def isToBeIgnored(self,entry:os.DirEntry):
if not entry.is_dir():
return False
if entry.name in self.ignoreDirList:
lgg.info(f" Ignoring {entry.name}",lgg.cR)
return True
return False
def buildExtDicts(self,sdir):
lgg.info(f" buildExtDicts dir:{sdir}",lgg.cP)
self.extDicts = {}
for entry in self.getFiles(sdir):
if entry.is_dir():
lgg.info(f" dir:{entry.path}",lgg.cB)
elif entry.is_file():
# lgg.info(f" file:{entry.path}",lgg.cC)
self.digestEntry(entry)
return
def getFiles(self,base_dir):
for entry in os.scandir(base_dir):
if entry.is_file():
yield entry
elif entry.is_dir():
lgg.info(f" Directory {entry.name}",lgg.cC)
if not self.isToBeIgnored(entry):
yield from self.getFiles(entry.path)
else:
print(f"Neither a file, nor a dir: {entry.path}")
def buildClsDicts(self):
lgg.info(f" buildClsDicts",lgg.cP)
self.clsDicts = {}
for extkey in self.extDicts.keys():
exd = self.extDicts[extkey]
clskey = self.getClass(extkey)
if not clskey in self.clsDicts:
self.clsDicts[clskey] = { "num":0, "bytes":0 }
cld = self.clsDicts[clskey]
cld["num"] += exd["num"]
cld["bytes"] += exd["bytes"]
def digestEntry(self, entry:os.DirEntry ):
_,ext = os.path.splitext(entry.name)
if not ext in self.extDicts:
self.extDicts[ext] = { "num":0, "bytes":0,"maxbytes":0,"maxname":"" }
exd = self.extDicts[ext]
exd["num"] += 1
esize = entry.stat().st_size
exd["bytes"] += esize
if esize>exd["maxbytes"]:
exd["maxbytes"] = esize
exd["maxname"] = entry.path
if esize>10e6:
self.bigFileList.append(entry)
emb = "%.3f" % round(esize/1e6,3)
lgg.info(f" big file: {emb} mb - {entry.path}")
def main(self):
sdir = self.args.sdir
lgg.info(f"FileClassing {sdir}",lgg.cY)
stime = timeit.time.time()
self.buildExtDicts(sdir)
self.dumpExtDicts()
self.buildClsDicts()
self.dumpClsDicts()
# (ovfiles,ovbytes) = copyFromTo(sdir,ddir,execute)
elap = timeit.time.time()-stime
#exword = "" if execute else "Would have "
#lgg.info(f"{exword} Overwritten files:{ovfiles}/{tfiles} overwritenbytes:{ovbytes} secs:{round(elap,3)} ",lgg.cY)
lgg.info(f"file class done - secs:{round(elap,3)} ",lgg.cY)
def copyFromTo(sdir, ddir, execute):
overwrittenfiles = 0
overwrittenbytes = 0
i = 0
for (fdname, fname) in listfiles:
sfname = f"{sdir}/{fdname}{fname}"
dfname = f"{ddir}/{fdname}{fname}"
dddir = f"{ddir}/{fdname}"
fclr = lgg.cC
ovbytes = 0
ddiristhere = os.path.exists(dddir)
if not ddiristhere:
if execute:
lgg.info(f"Creating directory {dddir}", lgg.cR)
os.makedirs(dddir)
else:
lgg.info(f"Would have created directory {dddir}", lgg.cR)
if os.path.exists(dfname):
fclr = lgg.cG
ftats = os.stat(dfname)
ovbytes = ftats.st_size
overwrittenbytes += ovbytes
overwrittenfiles += 1
execword = "execute" if execute else "fake"
lgg.info(
f"{i}: {execword} copy from {sfname} to {dfname} overwrittenbytes:{ovbytes}",
fclr)
if execute:
shutil.copyfile(sfname, dfname)
i += 1
return (overwrittenfiles, overwrittenbytes)
def dumpClsDicts(self):
lgg.info(f" dumpClsDicts",lgg.cP)
nfiles = 0
nbytes = 0
for clskey in self.clsDicts.keys():
cld = self.clsDicts[clskey]
nfilescls = cld["num"]
nbytescls = cld["bytes"]
nfiles += nfilescls
nbytes += nbytescls
mbytes = "%.3f" % round(nbytescls/1e6,3)
lgg.info(f" {clskey:>6} - num:{nfilescls:>4} tot-mb:{mbytes:>9}",lgg.cG)
mbytes = "%.3f" % round(nbytes / 1e6,3)
lgg.info(f"totals - files{nfiles} bytes:{nbytes} mb:{mbytes}",lgg.cG)
def dumpExtDicts(self):
lgg.info(f" dumpExtDicts",lgg.cP) cd
nfiles = 0
nbytes = 0
mxkeylen = self.getLongestExtKeyLength()
sortedExtDict = self.getSortedExtDict("bytes")
#for extkey in self.extDicts.keys():
for extkey in sortedExtDict:
exd = self.extDicts[extkey]
nfilesext = exd["num"]
nbytesext = exd["bytes"]
maxbytesext = exd["maxbytes"]
avgbytesext = int(maxbytesext/nfilesext)
cls = self.getClass(extkey)
nfiles += nfilesext
nbytes += nbytesext
mbytes = "%.3f" % round(nbytesext/1e6,3)
extkeypad = extkey.rjust(mxkeylen)
lgg.info(f" {extkeypad} {cls:>6} - num:{nfilesext:>4} size-max:{maxbytesext:>10} avg:{avgbytesext:>10} tot-mb:{mbytes:>9}",lgg.cB)
mbytes = "%.3f" % round(nbytes / 1e6,3)
lgg.info(f"totals - files{nfiles} bytes:{nbytes} mb:{mbytes}",lgg.cB)
ovbytes = ftats.st_size
overwrittenbytes += ovbytes
overwrittenfiles += 1
execword = "execute" if execute else "fake"
lgg.info(
f"{i}: {execword} copy from {sfname} to {dfname} overwrittenbytes:{ovbytes}",
fclr)
if execute:
shutil.copyfile(sfname, dfname)
i += 1
return (overwrittenfiles, overwrittenbytes)
sdir = args.sdir
ddir = args.ddir
execute = args.exec
lgg.info(
f"Copying {len(listfiles)} files from {sdir} to {ddir} execute:{execute}",
lgg.cY)
start = timeit.timeit()
(ovfiles, ovbytes) = copyFromTo(sdir, ddir, execute)
elap = timeit.timeit() - start
tfiles = len(listfiles)
exword = "" if execute else "Would have "
lgg.info(
f"{exword} Overwritten files:{ovfiles}/{tfiles} overwritenbytes:{ovbytes} secs:{round(elap,3)} ",
lgg.cY)