vli.h 6.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166
  1. /* SPDX-License-Identifier: 0BSD */
  2. /**
  3. * \file lzma/vli.h
  4. * \brief Variable-length integer handling
  5. * \note Never include this file directly. Use <lzma.h> instead.
  6. *
  7. * In the .xz format, most integers are encoded in a variable-length
  8. * representation, which is sometimes called little endian base-128 encoding.
  9. * This saves space when smaller values are more likely than bigger values.
  10. *
  11. * The encoding scheme encodes seven bits to every byte, using minimum
  12. * number of bytes required to represent the given value. Encodings that use
  13. * non-minimum number of bytes are invalid, thus every integer has exactly
  14. * one encoded representation. The maximum number of bits in a VLI is 63,
  15. * thus the vli argument must be less than or equal to UINT64_MAX / 2. You
  16. * should use LZMA_VLI_MAX for clarity.
  17. */
  18. /*
  19. * Author: Lasse Collin
  20. */
  21. #ifndef LZMA_H_INTERNAL
  22. # error Never include this file directly. Use <lzma.h> instead.
  23. #endif
  24. /**
  25. * \brief Maximum supported value of a variable-length integer
  26. */
  27. #define LZMA_VLI_MAX (UINT64_MAX / 2)
  28. /**
  29. * \brief VLI value to denote that the value is unknown
  30. */
  31. #define LZMA_VLI_UNKNOWN UINT64_MAX
  32. /**
  33. * \brief Maximum supported encoded length of variable length integers
  34. */
  35. #define LZMA_VLI_BYTES_MAX 9
  36. /**
  37. * \brief VLI constant suffix
  38. */
  39. #define LZMA_VLI_C(n) UINT64_C(n)
  40. /**
  41. * \brief Variable-length integer type
  42. *
  43. * Valid VLI values are in the range [0, LZMA_VLI_MAX]. Unknown value is
  44. * indicated with LZMA_VLI_UNKNOWN, which is the maximum value of the
  45. * underlying integer type.
  46. *
  47. * lzma_vli will be uint64_t for the foreseeable future. If a bigger size
  48. * is needed in the future, it is guaranteed that 2 * LZMA_VLI_MAX will
  49. * not overflow lzma_vli. This simplifies integer overflow detection.
  50. */
  51. typedef uint64_t lzma_vli;
  52. /**
  53. * \brief Validate a variable-length integer
  54. *
  55. * This is useful to test that application has given acceptable values
  56. * for example in the uncompressed_size and compressed_size variables.
  57. *
  58. * \return True if the integer is representable as a VLI or if it
  59. * indicates an unknown value. False otherwise.
  60. */
  61. #define lzma_vli_is_valid(vli) \
  62. ((vli) <= LZMA_VLI_MAX || (vli) == LZMA_VLI_UNKNOWN)
  63. /**
  64. * \brief Encode a variable-length integer
  65. *
  66. * This function has two modes: single-call and multi-call. Single-call mode
  67. * encodes the whole integer at once; it is an error if the output buffer is
  68. * too small. Multi-call mode saves the position in *vli_pos, and thus it is
  69. * possible to continue encoding if the buffer becomes full before the whole
  70. * integer has been encoded.
  71. *
  72. * \param vli Integer to be encoded
  73. * \param[out] vli_pos How many VLI-encoded bytes have already been written
  74. * out. When starting to encode a new integer in
  75. * multi-call mode, *vli_pos must be set to zero.
  76. * To use single-call encoding, set vli_pos to NULL.
  77. * \param[out] out Beginning of the output buffer
  78. * \param[out] out_pos The next byte will be written to out[*out_pos].
  79. * \param out_size Size of the out buffer; the first byte into
  80. * which no data is written to is out[out_size].
  81. *
  82. * \return Slightly different return values are used in multi-call and
  83. * single-call modes.
  84. *
  85. * Single-call (vli_pos == NULL):
  86. * - LZMA_OK: Integer successfully encoded.
  87. * - LZMA_PROG_ERROR: Arguments are not sane. This can be due
  88. * to too little output space; single-call mode doesn't use
  89. * LZMA_BUF_ERROR, since the application should have checked
  90. * the encoded size with lzma_vli_size().
  91. *
  92. * Multi-call (vli_pos != NULL):
  93. * - LZMA_OK: So far all OK, but the integer is not
  94. * completely written out yet.
  95. * - LZMA_STREAM_END: Integer successfully encoded.
  96. * - LZMA_BUF_ERROR: No output space was provided.
  97. * - LZMA_PROG_ERROR: Arguments are not sane.
  98. */
  99. extern LZMA_API(lzma_ret) lzma_vli_encode(lzma_vli vli, size_t *vli_pos,
  100. uint8_t *out, size_t *out_pos, size_t out_size) lzma_nothrow;
  101. /**
  102. * \brief Decode a variable-length integer
  103. *
  104. * Like lzma_vli_encode(), this function has single-call and multi-call modes.
  105. *
  106. * \param[out] vli Pointer to decoded integer. The decoder will
  107. * initialize it to zero when *vli_pos == 0, so
  108. * application isn't required to initialize *vli.
  109. * \param[out] vli_pos How many bytes have already been decoded. When
  110. * starting to decode a new integer in multi-call
  111. * mode, *vli_pos must be initialized to zero. To
  112. * use single-call decoding, set vli_pos to NULL.
  113. * \param in Beginning of the input buffer
  114. * \param[out] in_pos The next byte will be read from in[*in_pos].
  115. * \param in_size Size of the input buffer; the first byte that
  116. * won't be read is in[in_size].
  117. *
  118. * \return Slightly different return values are used in multi-call and
  119. * single-call modes.
  120. *
  121. * Single-call (vli_pos == NULL):
  122. * - LZMA_OK: Integer successfully decoded.
  123. * - LZMA_DATA_ERROR: Integer is corrupt. This includes hitting
  124. * the end of the input buffer before the whole integer was
  125. * decoded; providing no input at all will use LZMA_DATA_ERROR.
  126. * - LZMA_PROG_ERROR: Arguments are not sane.
  127. *
  128. * Multi-call (vli_pos != NULL):
  129. * - LZMA_OK: So far all OK, but the integer is not
  130. * completely decoded yet.
  131. * - LZMA_STREAM_END: Integer successfully decoded.
  132. * - LZMA_DATA_ERROR: Integer is corrupt.
  133. * - LZMA_BUF_ERROR: No input was provided.
  134. * - LZMA_PROG_ERROR: Arguments are not sane.
  135. */
  136. extern LZMA_API(lzma_ret) lzma_vli_decode(lzma_vli *vli, size_t *vli_pos,
  137. const uint8_t *in, size_t *in_pos, size_t in_size)
  138. lzma_nothrow;
  139. /**
  140. * \brief Get the number of bytes required to encode a VLI
  141. *
  142. * \param vli Integer whose encoded size is to be determined
  143. *
  144. * \return Number of bytes on success (1-9). If vli isn't valid,
  145. * zero is returned.
  146. */
  147. extern LZMA_API(uint32_t) lzma_vli_size(lzma_vli vli)
  148. lzma_nothrow lzma_attr_pure;