From 106ffe4d2f43f7001252c980277e279d289191cf Mon Sep 17 00:00:00 2001
From: George Marchment <georgemarchment@yahoo.fr>
Date: Wed, 16 Apr 2025 10:55:38 +0200
Subject: [PATCH] Updated the selection of random processes + updated the
extarction of paths and files (there was an issue with the groups)
---
src/constant.py | 6 ++++--
src/process.py | 11 ++++++-----
src/workflow.py | 39 +++++++++++++++++++++++----------------
3 files changed, 33 insertions(+), 23 deletions(-)
diff --git a/src/constant.py b/src/constant.py
index b46c2b5..400e439 100644
--- a/src/constant.py
+++ b/src/constant.py
@@ -123,8 +123,10 @@ END_PIPE_OPERATOR = r"\s*(\s*\|\s*\w+)+"
# PROCESS
#--------------------------
-FILE = r'file +(\w+) *\n|file *\( *(\w+) *\) *\n'
-PATH = r'path +(\w+) *\n|path *\( *(\w+) *\) *\n'
+FILE1 = r'file +(\w+) *\n'
+FILE2 = r'file *\( *(\w+) *\) *\n'
+PATH1 = r'path +(\w+) *\n'
+PATH2 = r'path *\( *(\w+) *\) *\n'
FROM = r'[^\w]from ([^\n]+)\n'
INPUT = r"\n\s*input *:"
INTO = r'into +([\w, ]+)'
diff --git a/src/process.py b/src/process.py
index 8d4c181..6beb287 100644
--- a/src/process.py
+++ b/src/process.py
@@ -289,7 +289,7 @@ class Process(Nextflow_Building_Blocks):
placed = False
#Case there is a single channel as an input -> doesn't use from to import channel -> uses file (see https://github.com/nextflow-io/nextflow/blob/45ceadbdba90b0b7a42a542a9fc241fb04e3719d/docs/process.rst)
- patterns = [constant.FILE, constant.PATH]
+ patterns = [constant.FILE1, constant.FILE2, constant.PATH1, constant.PATH2]
for pattern in patterns:
for match in re.finditer(pattern, line+"\n"):
#In the first case it's "file ch" in the second "file (ch)"
@@ -423,10 +423,11 @@ class Process(Nextflow_Building_Blocks):
tab.append(self.outputs[-1])
self.outputs_per_line.append(tab)
- pattern = constant.FILE
- for match in re.finditer(pattern, code):
- add_channel(match.group(1))
- self.outputs_per_line.append([self.outputs[-1]])
+ patterns = [constant.FILE1, constant.FILE2]
+ for pattern in patterns:
+ for match in re.finditer(pattern, code):
+ add_channel(match.group(1))
+ self.outputs_per_line.append([self.outputs[-1]])
#Function that extracts the inputs from a process (for DSLS workflows)
def initialise_outputs_DSL2(self):
diff --git a/src/workflow.py b/src/workflow.py
index 144cf2f..0d0abe0 100644
--- a/src/workflow.py
+++ b/src/workflow.py
@@ -431,12 +431,13 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
#So that means there are the 2 nodes which form the cycle edge in the relevant processes
#-> it means we need to regenerated relevant processes
searching = True
- break
-
- name_select = []
- for p in sampled:
- name_select.append(p.get_alias())
- return name_select
+ if(not searching):
+ name_select = []
+ for p in sampled:
+ name_select.append(p.get_alias())
+ return name_select
+ else:
+ print("here")
#This methods generates a random set of processes to consider as relavant
@@ -1036,21 +1037,27 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
def get_object(address):
address = int(re.findall(r"\dx\w+", address)[0], base=16)
return ctypes.cast(address, ctypes.py_object).value
- #Check that there are no cycles which will break the creation of the user view:
- edges_create_cycles = self.graph.get_edges_that_create_cycle()
- edges_create_cycles_objects = []
- for e in edges_create_cycles:
- edges_create_cycles_objects.append((get_object(e[0]), get_object(e[1])))
- for e in edges_create_cycles_objects:
- n1 = e[0].get_alias()
- n2 = e[1].get_alias()
- if(n1 in relevant_processes and n2 in relevant_processes):
- raise BioFlowInsightError(f"The processes '{n1}' and '{n2}' cannot both be relevant processes since there is a dependency apparant in the workflow between the 2")
+
ternary_operation_dico = self.ternary_operation_dico
map_element_dico = self.map_element_dico
if(self.duplicate):
+
+ #Check that there are no cycles which will break the creation of the user view:
+ edges_create_cycles = self.graph.get_edges_that_create_cycle()
+ #if(len(edges_create_cycles)>0):
+ # #TODO -> add link for further details for the user
+ # raise BioFlowInsightError("An indirect cycle was detected in the workflow. This happens often when a same channel is used multiple times in the workflow. This is a norm that we do not recommend. Try rewritting the workflow in a different way. See [link] for more details", type="Cycle detected")
+ edges_create_cycles_objects = []
+ for e in edges_create_cycles:
+ edges_create_cycles_objects.append((get_object(e[0]), get_object(e[1])))
+ for e in edges_create_cycles_objects:
+ n1 = e[0].get_alias()
+ n2 = e[1].get_alias()
+ if(n1 in relevant_processes and n2 in relevant_processes):
+ raise BioFlowInsightError(f"The processes '{n1}' and '{n2}' cannot both be relevant processes since there is a dependency apparant in the workflow between the 2")
+
#First check if there are any duplicate operations
#That method is in the "get_order_execution_executors" method -> so we just run that first
self.check_multiple_subworkflow()
--
GitLab