I have been following an example for using the XPath not ancestor selector in the form of .//x[not(ancestor::w:tbl)] but it is not behaving as I expect.
I am parsing a Word DOCX file that has a table in it. I am using the python lxml library to parse it as XML. I want to get paragraph items that do not have a table element anywhere in their ancestor tree.
I type the following in console:
selector = './/w:p[not(ancestor::w:drawing)][not(ancestor::w:tbl)][not(ancestor::v:textbox)][not(ancestor::wps:wsp)][not(ancestor::mc:Fallback)]
nsDict = {k:v for k,v in doc.nsmap.items() if k}
paragraphs = doc.xpath(selector,namespaces=nsDict)
for p in paragraphs:
print(bool(p.xpath(".//ancestor::w:tbl",namespaces=nsDict)))
>>>>False
>>>>False
>>>>False
>>>>False
>>>>False
>>>>False
>>>>True
>>>>False
The expected behavior is that the paragraph xpath selector is mutually exclusive with the parent doc element-level xpath selector. The paragraph boolean check should always be False.
How can I amend my initial selector so that no elements are picked up that have w:tbl as ancestors?
Your initial XPath is fine; it is your testing XPath that is faulty.
Your testing XPath,
.//ancestor::w:tbl
does not select w:tbl ancestors at the current node; it selects w:tbl ancestors of any of the descendents of the current node.
The case where a paragraph has no ancestors that are tables but does have a descendent table that contains a paragraph would yield True for your test, for example.
Change it instead to
ancestor::w:tbl
to select the w:tbl ancestors of the current node.