From 91d83baa61a92eb4f53a4e1ac716ede41542c1b9 Mon Sep 17 00:00:00 2001
From: George Marchment <georgemarchment@yahoo.fr>
Date: Mon, 2 Dec 2024 11:42:58 +0100
Subject: [PATCH] Added v1 of condition extraction -> only works for calls for
now
---
src/call.py | 2 +
src/condition.py | 26 ++++++++
src/executor.py | 1 +
src/main_DSL2.py | 3 +
src/nextflow_building_blocks.py | 6 ++
src/nextflow_file.py | 2 +
src/operation.py | 3 +-
src/outils.py | 107 +++++++++++++++++++++++++++++++-
8 files changed, 147 insertions(+), 3 deletions(-)
create mode 100644 src/condition.py
diff --git a/src/call.py b/src/call.py
index 4e53b82..99e9f17 100644
--- a/src/call.py
+++ b/src/call.py
@@ -3,6 +3,7 @@ import json
from .code_ import Code
+from .condition import Condition
from .outils import get_next_param
from .executor import Executor
from .bioflowinsighterror import BioFlowInsightError
@@ -13,6 +14,7 @@ class Call(Executor):
def __init__(self, code, origin, OG_code = ''):
self.code = Code(code = code, origin = self)
self.origin = origin
+ self.condition = Condition(self)
self.called = []
self.first_element_called = None
self.parameters = []#These are in the order
diff --git a/src/condition.py b/src/condition.py
new file mode 100644
index 0000000..22c4884
--- /dev/null
+++ b/src/condition.py
@@ -0,0 +1,26 @@
+
+from .outils import extract_conditions
+
+class Condition:
+ def __init__(self, origin):
+ self.origin = origin
+ self.conditions = []
+ self.initialise()
+
+
+ def initialise(self):
+ thing_defined = self.origin.get_code()
+ code = self.origin.get_workflow_code()
+ #print(f"'{thing_defined}'")
+ #print(f"'{code}'")
+
+
+ conditions_dico = extract_conditions(code)
+ pos = code.find(thing_defined)
+ for c in conditions_dico:
+ condition_extend = conditions_dico[c]
+ if(condition_extend[0]<pos and pos<condition_extend[1]):
+ self.conditions.append(c)
+ print(thing_defined, self.conditions)
+ print()
+
diff --git a/src/executor.py b/src/executor.py
index 8f099ab..082995b 100644
--- a/src/executor.py
+++ b/src/executor.py
@@ -21,6 +21,7 @@ class Executor(Nextflow_Building_Blocks):
self.origin = origin
self.code = Code(code = code, origin = self)
+
#---------------------------------
#AUXILIARY METHODS FOR ALL CLASSES
diff --git a/src/main_DSL2.py b/src/main_DSL2.py
index dd1ed4b..ecde61b 100644
--- a/src/main_DSL2.py
+++ b/src/main_DSL2.py
@@ -15,6 +15,9 @@ class Main_DSL2(Nextflow_Building_Blocks):
def get_channels(self):
return self.channels
+
+ def get_workflow_code(self):
+ return self.get_code()
def get_type(self):
return "Main DSL2"
diff --git a/src/nextflow_building_blocks.py b/src/nextflow_building_blocks.py
index 0c44921..15e7d5d 100644
--- a/src/nextflow_building_blocks.py
+++ b/src/nextflow_building_blocks.py
@@ -32,6 +32,9 @@ class Nextflow_Building_Blocks:
def get_code(self, get_OG = False):
return self.code.get_code(get_OG = get_OG)
+ def get_origin(self):
+ return self.origin
+
def get_output_dir(self):
return self.origin.get_output_dir()
@@ -124,6 +127,9 @@ class Nextflow_Building_Blocks:
def get_processes(self):
return self.processes
+
+ def get_workflow_code(self):
+ return self.origin.get_workflow_code()
#----------------------
#CHANNELS
diff --git a/src/nextflow_file.py b/src/nextflow_file.py
index 9299da5..9a7e0b0 100644
--- a/src/nextflow_file.py
+++ b/src/nextflow_file.py
@@ -49,6 +49,8 @@ class Nextflow_File(Nextflow_Building_Blocks):
name = self.get_file_address().split('/')[-1]
return name[:-3]
+ def get_workflow_code(self):
+ return self.get_code()
def check_file_correctness(self):
code = self.get_code()
diff --git a/src/operation.py b/src/operation.py
index b329a45..8fc7bf4 100644
--- a/src/operation.py
+++ b/src/operation.py
@@ -9,6 +9,7 @@ import warnings
import re
from .outils import get_end_operator, get_end_call, get_curly_count
from .code_ import Code
+from .condition import Condition
from .executor import Executor
from .bioflowinsighterror import BioFlowInsightError
from . import constant
@@ -744,8 +745,6 @@ class Operation(Executor):
self.initialise_origins()
self.initialise_gives()
-
-
self.write_summary(self.get_output_dir() / "debug/operations.nf")
def check_if_empty_call(self):
diff --git a/src/outils.py b/src/outils.py
index ad30ce7..a207b25 100644
--- a/src/outils.py
+++ b/src/outils.py
@@ -914,4 +914,109 @@ def check_file_exists(address, origin):
def is_git_directory(path = '.'):
- return subprocess.call(['git', '-C', path, 'status'], stderr=subprocess.STDOUT, stdout = open(os.devnull, 'w')) == 0
