A Modern Fortran-based Parser

Status
Info
License

This repository contains a modern Fortran-based input file parser which can be used within an MPI-enabled Fortran program. It uses a heavily modified libfdf library. More about this library in the Licence file.

---

Get the code

The library is included in this repository as a submodule. To clone the entire project, do

git clone git@github.com:neelravi/mpi-libfdf-parser.git

Compilation

The project contains two folders (a) modified-libfdf and (b) parser.

Compile and install the modified-libfdf using the following set of commands

./configure --prefix=/usr/local FC=ifort CC=icc

make

sudo make install

In the parser folder, link the modified libfdf library with the interface Fortran file.

ifort -c m_periodic_table.F90 m_keywords.F90

ifort interface.F90 m_keywords.F90 m_periodic_table.F90 /usr/local/lib/libfdf.a

Integrate parser in your code

Just include the interface.F90 file and the keyword declaration module files in the existing Makefile of your code.

---

Features of the parser (including inheritance from libfdf)

• Blank lines and lines starting with # are ignored.

• The order of keyword-value pairs does not matter

• Spaces and tabs are ignored; keyword-value pairs are parsed in a free-form.

• Keywords are case insensitive.

• A default value can be set for keywords not present in the input file.

• Large data can be parsed using the %block structure.

• Multiple keyword-value pairs can be clubbed together in a %module structure.

Syntax

1. Include another input file for parser to read using:

   %include global.inp

2. Include a data file for parser to read using:

   load label filename

3. Here, depending upon the label, parser will provide the filename. For example,

   load basis cc-pvtz.gbs

4. Read molecular coordinates directly from the input file using

   %block molecule
   12
   #benzene comment
   C    0.00000    1.40272  0
   H    0.00000    2.49029  0
   C   -1.21479    0.70136  0
   H   -2.15666    1.24515  0
   C   -1.21479   -0.70136  0
   H   -2.15666   -1.24515  0
   C    0.00000   -1.40272  0
   H    0.00000   -2.49029  0
   C    1.21479   -0.70136  0
   H    2.15666   -1.24515  0
   C    1.21479    0.70136  0
   H    2.15666    1.24515  0
   %endblock

5. Read molecular coordinates from an external .xyz file using

   %block molecule < benzene.xyz

6. Group certain keywords using the %module construct

   %module DMC
     tau     =   0.04
     etrial  = -15 Ha
   %endmodule

7. Logical variables accept true, TRUE, T, .true. as valid keywords for True. The fdf_boolean function can also take "1" as true and "0" as false from the input.

   optimize_wavefunction true

8. Single and Double precision numbers along with numbers in scientific format can be read using the fdf_get() function. The second number in the bracket denotes the default value.

   energy_tol = fdf_get('energy_tol', 0.00001d0)

9. Floats/integers/strings/booleans can be parsed generically using the interface fdf_get() function. Strings are limited to 132 characters per line.

   sr_tau = fdf_get('sr_tau', 0.025d0)

   nspin1 = fdf_get('nspin1', 1)

   opt_method = fdf_get('opt_method', "sr_n")

   multiple_adiag = fdf_get('multiple_adiag', .false.)

10. Units can be specified to variables. Unit conversion is possible at the parsing.

    If the input file has etrial = -15 Ha entry, the etrial variable can be assigned values using the fdf_physical function with unit conversion.

    etrial = fdf_physical('etrial', -20.d0, 'eV')

11. List of public functions available for parsing the data:

    Initiate

    fdf_init

    fdf_shutdown

    Single data

    fdf_get

    fdf_integer

    fdf_single

    fdf_double

    fdf_string

    fdf_boolean

    fdf_physical

    fdf_convfac

    fdf_load_filename

    Lists (data enclosed in [])

    fdf_islist

    fdf_islinteger

    fdf_islreal

    fdf_list

    fdf_linteger

    fdf_ldouble

    Returns the string associated with a mark line

    fdf_getline

    Test if a label is defined

    fdf_defined

    fdf_isphysical

    fdf_isblock

    fdf_load_defined

    Allow to overwrite things in the FDF

    fdf_overwrite

    fdf_removelabel

    fdf_addline

    Block reading (processing each line of data)

    fdf_block, fdf_block_linecount

    fdf_bline, fdf_bbackspace, fdf_brewind, fdf_bclose

    fdf_bnintegers, fdf_bnreals, fdf_bnvalues, fdf_bnnames, fdf_bntokens

    fdf_bintegers, fdf_breals, fdf_bvalues, fdf_bnames, fdf_btokens

    fdf_bboolean, fdf_bphysical

    fdf_bnlists, fdf_bnilists, fdf_bnrlists, fdf_bnvlists

    fdf_bilists, fdf_brlists, fdf_bvlists

    Match, search over blocks, and destroy block structure

    fdf_bmatch, fdf_bsearch, fdf_substring_search

    fdf_setoutput, fdf_setdebug

---

Demonstration

In the parser folder, we have included a sample interface.F90 and m_keywords.F90 files. In the interface.F90 file, we have demonstrated how keyword-values pairs, simple data blocks, and data from external files can be read easily.

Contributors

2021- Ravindra Shinde (@neelravi) r.l.shinde@utwente.nl

2007-2014 Based on the LibFDF library originally designed by Alberto Garcia and Jose Soler, and reimplemented by Raul de la Cruz (Barcelona Supercomputer Center).