FortranParser

Fortran 2008 parser of mathematical expressions, based on Roland Schmehl fparser

Table of contents

• Changes
• Compilation
• Basic usage
• Usage in combination with JSON-Fortran
• Error handling
• Function string syntax
• Notes
• Credits

Changes

• What's new in version 2.0:

  • Renamed fparser to FortranParser
  • Changed approach to OOP, now everything happens by using the EquationParser class

• What's new in version 1.1: (thanks to Wilton P. Silva and Juha Mäkipelto for the bug reports)

  • EXP failed: Corrected typo in alphabet string in subroutine LowCase.
  • Expression containing binary operator with precedence over unary minus (e.g. "-x^2") failed and has been corrected in subroutines IsBinaryOp and CompileSubstr.
  • Leading plus (e.g. "+x") is now understood by correcting subroutines CompileSubstr and CheckSyntax
  • Multiple operators produce error message in subroutine CheckSyntax

Compilation

To compile FortranParser you need CMake. If CMake is installed on your machine, you can compile and install FortranParser with the following commands

• Clone the GitHub repository

git clone https://github.com/jacopo-chevallard/FortranParser.git

this command will clone into a new directory FortranParser the current master branch.

• Move into the newly created FortranParser directory, create a build directory and move into it

mkdir build ; cd build

• Run CMake

 cmake -DCMAKE_INSTALL_PREFIX=<install_dir> ..

where <install_dir> is your installation directory. The FortranParser library will be installed in <install_dir>/lib, the *mod files in <install_dir>/include. The above CMake command include the compilation of some tests. This can be avoided by passing the option -DENABLE_TESTING=OFF to the cmake command.

• Compile and install the FortranParser library

make install

Basic usage

Step 0 - Module Import

In all program units where you want to use the function parser procedures and variables you must import the module by:

USE FortranParser, only : EquationParser

This command imports only the public class EquationParser, which has only two public methods, the class constructor, and the method evaluate

Step 1 - Constructor and function parsing

An instance of the EquationParser class is created with the following syntax

  use FortranParser, only : EquationParser

  implicit none

  type(EquationParser) :: eqParser
  character(len=100)    :: stringEquation
  character(len=10)    :: variables(3)

  stringEquation = '10 + 3*x - 5*x*y + exp(-z**2)'
  variables = ['x', 'y', 'z']

  eqParser = EquationParser(stringEquation, variables)

The constructor deals with the parsing (checking and compilation) into the bytecode.

Step 2 - Function evaluation

The function value is evaluated for a specific set of variable values by calling the method

  value = eqParser%evaluate(varValues)

where varValues is 1-dimensional array containing the variable values.

Usage in combination with JSON-Fortran

FortranParser in combination with JSON-Fortran opens an easy-to-use way of reading and evaluating mathematical expressions at runtime. If you have both packages installed, than you can use:

  use FortranParser, only : EquationParser
  use json_module

  implicit none

  type(json_file) :: json
  type(EquationParser) :: eqParser

  jsonString = '{"variables" : ["x", "y"], "function" : "2*x+sin(y**2)"}'
  call json%load_from_string(jsonString)

  call json%get('function', func, found)
  call json%get('variables', vars, found)

  eqParser = EquationParser(func, vars)

  value = eqParser%evaluate([2., 5.])
  value = eqParser%evaluate([-2., -5.])

Error handling

An error in the function parsing step leads to a detailed error message (Type and position of error) and program termination.

An error during function evaluation returns a function value of 0.0 and trigger an error message from the bytecode-interpreter.

Function string syntax

Although they have to be passed as array elements of the same declared length (Fortran 90 restriction), the variable names can be of arbitrary actual length for the parser. Parsing for variables is case sensitive.

The syntax of the function string is similar to the Fortran convention. Mathematical Operators recognized are +, -, *, /, ** or alternatively ^, whereas symbols for brackets must be ().

The function parser recognizes the (single argument) Fortran 90 intrinsic functions

• abs
• exp
• log10
• log
• sqrt
• sinh
• cosh
• tanh
• sin
• cos
• tan
• asin
• acos
• atan

Parsing for intrinsic functions is case INsensitive.

Operations are evaluated in the correct order:

• () expressions in brackets first
• -A unary minus (or plus)
• A**B A^B exponentiation (A raised to the power B)
• A*B A/B multiplication and division
• A+B A-B addition and subtraction

The function string can contain integer or real constants. To be recognized as explicit constants these must conform to the format

[+|-][nnn][.nnn][e|E|d|D[+|-]nnn]

where nnn means any number of digits. The mantissa must contain at least one digit before or following an optional decimal point. Valid exponent identifiers are 'e', 'E', 'd' or 'D'. If they appear they must be followed by a valid exponent!

Notes

• The precision of real numbers can be adapted to the calling program by adjusting the KIND parameter rn in the external module parameters.

• The package compilation is based on CMake

• The package contains some test programs to demonstrate implementation and performance of the function parser.

Credits

The original fparser, by Roland Schmehl can be found at http://fparser.sourceforge.net.

The function parser concept is based on a C++ class library written by Juha Nieminen warp@iki.fi available from:

http://warp.povusers.org/FunctionParser/