\ No newline at end of file
+ return subprocess.call(['git', '-C', path, 'status'], stderr=subprocess.STDOUT, stdout = open(os.devnull, 'w')) == 0
+
+
+#Function that extracts the conditions defined in some code
+def extract_conditions(code):
+
+ conditions_dico = {}
+
+ start = 0
+
+ curly_count, parenthese_count = 0, 0
+ quote_single, quote_double = False, False
+ triple_single, triple_double = False, False
+
+
+ while(start<len(code)):
+ checked_triple = False
+ if(start+3<=len(code)):
+ if(code[start:start+3]=="'''" and not quote_single and not quote_double and not triple_single and not triple_double):
+ triple_single = True
+ start+=3
+ checked_triple = True
+ elif(code[start:start+3]=="'''" and not quote_single and not quote_double and triple_single and not triple_double):
+ triple_single = False
+ start+=3
+ checked_triple = True
+
+ if(code[start:start+3]=='"""' and not quote_single and not quote_double and not triple_single and not triple_double):
+ triple_double = True
+ start+=3
+ checked_triple = True
+ elif(code[start:start+3]=='"""' and not quote_single and not quote_double and not triple_single and triple_double):
+ triple_double = False
+ start+=3
+ checked_triple = True
+
+ if(not checked_triple):
+ if(code[start]=="{" and not quote_single and not quote_double and not triple_single and not triple_double):
+ curly_count+=1
+ elif(code[start]=="}" and not quote_single and not quote_double and not triple_single and not triple_double):
+ curly_count-=1
+
+ if(code[start]=="(" and not quote_single and not quote_double and not triple_single and not triple_double):
+ parenthese_count+=1
+ elif(code[start]==")" and not quote_single and not quote_double and not triple_single and not triple_double):
+ parenthese_count-=1
+
+ if(code[start]=="'" and not quote_single and not quote_double and not triple_single and not triple_double):
+ if(code[start-1]!="\\" or (code[start-1]=="\\" and code[start-2]=="\\")):
+ quote_single=True
+ elif(code[start]=="'" and quote_single and not quote_double and not triple_single and not triple_double):
+ if(code[start-1]!="\\" or (code[start-1]=="\\" and code[start-2]=="\\")):
+ quote_single=False
+
+ if(code[start]=='"' and not quote_single and not quote_double and not triple_single and not triple_double):
+ if(code[start-1]!="\\" or (code[start-1]=="\\" and code[start-2]=="\\")):
+ quote_double=True
+ elif(code[start]=='"' and not quote_single and quote_double and not triple_single and not triple_double):
+ if(code[start-1]!="\\" or (code[start-1]=="\\" and code[start-2]=="\\")):
+ quote_double=False
+
+ #TODO add "else if" compatibaliaty
+ #TODO Right now -> support only for if/else written with curlies -> not on single line
+
+ def adding_inside(conditions_dico, code, start_inside, end_inside):
+ temp_dico = extract_conditions(code[start_inside:end_inside])
+ for c in temp_dico:
+ temp = temp_dico[c]
+ temp_dico[c] = (temp[0] + start_inside, temp[1] + start_inside)
+ #Merging the 2 dicos
+ conditions_dico = conditions_dico | temp_dico
+ return conditions_dico
+ #Just because there is an 'if' doesn't necessarily mean there is an if bloc
+ found_if_bloc = False
+ if(code[start:start+2]=="if" and [quote_single, quote_double, triple_single, triple_double]==[False, False, False, False]):
+ for match in re.finditer(r"if *\((.+)\)\s*\{", code[start:]):
+ if(match.span(0)[0]==0):
+ found_if_bloc = True
+ condition = match.group(1)
+ end = extract_curly(code, match.span(0)[1]+start)#Here we nedd to add the start index since we're only working on a subpart of code
+ conditions_dico[condition] = (start,end)
+ start_inside, end_inside = match.span(0)[1]+start, end-1
+ conditions_dico = adding_inside(conditions_dico, code, start_inside, end_inside)
+ break
+ #Try to find an else corresponding
+ if(found_if_bloc and code[end:].strip()[:4]=="else"):
+ #print("corresponding else")
+ for match in re.finditer(r"else\s*{", code[end:]):
+ start_else, end_else = match.span(0)
+ start_else+=end
+ end_else = extract_curly(code, end_else+end)
+ conditions_dico[f"neg({condition})"] = (start_else,end_else)
+ start_inside, end_inside = match.span(0)[1]+start, end_else-1
+ conditions_dico = adding_inside(conditions_dico, code, start_inside, end_inside)
+ #print(code[start_else:end_else])
+ break
+ #Case we need to jump to the end of the else
+ start = end_else-1
+ else:
+ if(found_if_bloc):
+ #Case we need to jump to the end of the if
+ start = end-1
+
+ start+=1
+ return conditions_dico
+
--
GitLab