195ec533aSDimitry Andric //===-- TarWriter.cpp - Tar archive file creator --------------------------===//
295ec533aSDimitry Andric //
395ec533aSDimitry Andric //                     The LLVM Compiler Infrastructure
495ec533aSDimitry Andric //
595ec533aSDimitry Andric // This file is distributed under the University of Illinois Open Source
695ec533aSDimitry Andric // License. See LICENSE.TXT for details.
795ec533aSDimitry Andric //
895ec533aSDimitry Andric //===----------------------------------------------------------------------===//
995ec533aSDimitry Andric //
1095ec533aSDimitry Andric // TarWriter class provides a feature to create a tar archive file.
1195ec533aSDimitry Andric //
1295ec533aSDimitry Andric // I put emphasis on simplicity over comprehensiveness when implementing this
1395ec533aSDimitry Andric // class because we don't need a full-fledged archive file generator in LLVM
1495ec533aSDimitry Andric // at the moment.
1595ec533aSDimitry Andric //
1695ec533aSDimitry Andric // The filename field in the Unix V7 tar header is 100 bytes. Longer filenames
1795ec533aSDimitry Andric // are stored using the PAX extension. The PAX header is standardized in
1895ec533aSDimitry Andric // POSIX.1-2001.
1995ec533aSDimitry Andric //
2095ec533aSDimitry Andric // The struct definition of UstarHeader is copied from
2195ec533aSDimitry Andric // https://www.freebsd.org/cgi/man.cgi?query=tar&sektion=5
2295ec533aSDimitry Andric //
2395ec533aSDimitry Andric //===----------------------------------------------------------------------===//
2495ec533aSDimitry Andric 
2595ec533aSDimitry Andric #include "llvm/Support/TarWriter.h"
2695ec533aSDimitry Andric #include "llvm/ADT/StringRef.h"
2795ec533aSDimitry Andric #include "llvm/Support/FileSystem.h"
2895ec533aSDimitry Andric #include "llvm/Support/MathExtras.h"
2924e2fe98SDimitry Andric #include "llvm/Support/Path.h"
3095ec533aSDimitry Andric 
3195ec533aSDimitry Andric using namespace llvm;
3295ec533aSDimitry Andric 
3395ec533aSDimitry Andric // Each file in an archive must be aligned to this block size.
3495ec533aSDimitry Andric static const int BlockSize = 512;
3595ec533aSDimitry Andric 
3695ec533aSDimitry Andric struct UstarHeader {
3795ec533aSDimitry Andric   char Name[100];
3895ec533aSDimitry Andric   char Mode[8];
3995ec533aSDimitry Andric   char Uid[8];
4095ec533aSDimitry Andric   char Gid[8];
4195ec533aSDimitry Andric   char Size[12];
4295ec533aSDimitry Andric   char Mtime[12];
4395ec533aSDimitry Andric   char Checksum[8];
4495ec533aSDimitry Andric   char TypeFlag;
4595ec533aSDimitry Andric   char Linkname[100];
4695ec533aSDimitry Andric   char Magic[6];
4795ec533aSDimitry Andric   char Version[2];
4895ec533aSDimitry Andric   char Uname[32];
4995ec533aSDimitry Andric   char Gname[32];
5095ec533aSDimitry Andric   char DevMajor[8];
5195ec533aSDimitry Andric   char DevMinor[8];
5295ec533aSDimitry Andric   char Prefix[155];
5395ec533aSDimitry Andric   char Pad[12];
5495ec533aSDimitry Andric };
5595ec533aSDimitry Andric static_assert(sizeof(UstarHeader) == BlockSize, "invalid Ustar header");
5695ec533aSDimitry Andric 
57f1a29dd3SDimitry Andric static UstarHeader makeUstarHeader() {
58f1a29dd3SDimitry Andric   UstarHeader Hdr = {};
59f1a29dd3SDimitry Andric   memcpy(Hdr.Magic, "ustar", 5); // Ustar magic
60f1a29dd3SDimitry Andric   memcpy(Hdr.Version, "00", 2);  // Ustar version
61f1a29dd3SDimitry Andric   return Hdr;
62f1a29dd3SDimitry Andric }
63f1a29dd3SDimitry Andric 
6495ec533aSDimitry Andric // A PAX attribute is in the form of "<length> <key>=<value>\n"
6595ec533aSDimitry Andric // where <length> is the length of the entire string including
6695ec533aSDimitry Andric // the length field itself. An example string is this.
6795ec533aSDimitry Andric //
6895ec533aSDimitry Andric //   25 ctime=1084839148.1212\n
6995ec533aSDimitry Andric //
7095ec533aSDimitry Andric // This function create such string.
7195ec533aSDimitry Andric static std::string formatPax(StringRef Key, StringRef Val) {
7295ec533aSDimitry Andric   int Len = Key.size() + Val.size() + 3; // +3 for " ", "=" and "\n"
7395ec533aSDimitry Andric 
7495ec533aSDimitry Andric   // We need to compute total size twice because appending
7595ec533aSDimitry Andric   // a length field could change total size by one.
