def fork(req, *opts):
"""
Make a copy of the working tree and process the arguments as a pipleline. This essentially resets the working
tree and allows a new plumbing to run. Useful for producing multiple outputs from a single source.
:param req: The request
:param opts: Options (unused)
:return: None
**Examples**
.. code-block:: yaml
- select # select all entities
- fork:
- certreport
- publish:
output: "/tmp/annotated.xml"
- fork:
- xslt:
stylesheet: tidy.xml
- publish:
output: "/tmp/clean.xml"
The second fork in this example is strictly speaking not necessary since the main plumbing is still active
but it may help to structure your plumbings this way.
**Merging**
Normally the result of the "inner" plumbing is disgarded - unless published or emit:ed to a calling client
in the case of the MDX server - but by adding 'merge' to the options with an optional 'merge strategy' the
behaviour can be changed to merge the result of the inner pipeline back to the parent working document.
The default merge strategy is 'replace_existing' which replaces each EntityDescriptor found in the resulting
document in the parent document (using the entityID as a pointer). Any python module path ('a.mod.u.le.callable')
ending in a callable is accepted. If the path doesn't contain a '.' then it is assumed to reference one of the
standard merge strategies in pyff.merge_strategies.
For instance the following block can be used to set an attribute on a single entity:
.. code-block:: yaml
- fork merge:
- select: http://sp.example.com/shibboleth-sp
- setattr:
attribute: value
Note that unless you have a select statement before your fork merge you'll be merging into an empty
active document which with the default merge strategy of replace_existing will result in an empty
active document. To avoid this do a select before your fork, thus:
.. code-block:: yaml
- select
- fork merge:
- select: http://sp.example.com/shibboleth-sp
- setattr:
attribute: value
"""
nt = None
if req.t is not None:
nt = deepcopy(req.t)
ip = Plumbing(pipeline=req.args, pid="{}.fork".format(req.plumbing.pid))
# ip.process(req.md,t=nt)
ireq = Plumbing.Request(ip, req.md, nt)
ip._process(ireq)
if req.t is not None and ireq.t is not None and len(root(ireq.t)) > 0:
if 'merge' in opts:
sn = "pyff.merge_strategies.replace_existing"
if opts[-1] != 'merge':
sn = opts[-1]
req.md.merge(req.t, ireq.t, strategy_name=sn)
