calling .save() or is_present() methods of Resource raise exception

[mromanello]

For the full description see issue on upstream repo.

Here you can see an example of how the code fails (this refers to the Python3 branch of hucitlib and uses your py3-compliant version of surf; however, I reproduce the same error with surf==1.2.0 and hucitlib master branch. With surf==1.1.9 all works fine).

PS: the fact that the surf repo does not have releases makes it difficult for me to do some debugging with a locally modified version of 1.1.9 and 1.2.0, and compare the two behaviours.

[mromanello]

Just a brief update: I've found the problematic line and how to make it work, but I'm not 100% sure about the logic.

The problematic line is this if/else block:

surfrdf/surf/query/__init__.py

Lines 411 to 435 in 79ffca6

	if isinstance(statement, (list, tuple)):
	try:
	s, p, o = statement
	except:
	raise ValueError('''Statement of type [list, tuple] does not
	have all the (s,p,o) members (the length of the
	supplied arguemnt must be at least 3)''')
	if isinstance(s, (URIRef, BNode)) or (isinstance(s, basestring) and s.startswith('?')):
	pass
	else:
	raise ValueError('The subject is not a valid variable type')
	if isinstance(p, URIRef) or (isinstance(p, basestring) and p.startswith('?')):
	pass
	else:
	raise ValueError('The predicate is not a valid variable type')
	if isinstance(o, (URIRef, BNode, Literal)) or (isinstance(o, basestring) and o.startswith('?')):
	pass
	else:
	raise ValueError(u'The object is not a valid variable type: {0:s}'.format(o))
	return True
	else:
	raise ValueError('Statement type not in {0:s}'.format)

Take a statement like

[('?s', '?p', '?o'), u'(?s = <http://purl.org/hucit/kb/authors/927/cts_urn>)']

In the version of surf that still works (1.1.9) the type is checked using

if type(statement) in [list, tuple]

which for the statement above returns False.

In version 1.2.0 the line has been changed into

isinstance(statement, (list, tuple))

which now returns True for the same statement. If I change that line to how it previously was, all works fine, but I do not understand why this difference between isinstance and type.

PS: the type of statement is NamedGroup, which is a subclass of Group, which is a subclass of list. Perhaps what's happening is that using isinstance here has the side effect of capturing this inheritance between classes (which is AFAIU unintentional)?

[PonteIneptique]

The big difference between the two is that isinstance takes into account inheritance. Might have been an oversight from the developer or myself. Le mer. 13 févr. 2019 à 10:12 PM, Matteo Romanello <[email protected]> a écrit :

…

Just a brief update: I've found the problematic line and how to make it work, but I'm not 100% sure about the logic. The problematic line is this if/else block: https://github.com/PonteIneptique/surfrdf/blob/79ffca6bd71cf5b2558aa08373cf71572fa1e64a/surf/query/__init__.py#L411-L435 Take a statement like [('?s', '?p', '?o'), u'(?s = <http://purl.org/hucit/kb/authors/927/cts_urn>)'] In the version of surf that still works (1.1.9) the type is checked using if type(statement) in [list, tuple] which for the statement above returns False. In version 1.2.0 the line has been changed into isinstance(statement, (list, tuple)) which now returns True for the same statement. If I change that line to how it previously was, all works fine, but I do not understand why this difference between isinstance and type. PS: the type of statement is NamedGroup, which is a subclass of Group, which is a subclass of list. Perhaps what's happening is that using isinstance here has the side effect of capturing this inheritance between classes (which is AFAIU unintentional)? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB1yZt1xRhQBxpJ_EqpnymdAfX0svyjPks5vNH_GgaJpZM4a538G> .