7695ec533aSDimitry Andric   int Total = Len + Twine(Len).str().size();
7795ec533aSDimitry Andric   Total = Len + Twine(Total).str().size();
7895ec533aSDimitry Andric   return (Twine(Total) + " " + Key + "=" + Val + "\n").str();
7995ec533aSDimitry Andric }
8095ec533aSDimitry Andric 
8195ec533aSDimitry Andric // Headers in tar files must be aligned to 512 byte boundaries.
8295ec533aSDimitry Andric // This function forwards the current file position to the next boundary.
8395ec533aSDimitry Andric static void pad(raw_fd_ostream &OS) {
8495ec533aSDimitry Andric   uint64_t Pos = OS.tell();
8595ec533aSDimitry Andric   OS.seek(alignTo(Pos, BlockSize));
8695ec533aSDimitry Andric }
8795ec533aSDimitry Andric 
8895ec533aSDimitry Andric // Computes a checksum for a tar header.
8995ec533aSDimitry Andric static void computeChecksum(UstarHeader &Hdr) {
9095ec533aSDimitry Andric   // Before computing a checksum, checksum field must be
9195ec533aSDimitry Andric   // filled with space characters.
9295ec533aSDimitry Andric   memset(Hdr.Checksum, ' ', sizeof(Hdr.Checksum));
9395ec533aSDimitry Andric 
9495ec533aSDimitry Andric   // Compute a checksum and set it to the checksum field.
9595ec533aSDimitry Andric   unsigned Chksum = 0;
9695ec533aSDimitry Andric   for (size_t I = 0; I < sizeof(Hdr); ++I)
9795ec533aSDimitry Andric     Chksum += reinterpret_cast<uint8_t *>(&Hdr)[I];
9895ec533aSDimitry Andric   snprintf(Hdr.Checksum, sizeof(Hdr.Checksum), "%06o", Chksum);
9995ec533aSDimitry Andric }
10095ec533aSDimitry Andric 
10195ec533aSDimitry Andric // Create a tar header and write it to a given output stream.
10295ec533aSDimitry Andric static void writePaxHeader(raw_fd_ostream &OS, StringRef Path) {
10395ec533aSDimitry Andric   // A PAX header consists of a 512-byte header followed
10495ec533aSDimitry Andric   // by key-value strings. First, create key-value strings.
10595ec533aSDimitry Andric   std::string PaxAttr = formatPax("path", Path);
10695ec533aSDimitry Andric 
10795ec533aSDimitry Andric   // Create a 512-byte header.
108f1a29dd3SDimitry Andric   UstarHeader Hdr = makeUstarHeader();
10995ec533aSDimitry Andric   snprintf(Hdr.Size, sizeof(Hdr.Size), "%011zo", PaxAttr.size());
11095ec533aSDimitry Andric   Hdr.TypeFlag = 'x'; // PAX magic
11195ec533aSDimitry Andric   computeChecksum(Hdr);
11295ec533aSDimitry Andric 
11395ec533aSDimitry Andric   // Write them down.
11495ec533aSDimitry Andric   OS << StringRef(reinterpret_cast<char *>(&Hdr), sizeof(Hdr));
11595ec533aSDimitry Andric   OS << PaxAttr;
11695ec533aSDimitry Andric   pad(OS);
11795ec533aSDimitry Andric }
11895ec533aSDimitry Andric 
1192cab237bSDimitry Andric // Path fits in a Ustar header if
1202cab237bSDimitry Andric //
1212cab237bSDimitry Andric // - Path is less than 100 characters long, or
1222cab237bSDimitry Andric // - Path is in the form of "<prefix>/<name>" where <prefix> is less
1232cab237bSDimitry Andric //   than or equal to 155 characters long and <name> is less than 100
1242cab237bSDimitry Andric //   characters long. Both <prefix> and <name> can contain extra '/'.
1252cab237bSDimitry Andric //
1262cab237bSDimitry Andric // If Path fits in a Ustar header, updates Prefix and Name and returns true.
1272cab237bSDimitry Andric // Otherwise, returns false.
