def getTextualNode(self, subreference=None, simple=False):
""" Finds a passage in the current text
:param subreference: Identifier of the subreference / passages
:type subreference: Union[list, CtsReference]
:param simple: If set to true, retrieves nodes up to the given one, cleaning non required siblings.
:type simple: boolean
:rtype: CapitainsCtsPassage, ContextPassage
:returns: Asked passage
"""
if subreference is None:
return self._getSimplePassage()
if not isinstance(subreference, CtsReference):
if isinstance(subreference, str):
subreference = CtsReference(subreference)
elif isinstance(subreference, list):
subreference = CtsReference(".".join(subreference))
if len(subreference.start) > self.citation.root.depth:
raise CitationDepthError("URN is deeper than citation scheme")
if simple is True:
return self._getSimplePassage(subreference)
if not subreference.is_range():
start = end = subreference.start.list
else:
start, end = subreference.start.list, subreference.end.list
citation_start = self.citation.root[len(start) - 1]
citation_end = self.citation.root[len(end) - 1]
start, end = citation_start.fill(passage=start), citation_end.fill(
passage=end)
start, end = normalizeXpath(start.split("/")[2:]), normalizeXpath(
end.split("/")[2:])
xml = self.textObject.xml
if isinstance(xml, etree._Element):
root = copyNode(xml)
else:
root = copyNode(xml.getroot())
root = passageLoop(xml, root, start, end)
if self.urn:
urn = URN("{}:{}".format(self.urn, subreference))
else:
urn = None
return CapitainsCtsPassage(urn=urn,
resource=root,
text=self,
citation=citation_start,
reference=subreference)
def getValidReff(self,
level: int = 1,
reference: CtsReference = None,
_debug: bool = False) -> CtsReferenceSet:
""" Retrieve valid passages directly
:param level: Depth required. If not set, should retrieve first encountered level (1 based)
:type level: int
:param reference: CapitainsCtsPassage Reference
:type reference: CtsReference
:param _debug: Check on passages duplicates
:type _debug: bool
:returns: List of levels
.. note:: GetValidReff works for now as a loop using CapitainsCtsPassage, subinstances of CtsTextMetadata, to retrieve the valid \
informations. Maybe something is more powerfull ?
"""
depth = 0
xml = self.textObject.xml
if reference:
if isinstance(reference, CtsReference):
if not reference.is_range():
passages = [reference.start.list]
depth = len(passages[0])
if level == 0:
level = None
if _debug:
warnings.warn(
"Using level=0 with a Non-range Reference is invalid. Autocorrected to 1"
)
else:
xml = self.getTextualNode(subreference=reference)
common = []
for index, part in enumerate(reference.start.list):
if index <= reference.end.depth:
if part == reference.end.list[index]:
common.append(part)
else:
break
else:
break
passages = [common]
depth = len(common)
if level is None:
level = reference.start.depth + depth
elif level == 1:
level = reference.start.depth + 1
elif level == 0:
level = reference.start.depth
else:
raise TypeError()
else:
passages = [[]]
if level is None:
level = 1
if level <= len(passages[0]) and reference is not None:
level = len(passages[0]) + 1
if level > len(self.citation.root):
raise CitationDepthError("The required level is too deep")
nodes = [None] * (level - depth)
citations = [citation for citation in self.citation.root]
while len(nodes) >= 1:
passages = [
refs + [
node.get(
current_citation.attribute.replace(
"xml:", "{http://www.w3.org/XML/1998/namespace}"))
] for xpath_result, refs, current_citation in [(
xml.xpath(citations[len(filling) - 1].fill(filling),
namespaces=XPATH_NAMESPACES), refs,
citations[len(filling) -
1]) for filling, refs in [(refs + [None], refs)
for refs in passages]]
for node in xpath_result
]
nodes.pop(0)
if len(passages) == 0:
msg = "Unknown reference {}".format(reference)
raise KeyError(msg)
passages = [".".join(passage) for passage in passages]
if _debug:
duplicates = set()
seen = set()
for n in passages:
if n in seen:
duplicates.add(n)
else:
seen.add(n)
if len(duplicates) > 0:
message = ", ".join(duplicates)
warnings.warn(message, DuplicateReference)
del duplicates
empties = [n for n in passages if n.rstrip('.') != n or n == '']
if len(empties) > 0:
message = '{} empty reference(s) at citation level {}'.format(
len(empties), level)
warnings.warn(message, EmptyReference)
references = CtsReferenceSet([CtsReference(reff) for reff in passages],
citation=self.citation.root[level - 1],
level=level)
return references