def CombineInputVectors(inputVectorFilenames, outputVectorFilename):
outFile = open(outputVectorFilename, "w")
for inFilename in inputVectorFilenames:
inFile = open(inFilename, "r")
outFile.write(inFile.read())
outFile.close()
from ConfigurationParser import getTVCs
TVC = getTVCs(inputVectorFilenames)
newTVC = sum(TVC.values())
inHdrFile = open(inputVectorFilenames[0] + ".hdr", "r")
import os
outHdrFilename = os.path.abspath(outputVectorFilename) + ".hdr"
outHdrFile = open(outHdrFilename, "w")
outHdrFile.write(inHdrFile.readline())
outHdrFile.write(inHdrFile.readline())
outHdrFile.write("TVC {value}".format(value=newTVC))
outHdrFile.close()
outFileBase = outputVectorFilename
outShuffleFileBase = outFileBase + "_ShuffledANN"
shuffleVector = [
"ShuffleVectorsModule "
+ " --outputVectorFileBaseName "
+ outShuffleFileBase
+ " --inputVectorFileBaseName "
+ outFileBase
+ " --resampleProportion 1 "
]
import sys
import subprocess
try:
subprocess.call(shuffleVector, shell=True)
except:
print(
(
"""ERROR
fail to run {str}""".format(
str=shuffleVector
)
)
)
sys.exit()
return os.path.abspath(outShuffleFileBase), outShuffleFileBase + ".hdr"
def BalanceInputVectors(inputVectorFilenames):
## read header file
from ConfigurationParser import getTVCs
TVC = getTVCs(inputVectorFilenames)
print(TVC)
print(TVC)
print(TVC)
print(TVC)
print(TVC)
import operator
maxFile = max(iter(list(TVC.items())), key=operator.itemgetter(1))[0]
# maxFile = max(TVC)
print(maxFile)
print(maxFile)
print(maxFile)
print(maxFile)
maxTVC = TVC[maxFile]
# print( "{file} = {tvc}".format( file=maxFile, tvc=TVC[maxFile]))
outputVectorFilenames = {}
outputVectorHdrFilenames = {}
for inFile in inputVectorFilenames:
outputVectorFilenames[inFile] = inFile + "_upsampled.txtANN"
outputVectorHdrFilenames[inFile] = inFile + "_upsampled.txtANN.hdr"
## upsample all other files
import subprocess
import os
for inputVectorFile in inputVectorFilenames:
upsampleCMD = [
"ShuffleVectorsModule " + " --outputVectorFileBaseName " +
outputVectorFilenames[inputVectorFile] +
" --inputVectorFileBaseName " + inputVectorFile +
" --resampleProportion " +
str(old_div(float(maxTVC), float(TVC[inputVectorFile])))
]
print(("HACK: UPSAMPPLING: {0}".format(upsampleCMD)))
subprocess.call(upsampleCMD, shell=True)
outputVectorFilenames[inFile] = os.path.abspath(
outputVectorFilenames[inFile])
## return list of upsampled file names
return list(outputVectorFilenames.values()), list(
outputVectorHdrFilenames.values())
def BalanceInputVectors(inputVectorFilenames):
## read header file
from ConfigurationParser import getTVCs
TVC = getTVCs(inputVectorFilenames)
print(TVC)
print(TVC)
print(TVC)
print(TVC)
print(TVC)
import operator
maxFile = max(iter(list(TVC.items())), key=operator.itemgetter(1))[0]
# maxFile = max(TVC)
print(maxFile)
print(maxFile)
print(maxFile)
print(maxFile)
maxTVC = TVC[maxFile]
# print( "{file} = {tvc}".format( file=maxFile, tvc=TVC[maxFile]))
outputVectorFilenames = {}
outputVectorHdrFilenames = {}
for inFile in inputVectorFilenames:
outputVectorFilenames[inFile] = inFile + "_upsampled.txtANN"
outputVectorHdrFilenames[inFile] = inFile + "_upsampled.txtANN.hdr"
## upsample all other files
import subprocess
import os
for inputVectorFile in inputVectorFilenames:
upsampleCMD = [
"ShuffleVectorsModule "
+ " --outputVectorFileBaseName "
+ outputVectorFilenames[inputVectorFile]
+ " --inputVectorFileBaseName "
+ inputVectorFile
+ " --resampleProportion "
+ str(old_div(float(maxTVC), float(TVC[inputVectorFile])))
]
print(("HACK: UPSAMPPLING: {0}".format(upsampleCMD)))
subprocess.call(upsampleCMD, shell=True)
outputVectorFilenames[inFile] = os.path.abspath(outputVectorFilenames[inFile])
## return list of upsampled file names
return list(outputVectorFilenames.values()), list(outputVectorHdrFilenames.values())
def CombineInputVectors(inputVectorFilenames, outputVectorFilename):
outFile = open(outputVectorFilename, "w")
for inFilename in inputVectorFilenames:
inFile = open(inFilename, "r")
outFile.write(inFile.read())
outFile.close()
from ConfigurationParser import getTVCs
TVC = getTVCs(inputVectorFilenames)
newTVC = sum(TVC.values())
inHdrFile = open(inputVectorFilenames[0] + ".hdr", "r")
import os
outHdrFilename = os.path.abspath(outputVectorFilename) + ".hdr"
outHdrFile = open(outHdrFilename, "w")
outHdrFile.write(inHdrFile.readline())
outHdrFile.write(inHdrFile.readline())
outHdrFile.write("TVC {value}".format(value=newTVC))
outHdrFile.close()
outFileBase = outputVectorFilename
outShuffleFileBase = outFileBase + "_ShuffledANN"
shuffleVector = [
"ShuffleVectorsModule " + " --outputVectorFileBaseName " +
outShuffleFileBase + " --inputVectorFileBaseName " + outFileBase +
" --resampleProportion 1 "
]
import sys
import subprocess
try:
subprocess.call(shuffleVector, shell=True)
except:
print(("""ERROR
fail to run {str}""".format(str=shuffleVector)))
sys.exit()
return os.path.abspath(outShuffleFileBase), outShuffleFileBase + ".hdr"