Fortran hash table {#mainpage}

The module dictionary_m implements a "<string,string>" hash table based on the djb2 hash function.

program example_from_readme
  use dictionary_m
  implicit none

  type(dictionary_t) :: d

  call d%init(1024)

  call d%set('one', 'one')
  call d%set('pi', '3.14159')

  write(*,*) 'one', ' = ', d%get('one')
  write(*,*) 'pi', ' = ', d%get('pi')

end program example_from_readme

Author: Pierre de Buyl

License: BSD

Project goals

• Demonstrate a hash table in Fortran.
• Use standard Fortran 2008.
• Single file implementation.
• No dependency.
• No preprocessing.
• Not design a generic container, dictionary_m just stores strings. It is of course possible to replace the "value" entry in the type entry_t for a single-datatype dictionary.

The hash table is based on plain allocatable arrays and the base data is stored in (len=:), allocatable character variables. Buckets are extended arbitrarily but reallocating, thus collisions will slow down this implementation with respect to others using better data structures.

The goal is not to beat other implementations performance-wise but to provide the convenience for Fortran programmers to store configuration settings, or other auxiliary data easily.

Installation

dictionary_m consists of a single Fortran file. You can just drop src/dictionary_m.f90 in your Fortran project.