DAMASK_EICMD/processing/pre/geom_canvas.py

194 lines
8.7 KiB
Python
Raw Normal View History

#!/usr/bin/env python
# -*- coding: UTF-8 no BOM -*-
import os,re,sys,math,string
import numpy as np
from optparse import OptionParser
import damask
scriptID = string.replace('$Id$','\n','\\n')
2014-11-18 21:01:39 +05:30
scriptName = os.path.splitext(scriptID.split()[1])[0]
# --------------------------------------------------------------------
2013-05-15 22:53:21 +05:30
# MAIN
# --------------------------------------------------------------------
identifiers = {
'grid': ['a','b','c'],
'size': ['x','y','z'],
'origin': ['x','y','z'],
}
mappings = {
'grid': lambda x: int(x),
'size': lambda x: float(x),
'origin': lambda x: float(x),
'homogenization': lambda x: int(x),
'microstructures': lambda x: int(x),
}
parser = OptionParser(option_class=damask.extendableOption, usage='%prog options [file[s]]', description = """
Changes the (three-dimensional) canvas of a spectral geometry description.
""", version = scriptID)
parser.add_option('-g', '--grid', dest='grid', nargs = 3, metavar=' '.join(['string']*3),
help='a,b,c grid of hexahedral box [unchanged]')
parser.add_option('-o', '--offset', dest='offset', type='int', nargs = 3, metavar=' '.join(['int']*3),
help='a,b,c offset from old to new origin of grid %default')
parser.add_option('-f', '--fill', dest='fill', type='int', metavar = 'int',
help='(background) canvas grain index. "0" selects maximum microstructure index + 1 [%default]')
parser.set_defaults(grid = ['0','0','0'])
parser.set_defaults(offset = (0,0,0))
parser.set_defaults(fill = 0)
(options, filenames) = parser.parse_args()
2013-05-15 22:53:21 +05:30
#--- setup file handles --------------------------------------------------------------------------
files = []
if filenames == []:
files.append({'name':'STDIN',
'input':sys.stdin,
'output':sys.stdout,
'croak':sys.stderr,
})
else:
for name in filenames:
if os.path.exists(name):
files.append({'name':name,
'input':open(name),
'output':open(name+'_tmp','w'),
'croak':sys.stdout,
})
#--- loop over input files ------------------------------------------------------------------------
for file in files:
2014-11-18 21:01:39 +05:30
file['croak'].write('\033[1m' + scriptName + '\033[0m: ' + (file['name'] if file['name'] != 'STDIN' else '') + '\n')
2015-02-14 16:53:50 +05:30
table = damask.ASCIItable(file['input'],file['output'],labels = False)
table.head_read()
#--- interpret header ----------------------------------------------------------------------------
info = {
'grid': np.zeros(3,'i'),
'size': np.zeros(3,'d'),
'origin': np.zeros(3,'d'),
'homogenization': 0,
2013-06-30 18:05:04 +05:30
'microstructures': 0,
}
2013-05-15 22:53:21 +05:30
newInfo = {
'grid': np.zeros(3,'i'),
'origin': np.zeros(3,'d'),
2013-05-15 22:53:21 +05:30
'microstructures': 0,
}
extra_header = []
2015-02-14 16:53:50 +05:30
for header in table.info:
headitems = map(str.lower,header.split())
if len(headitems) == 0: continue # skip blank lines
if headitems[0] in mappings.keys():
if headitems[0] in identifiers.keys():
for i in xrange(len(identifiers[headitems[0]])):
info[headitems[0]][i] = \
mappings[headitems[0]](headitems[headitems.index(identifiers[headitems[0]][i])+1])
else:
info[headitems[0]] = mappings[headitems[0]](headitems[1])
else:
extra_header.append(header)
2013-05-15 22:53:21 +05:30
file['croak'].write('grid a b c: %s\n'%(' x '.join(map(str,info['grid']))) + \
'size x y z: %s\n'%(' x '.join(map(str,info['size']))) + \
'origin x y z: %s\n'%(' : '.join(map(str,info['origin']))) + \
'homogenization: %i\n'%info['homogenization'] + \
'microstructures: %i\n'%info['microstructures'])
if np.any(info['grid'] < 1):
2013-05-15 22:53:21 +05:30
file['croak'].write('invalid grid a b c.\n')
continue
if np.any(info['size'] <= 0.0):
2013-05-15 22:53:21 +05:30
file['croak'].write('invalid size x y z.\n')
continue
2013-05-15 22:53:21 +05:30
#--- read data ------------------------------------------------------------------------------------
microstructure = np.zeros(info['grid'].prod(),'i') # initialize as flat array
i = 0
2015-02-14 16:53:50 +05:30
while table.data_read():
items = table.data
