mirror of
https://github.com/fjosw/pyerrors.git
synced 2025-05-14 19:43:41 +02:00
365 lines
16 KiB
Python
365 lines
16 KiB
Python
import os
|
|
import fnmatch
|
|
import re
|
|
import numpy as np # Thinly-wrapped numpy
|
|
from ..obs import Obs
|
|
from . import utils
|
|
|
|
|
|
def read_sfcf(path, prefix, name, quarks='.*', corr_type='bi', noffset=0, wf=0, wf2=0, version="1.0c", cfg_separator="n", **kwargs):
|
|
"""Read sfcf c format from given folder structure.
|
|
|
|
Parameters
|
|
----------
|
|
path : str
|
|
Path to the sfcf files.
|
|
prefix : str
|
|
Prefix of the sfcf files.
|
|
name : str
|
|
Name of the correlation function to read.
|
|
quarks : str
|
|
Label of the quarks used in the sfcf input file. e.g. "quark quark"
|
|
for version 0.0 this does NOT need to be given with the typical " - "
|
|
that is present in the output file,
|
|
this is done automatically for this version
|
|
corr_type : str
|
|
Type of correlation function to read. Can be
|
|
- 'bi' for boundary-inner
|
|
- 'bb' for boundary-boundary
|
|
- 'bib' for boundary-inner-boundary
|
|
noffset : int
|
|
Offset of the source (only relevant when wavefunctions are used)
|
|
wf : int
|
|
ID of wave function
|
|
wf2 : int
|
|
ID of the second wavefunction
|
|
(only relevant for boundary-to-boundary correlation functions)
|
|
im : bool
|
|
if True, read imaginary instead of real part
|
|
of the correlation function.
|
|
names : list
|
|
Alternative labeling for replicas/ensembles.
|
|
Has to have the appropriate length
|
|
ens_name : str
|
|
replaces the name of the ensemble
|
|
version: str
|
|
version of SFCF, with which the measurement was done.
|
|
if the compact output option (-c) was specified,
|
|
append a "c" to the version (e.g. "1.0c")
|
|
if the append output option (-a) was specified,
|
|
append an "a" to the version
|
|
cfg_separator : str
|
|
String that separates the ensemble identifier from the configuration number (default 'n').
|
|
replica: list
|
|
list of replica to be read, default is all
|
|
files: list
|
|
list of files to be read per replica, default is all.
|
|
for non-compact output format, hand the folders to be read here.
|
|
check_configs:
|
|
list of list of supposed configs, eg. [range(1,1000)]
|
|
for one replicum with 1000 configs
|
|
"""
|
|
if kwargs.get('im'):
|
|
im = 1
|
|
part = 'imaginary'
|
|
else:
|
|
im = 0
|
|
part = 'real'
|
|
if "replica" in kwargs:
|
|
reps = kwargs.get("replica")
|
|
if corr_type == 'bb':
|
|
b2b = True
|
|
single = True
|
|
elif corr_type == 'bib':
|
|
b2b = True
|
|
single = False
|
|
else:
|
|
b2b = False
|
|
single = False
|
|
compact = True
|
|
appended = False
|
|
known_versions = ["0.0", "1.0", "2.0", "1.0c", "2.0c", "1.0a", "2.0a"]
|
|
|
|
if version not in known_versions:
|
|
raise Exception("This version is not known!")
|
|
if (version[-1] == "c"):
|
|
appended = False
|
|
compact = True
|
|
version = version[:-1]
|
|
elif (version[-1] == "a"):
|
|
appended = True
|
|
compact = False
|
|
version = version[:-1]
|
|
else:
|
|
compact = False
|
|
appended = False
|
|
read = 0
|
|
T = 0
|
|
start = 0
|
|
ls = []
|
|
if "replica" in kwargs:
|
|
ls = reps
|
|
else:
|
|
for (dirpath, dirnames, filenames) in os.walk(path):
|
|
if not appended:
|
|
ls.extend(dirnames)
|
|
else:
|
|
ls.extend(filenames)
|
|
break
|
|
if not ls:
|
|
raise Exception('Error, directory not found')
|
|
# Exclude folders with different names
|
|
for exc in ls:
|
|
if not fnmatch.fnmatch(exc, prefix + '*'):
|
|
ls = list(set(ls) - set([exc]))
|
|
|
|
if not appended:
|
|
if len(ls) > 1:
