CIRCL
/
PyCIRCLean
mirror of https://github.com/CIRCL/PyCIRCLean
2
0
Fork 0
Code Issues Releases Wiki Activity

Fix description for ooxml files

pull/21/head
Dan Puttick 2017-08-04 16:31:52 -04:00
parent 17b453af20
commit e20035e325
1 changed files with 2 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
bin/filecheck.py
View File

@ -219,12 +219,10 @@ class File(FileBase):
for mt in Config.mimes_rtf: for mt in Config.mimes_rtf:
if mt in self.subtype: if mt in self.subtype:
self.add_description('Rich Text (rtf) file') self.add_description('Rich Text (rtf) file')
# TODO: need a way to convert it to plain text
self.force_ext('.txt') self.force_ext('.txt')
return return
for mt in Config.mimes_ooxml: for mt in Config.mimes_ooxml:
if mt in self.subtype: if mt in self.subtype:
self.add_description('OOXML (openoffice) file')
self._ooxml() self._ooxml()
return return
self.add_description('Plain text file') self.add_description('Plain text file')
@ -240,7 +238,6 @@ class File(FileBase):
def _executables(self): def _executables(self):
"""Process an executable file.""" """Process an executable file."""
# LOG: change the processing_type property to some other name or include in file_string
self.make_dangerous('Executable file') self.make_dangerous('Executable file')
def _winoffice(self): def _winoffice(self):
@ -275,6 +272,7 @@ class File(FileBase):
def _ooxml(self): def _ooxml(self):
"""Process an ooxml file.""" """Process an ooxml file."""
self.add_description('OOXML (openoffice) file')
try: try:
doc = officedissector.doc.Document(self.src_path) doc = officedissector.doc.Document(self.src_path)
except Exception: except Exception:
@ -292,6 +290,7 @@ class File(FileBase):
if len(doc.features.embedded_packages) > 0: if len(doc.features.embedded_packages) > 0:
self.make_dangerous('Ooxml file with embedded packages') self.make_dangerous('Ooxml file with embedded packages')
def _libreoffice(self): def _libreoffice(self):
"""Process a libreoffice file.""" """Process a libreoffice file."""
# As long as there is no way to do a sanity check on the files => dangerous # As long as there is no way to do a sanity check on the files => dangerous
@ -312,7 +311,6 @@ class File(FileBase):
"""Process a PDF file.""" """Process a PDF file."""
xmlDoc = PDFiD(self.src_path) xmlDoc = PDFiD(self.src_path)
oPDFiD = cPDFiD(xmlDoc, True) oPDFiD = cPDFiD(xmlDoc, True)
# TODO: are there other pdf characteristics which should be dangerous?
if oPDFiD.encrypt.count > 0: if oPDFiD.encrypt.count > 0:
self.make_dangerous('Encrypted pdf') self.make_dangerous('Encrypted pdf')
if oPDFiD.js.count > 0 or oPDFiD.javascript.count > 0: if oPDFiD.js.count > 0 or oPDFiD.javascript.count > 0:
@ -437,7 +435,6 @@ class File(FileBase):
using PIL.Image, saves it to the temporary directory, and copies it to using PIL.Image, saves it to the temporary directory, and copies it to
the destination. the destination.
""" """
# TODO: make sure this method works for png, gif, tiff
if self.has_metadata: if self.has_metadata:
self.extract_metadata() self.extract_metadata()
tempdir_path = self.make_tempdir() tempdir_path = self.make_tempdir()