DAG walking isn’t slow if your algorithm is well-written, and it’s really
the only solution for pairing output paths with sets of input dependencies.
I use it in conjunction with some group flattening logic to gather
dependency tracking information from scripts.
Here's a basic generator for depth-first traversal:
def iterDependencies(node):
mask = nuke.INPUTS | nuke.HIDDEN_INPUTS
stack = node.dependencies(mask)
seen = set()
while stack:
node = stack.pop()
if node in seen:
continue
yield node
seen.add(node)
stack.extend(node.dependencies(mask))
Implementing a solution outside of Nuke is possible as well, but because
Nuke writes knob values sparsely, you have to provide one important piece of
information yourself: The default number of inputs a given node class has.
Once you have that, it’s relatively simple to parse a Nuke script in the
same stack-based manner Nuke does and rebuild a DAG in whatever format you’d
like. Honestly, though, it would be much simpler to just use the nuke module
directly as Ivan suggests.
-Nathan
From: Jep Hill
Sent: Sunday, January 26, 2014 12:25 PM
To: [email protected]
Subject: [Nuke-python] [nuke -t] oddness when opening scripts
witherrorsininteractive mode
Thanks for checking me on this guys. Ivan, I am getting the same print out
now... the runtime exception does indeed work. I realized I hadn't tested it
properly -- DERP.
the "print nuke.allNodes()" was just a quick test call I threw in there --
in my code, I was actually running a block that included
"nuke.selectConnectedNodes()" -- that spits out an unbelievable amount of
junk when run on a script with "issues" -- any one know how to suppress all
that garbage when run via shell? We should be able to run that in "quiet"
mode.
Ivan, can you confirm you're getting the same behavior with
nuke.selectConnectedNodes when running your "gizmos missing" script through
something like:
##### process2.py ########## import nuke import os def main(): nk =
os.path.abspath(sys.argv[1]) try: nuke.scriptOpen(nk) except RuntimeError:
print "A RuntimeError occurred: Ignoring..." pass # Moving on...
activeWrites = [ x['selected'].setValue(True) for x in
nuke.allNodes('Write') if not x['disable'].value() ] if activeWrites:
nuke.selectConnectedNodes() files = list(set([ x['file'].value() for x in
nuke.selectedNodes('Read') ])) print "\n%s\n" % '\n'.join(files)if __name__
== '__main__': main() ######################
nuke -t process2.py missing_gizmos.nk
It may be that this behavior is just inherent with the
nuke.selectConnectedNdoes method -- the larger issue I'm looking to solve
(lot's of posts on this issue -- not sure what the latest conventional
wisdom is as of 2014)
Basically, I'd like to optimize selectConnectedNodes to be able to find out
the files *actually used* by the active (non-disabled) write nodes to
replace a "tree walker" which is very slow. Ultimately I'd like to implement
a solution OUTSIDE of nuke that just parses the script file however, the .nk
file leaves no clues (that I've found) that would suggest which Read nodes
are actually used in any of the script trees. Since a shell based parse
doesn't seem to be an option, and python seems overly slow, perhaps C++ is a
better route to take?
Thanks again...
Cheers,Jep
_______________________________________________
Nuke-python mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-python
_______________________________________________
Nuke-python mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-python