|
|
# New version, to cope with ids, etc.
|
|
ls.sort(key=lambda x: int(re.findall(r'\d+', x[len(prefix):])[0]))
|
|
replica = len(ls)
|
|
else:
|
|
replica = len([file.split(".")[-1] for file in ls]) // len(set([file.split(".")[-1] for file in ls]))
|
|
print('Read', part, 'part of', name, 'from', prefix[:-1],
|
|
',', replica, 'replica')
|
|
if 'names' in kwargs:
|
|
new_names = kwargs.get('names')
|
|
if len(new_names) != len(set(new_names)):
|
|
raise Exception("names are not unique!")
|
|
if len(new_names) != replica:
|
|
raise Exception('Names does not have the required length', replica)
|
|
else:
|
|
new_names = []
|
|
if not appended:
|
|
for entry in ls:
|
|
try:
|
|
idx = entry.index('r')
|
|
except Exception:
|
|
raise Exception("Automatic recognition of replicum failed, please enter the key word 'names'.")
|
|
|
|
if 'ens_name' in kwargs:
|
|
new_names.append(kwargs.get('ens_name') + '|' + entry[idx:])
|
|
else:
|
|
new_names.append(entry[:idx] + '|' + entry[idx:])
|
|
else:
|
|
|
|
for exc in ls:
|
|
if not fnmatch.fnmatch(exc, prefix + '*.' + name):
|
|
ls = list(set(ls) - set([exc]))
|
|
ls.sort(key=lambda x: int(re.findall(r'\d+', x)[-1]))
|
|
for entry in ls:
|
|
myentry = entry[:-len(name) - 1]
|
|
try:
|
|
idx = myentry.index('r')
|
|
except Exception:
|
|
raise Exception("Automatic recognition of replicum failed, please enter the key word 'names'.")
|
|
|
|
if 'ens_name' in kwargs:
|
|
new_names.append(kwargs.get('ens_name') + '|' + myentry[idx:])
|
|
else:
|
|
new_names.append(myentry[:idx] + '|' + myentry[idx:])
|
|
idl = []
|
|
if not appended:
|
|
for i, item in enumerate(ls):
|
|
sub_ls = []
|
|
if "files" in kwargs:
|
|
sub_ls = kwargs.get("files")
|
|
sub_ls.sort(key=lambda x: int(re.findall(r'\d+', x)[-1]))
|
|
else:
|
|
for (dirpath, dirnames, filenames) in os.walk(path + '/' + item):
|
|
if compact:
|
|
sub_ls.extend(filenames)
|
|
else:
|
|
sub_ls.extend(dirnames)
|
|
break
|
|
if compact:
|
|
for exc in sub_ls:
|
|
if not fnmatch.fnmatch(exc, prefix + '*'):
|
|
sub_ls = list(set(sub_ls) - set([exc]))
|
|
sub_ls.sort(key=lambda x: int(re.findall(r'\d+', x)[-1]))
|
|
else:
|
|
for exc in sub_ls:
|
|
if not fnmatch.fnmatch(exc, 'cfg*'):
|
|
sub_ls = list(set(sub_ls) - set([exc]))
|
|
sub_ls.sort(key=lambda x: int(x[3:]))
|
|
rep_idl = []
|
|
no_cfg = len(sub_ls)
|
|
for cfg in sub_ls:
|
|
try:
|
|
if compact:
|
|
rep_idl.append(int(cfg.split(cfg_separator)[-1]))
|
|
else:
|
|
rep_idl.append(int(cfg[3:]))
|
|
except Exception:
|
|
raise Exception("Couldn't parse idl from directroy, problem with file " + cfg)
|
|
rep_idl.sort()
|
|
# maybe there is a better way to print the idls
|
|
print(item, ':', no_cfg, ' configurations')
|
|
idl.append(rep_idl)