if len(items) > 2:
if items[1].lower() == 'of': items = [int(items[2])]*int(items[0])
elif items[1].lower() == 'to': items = xrange(int(items[0]),1+int(items[2]))
else: items = map(int,items)
else: items = map(int,items)
s = len(items)
microstructure[i:i+s] = items
i += s
#--- do work ------------------------------------------------------------------------------------
newInfo['grid'] = np.array(int(o*float(n.translate(None,'xX'))) if [n[-1].lower() == 'x' else int(n.translate(None,'xX'))}[n[-1].lower() == 'x'] for o,n in zip(info['grid'],options.grid)],'i')
newInfo['grid'] = np.where(newInfo['grid'] <= 0 , info['grid'],newInfo['grid'])
microstructure = microstructure.reshape(info['grid'],order='F')
microstructure_cropped = np.zeros(newInfo['grid'],'i')
2013-05-15 22:53:21 +05:30
microstructure_cropped.fill({True:options.fill,False:microstructure.max()+1}[options.fill>0])
xindex = list(set(xrange(options.offset[0],options.offset[0]+newInfo['grid'][0])) & \
set(xrange(info['grid'][0])))
yindex = list(set(xrange(options.offset[1],options.offset[1]+newInfo['grid'][1])) & \
set(xrange(info['grid'][1])))
zindex = list(set(xrange(options.offset[2],options.offset[2]+newInfo['grid'][2])) & \
set(xrange(info['grid'][2])))
translate_x = [i - options.offset[0] for i in xindex]
translate_y = [i - options.offset[1] for i in yindex]
translate_z = [i - options.offset[2] for i in zindex]
microstructure_cropped[min(translate_x):(max(translate_x)+1),\
min(translate_y):(max(translate_y)+1),\
min(translate_z):(max(translate_z)+1)] \
= microstructure[min(xindex):(max(xindex)+1),\
min(yindex):(max(yindex)+1),\
min(zindex):(max(zindex)+1)]
newInfo['size'] = info['size']/info['grid']*newInfo['grid']
2013-05-15 22:53:21 +05:30
newInfo['origin'] = info['origin']+info['size']/info['grid']*options.offset
newInfo['microstructures'] = microstructure_cropped.max()
2013-05-15 22:53:21 +05:30
#--- report ---------------------------------------------------------------------------------------
if (any(newInfo['grid'] != info['grid'])):
file['croak'].write('--> grid a b c: %s\n'%(' x '.join(map(str,newInfo['grid']))))
if (any(newInfo['size'] != info['size'])):
file['croak'].write('--> size x y z: %s\n'%(' x '.join(map(str,newInfo['size']))))
if (any(newInfo['origin'] != info['origin'])):
file['croak'].write('--> origin x y z: %s\n'%(' : '.join(map(str,newInfo['origin']))))
2013-05-15 22:53:21 +05:30
if (newInfo['microstructures'] != info['microstructures']):
file['croak'].write('--> microstructures: %i\n'%newInfo['microstructures'])
if np.any(newInfo['grid'] < 1):
2013-05-15 22:53:21 +05:30
file['croak'].write('invalid new grid a b c.\n')
continue
if np.any(newInfo['size'] <= 0.0):
2013-05-15 22:53:21 +05:30
file['croak'].write('invalid new size x y z.\n')
continue
#--- write header ---------------------------------------------------------------------------------
2015-02-14 16:53:50 +05:30
table.labels_clear()
table.info_clear()
table.info_append(extra_header+[
2014-01-20 20:11:56 +05:30
scriptID + ' ' + ' '.join(sys.argv[1:]),
"grid\ta %i\tb %i\tc %i"%(newInfo['grid'][0],newInfo['grid'][1],newInfo['grid'][2],),
"size\tx %f\ty %f\tz %f"%(newInfo['size'][0],newInfo['size'][1],newInfo['size'][2],),
"origin\tx %f\ty %f\tz %f"%(newInfo['origin'][0],newInfo['origin'][1],newInfo['origin'][2],),
"homogenization\t%i"%info['homogenization'],
2013-06-30 18:05:04 +05:30
"microstructures\t%i"%(newInfo['microstructures']),
])
2015-02-14 16:53:50 +05:30
table.head_write()
table.output_flush()
2013-05-15 22:53:21 +05:30
# --- write microstructure information ------------------------------------------------------------
formatwidth = int(math.floor(math.log10(microstructure_cropped.max())+1))
2015-02-14 16:53:50 +05:30
table.data = microstructure_cropped.reshape((newInfo['grid'][0],newInfo['grid'][1]*newInfo['grid'][2]),order='F').transpose()
table.data_writeArray('%%%ii'%(formatwidth),delimiter=' ')
2013-05-15 22:53:21 +05:30
#--- output finalization --------------------------------------------------------------------------
if file['name'] != 'STDIN':
table.input_close()
table.output_close()
os.rename(file['name']+'_tmp',file['name'])