From 64e13dcd5c27aa9f347a60de37f39636c6cf497c Mon Sep 17 00:00:00 2001
From: George Marchment <georgemarchment@yahoo.fr>
Date: Fri, 3 Jan 2025 11:53:29 +0100
Subject: [PATCH] Started working on a good version for the detection of the
broken subworkflows
---
src/channel.py | 2 +-
src/workflow.py | 118 ++++++++++++++++++++++++++++++++++++++++--------
2 files changed, 99 insertions(+), 21 deletions(-)
diff --git a/src/channel.py b/src/channel.py
index 1e9cdfb..c51a205 100644
--- a/src/channel.py
+++ b/src/channel.py
@@ -23,7 +23,7 @@ class Channel(Nextflow_Building_Blocks):
self.sink = []
- def get_code(self):
+ def get_code(self, get_OG = True):
"""Method that returns the channels code
Keyword arguments:
diff --git a/src/workflow.py b/src/workflow.py
index 9dcb913..2b6e538 100644
--- a/src/workflow.py
+++ b/src/workflow.py
@@ -810,6 +810,46 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
raise Exception("This shouldn't happen")
return code
+ #I do not recommand that the dev uses the same name for the channels inside and outside the channels
+ #Since the local channels become local at the upper level
+ def rewrite_subworkflow_call(seld, code, subworklfow):
+ print(subworklfow.get_name())
+ #Remove the defintion from the code
+ code = code.replace(subworklfow.get_code(), "")
+ OG_call = subworklfow.get_call()
+ if(len(OG_call)>1):
+ raise Exception("This shouldn't happen")
+ OG_call = OG_call[0]
+ OG_body = subworklfow.get_work()
+
+ #REPLACE HEADER TAKES
+ subworkflow_takes = subworklfow.get_takes()
+ parameters = OG_call.get_parameters()
+ if(len(subworkflow_takes)!=len(parameters)):
+ raise Exception("This shouldn't happen -> the same number of parameters should be kept")
+ #This is to replace the paramters and the takes
+ new_header = ""
+ for i in range(len(parameters)):
+ param = parameters[i]
+ takes = subworkflow_takes[i].get_gives()[0]
+ if(takes.get_code()!=param.get_code(get_OG = True)):
+ new_header+=f"{takes.get_code()} = {param.get_code(get_OG = True)}"
+
+ code = code.replace(OG_call.get_code(get_OG = True), f"{new_header}\n\n{OG_body}")
+
+ #REPLACE THE EMITS
+ emits = subworklfow.get_emit()
+ for o in emits:
+ origin = o.get_origins()[0]
+ gives = o.get_gives()
+ #TODO finish this -> i think that the gives is empty
+ print(o.get_type())
+ print(o.get_code(), origin.get_code(), gives)
+
+
+ #print(code)
+ return code
+
#Conert workflow to user_view only makes sense when the option duplicate is activated -> otherwise is doesn't make sense + it makes the analysis way more complicated
@@ -832,7 +872,7 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
self.nextflow_file.generate_user_view(relevant_processes = relevant_processes, processes_2_remove = [])
clusters = self.nextflow_file.graph.get_clusters_from_user_view()
- """#DETERMING WHICH SUBWORKFLOWS ARE BROKEN WITH THE CLUSTER
+ #DETERMING WHICH SUBWORKFLOWS ARE BROKEN WITH THE CLUSTER
#Creating the clusters with calls instead of processes or subworkflows
set_clusters_with_calls = []
for c in clusters:
@@ -847,30 +887,68 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
raise Exception("This shouldn't happen")
tab.append(call[0])
set_clusters_with_calls.append(set(tab))
- print(set_clusters_with_calls)
+ #Getting subworkflows to executors
subworkflow_2_executors = {}
for sub in self.get_subworkflows_called():
dico = {}
sub.get_all_executors(dico)
- subworkflow_2_executors[sub.get_name()] = set(list(dico.keys()))
- print(subworkflow_2_executors)
-
- def check_subworkflow_intact(original_sets, new_sets):
- results = {}
- for sub in original_sets:
- original = original_sets[sub]
- #temp = []
- #for new in new_sets:
- # print(original, new, new.issubset(original))
- # #temp.append(original.issubset(new))
- #print(temp)
- intact = all(new.issubset(original) for new in new_sets)
- results[sub] = intact
- return results
-
- print(check_subworkflow_intact(subworkflow_2_executors, set_clusters_with_calls))
- 1/0"""
+ temp = set(list(dico.keys()))
+ subworkflow_2_executors[sub] = []
+ for ele in temp:
+ #Cannot add calls to subworkflows -> they are already present by definition
+ if(ele.get_type()=="Call" and ele.get_first_element_called().get_type()=="Subworkflow"):
+ None
+ #We don't add it
+ else:
+ subworkflow_2_executors[sub].append(ele)
+ #subworkflow_2_executors[sub.get_name()] = set(list(dico.keys()))
+
+ def get_workflows_broken(subworkflow_2_executors, set_clusters_with_calls):
+ broken_subworkflows = []
+ for sub in subworkflow_2_executors:
+ #You want this list (set) to be equal to subworkflow_2_executors[sub]
+ elements_in_sub_with_clusters = []
+ for cluster in set_clusters_with_calls:
+ if(len(elements_in_sub_with_clusters)>len(subworkflow_2_executors[sub])):
+ break
+ for c in cluster:
+ if(len(elements_in_sub_with_clusters)>len(subworkflow_2_executors[sub])):
+ break
+ if(c in subworkflow_2_executors[sub]):
+ elements_in_sub_with_clusters+=list(cluster)
+ break
+ if(set(elements_in_sub_with_clusters)==set(subworkflow_2_executors[sub])):
+ None
+ #This means that the subworkflow is Intact
+ else:
+ #This means that the subworkflow is broken
+ broken_subworkflows.append(sub)
+ return broken_subworkflows
+
+ broken_subworkflows = get_workflows_broken(subworkflow_2_executors, set_clusters_with_calls)
+
+ #Rewrite broken subworkflows
+ for sub in broken_subworkflows:
+ code = self.rewrite_subworkflow_call(code, sub)
+
+ 1/0
+
+ #def check_subworkflow_intact(original_sets, new_sets):
+ # results = {}
+ # for sub in original_sets:
+ # original = original_sets[sub]
+ # #temp = []
+ # #for new in new_sets:
+ # # print(original, new, new.issubset(original))
+ # # #temp.append(original.issubset(new))
+ # #print(temp)
+ # intact = all(new.issubset(original) for new in new_sets)
+ # results[sub] = intact
+ # return results
+
+ #print(check_subworkflow_intact(subworkflow_2_executors, set_clusters_with_calls))
+ 1/0
#cluster_2_subworkflows = []
#print(clusters)
--
GitLab