1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653
|
# shelx.py
# BL python adaption of shelx.scm by Paul Emsley
#
# Copyright (C) 2005, 2006, 2007 by Bernhard Lohkamp
# Copyright (C) 2007, 2008 by Bernhard Lohkamp, The University of York
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
# Provide functions to handle shelx fcf file (which I believe to be old-style
# cif rather than mmCIF).
#
# So, in Coot, if it sees the extention of .fcf then it calls the
# fcf handler and the handler does the conversion, writes a new
# mmCIF file and then reads it.
#
import coot
from types import *
import numbers
def handle_shelx_fcf_file_old(filename):
import os
filename = os.path.normpath(filename)
# OK, so we have been passed a shelx fcf file.
# We need to run this through the awk filter.
# BL says: I guess in python it's easier to do it within python
output_mmCIF_file_name = filename + ".cif"
convert_shelx_fcf_to_cif(filename,output_mmCIF_file_name)
ret = coot.auto_read_cif_data_with_phases(output_mmCIF_file_name)
return ret
# read_cif_data_with_phases(output_mmCIF_file_name)
# coot.read_cif_data_with_phases_fo_fc(output_mmCIF_file_name)
def handle_shelx_fcf_file(filename):
return coot.read_small_molecule_data_cif(filename)
def convert_shelx_fcf_to_cif(fcf_filename,cif_filename):
import string, math
fin = False
fout = False
intable = 0
try:
fin = open(fcf_filename,'r')
except IOError:
print("BL ERROR:: Cannot read ", fcf_filename)
try:
fout = file(cif_filename,"w")
except IOError:
print("BL ERROR:: Cannot write ", cif_filename)
if (fin and fout):
lines = fin.readlines()
for line in lines:
if "_shelx_title" in line:
# dont write
pass
elif "_shelx_refln_list_code" in line:
# dont write
pass
elif "_shelx_F_calc_maximum" in line:
# dont write
pass
elif "_exptl_crystal_F_000" in line:
# dont write
pass
elif "_reflns_d_resolution_high" in line:
# dont write
pass
elif "_shelx_refln_list_code" in line:
# dont write
pass
elif "_shelx_F_calc_maximum" in line:
# dont write
pass
elif "_exptl_crystal_F_000" in line:
# dont write
pass
elif "_reflns_d_resolution_high" in line:
# dont write
pass
else:
if "_refln_F_squared_meas" in line:
line = line.replace("_refln_F_squared_meas"," _refln.F_meas")
if "_refln_F_squared_sigma" in line:
line = line.replace("_refln_F_squared_sigma"," _refln.F_meas_sigma")
# symm equivs:
if " _symmetry_equiv_pos_as_xyz" in line:
line = line.replace(" _symmetry_equiv_pos_as_xyz","_symmetry_equiv.pos_as_xyz")
if "_cell_length_a" in line:
line = line.replace("_cell_length_a","_cell.length_a ")
if "_refln_F_squared_meas" in line:
line = line.replace("_refln_F_squared_meas"," _refln.F_meas")
if "_refln_F_squared_sigma" in line:
line = line.replace("_refln_F_squared_sigma"," _refln.F_meas_sigma")
# symm equivs:
if "_symmetry_equiv_pos_as_xyz" in line:
line = line.replace("_symmetry_equiv_pos_as_xyz","_symmetry_equiv.pos_as_xyz")
if "_cell_length_a" in line:
line = line.replace("_cell_length_a","_cell.length_a ")
if "_cell_length_b" in line:
line = line.replace("_cell_length_b","_cell.length_b ")
if "_cell_length_c" in line:
line = line.replace("_cell_length_c","_cell.length_c ")
if "_cell_angle_alpha" in line:
line = line.replace("_cell_angle_alpha","_cell.angle_alpha ")
if "_cell_angle_beta" in line:
line = line.replace("_cell_angle_beta","_cell.angle_beta ")
if "_cell_angle_gamma" in line:
line = line.replace("_cell_angle_gamma","_cell.angle_gamma ")
if "_refln_index_h" in line:
line = line.replace("_refln_index_h"," _refln.index_h")
if "_cell_length_b" in line:
line = line.replace("_cell_length_b","_cell.length_b ")
if "_cell_length_c" in line:
line = line.replace("_cell_length_c","_cell.length_c ")
if "_cell_angle_alpha" in line:
line = line.replace("_cell_angle_alpha","_cell.angle_alpha ")
if "_cell_angle_beta" in line:
line = line.replace("_cell_angle_beta","_cell.angle_beta ")
if "_cell_angle_gamma" in line:
line = line.replace("_cell_angle_gamma","_cell.angle_gamma ")
if "_refln_index_h" in line:
line = line.replace("_refln_index_h"," _refln.index_h")
if "_refln_index_k" in line:
line = line.replace("_refln_index_k"," _refln.index_k")
if "_refln_index_l" in line:
line = line.replace("_refln_index_l"," _refln.index_l")
if "_refln_F_calc" in line:
line = line.replace("_refln_F_calc"," _refln.F_calc")
if "_refln_phase_calc" in line:
line = line.replace("_refln_phase_calc"," _refln.phase_calc")
intable = 1
if (intable == 1 and len(string.split(line))>5):
col = string.split(line)
col[3] = str(math.sqrt(float(col[3])))
if (float(col[3]) > 0.0):
col[4] = str(0.5 * (float(col[4]) + 0.0) / float(col[3]))
line = string.join(col) + "\n"
fout.write(line)
fin.close()
fout.close()
else:
print("BL ERROR:: IO error in conversion!")
