MAT2 is now cleaning revisions from odt files!

.gitlab-ci.yml

@@ -6,10 +6,10 @@ stages:
 
 bandit:
   stage: linting
-  script:
+  script:  # TODO: remove B405 and B314
   - apt-get -qqy update
   - apt-get -qqy install --no-install-recommends python3-bandit
-  - bandit -r ./libmat2 --format txt --skip B101,B404,B603
+  - bandit -r ./libmat2 --format txt --skip B101,B404,B603,B405,B314
 
 pyflakes:
   stage: linting

libmat2/office.py

@@ -4,8 +4,10 @@ import shutil
 import tempfile
 import datetime
 import zipfile
+import xml.etree.ElementTree as ET
 from typing import Dict, Set, Pattern
 
+
 from . import abstract, parser_factory
 
 # Make pyflakes happy
@@ -13,7 +15,12 @@ assert Set
 assert Pattern
 
 class ArchiveBasedAbstractParser(abstract.AbstractParser):
+    # Those are the files that have a format that _isn't_
+    # supported by MAT2, but that we want to keep anyway.
     files_to_keep = set()  # type: Set[str] 
+
+    # Those are the files that we _do not_ want to keep,
+    # no matter if they are supported or not.
     files_to_omit = set() # type: Set[Pattern] 
 
     def __init__(self, filename):
@@ -23,6 +30,11 @@ class ArchiveBasedAbstractParser(abstract.AbstractParser):
         except zipfile.BadZipFile:
             raise ValueError
 
+    def _specific_cleanup(self, full_path:str) -> bool:
+        """ This method can be used to apply specific treatment
+        to files present in the archive."""
+        return True
+
     def _clean_zipinfo(self, zipinfo: zipfile.ZipInfo) -> zipfile.ZipInfo:
         zipinfo.create_system = 3  # Linux
         zipinfo.comment = b''
@@ -56,15 +68,19 @@ class ArchiveBasedAbstractParser(abstract.AbstractParser):
             for item in zin.infolist():
                 if item.filename[-1] == '/':  # `is_dir` is added in Python3.6
                     continue  # don't keep empty folders
-                elif item.filename in self.files_to_keep:
-                    item = self._clean_zipinfo(item)
-                    zout.writestr(item, zin.read(item))
-                    continue
-                elif any(map(lambda r: r.search(item.filename), self.files_to_omit)):
-                    continue
 
                 zin.extract(member=item, path=temp_folder)
                 full_path = os.path.join(temp_folder, item.filename)
+
+                self._specific_cleanup(full_path)
+
+                if item.filename in self.files_to_keep:
+                    # those files aren't supported, but we want to add them anyway
+                    pass
+                elif any(map(lambda r: r.search(item.filename), self.files_to_omit)):
+                    continue
+                else:
+                    # supported files that we want to clean then add
                     tmp_parser, mtype = parser_factory.get_parser(full_path)  # type: ignore
                     if not tmp_parser:
                         shutil.rmtree(temp_folder)
@@ -72,10 +88,11 @@ class ArchiveBasedAbstractParser(abstract.AbstractParser):
                         print("%s's format (%s) isn't supported" % (item.filename, mtype))
                         return False
                     tmp_parser.remove_all()
+                    os.rename(tmp_parser.output_filename, full_path)
 
                 zinfo = zipfile.ZipInfo(item.filename)  # type: ignore
                 clean_zinfo = self._clean_zipinfo(zinfo)
-                with open(tmp_parser.output_filename, 'rb') as f:
+                with open(full_path, 'rb') as f:
                     zout.writestr(clean_zinfo, f.read())
 
         shutil.rmtree(temp_folder)
@@ -149,6 +166,37 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
             '^Thumbnails/',
     }))
 
+
+    def __remove_revisions(self, full_path:str) -> bool:
+        def parse_map(f):  # etree support for ns is a bit rough
+            ns_map = dict()
+            for event, (k, v) in ET.iterparse(f, ("start-ns", )):
+                if event == "start-ns":
+                    ns_map[k] = v
+            return ns_map
+
+        ns = parse_map(full_path)
+        if 'office' not in ns.keys():  # no revisions in the current file
+            return True
+
+        # Register the namespaces
+        for k,v in ns.items():
+            ET.register_namespace(k, v)
+
+        tree = ET.parse(full_path)
+        for text in tree.getroot().iterfind('.//office:text', ns):
+            for changes in text.iterfind('.//text:tracked-changes', ns):
+                text.remove(changes)
+
+        tree.write(full_path, xml_declaration = True)
+
+        return True
+
+    def _specific_cleanup(self, full_path:str) -> bool:
+        if os.path.basename(full_path) == 'content.xml':
+            return self.__remove_revisions(full_path)
+        return True
+
     def get_meta(self) -> Dict[str, str]:
         """
         Yes, I know that parsing xml with regexp ain't pretty,

tests/test_libmat2.py

@@ -122,6 +122,27 @@ class TestRemovingThumbnails(unittest.TestCase):
 
         os.remove('./tests/data/clean.cleaned.odt')
 
+
+class TestRevisionsCleaning(unittest.TestCase):
+    def test_libreoffice(self):
+        with zipfile.ZipFile('./tests/data/revision.odt') as zipin:
+            c = zipin.open('content.xml')
+            r = c.read()
+            self.assertIn(b'tracked-changes', r)
+
+        shutil.copy('./tests/data/revision.odt', './tests/data/clean.odt')
+        p = office.LibreOfficeParser('./tests/data/clean.odt')
+        self.assertTrue(p.remove_all())
+
+        with zipfile.ZipFile('./tests/data/clean.cleaned.odt') as zipin:
+            c = zipin.open('content.xml')
+            r = c.read()
+            self.assertNotIn(b'tracked-changes', r)
+
+        os.remove('./tests/data/clean.odt')
+        os.remove('./tests/data/clean.cleaned.odt')
+
+
 class TestDeepCleaning(unittest.TestCase):
     def __check_deep_meta(self, p):
         tempdir = tempfile.mkdtemp()