libCSD is a C++ library providing some different techniques for managing string dictionaries in compressed space. These approaches are inspired on the paper:

========================================================================== "Compressed String Dictionaries" Nieves R. Brisaboa, Rodrigo Cánovas, Francisco Claude, Miguel A. Martínez-Prieto, and Gonzalo Navarro 10th Symposium on Experimental Algorithms (SEA'2011), p.136-147, 2011.

The current release is a beta version providing full functionality for string location and ID extraction. Besides, some techniques support queries by prefix and substrings. The set of techniques is described as follows:

• "Hash": implements four different techniques which combines hashing and Huffman / Re-Pair compression.
• "FC": implements two different techniques based on Front-Coding compression. PFC is a straightforward byte-oriented implementation, and HTFC performs Hu-Tucker compression over the headers while supports Huffman / Re-Pair compression for the internal strings.
• "RPDAC": uses Re-Pair for compressing the strings and.
• "FMI" : self-indexes the dictionary and provides all types of queries using the FM-Index facilities.
• "XBW" : obtains a compressed trie of the dictionary and transforms it to support all types of queries in highly compressed space.

The techniques support different parameterizations to achieve varied and competitive space/time tradeoffs.

Compiling the library

1) For building the libcds library, type "make" within the libcds folder. 2) Once built libcds, type "make" within the root folder and libCSD is ready.

Building a dictionary

The library provides a simple command-line script for building dictionaries:

./Build

• The first parameter chooses the of dictionary to be built.
• The set of provide specific configuration values.
• The parameter locates the original dictionary file.
• The parameter tells the destination path.

It is worth noting that "in" dictionary file must be lexicographically sorted and strings must be ended with the '\0' ASCII char.

Examples:

./Build 1 h 10 geonames dicts/geo.10

Builds a HASH dictionary for "geonames" and stores it as "dicts/geo.10". The dictionary uses a hash table which hash 10% more cells than strings in "geonames" and compresses them using Huffman (h).

./Build 4 r 16 geonames dicts/geo.16

Builds a HTFC (Hu-Tucker Front-Coding) dictionary for "geonames" and stores it as "dicts/geo.16". The dictionary uses buckets of 16 strings and compresses them using Re-Pair.

Testing a dictionary

The library also provides a command-line script for testing purposes:

./Test

• This script supports four different :

  • 'r' is used for running the test chosen in :
    • 'l' (for testing locate), 'e' (extract).
    • 'pl' (prefix location), 'pe' (prefix extraction).
    • 'sl' (substring location), 'pe' (substring extraction).
  • 'g' is used for generating a basic testbed comprising valid strings for locate and valid IDs for extract.
  • 'p' is used for generating a prefix testbed comprising 100,000 query patterns of some different lengths. The parameter tells the a reference string length, and the testbed obtain different files with patterns longer and shorter than this length.
  • 's' is used for generating a substring testbed with the features described above.

• The parameter locates the file storing the compressed string dictionary.

• The last parameter locates the file comprising the testbed (for 'r' mode) or the destination path for saving the corresponding testbed (modes 'g', 'p', or 's').

Examples:

./Test r l dicts/geo.10 tests/geo.strings

Uses the testbed "tests/geo.strings" for analyzing how the dictionary stored at "dicts/geo.10" behaves for 'locate'.

./Test g 100000 dicts/geo.10 tests/geo

Uses the dictionary stored at "dicts/geo.10" for generating a basic testbed of 100,000 strings and IDs which are stored as "tests/geo".

./Test p 16 dicts/geo.10 tests/geo

Uses the dictionary stored at "dicts/geo.10" for generating a prefix testbed comprising patterns of lengths longer and shorter than 16 chars. The resulting pattern set are stored at "tests/geo".

If you find bugs or have any issue with library, please ask us. Enjoy the library and if you find it useful for your research, please cite our paper:

@article{MPBCCN16, author = "M.A. Mart{\'{\i}}nez-Prieto and N. Brisaboa and R. C{\'a}novas
and F. Claude and G. Navarro", title = "Practical Compressed String Dictionaries", booktitle = "Information Systems", year = "2016", volume = {56}, pages = "73--108" }