global coot_shelxl_dir
coot_shelxl_dir = "coot-shelxl"
def remove_time_extensions(str):
# e.g. remove -2007-04-27_1326.24 from 03srv164-2007-04-27_1326.24
import re
pattern = "-20[0-9][0-9]-[01][0-9]-[0-3][0-9]_[0-2][0-9][0-5][-0-9].[0-5][-0-9]"
# we replace pattern with ''
result = re.sub(pattern,"",str)
return result
# BL says: shelxl_refine can have 3 arguments here:
#
# first: imol
#
# second: orig hkl file
#
# third: shelxh_flag , i.e. use shelxh instead of shelxl (in absenc of hkl
# inputfile this has to be written as: shelxh_flag=YOURSETTING
# True: use shelxh
# False: use shelxl (default)
#
def shelxl_refine(imol, hkl_file_in_maybe=False, shelxh_flag=False):
func = lambda ins_file_name: coot.write_shelx_ins_file(imol, ins_file_name)
shelxl_refine_inner(imol, hkl_file_in_maybe, func, shelxh_flag)
# shelxl refinement with input text
#
def shelxl_refine_primitive(imol, ins_text, hkl_file_in_maybe=False, shelxh_flag=False):
func = lambda ins_file_name: coot_utils.save_string_to_file(ins_text, ins_file_name, True)
shelxl_refine_inner(imol, hkl_file_in_maybe, func, shelxh_flag)
def shelxl_refine_inner(imol, hkl_file_in_maybe, func, shelxh_flag):
import os
import shutil
global coot_shelxl_dir
# First write out the imol-th molecule in shelx format.
# What should the filename be?
# Let's say that we read "abc.res"
# We want to create "abc-coot.ins". Ah, but what about the data file?
# Urg.
#
# This, as it stands, doesn't properly deal with incrementing the
# filenames.
if (shelxh_flag):
shelxl_exe = coot_utils.find_exe("shelxh", "PATH")
else:
shelxl_exe = coot_utils.find_exe("shelxl", "PATH")
if (not shelxl_exe):
print("WARNING:: can't find executable shelxl/h")
else:
orig_hkl_file = False
if (hkl_file_in_maybe and os.path.isfile(hkl_file_in_maybe)):
orig_hkl_file = hkl_file_in_maybe
dir = os.path.join(os.path.abspath("."), coot_shelxl_dir)
if (not os.path.isdir(dir)):
try:
os.mkdir(dir)
except:
pass
if (not os.path.isdir(dir)):
# We dont have shelxl dir and couldnt make it so exit!
print("Failed to make shelxl directory ", dir)
else:
# run shelx
stub = os.path.join(dir,(coot_utils.strip_path(coot_utils.strip_extension(coot.molecule_name(imol))) + "-" + coot_utils.unique_date_time_str()))
# BL says: shelxl only excepts filename <= 80 char, so check it before
if (len(stub) > 80):
print("BL WARNING:: filename %s too long! Has %s characters, only 80 are allowed!" %(stub, len(stub)))
else:
ins_filename = stub + ".ins"
res_filename = stub + ".res"
lst_filename = stub + ".lst"
fcf_filename = stub + ".fcf"
log_filename = stub + ".log"
hkl_filename = stub + ".hkl"
