def get(self, ifile):
self.bsdata = ifile
if isinstance(ifile, FileInfo):
self.bsdata = BeautifulSoup(open(ifile.path).read(),
features="lxml")
elif isinstance(ifile, (PosixPath, Path)):
self.bsdata = BeautifulSoup(open(ifile).read(), features="lxml")
elif isinstance(ifile, str):
finfo = FileInfo(ifile)
if finfo.isFile():
if finfo.ext == ".p":
self.bsdata = BeautifulSoup(PickleIO().get(ifile),
features="lxml")
elif finfo.ext in [".html", ".htm"]:
self.bsdata = BeautifulSoup(open(ifile).read(),
features="lxml")
else:
raise ValueError(
"File extention [{0}] is not recognized".format(
finfo.ext))
else:
self.bsdata = BeautifulSoup(StringIO(ifile), features="lxml")
elif isinstance(ifile, bytes):
self.bsdata = BeautifulSoup(ifile, features="lxml")
else:
raise ValueError(
"Can't create BeautifulSoup() from [{0}] input".format(
type(ifile)))
return self.bsdata
def get(self, ifile, **kwargs):
finfo = FileInfo(ifile)
extIO = self.getIO(finfo)
if extIO is None:
raise ValueError("Unknown file extension: [{0}] [{1}]".format(
ifile, finfo.ext))
if finfo.exists():
retval = extIO.get(finfo.str, **kwargs)
if self.debug:
tsGet.stop()
return retval
else:
print("[{0}] does not exist. Returning None".format(ifile))
return None
def __init__(self, ifile, now=None, lastModified=None):
self.lastModified = lastModified
self.err = False
self.now = Timestamp.today().round('s') if now is None else now
###########################################################################
# Find Mod Times
###########################################################################
#modTime = datetime.fromtimestamp(getmtime(ifile))
modTime = FileInfo(ifile).time()
modTime = Timestamp(modTime) if modTime is not None else None
###########################################################################
# Compute Time Deltas
###########################################################################
if modTime is None:
deltaLastMod = None
deltaNow = None
else:
deltaLastMod = self.lastModified - modTime if self.lastModified is not None else None
deltaNow = self.now - modTime
###########################################################################
# Set Variables
###########################################################################
self.modTime = modTime
self.deltaLastMod = Timedelta(deltaLastMod).round(
's') if deltaLastMod is not None else None
self.deltaNow = Timedelta(deltaNow).round(
's') if deltaNow is not None else None
def getFilesByRecency(self, expr, debug=False):
self.setFileTimeDelta(expr)
self.setFileModTimes()
if debug:
self.ftd.show()
N = len(self.fileModTimes)
for i, (ifile, fmtd) in enumerate(self.fileModTimes.items()):
print("{0}/{1}\t{2: <60}{3} {4} {5}".format(
i, N,
FileInfo(ifile).basename, fmtd.modTime, fmtd.getDeltaNow(),
self.ftd.isTrue(fmtd.getDeltaNow())))
retval = [
ifile for ifile, fmtd in self.fileModTimes.items()
if self.ftd.isTrue(fmtd.getDeltaNow())
]
return retval
def save(self, ifile, idata):
finfo = FileInfo(ifile)
extIO = self.getIO(finfo)
extIO.save(finfo.str, idata)
def setLastModTimeFromFile(self, ifile):
self.lastModified = FileInfo(ifile).time()
self.lastModified = Timestamp(
self.lastModified) if self.lastModified is not None else None