Updated extracting inputs DSL1 + started fixing a thing which was quite...

Updated extracting inputs DSL1 + started fixing a thing which was quite complicated -> added error see at what extend it is used (it's when there is an ambiguous emit)

src/constant.py

@@ -123,7 +123,8 @@ END_PIPE_OPERATOR = r"\s*(\s*\|\s*\w+)+"
 
 #         PROCESS
 #--------------------------
-FILE = r'file +(\w+) *\n|file +\( *(\w+) *\) *\n'
+FILE = r'file +(\w+) *\n|file *\( *(\w+) *\) *\n'
+PATH = r'path +(\w+) *\n|path *\( *(\w+) *\) *\n'
 FROM = r' from ([^\n]+)\n'
 INPUT = r"\n\s*input *:"
 INTO = r'into +([\w, ]+)'

src/emitted.py

@@ -16,6 +16,9 @@ class Emitted(Channel):
         self.source.append(emitted_by)
         self.emits = None #->this is the channel it's emits -> in the case of a subworkflow
 
+    def get_all_conditions(self):
+        return self.origin.get_all_conditions()
+
     def simplify_code(self):
         code = self.get_code(get_OG=True)
         thing_which_emits = self.emitted_by.get_first_element_called()

src/operation.py

@@ -912,15 +912,47 @@ class Operation(Executor):
             #    return f"{head}\noperation_{operation_id}_{index} = {lines[-1]}"
 
         to_add = []
+        dico_origin_2_replace = {}
         for o in self.origins:
+            try:
+                tmp = dico_origin_2_replace[o.get_code(get_OG=True)]
+            except:
+                dico_origin_2_replace[o.get_code(get_OG=True)] = []
             if(o.get_type()=="Call"):
                 #If it's not a function -> then we rewrite it
                 if(o.get_first_element_called().get_type()!="Function"):
                     to_add.append(add_origin_equals(o, index))
-                    code = code.replace(o.get_code(get_OG=True), f"operation_{operation_id}_{index}")
+                    dico_origin_2_replace[o.get_code(get_OG=True)].append(f"operation_{operation_id}_{index}")
             elif(o.get_type()=="Emitted"):
-                code = code.replace(o.get_code(get_OG=True), o.simplify_code())
+                dico_origin_2_replace[o.get_code(get_OG=True)].append(o)
             index += 1
+        
+        temporary_index = 1
+        for origin in dico_origin_2_replace:
+            if(len(dico_origin_2_replace[origin])<=1):
+                for val in dico_origin_2_replace[origin]:
+                    if(type(val)==str):
+                        code = code.replace(origin, val)
+                    else:
+                        code = code.replace(origin, val.simplify_code())
+            #Case there are mutiple origins then:
+            else:
+                #TODO -> fix this problem if you have the courage to -> see at what extend it is used
+                raise BioFlowInsightError("TODO (update): The output of a emit is ambiguous -> there are multiple instance where it exits")
+                #print(dico_origin_2_replace)
+                #for e in dico_origin_2_replace[origin]:
+                #    print(e.get_all_conditions())
+            temporary_index+=1
+
+        #for o in self.origins:
+        #    if(o.get_type()=="Call"):
+        #        #If it's not a function -> then we rewrite it
+        #        if(o.get_first_element_called().get_type()!="Function"):
+        #            to_add.append(add_origin_equals(o, index))
+        #            code = code.replace(o.get_code(get_OG=True), f"operation_{operation_id}_{index}")
+        #    elif(o.get_type()=="Emitted"):
+        #        code = code.replace(o.get_code(get_OG=True), o.simplify_code())
+        #    index += 1
 
         to_add.reverse()
         for c in to_add:

src/process.py

@@ -259,16 +259,17 @@ class Process(Nextflow_Building_Blocks):
         for line in code.split("\n"):
 
             #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)
-            pattern = constant.FILE
-            for match in re.finditer(pattern, line+"\n"):
-                #In the first case it's "file ch" in the second "file (ch)" 
-                try:
-                    extracted = match.group(1).strip()
-                except:
-                    extracted = match.group(2).strip()
+            patterns = [constant.FILE, constant.PATH]
+            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)" 
+                    try:
+                        extracted = match.group(1).strip()
+                    except:
+                        extracted = match.group(2).strip()
 
-                add_channel(extracted)
-                self.raw_input_names.append(extracted)
+                    add_channel(extracted)
+                    self.raw_input_names.append(extracted)
             
         
             #Case there are multiple channels as input (e.g. channel1.mix(channel2))

src/workflow.py

@@ -528,7 +528,11 @@ George Marchment, Bryan Brancotte, Marie Schmit, Frédéric Lemoine, Sarah Cohen
 
     def rewrite_and_initialise(self, code, processes_2_remove, render_graphs):
         temp_process_dependency_graph = self.graph.get_process_dependency_graph() 
-        temp_spec_graph = self.graph.full_dico    
+        temp_spec_graph = self.graph.full_dico
+
+        #Remove the "_GG_\d+"
+        #code = re.sub(r"_GG_\d+", "", code)
+
         #Write new code in temporary file
         temp_file = self.get_output_dir()/f"temp_{str(self)[-7:-2]}.nf"
         with open(temp_file, "w") as file: