diff --git a/src/code_.py b/src/code_.py
index 3c9fefd039f4cc5222be784c75f5cbe75cdfd7fe..349da56b56c09768b98c7a15ffea16a0285a9489 100644
--- a/src/code_.py
+++ b/src/code_.py
@@ -1,7 +1,8 @@
-from .outils import remove_comments, get_parenthese_count, get_curly_count, get_code_until_parenthese_count, extract_curly, get_next_element_caracter, get_code_until_character
+from .outils import remove_comments, get_parenthese_count, get_curly_count, get_code_until_parenthese_count, extract_curly, get_next_element_caracter, get_code_until_character, get_end_call
from .bioflowinsighterror import BioFlowInsightError
import re
from . import constant
+import numpy as np
class Code:
def __init__(self, code, origin, initialise):
@@ -20,7 +21,9 @@ class Code:
self.code_wo_comments = remove_comments(self.code)
self.code_wo_comments = re.sub(constant.DOUBLE_BACKSLAPSH_JUMP, ' ', self.code_wo_comments)
self.code_wo_comments = re.sub(constant.BACKSLAPSH_JUMP, ' ', self.code_wo_comments)
+ self.code_wo_comments = re.sub(r"(\n( |\t)*)+\.", '.', self.code_wo_comments)
self.code_wo_comments = self.code_wo_comments.replace("||", "$OR$")
+ self.code_wo_comments = self.turn_single_multiline_conditions_into_single(self.code_wo_comments)
self.code_wo_comments = self.turn_single_condition_into_multiline(self.code_wo_comments)
self.code_wo_comments = self.remove_things_inside_map(self.code_wo_comments )
self.code_wo_comments = self.rewrite_ternary_operation_to_normal_condition(self.code_wo_comments)
@@ -119,6 +122,18 @@ class Code:
if(char!="{" and temp_code[pos:pos+3] not in ['"""', "'''"]):
raise BioFlowInsightError(f"The condition '({extracted_condition}' was not extracted correctly. Make sure the condition follows the correct syntaxe.", type="Unable to extract condition")
+ if(code[start:start+4]=="else" and [quote_single, quote_double, triple_single, triple_double]==[False, False, False, False]):
+ pattern = r"(else *)(.+)\n"
+ temp_code = code[start:]
+ for match in re.finditer(pattern, temp_code):
+ if(match.span(0)[0]==0):
+ _, end_line = match.span(0)
+ all = match.group(0)
+ extarcted = match.group(2).strip()
+
+ if(extarcted!="" and extarcted[0] not in ["{"] and extarcted[-1] not in ["{", "}"]):
+ new = f"else {{\n{extarcted}\n}}\n"
+ to_replace.append((all, new))
@@ -247,8 +262,11 @@ class Code:
#This methods rewrite ternary operation into "normal" conditions
#variable = (condition) ? Expression2 : Expression3;
def rewrite_ternary_operation_to_normal_condition(self, code):
+ #Turning multi line ternry operations into single ternary operations
+ code = re.sub(r"(\n( |\t)*)+:", ' :', code)
pattern = r"\n *(def)? *(\w+) *\= *(([^?\n]+) *\? *([^:\n]+) *\: *([^\n]+))\n"
to_replace = []
+ already_searched = []
searching = True
@@ -265,42 +283,67 @@ class Code:
old = match.group(0)
old = old[1:]
dico_conditions = {}
-
+
def rewrite_ternary(exp, dico_conditions):
exp = exp.strip()
+ if(exp==""):
+ return "null"
if(exp[0]=="(" and exp[-1]==")"):
exp = exp[1:-1].strip()
+
try:
expression, end_condition = get_code_until_character(exp, "?")
except:
return exp
+
condition = exp[:end_condition].strip()
+ try:
+ _, end_potential_condition = get_code_until_character(expression, "?")
+
+ except:
+ end_potential_condition = np.inf
exp2, end_exp1 = get_code_until_character(expression, ":")
dico_true, dico_false = {}, {}
- dico_conditions[condition] = {"True": rewrite_ternary(expression[:end_exp1], dico_true), "False": rewrite_ternary(exp2, dico_false)}
+ #if(expression[:end_exp1]==""):
+ # raise BioFlowInsightError(f"The 'True' case in the ternary condition '{exp}' has no value.", type="Incomplete ternary operation")
+ #if(exp2==""):
+ # raise BioFlowInsightError(f"The 'False' case in the ternary condition '{exp}' has no value.", type="Incomplete ternary operation")
+
+ #Case there is a condition right after the first condition
+ if(end_potential_condition<end_exp1):
+ expression_false, end_exp1 = get_code_until_character(expression, ":", left_to_right=False)
+ dico_conditions[condition] = {"True": rewrite_ternary(expression[:end_exp1], dico_true), "False": rewrite_ternary(expression_false, dico_false)}
+ else:
+ dico_conditions[condition] = {"True": rewrite_ternary(expression[:end_exp1], dico_true), "False": rewrite_ternary(exp2, dico_false)}
return dico_conditions
-
- rewrite_ternary(exp, dico_conditions)
+ if(old not in already_searched):
+
+ try:
+ rewrite_ternary(exp, dico_conditions)
+ except:
+ already_searched.append(old)
+ searching = True
+ break
-
- def rewrite_dico_2_condition(var, dico_condition, num = 0):
- code = ''
- if(type(dico_condition)==str):
- return f"{var} = {dico_condition}\n"
- for condition in dico_condition:
- code = f"if({condition}) {{\n\t{rewrite_dico_2_condition(var, dico_condition[condition]['True'], num = num+1)}}} else {{\n\t{rewrite_dico_2_condition(var, dico_condition[condition]['False'], num = num+1)}}}\n"
- return code
-
- new = rewrite_dico_2_condition(f"{def_variable} {variable}", dico_conditions)+'\n'
+
+ def rewrite_dico_2_condition(var, dico_condition, num = 0):
+ code = ''
+ if(type(dico_condition)==str):
+ return f"{var} = {dico_condition}\n"
+ for condition in dico_condition:
+ code = f"if({condition}) {{\n\t{rewrite_dico_2_condition(var, dico_condition[condition]['True'], num = num+1)}}} else {{\n\t{rewrite_dico_2_condition(var, dico_condition[condition]['False'], num = num+1)}}}\n"
+ return code
+
+ new = rewrite_dico_2_condition(f"{def_variable} {variable}", dico_conditions)+'\n'
- to_replace.append((old, new))
- tmp = code
- code = code.replace(old, new, 1)
- if(old!=new and tmp==code):
- raise Exception("This shouldn't happen -> the code wasn't replaced")
- searching = True
- break
+ to_replace.append((old, new))
+ tmp = code
+ code = code.replace(old, new, 1)
+ if(old!=new and tmp==code):
+ raise Exception("This shouldn't happen -> the code wasn't replaced")
+ searching = True
+ break
for r in to_replace:
old, new = r
@@ -308,7 +351,95 @@ class Code:
#Check if there is still a ternary operation in this case we cannot analyse it
for match in re.finditer(pattern, code):
- raise BioFlowInsightError(f"Detected a multi ternary operation (a ternary operation in a ternary operation) in the file '{self.origin.get_file_address()}'. BioFlow-Insight does not support this, try defining it in a different way.", type="Multi ternary operation")
+ old = match.group(0)
+ old = old[1:]
+ if(old not in already_searched):
+ raise BioFlowInsightError(f"Detected a multi ternary operation (a ternary operation in a ternary operation) in the file '{self.origin.get_file_address()}'. BioFlow-Insight does not support this, try defining it in a different way.", type="Multi ternary operation")
+
+ return code
+
+ def turn_single_multiline_conditions_into_single(self, code):
+ start = 0
+
+ curly_count, parenthese_count = 0, 0
+ quote_single, quote_double = False, False
+ triple_single, triple_double = False, False
