Skip to content

Files

Latest commit

dustsignsdustsigns
Convert DCLib documentation to Markdown
Apr 1, 2019
aa10435 · Apr 1, 2019

History

History
38 lines (29 loc) · 2.81 KB

readme.md

File metadata and controls

38 lines (29 loc) · 2.81 KB

Overview

DCLib is a library which allows compressing and decompressing (referred to as encoding and decoding henceforth) data from buffers (see DCIOLib).

enc_dec.h: Allows listing and using all implemented encoders/decoders as well as their options. Life cycle: GetEncoder -> (optional option configuration, see below) -> enc_dec_t.encoder (for encoding) or enc_dec_t.decoder (for decoding) call on initialized input and output bit buffers. Optional option configuration: (optional) OptionNameExists -> (optional) EncoderSupportsOption -> GetOptionType -> GetAllowedOptionValueRange -> SetOptionValue<Type>.

Encoders/decoders

  • aggregate: Sums of num_values (option name) consecutive floating-point values (no decoder!).
  • bac: Performs binary arithmetic coding as implemented by Witten et al.
  • copy: Copies the input to the output, i.e., it performs no compression whatsoever. This encoder/decoder operates on blocks of blocksize (option name) bits size.
  • csv: Reads lines of comma-separated values and converts the strings in column number column (option name) of each line to a list of (binary) floating-point values when encoding; performs the reverse conversion when decoding and inserts blank columns if necessary.
  • diff: Encodes (signed) differences between consecutive (unsigned) values of valuesize (option name) bits size when encoding; reconstructs (unsigned) values of valuesize (option name) bits size from their consecutive (signed) differences when decoding
  • lzmh: Performs LZMH coding and decoding from Ringwelski et al. This is an integrated third-party implementation.
  • normalize: Converts floating-point values to (signed) integer values of valuesize (option name) bits size when encoding; performs the reverse conversion when decoding. To preserve decimal places after the decimal point, all values are multiplied by normalization_factor (option name) when encoding, and divided when decoding.
  • seg: Creates Exponential Golomb code words from values when encoding; reconstructs Exponential Golomb code words when decoding. All values are valuesize (option name) bits in size and signed.

Supported encoder input and output formats

Note: Decoder input and formats are reversed, if there is a decoder).

  • aggregate: binary float in, binary float out
  • bac: arbitrary in, binary out
  • copy: arbitrary in, arbitrary out
  • csv: ASCII float in, binary float out
  • diff: unsigned int in, signed int out
  • lzmh: ASCII float in, binary out
  • normalize: float in, signed int out
  • seg: signed int in, binary out

Notes on usage

  • GetEncoderNames requires a char* array with GetNumberOfEncoders fields.
  • When adding or renaming encoders/decoders or options, make sure the arrays remain sorted by name. Otherwise, the find operations will not work as expected.