ceph/src/rgw/rgw_tar.h

   1 // -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
   2 // vim: ts=8 sw=2 smarttab ft=cpp
   3
   4 #ifndef CEPH_RGW_TAR_H
   5 #define CEPH_RGW_TAR_H
   6
   7 #include <algorithm>
   8 #include <array>
   9 #include <cstring>
  10 #include <string_view>
  11 #include <tuple>
  12 #include <utility>
  13
  14 #include <boost/optional.hpp>
  15 #include <boost/range/adaptor/reversed.hpp>
  16
  17 namespace rgw {
  18 namespace tar {
  19
  20 static constexpr size_t BLOCK_SIZE = 512;
  21
  22
  23 static inline std::pair<class StatusIndicator,
  24                         boost::optional<class HeaderView>>
  25 interpret_block(const StatusIndicator& status, ceph::bufferlist& bl);
  26
  27
  28 class StatusIndicator {
  29   friend std::pair<class StatusIndicator,
  30                    boost::optional<class HeaderView>>
  31   interpret_block(const StatusIndicator& status, ceph::bufferlist& bl);
  32
  33   bool is_empty;
  34   bool is_eof;
  35
  36   StatusIndicator()
  37     : is_empty(false),
  38       is_eof(false) {
  39   }
  40
  41   StatusIndicator(const StatusIndicator& prev_status,
  42                   const bool is_empty)
  43   : is_empty(is_empty),
  44     is_eof(is_empty && prev_status.empty()) {
  45   }
  46
  47 public:
  48   bool empty() const {
  49     return is_empty;
  50   }
  51
  52   bool eof() const {
  53     return is_eof;
  54   }
  55
  56   static StatusIndicator create() {
  57     return StatusIndicator();
  58   }
  59 } /* class StatusIndicator */;
  60
  61
  62 enum class FileType : char {
  63   UNKNOWN = '\0',
  64
  65   /* The tar format uses ASCII encoding. */
  66   NORMAL_FILE = '0',
  67   DIRECTORY = '5'
  68 }; /* enum class FileType */
  69
  70 class HeaderView {
  71 protected:
  72   /* Everythng is char here (ASCII encoding), so we don't need to worry about
  73    * the struct padding. */
  74   const struct header_t {
  75     char filename[100];
  76     char __filemode[8];
  77     char __owner_id[8];
  78     char __group_id[8];
  79     char filesize[12];
  80     char lastmod[12];
  81     char checksum[8];
  82     char filetype;
  83     char __padding[355];
  84   } *header;
  85
  86   static_assert(sizeof(*header) == BLOCK_SIZE,
  87                 "The TAR header must be exactly BLOCK_SIZE length");
  88
  89   /* The label is far more imporant from what the code really does. */
  90   static size_t pos2len(const size_t pos) {
  91     return pos + 1;
  92   }
  93
  94 public:
  95   explicit HeaderView(const char (&header)[BLOCK_SIZE])
  96     : header(reinterpret_cast<const header_t*>(header)) {
  97   }
  98
  99   FileType get_filetype() const {
 100     switch (header->filetype) {
 101       case static_cast<char>(FileType::NORMAL_FILE):
 102         return FileType::NORMAL_FILE;
 103       case static_cast<char>(FileType::DIRECTORY):
 104         return FileType::DIRECTORY;
 105       default:
 106         return FileType::UNKNOWN;
 107     }
 108   }
 109
 110   std::string_view get_filename() const {
 111     return std::string_view(header->filename,
 112                              std::min(sizeof(header->filename),
 113                                       strlen(header->filename)));
 114   }
 115
 116   size_t get_filesize() const {
 117     /* The string_ref is pretty suitable here because tar encodes its
 118      * metadata in ASCII. */
 119     const std::string_view raw(header->filesize, sizeof(header->filesize));
 120
 121     /* We need to find where the padding ends. */
 122     const auto pad_ends_at = std::min(raw.find_last_not_of('\0'),
 123                                       raw.find_last_not_of(' '));
 124     const auto trimmed = raw.substr(0,
 125       pad_ends_at == std::string_view::npos ? std::string_view::npos
 126                                              : pos2len(pad_ends_at));
 127
 128     size_t sum = 0, mul = 1;
 129     for (const char c : boost::adaptors::reverse(trimmed)) {
 130       sum += (c - '0') * mul;
 131       mul *= 8;
 132     }
 133
 134     return sum;
 135   }
 136 }; /* class Header */
 137
 138
 139 static inline std::pair<StatusIndicator,
 140                         boost::optional<HeaderView>>
 141 interpret_block(const StatusIndicator& status, ceph::bufferlist& bl) {
 142   static constexpr std::array<char, BLOCK_SIZE> zero_block = {0, };
 143   const char (&block)[BLOCK_SIZE] = \
 144     reinterpret_cast<const char (&)[BLOCK_SIZE]>(*bl.c_str());
 145
 146   if (std::memcmp(zero_block.data(), block, BLOCK_SIZE) == 0) {
 147     return std::make_pair(StatusIndicator(status, true), boost::none);
 148   } else {
 149     return std::make_pair(StatusIndicator(status, false), HeaderView(block));
 150   }
 151 }
 152
 153 } /* namespace tar */
 154 } /* namespace rgw */
 155
 156 #endif /* CEPH_RGW_TAR_H */