def walkAndCalcReps(root_dir):
"""Clasify files in a given directory to prortrait or not.
Traverses all sub directories of the given root
Saves to db if portrait
"""
global processedFname
global repFname, failFname
processedFname = os.path.join(dataPath,processedFname)
repFname = os.path.join(dataPath,repFname)
failFname = os.path.join(dataPath,failFname)
processedList = []
if os.path.isfile(processedFname):
processedList = pickle.load(open(processedFname,'r'))
repDict = {}
if os.path.isfile(repFname):
repDict = pickle.load(open(repFname,'r'))
failDict = {}
if os.path.isfile(failFname):
failDict = pickle.load(open(failFname,'r'))
# print repFname
# print
dirs = [s for s in os.listdir(root_dir) if s.lower().startswith('ie')]
# print '********'
# print dirs
# print '********'
nDirs = len(dirs)
iDir = 0
ts = lt.resetTimer(nDirs,'Going over NLI dirs!')
for root, dirs, files in os.walk(root_dir, topdown=True):
# print "***" + root + "***"
# print files
ent = root.split('/')[-1]
if not ent.lower().startswith(r'ie'):
continue
print "-----" + ent + "-----" #+ str(ent.lower().startswith(r'ie'))
for name in files:
key = os.path.join(ent, name)
# print key
if key in processedList:
continue
# print "=====calculating rep====="
fname = os.path.join(root, name)
if not fname.endswith(('.jpg', '.png', '.tif')):
continue
rep, err = getImgRep(fname)
if rep is not None:
repDict[key]=rep
else:
failDict[key]=err
processedList.append(key)
pickle.dump(repDict,open(repFname,'w'),pickle.HIGHEST_PROTOCOL)
pickle.dump(failDict,open(failFname,'w'))
pickle.dump(processedList,open(processedFname,'w'))
lt.sampleTimer(iDir,ts)
iDir = iDir + 1
def walkAndCalcReps(root_dir):
"""Clasify files in a given directory to prortrait or not.
Traverses all sub directories of the given root
Saves to db if portrait
"""
processedFname = os.path.join(root_dir,'processed.pkl')
repFname = os.path.join(root_dir,'repsDBase.pkl')
repFnameOldFmt = os.path.join(root_dir,'repsDBaseOldFmt.pkl')
failFname = os.path.join(root_dir,'failDBase.pkl')
repFnameJson = os.path.join(root_dir,'repsDBase.json')
startTime = None
# processedFname = os.path.join(root_dir,processedFname)
# repFname = os.path.join(root_dir,repFname)
# failFname = os.path.join(root_dir,failFname)
processedList = []
if os.path.isfile(processedFname):
processedList = pickle.load(open(processedFname,'r'))
repDict = {}
if os.path.isfile(repFname):
repDict = pickle.load(open(repFname,'r'))
failDict = {}
if os.path.isfile(failFname):
failDict = pickle.load(open(failFname,'r'))
# print repFname
# print
# dirs = [s for s in os.listdir(root_dir) if s.lower().startswith('ie')]
# print '********'
# print dirs
# print '********'
# nDirs = len(dirs)
# iDir = 0
# first walk - to get the number of files
nFiles = 0
for root, dirs, files in os.walk(root_dir, topdown=True):
nFiles += len([f for f in files if f.lower().endswith(('.jpg', '.png', '.tif'))])
ts = lt.resetTimer(nFiles,'Analyzing images!', percentile=1.0)
iFile = 0
for root, dirs, files in os.walk(root_dir, topdown=True):
# print "***" + root + "***"
# print files
ent = root.split(root_dir)[-1]
if ent.startswith(r'/'):
ent = ent[1:]
print "-----" + ent + "-----" #+ str(ent.lower().startswith(r'ie'))
for name in files:
key = os.path.join(ent, name)
# print key
# if key in processedList:
#
# continue
# print "=====calculating rep====="
fname = os.path.join(root, name)
if not fname.endswith(('.jpg', '.png', '.tif')):
continue
rep, err = getImgRep(fname)
iFile += 1
lt.sampleTimer(iFile ,ts)
if rep is not None:
repDict[key]=rep
else:
failDict[key]=err
processedList.append(key)
pickle.dump(repDict,open(repFname,'w'),pickle.HIGHEST_PROTOCOL)
pickle.dump(repDict,open(repFnameOldFmt, 'w'))
pickle.dump(failDict,open(failFname,'w'))
pickle.dump(processedList,open(processedFname,'w'))
repDictJson = {k:list(v) for k,v in repDict.iteritems()}
f = open(repFnameJson,'w')
f.write(json.dumps(repDictJson))
f.close()