# BL says: non-unix OS python cant make a link. grr!
# So we have to make a copy of the passed hkl-file-in
# if we cannot create a link
# If we copied the file we maybe we should remove the
# copied file after use since it's large?!
#
# helper function to either link or copy file
# return True on sucess, False otherwise
def make_symlink(src_file, dst_file):
try:
print("make link from %s to %s" %(dst_file, src_file))
os.symlink(src_file, dst_file)
return True
except:
# failed to create link, try so copy
try:
print("copy %s to %s" %(src_file, dst_file))
shutil.copyfile(src_file, dst_file)
return True
except:
return False
# make a link to the passed hkl-file-in, if it was passed.
# (if not, we presume that this file (or the link) exists
# already. (Let's check that, and stop early if it doesn't
# exist)
if (orig_hkl_file):
symlink_target = False
if (len(orig_hkl_file) == 0):
# shouldnt happen!?
symlink_target = False
elif (hkl_filename[0:1] == "/" or hkl_filename[0:1] == "\\"):
# BL nore: maybe we shoudl use coot_utils.slash_start_qm here too?!
symlink_target = orig_hkl_file
else:
if (coot_utils.slash_start_qm(orig_hkl_file)):
symlink_target = orig_hkl_file
else:
symlink_target = "../" + orig_hkl_file
if symlink_target:
make_symlink(symlink_target, hkl_filename)
else:
# no input hklin or it wasnt an accessible file
#
# hklin was not given, let's generate a filename
# from the filename of the coordinates molecule
# imol, trial names are derived from the name of
# the coordinates molecule
trial_file_stub = coot_utils.strip_extension(coot.molecule_name(imol))
trial_hkl_file_name = trial_file_stub + ".hkl"
print("BL DEBUG:: looking for", trial_hkl_file_name)
if (os.path.isfile(trial_hkl_file_name)):
print("INFO:: hkl file %s found" %trial_hkl_file_name)
# BL says: I dont understand the following but I gues
# I have tp believe Paul (at least for now)
# if trial-hkl-file-name and hkl-filename
# are in the same directory (e.g
# shelx-coot), then (strange as it may
# seem) we *don't* put the directory name
# in the "file that exists" (the first
# argument - ie. the target). Bizarre but true.
hkl_filename_dir = os.path.abspath(hkl_filename)
trial_hkl_file_name_dir = os.path.abspath(trial_hkl_file_name)
print("BL DEBUG:: comparing path", hkl_filename_dir, trial_hkl_file_name_dir)
if (hkl_filename_dir == trial_hkl_file_name_dir):
# same dirs
# so lets strip off the dir of the target:
new_target = coot_utils.strip_path(trial_hkl_file_name)
make_symlink(new_target, hkl_filename)
else:
# different dirs, keep the dirs:
make_symlink(trial_hkl_file_name, hkl_filename)
else:
print("INFO:: hkl file %s does not exist!" %trial_hkl_file_name)
if (not os.path.isfile(hkl_filename)):
print("data (hkl) file %s not found - not running shelxl!" %hkl_filename)
else:
func(ins_filename)
# running shelxl creates stub.res
print("BL INFO:: Running shelxl as: ", shelxl_exe + " " + stub + " > " + log_filename)
shelx_status = coot_utils.popen_command(shelxl_exe, [stub], [], log_filename)
if (not shelx_status and
os.path.isfile(res_filename) and
os.path.isfile(fcf_filename)):
# it isn't a pdb file, but Coot knows what to do.
imol_res = coot.handle_read_draw_molecule_with_recentre(res_filename, 0)
handle_shelx_fcf_file(fcf_filename)
read_shelx_lst_file(lst_filename, imol_res)
else:
print("BL WARNING:: shelxl didnt run ok! No output files found!")
#handle_shelx_fcf_file("test.fcf")
#shelxl_refine(0,"./shelx_tut/lyso-tet.hkl")
# do-shelx-lst-file
#
# ie. create a interesting-things GUI for split (and other things?)
# in a shelx .lst file.
#
def read_shelx_lst_file(file_name, imol):
import os
import operator
# turn interesting-list into a GUI:
#
def gui_interesting_list(interesting_list):
if (interesting_list):
coot_gui.interesting_things_gui("Interesting Things from SHELX", interesting_list)
else:
print("INFO:: Nothing Interesting from LST file")
coot.add_status_bar_text("Nothing Interesting from LST file")
# chop off last char of string
def chop_end(s):
return s[:-1]
# return last char of string:
def last_char(s):
return s[-1]
# return a chain resno inscode atom-name altconf list
#
def make_atom_parts(atom_id):
list = []
ls = atom_id.rsplit("_",1)
if (len(ls)==2):
