This is related to the previous commit

nsids are random identifiers, usually used to ease merging
between documents, and can trivially be used for fingerprinting.

Brolf authored 6 years ago and georg committed 6 years ago

Closes #96

This should make easier to get more metadata from
archive-based file formats.

intrigeri authored 6 years ago and jvoisin committed 6 years ago

This should closes  #90

We didn't take the whitelist into account while
removing dangling files from [Content_types].xml

- Replace some class members by instance members
- Don't thread the cleaning process anymore for now

MS Office XML rsid is a "unique identifier used to track the editing session
when the physical character representing this section mark was last formatted."

See the following links for details:
- https://msdn.microsoft.com/en-us/library/office/documentformat.openxml.wordprocessing.previoussectionproperties.rsidrpr.aspx
- https://blogs.msdn.microsoft.com/brian_jones/2006/12/11/whats-up-with-all-those-rsids/.

In XML, the order of the attributes shouldn't be meaningful,
however, MS Office sorts attributes for a given XML tag
differently than LibreOffice.

Closes #60

Note: this changeset also ensures that clean.cleaned.docx is removed
up after the pytest is over.

the end result will be the same -- an abort -- but the user will get
to see all the warnings for a particular file, instead of getting them
one at a time.

This makes it easy to get a list of all files that mat2 doesn't know
how to handle, without having to choose -u keep or -u omit.

previously, encountering an unknown member meant that any parser of
this type would abort.

now, the user can set parser.unknown_member_policy to either 'omit' or
'keep' if they don't want the current action of 'abort'

note that this causes pylint to complain about branching depth for
remove_all() because of the nuanced error-handling.  I've disabled
this check.