1 /** 2 * \file lzma/vli.h 3 * \brief Variable-length integer handling 4 * 5 * In the .xz format, most integers are encoded in a variable-length 6 * representation, which is sometimes called little endian base-128 encoding. 7 * This saves space when smaller values are more likely than bigger values. 8 * 9 * The encoding scheme encodes seven bits to every byte, using minimum 10 * number of bytes required to represent the given value. Encodings that use 11 * non-minimum number of bytes are invalid, thus every integer has exactly 12 * one encoded representation. The maximum number of bits in a VLI is 63, 13 * thus the vli argument must be less than or equal to UINT64_MAX / 2. You 14 * should use LZMA_VLI_MAX for clarity. 15 */ 16 17 /* 18 * Author: Lasse Collin 19 * 20 * This file has been put into the public domain. 21 * You can do whatever you want with this file. 22 * 23 * See ../lzma.h for information about liblzma as a whole. 24 */ 25 26 #ifndef LZMA_H_INTERNAL 27 # error Never include this file directly. Use <lzma.h> instead. 28 #endif 29 30 31 /** 32 * \brief Maximum supported value of variable-length integer 33 */ 34 #define LZMA_VLI_MAX (UINT64_MAX / 2) 35 36 /** 37 * \brief VLI value to denote that the value is unknown 38 */ 39 #define LZMA_VLI_UNKNOWN UINT64_MAX 40 41 /** 42 * \brief Maximum supported length of variable length integers 43 */ 44 #define LZMA_VLI_BYTES_MAX 9 45 46 47 /** 48 * \brief VLI constant suffix 49 */ 50 #define LZMA_VLI_C(n) UINT64_C(n) 51 52 53 /** 54 * \brief Variable-length integer type 55 * 56 * This will always be unsigned integer. Valid VLI values are in the range 57 * [0, LZMA_VLI_MAX]. Unknown value is indicated with LZMA_VLI_UNKNOWN, 58 * which is the maximum value of the underlaying integer type. 59 * 60 * In future, even if lzma_vli is defined to be something other than uint64_t, 61 * it is guaranteed that 2 * LZMA_VLI_MAX will not overflow lzma_vli. 62 * This simplifies integer overflow detection. 63 */ 64 typedef uint64_t lzma_vli; 65 66 67 /** 68 * \brief Simple macro to validate variable-length integer 69 * 70 * This is useful to test that application has given acceptable values 71 * for example in the uncompressed_size and compressed_size variables. 72 * 73 * \return True if the integer is representable as VLI or if it 74 * indicates unknown value. 75 */ 76 #define lzma_vli_is_valid(vli) \ 77 ((vli) <= LZMA_VLI_MAX || (vli) == LZMA_VLI_UNKNOWN) 78 79 80 /** 81 * \brief Encode a variable-length integer 82 * 83 * This function has two modes: single-call and multi-call. Single-call mode 84 * encodes the whole integer at once; it is an error if the output buffer is 85 * too small. Multi-call mode saves the position in *vli_pos, and thus it is 86 * possible to continue encoding if the buffer becomes full before the whole 87 * integer has been encoded. 88 * 89 * \param vli Integer to be encoded 90 * \param vli_pos How many VLI-encoded bytes have already been written 91 * out. When starting to encode a new integer, *vli_pos 92 * must be set to zero. To use single-call encoding, 93 * set vli_pos to NULL. 94 * \param out Beginning of the output buffer 95 * \param out_pos The next byte will be written to out[*out_pos]. 96 * \param out_size Size of the out buffer; the first byte into 97 * which no data is written to is out[out_size]. 98 * 99 * \return Slightly different return values are used in multi-call and 100 * single-call modes. 101 * 102 * Single-call (vli_pos == NULL): 103 * - LZMA_OK: Integer successfully encoded. 104 * - LZMA_PROG_ERROR: Arguments are not sane. This can be due 105 * to too little output space; single-call mode doesn't use 106 * LZMA_BUF_ERROR, since the application should have checked 107 * the encoded size with lzma_vli_size(). 108 * 109 * Multi-call (vli_pos != NULL): 110 * - LZMA_OK: So far all OK, but the integer is not 111 * completely written out yet. 112 * - LZMA_STREAM_END: Integer successfully encoded. 113 * - LZMA_BUF_ERROR: No output space was provided. 114 * - LZMA_PROG_ERROR: Arguments are not sane. 115 */ 116 extern LZMA_API(lzma_ret) lzma_vli_encode(lzma_vli vli, size_t *vli_pos, 117 uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow; 118 119 120 /** 121 * \brief Decode a variable-length integer 122 * 123 * Like lzma_vli_encode(), this function has single-call and multi-call modes. 124 * 125 * \param vli Pointer to decoded integer. The decoder will 126 * initialize it to zero when *vli_pos == 0, so 127 * application isn't required to initialize *vli. 128 * \param vli_pos How many bytes have already been decoded. When 129 * starting to decode a new integer, *vli_pos must 130 * be initialized to zero. To use single-call decoding, 131 * set this to NULL. 132 * \param in Beginning of the input buffer 133 * \param in_pos The next byte will be read from in[*in_pos]. 134 * \param in_size Size of the input buffer; the first byte that 135 * won't be read is in[in_size]. 136 * 137 * \return Slightly different return values are used in multi-call and 138 * single-call modes. 139 * 140 * Single-call (vli_pos == NULL): 141 * - LZMA_OK: Integer successfully decoded. 142 * - LZMA_DATA_ERROR: Integer is corrupt. This includes hitting 143 * the end of the input buffer before the whole integer was 144 * decoded; providing no input at all will use LZMA_DATA_ERROR. 145 * - LZMA_PROG_ERROR: Arguments are not sane. 146 * 147 * Multi-call (vli_pos != NULL): 148 * - LZMA_OK: So far all OK, but the integer is not 149 * completely decoded yet. 150 * - LZMA_STREAM_END: Integer successfully decoded. 151 * - LZMA_DATA_ERROR: Integer is corrupt. 152 * - LZMA_BUF_ERROR: No input was provided. 153 * - LZMA_PROG_ERROR: Arguments are not sane. 154 */ 155 extern LZMA_API(lzma_ret) lzma_vli_decode(lzma_vli *vli, size_t *vli_pos, 156 const uint8_t *in, size_t *in_pos, size_t in_size) 157 lzma_nothrow; 158 159 160 /** 161 * \brief Get the number of bytes required to encode a VLI 162 * 163 * \return Number of bytes on success (1-9). If vli isn't valid, 164 * zero is returned. 165 */ 166 extern LZMA_API(uint32_t) lzma_vli_size(lzma_vli vli) 167 lzma_nothrow lzma_attr_pure; 168