# chain resno inscode atom-name altconf
# The resno part can contain an alt conf specifier
# e.g.: 34 or 34a. So let's try to make it a number
# simply. If that fails, then strip off the last char
# and try again. If *that* is successful, return the
# resno and the last char is the altconf specifier.
# BL says:: we try to make it to integer, otherwise exception!?
resno = ls[1]
try:
resno = int(resno)
chain_id = chain_id_for_shelxl_residue_number(imol, resno)
if (not chain_id):
print("couldn't find chain id for resno ", resno)
return ["", resno, "", ls[0], ""]
else:
return [chain_id, resno, "", ls[0], ""]
except: # handle the alt conf
s = ls[1]
p = chop_end(s)
alt_conf = last_char(s)
try:
n = int(p)
except:
return ["", 1, "", "blank", ""] # failure
chain_id = chain_id_for_shelxl_residue_number(imol,n)
if (not chain_id):
print("couldn't find chain id for resno ", resno)
return ["", n, "", ls[0], alt_conf]
else:
return [chain_id, n, "", ls[0], alt_conf]
# So we have a line in a table something like these:
#
# " 2.7499 2.5580 0.1919 0.0400 DANG CN_2011 C_2011"
# " 1.5156 0.5000 FLAT O_1012 CA_1012 N_1013 CA_1013"
#
# It seems that there are either 4 leading numbers or 2 leading
# numbers. Let's parse up the line here and return (list
# observed target error sigma restraint) where observed and
# target are #f if there number is missing (e.g. a "FLAT"
# restraint).
#
# There are 2 regimes we understand, either
# number number number number restraint-descr atom-1-desc atom-2-desc
# or
# number number restraint-descr atom-stuff [.. atom-stuff]
#
# Actually here we will just make a string list of the
# atom-stuffs (after the restraint-descr) and parse that
# elsewhere (in do-gui) to find the atoms. The actual
# meaning of the atom-stuffs depends on the restraint-descr.
#
# return #f on something we don't understand
#
def parse_dr_line(line):
import string
bits = line.split()
if (len(bits) <=3):
return False # failed to understand
else:
try:
n0 = float(bits[0])
except:
n0 = False
try:
n1 = float(bits[1])
except:
n1 = False
try:
n2 = float(bits[2])
except:
n2 = False
try:
n3 = float(bits[3])
except:
n3 = False
if (not n0 and not n1):
return False # failed to understand
else:
if (n2 and n3):
# a 4 number line
if (len(bits)<=6):
return False # failed to understand
else:
lst = [n0, n1, n2, n3, bits[4], bits[5:]]
return lst
else:
# a 2 number line
if (len(bits)<=3): # shouldnt actually happen as we checked earlier!!! Paul!?
return False # failed to understand
else:
lst = [False, False, n0, n1, bits[2], bits[3:]]
return lst
#
def do_gui(disagreeable_restraints_list, interesting_list):
#print "BL DEBUG:: DR: ", disagreeable_restraints_list
dis_res_button_list = []
for dr_list in disagreeable_restraints_list:
restraint_type = dr_list[4]
drl_dict = {"BUMP": 0, "DANG": 0, "FLAT": 0,
"SIMU": 1, "ISOR": 1}
if (restraint_type in drl_dict):
drl_index = drl_dict[restraint_type]
else:
drl_index = 0
atom_parts = make_atom_parts(dr_list[5][drl_index])
stats_string = "not sure what?"
n2 = dr_list[2]
n3 = dr_list[3]
if (isinstance(n2, numbers.Number) and isinstance(n3, numbers.Number) and (n2 != 0)):
z = n3 / abs(n2)
stats_string = " " + str(n2) + " [Z=" + str(z) +"]"
else:
stats_string = ""
if (drl_index == 0):
rt = restraint_type + " "
else:
rt = restraint_type + " " + dr_list[5][0]
button_label = "Disagreeable Restraint " + rt + " " + atom_parts[0] + " " + \
str(atom_parts[1]) + " " + atom_parts[3] + stats_string
interesting_thing = [button_label, imol] + atom_parts
dis_res_button_list.append(interesting_thing)
gui_interesting_list(interesting_list + dis_res_button_list)
# main body
file_name = os.path.abspath(file_name)
if not coot_utils.valid_model_molecule_qm(imol):
print("WARNING:: Molecule number %i not valid" %imol)
else:
if not os.path.isfile(file_name):
print("WARNING:: shelx lst file: %s does not exist" %file_name)
else:
fin = False
interesting_list = []
try:
fin = open(file_name,'r')
except:
print("BL ERROR:: Cannot read ", file_name)
if (fin):
lines = fin.readlines()
split_list = []
disagreeable_restraints_list = []
dr_count = 0
for line in lines:
if ("may be split into" in line):
parts = line.split()
if len(parts)>6:
atom_id = parts[3]
# e.g; atom_id: CD1_1392
atom_parts = make_atom_parts(atom_id)
button_label = "Atom " + str(atom_parts[1]) + " " + atom_parts[3] + " may be split?"
