def refreshMetadata(self):
self.allMetadata = []
self.matObject = mat.create_class_file(str(self.filePath), str(self.filePath), add2archive=True,
low_pdf_quality=True)
logging.debug("Refreshing metadata..")
metaDict = self.matObject.get_meta()
if metaDict is not None:
for key in metaDict.keys():
logging.debug("Found metadata header: " + key)
self.allMetadata.append([key, metaDict[key]])
if len(self.allMetadata) == 0:
self.hasMetadata = False
logging.debug("Done refreshing metadata..")
self.checkState()
return self.allMetadata
def initAllMetadata(self):
"""
Gets all metadata of the file using the MAT
library.
"""
if self.isFile:
self.allMetadata = []
self.matObject = mat.create_class_file(str(self.filePath), str(self.filePath), add2archive=True,
low_pdf_quality=True)
classType = self.matObject.__class__.__name__
self.type = classType.replace("Stripper", "")
if self.matObject is None:
logging.debug(self.filePath + " is unsupported!")
self.supported = False
self.checkState()
return self.allMetadata
metaDict = self.matObject.get_meta()
if metaDict is None or len(metaDict.keys()) == 0:
self.hasMetadata = False
else:
for key in metaDict.keys():
logging.debug("Found metadata header: " + key)
self.allMetadata.append([key, metaDict[key]])
else:
self.hasMetadata = False
logging.debug("Checking state")
self.checkState()
if self.type in ["JpegStripper", "PngStripper"]:
self.detectFaces()
logging.debug("Finished Extracting metadata")
return self.allMetadata
def metadata_handler(f):
return mat.create_class_file(f, False, add2archive=True)
def cleanPrintedPdf(self, outFile):
mobj = mat.create_class_file(str(outFile), str(outFile), add2archive=True,
low_pdf_quality=True)
mobj.remove_all()
def test_remove_empty(self):
'''Test removal with clean files'''
for clean, _ in self.file_list:
subprocess.call(['../mat', clean])
current_file = mat.create_class_file(clean, False, add2archive=True, low_pdf_quality=True)
self.assertTrue(current_file.is_clean())
def test_remove(self):
'''make sure that the cli remove all compromizing meta'''
for _, dirty in self.file_list:
subprocess.call(['../mat', dirty])
current_file = mat.create_class_file(dirty, False, add2archive=True, low_pdf_quality=True)
self.assertTrue(current_file.is_clean())