|
|
# here we have found all the files we need to look into.
|
|
if i == 0:
|
|
# here, we want to find the place within the file,
|
|
# where the correlator we need is stored.
|
|
# to do so, the pattern needed is put together
|
|
# from the input values
|
|
if version == "0.0":
|
|
pattern = "# " + name + " : offset " + str(noffset) + ", wf " + str(wf)
|
|
# if b2b, a second wf is needed
|
|
if b2b:
|
|
pattern += ", wf_2 " + str(wf2)
|
|
qs = quarks.split(" ")
|
|
pattern += " : " + qs[0] + " - " + qs[1]
|
|
file = open(path + '/' + item + '/' + sub_ls[0] + '/' + name, "r")
|
|
for k, line in enumerate(file):
|
|
if read == 1 and not line.strip() and k > start + 1:
|
|
break
|
|
if read == 1 and k >= start:
|
|
T += 1
|
|
if pattern in line:
|
|
read = 1
|
|
start = k + 1
|
|
print(str(T) + " entries found.")
|
|
file.close()
|
|
else:
|
|
pattern = 'name ' + name + '\nquarks ' + quarks + '\noffset ' + str(noffset) + '\nwf ' + str(wf)
|
|
if b2b:
|
|
pattern += '\nwf_2 ' + str(wf2)
|
|
# and the file is parsed through to find the pattern
|
|
if compact:
|
|
file = open(path + '/' + item + '/' + sub_ls[0], "r")
|
|
else:
|
|
# for non-compactified versions of the files
|
|
file = open(path + '/' + item + '/' + sub_ls[0] + '/' + name, "r")
|
|
|
|
content = file.read()
|
|
match = re.search(pattern, content)
|
|
if match:
|
|
start_read = content.count('\n', 0, match.start()) + 5 + b2b
|
|
end_match = re.search(r'\n\s*\n', content[match.start():])
|
|
T = content[match.start():].count('\n', 0, end_match.start()) - 4 - b2b
|
|
assert T > 0
|
|
print(T, 'entries, starting to read in line', start_read)
|
|
file.close()
|
|
else:
|
|
file.close()
|
|
raise Exception('Correlator with pattern\n' + pattern + '\nnot found.')
|
|
|
|
# we found where the correlator
|
|
# that is to be read is in the files
|
|
# after preparing the datastructure
|
|
# the correlators get parsed into...
|
|
deltas = []
|
|
for j in range(T):
|
|
deltas.append([])
|
|
|
|
for t in range(T):
|
|
deltas[t].append(np.zeros(no_cfg))
|
|
if compact:
|
|
for cfg in range(no_cfg):
|
|
with open(path + '/' + item + '/' + sub_ls[cfg]) as fp:
|
|
lines = fp.readlines()
|
|
# check, if the correlator is in fact
|
|
# printed completely
|
|
if (start_read + T > len(lines)):
|
|
raise Exception("EOF before end of correlator data! Maybe " + path + '/' + item + '/' + sub_ls[cfg] + " is corrupted?")
|
|
# and start to read the correlator.
|
|
# the range here is chosen like this,
|
|
# since this allows for implementing
|
|
# a security check for every read correlator later...
|
|
for k in range(start_read - 6, start_read + T):
|
|
if k == start_read - 5 - b2b:
|
|
if lines[k].strip() != 'name ' + name:
|
|
raise Exception('Wrong format', sub_ls[cfg])
|
|
if (k >= start_read and k < start_read + T):
|
|
floats = list(map(float, lines[k].split()))
|
|
deltas[k - start_read][i][cfg] = floats[-2:][im]
|
|
else:
|
|
for cnfg, subitem in enumerate(sub_ls):
|
|
with open(path + '/' + item + '/' + subitem + '/' + name) as fp:
|
|
# since the non-compatified files
|
|
# are typically not so long,
|
|
# we can iterate over the whole file.
|
|
# here one can also implement the chekc from above.
|
|
for k, line in enumerate(fp):
|
|
if (k >= start_read and k < start_read + T):
|
|
floats = list(map(float, line.split()))
|
|
if version == "0.0":
|
|
deltas[k - start][i][cnfg] = floats[im - single]
|
|
else:
|
|
deltas[k - start_read][i][cnfg] = floats[1 + im - single]
|
|
|
|
else:
|
|
if "files" in kwargs:
|
|
ls = kwargs.get("files")
|
|
else:
|
|
for exc in ls:
|
|
if not fnmatch.fnmatch(exc, prefix + '*.' + name):
|
|
ls = list(set(ls) - set([exc]))
|
|
ls.sort(key=lambda x: int(re.findall(r'\d+', x)[-1]))
|
|
pattern = 'name ' + name + '\nquarks ' + quarks + '\noffset ' + str(noffset) + '\nwf ' + str(wf)
|
|
if b2b:
|
|
pattern += '\nwf_2 ' + str(wf2)
|
|
for rep, file in enumerate(ls):
|
|
rep_idl = []
|
|
with open(path + '/' + file, 'r') as fp:
|
|
content = fp.readlines()
|
|
data_starts = []
|
|
for linenumber, line in enumerate(content):
|
|
if "[run]" in line:
|
|
data_starts.append(linenumber)
|
|
if len(set([data_starts[i] - data_starts[i - 1] for i in range(1, len(data_starts))])) > 1:
|
|
raise Exception("Irregularities in file structure found, not all runs have the same output length")
|
|
chunk = content[:data_starts[1]]
|
|
for linenumber, line in enumerate(chunk):
|
|
if line.startswith("gauge_name"):
|
|
gauge_line = linenumber
|
|
elif line.startswith("[correlator]"):
|
|
corr_line = linenumber
|
|
found_pat = ""
|
|
for li in chunk[corr_line + 1: corr_line + 6 + b2b]:
|
|
found_pat += li
|
|
if re.search(pattern, found_pat):
|
|
start_read = corr_line + 7 + b2b
|
|
break
|
|
endline = corr_line + 6 + b2b
|
|
while not chunk[endline] == "\n":
|
|
endline += 1
|
|
T = endline - start_read
|
|
if rep == 0:
|
|
deltas = []
|
|
for t in range(T):
|
|
deltas.append([])
|
|
for t in range(T):
|
|
deltas[t].append(np.zeros(len(data_starts)))
|
|
# all other chunks should follow the same structure
|
|
for cnfg in range(len(data_starts)):
|
|
start = data_starts[cnfg]
|
|
stop = start + data_starts[1]
|
|
chunk = content[start:stop]
|
|
try:
|
|
rep_idl.append(int(chunk[gauge_line].split(cfg_separator)[-1]))
|
|
except Exception:
|
|
raise Exception("Couldn't parse idl from directory, problem with chunk around line ", gauge_line)
|
|
|
|
found_pat = ""
|
|
for li in chunk[corr_line + 1:corr_line + 6 + b2b]:
|
|
found_pat += li
|
|
if re.search(pattern, found_pat):
|
|
for t, line in enumerate(chunk[start_read:start_read + T]):
|
|
floats = list(map(float, line.split()))
|
|
deltas[t][rep][cnfg] = floats[im + 1 - single]
|
|
idl.append(rep_idl)
|
|
|
|
if "check_configs" in kwargs:
|
|
print("Checking for missing configs...")
|
|
che = kwargs.get("check_configs")
|
|
if not (len(che) == len(idl)):
|
|
raise Exception("check_configs has to be the same length as replica!")
|
|
for r in range(len(idl)):
|
|
print("checking " + new_names[r])
|
|
utils.check_idl(idl[r], che[r])
|
|
print("Done")
|
|
result = []
|
|
for t in range(T):
|
|
result.append(Obs(deltas[t], new_names, idl=idl))
|
|
return result
|