split_list.append([button_label, imol] + atom_parts)
#split_list.append(atom_parts)
elif (" Observed Target Error Sigma Restraint" in line):
dr_count = 1
elif (dr_count == 1):
disagreeable_restraints_list = [] # reset the disagreeable-restraints-list
dr_count = 2
elif (dr_count == 2): # OK, in the disagreeable table
#print "BL DEBUG:: DR: ", line
dr_bits = parse_dr_line(line)
#print "BL DEBUG:: dr_bits: ", dr_bits,len(line)
if (type(dr_bits) is ListType):
disagreeable_restraints_list.append(dr_bits)
if (len(line) > 1):
dr_count = 2
else:
dr_count = 0
else:
pass
fin.close()
# print "BL DEBUG:: now run the gui with ", disagreeable_restraints_list, split_list
do_gui(disagreeable_restraints_list, split_list)
else:
print("BL ERROR:: unknown error opening file ", file_name)
#read_shelx_lst_file("test.lst",0)
# Read a shelx project (i.e. the .res file, the .lst file and the
# .fcf.cif file (if it exists (else read the .fcf file (if it
# exists)))
#
# If the file-name has an extension of .lst .res .ins .fcf .log .hkl
# then strip that off before adding new extensions.
#
def read_shelx_project(file_name):
import os
file_name = os.path.abspath(file_name)
extension = coot_utils.file_name_extension(file_name)
if (extension == "lst"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "ins"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "insh"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "log"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "hkl"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "res"):
file_stub = coot_utils.file_name_sans_extension(file_name)
elif (extension == "fcf"):
file_stub = coot_utils.file_name_sans_extension(file_name)
else:
file_stub = file_name
res_file_name = file_stub + ".res"
lst_file_name = file_stub + ".lst"
fcf_file_name = file_stub + ".fcf"
fcf_cif_file_name = file_stub + ".fcf.cif"
print(" file_name: ", file_name)
print(" file_stub: ", file_stub)
print(" extension: ", extension)
if (os.path.isfile(res_file_name)):
print("Read res file ", res_file_name)
imol_res = coot.handle_read_draw_molecule_with_recentre(res_file_name, 0)
else:
print(" No res file ", res_file_name)
imol_res = -1
if (not coot_utils.valid_model_molecule_qm(imol_res)):
print("WARNING:: Bad molecule from res file read.")
else:
if (os.path.isfile(fcf_cif_file_name)):
if (not os.path.isfile(fcf_file_name)):
print(" Read fcf-cif file ", fcf_cif_file_name)
coot.auto_read_cif_data_with_phases(fcf_cif_file_name)
else:
# OK both xxx.fcf and xxx.fcf.cif exist, we
# only want to read the xxx.fcf.cif if it is
# more recent than xxx.fcf (if it is not, we
# want to handle-shelx-fcf-file.
from stat import ST_MTIME
fcf_date = os.stat(fcf_file_name)[ST_MTIME]
fcf_cif_date = os.stat(fcf_cif_file_name)[ST_MTIME]
# print " fcf_date %s fcf_cif_date %s" %(time.ctime(fcf_date), time.ctime(fcf_cif_date))
if (fcf_date < fcf_cif_date):
coot.auto_read_cif_data_with_phases(fcf_cif_file_name)
else:
handle_shelx_fcf_file(fcf_file_name)
# xxx.fcf.cif did not exist:
else:
if (os.path.isfile(fcf_file_name)):
print(" Read fcf file ", fcf_file_name)
handle_chelx_fcf_file(fcf_file_name)
else:
print("BL WARNING:: no fcf and or cif files found!")
if (os.path.isfile(lst_file_name)):
print(" ::Read lst file ", lst_file_name)
read_shelx_lst_file(lst_file_name, imol_res)
else:
print(" ::No lst file ", lst_file_name)
|