df034_SaveGraph.py

## \file
## \ingroup tutorial_dataframe
## \notebook -nodraw
## Basic SaveGraph usage.
##
## This tutorial shows how to use the SaveGraph action.
## SaveGraph inspects the sequence of RDataFrame actions.
##
## \macro_code
## \macro_output
##
## \date January 2022
## \author Ivan Kabadzhov (CERN)

# First, an RDataFrame computation graph is created with Defines, Filters and methods such as Mean, Count, etc.
# After that, SaveGraph can be called either on the root RDataFrame object or on a specific node of the computation
# graph: in the first case, the graph returned will span the full computation graph, in the second case it will show
# only the branch of the computation graph that the node belongs to.
# If a filename is passed as second argument, the graph is saved to that file, otherwise it is returned as a string.

import ROOT

rd1 = ROOT.RDataFrame(2)

rd2 = rd1.Define("Root_def1", "1") \
         .Filter("Root_def1 < 2", "Main_Filter") \
         .Define("Root_def2", "1")

branch1 = rd2.Define("Branch_1_def", "1")
branch2 = rd2.Define("Branch_2_def", "1")

branch1_1 = branch1.Filter("Branch_1_def < 2", "Filter_1") \
                   .Define("Branch_1_1_def", "1") \
                   .Filter("1 == Branch_1_1_def % 2", "Filter_1_1") \
                   .Mean("Branch_1_1_def");

branch1_2 = branch1.Define("Branch_1_2_def", "1") \
                   .Filter("Branch_1_2_def < 2", "Filter_1_2") \
                   .Count()

branch2_1 = branch2.Filter("Branch_2_def < 2", "Filter_2") \
                   .Define("Branch_2_1_def", "1") \
                   .Define("Branch_2_2_def", "1") \
                   .Filter("1 == Branch_2_1_def % 2", "Filter_2_1") \
                   .Max("Branch_2_1_def")

branch2_2 = branch2.Count()

print(ROOT.RDF.SaveGraph(branch1_1))
ROOT.RDF.SaveGraph(rd1, "exampleGraph.dot")