andric if (PrintFileName)
9462cfcf62SDimitry Andric OS << FileName << ": ";
9562cfcf62SDimitry Andric switch (Radix) {
9662cfcf62SDimitry Andric case none:
9762cfcf62SDimitry Andric break;
9862cfcf62SDimitry Andric case octal:
9962cfcf62SDimitry Andric OS << format("%7o ", Offset);
10062cfcf62SDimitry Andric break;
10162cfcf62SDimitry Andric case hexadecimal:
10262cfcf62SDimitry Andric OS << format("%7x ", Offset);
10362cfcf62SDimitry Andric break;
10462cfcf62SDimitry Andric case decimal:
10562cfcf62SDimitry Andric OS << format("%7u ", Offset);
10662cfcf62SDimitry Andric break;
10762cfcf62SDimitry Andric }
10862cfcf62SDimitry Andric OS << L << '\n';
10962cfcf62SDimitry Andric };
11062cfcf62SDimitry Andric
11162cfcf62SDimitry Andric const char *B = Contents.begin();
11262cfcf62SDimitry Andric const char *P = nullptr, *E = nullptr, *S = nullptr;
11362cfcf62SDimitry Andric for (P = Contents.begin(), E = Contents.end(); P < E; ++P) {
11462cfcf62SDimitry Andric if (isPrint(*P) || *P == '\t') {
11562cfcf62SDimitry Andric if (S == nullptr)
11662cfcf62SDimitry Andric S = P;
11762cfcf62SDimitry Andric } else if (S) {
11862cfcf62SDimitry Andric print(S - B, StringRef(S, P - S));
11962cfcf62SDimitry Andric S = nullptr;
12062cfcf62SDimitry Andric }
12162cfcf62SDimitry Andric }
12262cfcf62SDimitry Andric if (S)
12362cfcf62SDimitry Andric print(S - B, StringRef(S, E - S));
12462cfcf62SDimitry Andric }
12562cfcf62SDimitry Andric
main(int argc,char ** argv)12662cfcf62SDimitry Andric int main(int argc, char **argv) {
12762cfcf62SDimitry Andric InitLLVM X(argc, argv);
128fe6060f1SDimitry Andric BumpPtrAllocator A;
129fe6060f1SDimitry Andric StringSaver Saver(A);
130fe6060f1SDimitry Andric StringsOptTable Tbl;
13104eeddc0SDimitry Andric ToolName = argv[0];
132fe6060f1SDimitry Andric opt::InputArgList Args =
133fe6060f1SDimitry Andric Tbl.parseArgs(argc, argv, OPT_UNKNOWN, Saver,
134fe6060f1SDimitry Andric [&](StringRef Msg) { reportCmdLineError(Msg); });
135fe6060f1SDimitry Andric if (Args.hasArg(OPT_help)) {
136fe6060f1SDimitry Andric Tbl.printHelp(
137fe6060f1SDimitry Andric outs(),
138fe6060f1SDimitry Andric (Twine(ToolName) + " [options] <input object files>").str().c_str(),
139fe6060f1SDimitry Andric "llvm string dumper");
140fe6060f1SDimitry Andric // TODO Replace this with OptTable API once it adds extrahelp support.
141fe6060f1SDimitry Andric outs() << "\nPass @FILE as argument to read options from FILE.\n";
142fe6060f1SDimitry Andric return 0;
143fe6060f1SDimitry Andric }
144fe6060f1SDimitry Andric if (Args.hasArg(OPT_version)) {
145fe6060f1SDimitry Andric outs() << ToolName << '\n';
146fe6060f1SDimitry Andric cl::PrintVersionMessage();
147fe6060f1SDimitry Andric return 0;
148fe6060f1SDimitry Andric }
14962cfcf62SDimitry Andric
150fe6060f1SDimitry Andric parseIntArg(Args, OPT_bytes_EQ, MinLength);
151fe6060f1SDimitry Andric PrintFileName = Args.hasArg(OPT_print_file_name);
152fe6060f1SDimitry Andric StringRef R = Args.getLastArgValue(OPT_radix_EQ);
153fe6060f1SDimitry Andric if (R.empty())
154fe6060f1SDimitry Andric Radix = none;
155fe6060f1SDimitry Andric else if (R == "o")
156fe6060f1SDimitry Andric Radix = octal;
157fe6060f1SDimitry Andric else if (R == "d")
158fe6060f1SDimitry Andric Radix = decimal;
159fe6060f1SDimitry Andric else if (R == "x")
160fe6060f1SDimitry Andric Radix = hexadecimal;
161fe6060f1SDimitry Andric else
162fe6060f1SDimitry Andric reportCmdLineError("--radix value should be one of: '' (no offset), 'o' "
163fe6060f1SDimitry Andric "(octal), 'd' (decimal), 'x' (hexadecimal)");
164fe6060f1SDimitry Andric
16562cfcf62SDimitry Andric if (MinLength == 0) {
16662cfcf62SDimitry Andric errs() << "invalid minimum string length 0\n";
16762cfcf62SDimitry Andric return EXIT_FAILURE;
16862cfcf62SDimitry Andric }
16962cfcf62SDimitry Andric
170fe6060f1SDimitry 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