+
+ to_replace = []
+ timeout = 0
+ while(start<len(code) and timeout < constant.WHILE_UPPER_BOUND):
+ 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
+
+
+ if(code[start:start+2]=="if" and [quote_single, quote_double, triple_single, triple_double]==[False, False, False, False]):
+ found_if_bloc = False
+ for match in re.finditer(r"if *\(", code[start:]):
+ if(match.span(0)[0]==0):
+ start_if, end_if = match.span(0)
+ txt = get_end_call(code[start:], start_if, end_if)
+ if('\n' in txt):
+ to_replace.append(txt)
+ end = start+len(txt)
+ found_if_bloc = True
+ break
+ else:
+ break
+ if(found_if_bloc):
+ #Case we need to jump to the end of the if
+ start = end-1
+
+ start+=1
+ timeout+=1
+ if(timeout>=constant.WHILE_UPPER_BOUND):
+ raise BioFlowInsightError("The WHILE_UPPER_BOUND was exceeded. BioFlow-Insight was unable to extarct the conditions. Make sure the workflow uses correct Nextflow syntaxe (https://www.nextflow.io/docs/latest/index.html).", type="Unable to extract conditions")
+ for r in to_replace:
+ temp = code
+ code = code.replace(r, r.replace('\n', " "))
+ if(code==temp):
+ raise Exception("Not updated!")
return code
def rewrite_jump_dot(self, code):
diff --git a/src/outils.py b/src/outils.py
index a80f07c22e3c980577fa7842c489d80b1417130a..d3b8a5d7c2174fa25429dd710315a76df21ff124 100644
--- a/src/outils.py
+++ b/src/outils.py
@@ -1668,7 +1668,7 @@ def extract_tools(script, extract_general_tools = False):
return list(set(tools))
-def get_code_until_character(code, char):
+def get_code_until_character_2(code, char):
start = 0
@@ -1726,3 +1726,78 @@ def get_code_until_character(code, char):
return code[start+1:], start
start+=1
raise Exception("")
+
+
+def get_code_until_character(code, char, left_to_right = True):
+
+ def addition(variable, val):
+ return variable + val
+
+ def substraction(variable, val):
+ return variable - val
+
+ if(left_to_right):
+ fun = addition
+ start = 0
+ else:
+ fun = substraction
+ start = -1
+
+ 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(fun(start, 3)<=len(code)):
+ if(code[start:fun(start, 3)]=="'''" and not quote_single and not quote_double and not triple_single and not triple_double):
+ triple_single = True
+ start=fun(start, 3)
+ checked_triple = True
+ elif(code[start:fun(start, 3)]=="'''" and not quote_single and not quote_double and triple_single and not triple_double):
+ triple_single = False
+ start=fun(start, 3)
+ checked_triple = True
+
+ if(code[start:fun(start, 3)]=='"""' and not quote_single and not quote_double and not triple_single and not triple_double):
+ triple_double = True
+ start=fun(start, 3)
+ checked_triple = True
+ elif(code[start:fun(start, 3)]=='"""' and not quote_single and not quote_double and not triple_single and triple_double):
+ triple_double = False
+ start=fun(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[substraction(start, 1)]!="\\" or (code[substraction(start, 1)]=="\\" and code[substraction(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[substraction(start, 1)]!="\\" or (code[substraction(start, 1)]=="\\" and code[substraction(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[substraction(start, 1)]!="\\" or (code[substraction(start, 1)]=="\\" and code[substraction(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[substraction(start, 1)]!="\\" or (code[substraction(start, 1)]=="\\" and code[substraction(start, 2)]=="\\")):
+ quote_double=False
+ if(code[start]==char and parenthese_count==0 and curly_count ==0 and [quote_single, quote_double, triple_single, triple_double]==[False, False, False, False]):
+ if(left_to_right):
+ return code[start+1:], start
+ else:
+ return code[start+1:], start
+
+ start=fun(start, 1)
+ raise Exception("")