make_diagnostic_fields.py

#!/usr/bin/env python3

import glob
import re
import sys
import hashlib

def Error(msg):
  sys.stderr.write("Diagnostics error: " + str(msg) + "\n")
  sys.stderr.flush()
  sys.exit(1)

baseName = "Diagnostic_Fields_New"
disabledDiags = ["Diagnostic_Source_Fields.F90", \
  "Diagnostic_Fields_Interfaces.F90"]

inputFilename = baseName + ".F90.in"
outputFilename = baseName + ".F90"

# get sha1 digest of existing generated file.  Can't use 'rw' here
# because it updates the modtime of the file, which we're trying to
# avoid doing.
orig=hashlib.sha1()
try:
    f=open(outputFilename, 'r')
    orig.update(f.read().encode("utf8"))
except IOError:
    pass
else:
    f.close()

# Valid arguments
diagnosticArguments = ["states", "state", "s_field", "v_field", "t_field", \
  "current_time", "dt", "state_index"]
# Additional code used to form above arguments
diagnosticArgumentsCode = """
    real :: current_time, dt
    type(state_type), pointer :: state => null()

    state => states(state_index)

    call get_option("/timestepping/current_time", current_time)
    call get_option("/timestepping/timestep", dt)
"""

# Initialise automagic code
useModulesCode = ""
singleStateScalarDiagnosticsCode = ""
multipleStateScalarDiagnosticsCode = ""
singleStateVectorDiagnosticsCode = ""
multipleStateVectorDiagnosticsCode = ""
singleStateTensorDiagnosticsCode = ""
multipleStateTensorDiagnosticsCode = ""

# Parse fortran files and search for diagnostic algorithms

moduleRe = re.compile(r"^\s*module\s+(\w+)\s*$", re.IGNORECASE | re.MULTILINE)
subroutineRe = re.compile(r"^\s*subroutine\s+(\w+)\(?([\w,\s]*)\)?\s*$", re.IGNORECASE | re.MULTILINE)

diagFiles = glob.glob("*.F90")
for file in [inputFilename, outputFilename] + disabledDiags:
  try:
    diagFiles.remove(file)
  except ValueError:
    pass
    
for file in diagFiles:
  fileHandle = open(file, "r")
  code = fileHandle.read()
  fileHandle.close()  
  
  modules = moduleRe.findall(code)
  for module in modules:
    useModulesCode += "  use " + module + "\n"
  
  subroutines = subroutineRe.findall(code)
  for subroutine in subroutines:
    name, args = subroutine[0].lower(), subroutine[1].lower()
    
    if not name.startswith("calculate_"):
      continue
    
    alg = name[10:]
    
    newArgs = []
    for arg in args.split(","):
      arg = arg.strip()
      if len(arg) > 0:
        newArgs.append(arg)
    args = newArgs
    
    for arg in args:
      if not arg in diagnosticArguments:
        Error("For subroutine \"" + name + "\", invalid argument \"" + arg + "\"")
    
    if "state" in args:
      if "states" in args:
        Error("For subroutine \"" + name + "\", expected exactly one type of state argument")
      singleState = True
    elif "states" in args:
      singleState = False
    else:
      singleState = True
     
    if "s_field" in args:
      if "v_field" in args or "t_field" in args:
        Error("For subroutine \"" + name + "\", expected exactly one type of diagnostic field argument")
      rank = 0
    elif "v_field" in args:
      if "s_field" in args or "t_field" in args:
        Error("For subroutine \"" + name + "\", expected exactly one type of diagnostic field argument")
      rank = 1
    elif "t_field" in args:
      if "s_field" in args or "v_field" in args:
        Error("For subroutine \"" + name + "\", expected exactly one type of diagnostic field argument")
      rank = 2
    else:
      Error("For subroutine \"" + name + "\", expected exactly one type of diagnostic field argument")
     
    algCode = "      case(\"" + alg + "\")\n" + \
      "        call calculate_" + alg
    algCode += "("
    for i, arg in enumerate(args):
      algCode += arg
      if i < len(args) - 1:
        algCode += ", "
      else:
        algCode += ")\n"
  
    if singleState:
      if rank == 0:
        singleStateScalarDiagnosticsCode += algCode
      elif rank == 1:
        singleStateVectorDiagnosticsCode += algCode
      elif rank == 2:
        singleStateTensorDiagnosticsCode += algCode
      else:
        Error("Unexpected diagnostic field type")
    else:
      if rank == 0:
        multipleStateScalarDiagnosticsCode += algCode
      elif rank == 1:
        multipleStateVectorDiagnosticsCode += algCode
      elif rank == 2:
        multipleStateTensorDiagnosticsCode += algCode
      else:
        Error("Unexpected diagnostic field type")
         
# Read input
fileHandle = open(inputFilename, "r")
outputCode = fileHandle.read()
fileHandle.close()

# Insert automagic code
outputCode = outputCode.replace("USE_MODULES", useModulesCode)
outputCode = outputCode.replace("DIAGNOSTIC_ARGUMENTS", diagnosticArgumentsCode)
outputCode = outputCode.replace("SINGLE_STATE_SCALAR_DIAGNOSTICS", singleStateScalarDiagnosticsCode)
outputCode = outputCode.replace("MULTIPLE_STATE_SCALAR_DIAGNOSTICS", multipleStateScalarDiagnosticsCode)
outputCode = outputCode.replace("SINGLE_STATE_VECTOR_DIAGNOSTICS", singleStateVectorDiagnosticsCode)
outputCode = outputCode.replace("MULTIPLE_STATE_VECTOR_DIAGNOSTICS", multipleStateVectorDiagnosticsCode)
outputCode = outputCode.replace("SINGLE_STATE_TENSOR_DIAGNOSTICS", singleStateTensorDiagnosticsCode)
outputCode = outputCode.replace("MULTIPLE_STATE_TENSOR_DIAGNOSTICS", multipleStateTensorDiagnosticsCode)

# Write the output
new=hashlib.sha1()
new.update(outputCode.encode("utf8"))

# Only write file if sha1sums differ
if new.digest() != orig.digest():
    try:
        f=open(outputFilename, 'w')
        f.write(outputCode)
    except IOError:
        # Fixme, this should fail better
        pass
    else:
        f.close()