1282cab237bSDimitry Andric static bool splitUstar(StringRef Path, StringRef &Prefix, StringRef &Name) {
1292cab237bSDimitry Andric   if (Path.size() < sizeof(UstarHeader::Name)) {
1302cab237bSDimitry Andric     Prefix = "";
1312cab237bSDimitry Andric     Name = Path;
1322cab237bSDimitry Andric     return true;
13324e2fe98SDimitry Andric   }
13424e2fe98SDimitry Andric 
1352cab237bSDimitry Andric   size_t Sep = Path.rfind('/', sizeof(UstarHeader::Prefix) + 1);
1362cab237bSDimitry Andric   if (Sep == StringRef::npos)
1372cab237bSDimitry Andric     return false;
1382cab237bSDimitry Andric   if (Path.size() - Sep - 1 >= sizeof(UstarHeader::Name))
1392cab237bSDimitry Andric     return false;
1402cab237bSDimitry Andric 
1412cab237bSDimitry Andric   Prefix = Path.substr(0, Sep);
1422cab237bSDimitry Andric   Name = Path.substr(Sep + 1);
1432cab237bSDimitry Andric   return true;
14424e2fe98SDimitry Andric }
14524e2fe98SDimitry Andric 
14695ec533aSDimitry Andric // The PAX header is an extended format, so a PAX header needs
14795ec533aSDimitry Andric // to be followed by a "real" header.
1482cab237bSDimitry Andric static void writeUstarHeader(raw_fd_ostream &OS, StringRef Prefix,
1492cab237bSDimitry Andric                              StringRef Name, size_t Size) {
150f1a29dd3SDimitry Andric   UstarHeader Hdr = makeUstarHeader();
15124e2fe98SDimitry Andric   memcpy(Hdr.Name, Name.data(), Name.size());
15295ec533aSDimitry Andric   memcpy(Hdr.Mode, "0000664", 8);
15395ec533aSDimitry Andric   snprintf(Hdr.Size, sizeof(Hdr.Size), "%011zo", Size);
15424e2fe98SDimitry Andric   memcpy(Hdr.Prefix, Prefix.data(), Prefix.size());
15595ec533aSDimitry Andric   computeChecksum(Hdr);
15695ec533aSDimitry Andric   OS << StringRef(reinterpret_cast<char *>(&Hdr), sizeof(Hdr));
15795ec533aSDimitry Andric }
15895ec533aSDimitry Andric 
15995ec533aSDimitry Andric // Creates a TarWriter instance and returns it.
16095ec533aSDimitry Andric Expected<std::unique_ptr<TarWriter>> TarWriter::create(StringRef OutputPath,
16195ec533aSDimitry Andric                                                        StringRef BaseDir) {
16295ec533aSDimitry Andric   int FD;
16395ec533aSDimitry Andric   if (std::error_code EC = openFileForWrite(OutputPath, FD, sys::fs::F_None))
16495ec533aSDimitry Andric     return make_error<StringError>("cannot open " + OutputPath, EC);
16595ec533aSDimitry Andric   return std::unique_ptr<TarWriter>(new TarWriter(FD, BaseDir));
16695ec533aSDimitry Andric }
16795ec533aSDimitry Andric 
16895ec533aSDimitry Andric TarWriter::TarWriter(int FD, StringRef BaseDir)
16995ec533aSDimitry Andric     : OS(FD, /*shouldClose=*/true, /*unbuffered=*/false), BaseDir(BaseDir) {}
17095ec533aSDimitry Andric 
17195ec533aSDimitry Andric // Append a given file to an archive.
17295ec533aSDimitry Andric void TarWriter::append(StringRef Path, StringRef Data) {
17395ec533aSDimitry Andric   // Write Path and Data.
1742cab237bSDimitry Andric   std::string Fullpath = BaseDir + "/" + sys::path::convert_to_slash(Path);
1752cab237bSDimitry Andric 
1762cab237bSDimitry Andric   // We do not want to include the same file more than once.
1772cab237bSDimitry Andric   if (!Files.insert(Fullpath).second)
1782cab237bSDimitry Andric     return;
1792cab237bSDimitry Andric 
1802cab237bSDimitry Andric   StringRef Prefix;
1812cab237bSDimitry Andric   StringRef Name;
1822cab237bSDimitry Andric   if (splitUstar(Fullpath, Prefix, Name)) {
1832cab237bSDimitry Andric     writeUstarHeader(OS, Prefix, Name, Data.size());
18495ec533aSDimitry Andric   } else {
1852cab237bSDimitry Andric     writePaxHeader(OS, Fullpath);
1862cab237bSDimitry Andric     writeUstarHeader(OS, "", "", Data.size());
18795ec533aSDimitry Andric   }
18895ec533aSDimitry Andric 
18995ec533aSDimitry Andric   OS << Data;
19095ec533aSDimitry Andric   pad(OS);
19195ec533aSDimitry Andric 
19295ec533aSDimitry Andric   // POSIX requires tar archives end with two null blocks.
19395ec533aSDimitry Andric   // Here, we write the terminator and then seek back, so that
19495ec533aSDimitry Andric   // the file being output is terminated correctly at any moment.
19595ec533aSDimitry Andric   uint64_t Pos = OS.tell();
19695ec533aSDimitry Andric   OS << std::string(BlockSize * 2, '\0');
19795ec533aSDimitry Andric   OS.seek(Pos);
19895ec533aSDimitry Andric   OS.flush();
19995ec533aSDimitry Andric }
200