Got a one to one mapping between the output of the view user + rewrite -> the...

Got a one to one mapping between the output of the view user + rewrite -> the only expection is that sometimes the userview graph may have extra edges cause it's possible that i group together 2 operations (which isn't shown in the process dependency graph)

src/graph.py

@@ -430,7 +430,7 @@ class Graph():
     #GENERATE USER VIEW
     #============================
 
-    def get_user_view_graph(self, relevant_processes = [], use_process_dependency_graph = False):
+    def get_user_view_graph(self, relevant_processes = [], use_process_dependency_graph = False, alias_2_tools = {}):
         #For now i'm only gonna work from the flattened dico
         if(use_process_dependency_graph):
             self.initialise_flattened_dico(self.dico_process_dependency_graph)
@@ -438,7 +438,7 @@ class Graph():
             self.initialise_flattened_dico(self.full_dico)
         dico = remove_artificial_nodes(self.dico_flattened)
 
-        self.user_view, self.new_nodes_user_view = relev_user_view_builder(dico, relevant_modules=relevant_processes)
+        self.user_view, self.new_nodes_user_view = relev_user_view_builder(dico, relevant_modules=relevant_processes, alias_2_tools = alias_2_tools)
 
         with open(self.get_output_dir()/ "graphs/user_view.json", 'w') as output_file :
             json.dump(self.user_view, output_file, indent=4)
@@ -451,9 +451,9 @@ class Graph():
         #return self.user_view, user_view_with_subworkflows
         return self.user_view
     
-    def generate_user_view(self, relevant_processes = [], render_graphs = True, use_process_dependency_graph = False):
+    def generate_user_view(self, relevant_processes = [], render_graphs = True, use_process_dependency_graph = False, alias_2_tools = {}):
         #user_view, user_view_with_subworkflows = self.get_user_view_graph(relevant_processes = relevant_processes)
-        user_view = self.get_user_view_graph(relevant_processes = relevant_processes, use_process_dependency_graph = use_process_dependency_graph)
+        user_view = self.get_user_view_graph(relevant_processes = relevant_processes, use_process_dependency_graph = use_process_dependency_graph, alias_2_tools = alias_2_tools)
         #self.user_view_with_subworkflows = user_view_with_subworkflows
         generate_graph(self.get_output_dir()/'graphs'/"user_view", user_view, label_edge=True, label_node=True, render_graphs = render_graphs, root = False, relevant_nodes = copy.deepcopy(relevant_processes))
         #generate_graph(self.get_output_dir()/'graphs'/"user_view_with_subworkflows", user_view_with_subworkflows, label_edge=True, label_node=True, render_graphs = render_graphs, root = False, relevant_nodes = copy.deepcopy(relevant_processes))

src/outils_graph.py

 import graphviz
 import copy
 import numpy as np
+import os
+import json
 
 process_id = "src.process.Process"
 operation_id = "<src.operation.Operation"
@@ -808,18 +810,53 @@ def get_names_tab(dico, tab):
         final.append(names)
     return final
 
-def get_name_new_node(new_nodes, relevant_modules):
+def get_name_new_node(new_nodes, relevant_modules, tag, alias_2_tools):
     for r in relevant_modules:
         for new in new_nodes:
             if(r in new):
                 return r
-    #Arbitrary choice of choosing the name with the longest name
-    longest_name = new_nodes[0][0]
-    for name in new_nodes:
-        if(len(longest_name)<len(name[0])):
-            longest_name = name[0]
-
-    return longest_name
+    if(alias_2_tools=={}):
+        #Arbitrary choice of choosing the name with the longest name
+        longest_name = new_nodes[0][0]
+        for name in new_nodes:
+            if(len(longest_name)<len(name[0])):
+                longest_name = name[0]
+        return longest_name
+    else:
+        #We choose the process which has the "rarest" tool process
+        names = []
+        for name in new_nodes:
+            process_name = name[0].split(tag)[0]
+            if(process_name!=""):
+                names.append(process_name)
+        if(names==[]):
+            return new_nodes[0][0]
+        else:
+            OG_path = os.getcwd()
+            #Change working directory to the one of the file
+            os.chdir("/".join((str(__file__).split("/")[:-1])))
+            with open("../ressources/tool_2_nb_usage.json", 'r') as file:
+                tool_2_nb_usage = json.load(file)
+            os.chdir(OG_path)
+
+            min_tool, min_process = np.inf, names[0]
+            for alias in names:
+                try:
+                    tools = alias_2_tools[alias]
+                except:
+                    tools = [""]
+                for t in tools:
+                    try:
+                        val = tool_2_nb_usage[t]
+                        if(t in ['python', 'r', 'perl', 'julia']):#Cause in this case it is a custom script -> one should hope that it is important in this case
+                            val = 1
+                    except:
+                        val = 1
+                    if(val<min_tool):
+                        min_tool = val
+                        min_process = alias
+            return min_process
+        
 
 def check_same_elements(list1, list2):
     return set(list1)==set(list2)
@@ -835,7 +872,7 @@ def get_color_node(node, new_nodes):
     prop = 256- int(127*len(node)/max)
     return rgb_to_hex(prop, prop, prop)
 
-def relev_user_view_builder(dico_param, relevant_modules):
+def relev_user_view_builder(dico_param, relevant_modules, alias_2_tools):
     import time
     dico = copy.deepcopy(dico_param)
     tag = str(time.time())
@@ -963,7 +1000,7 @@ def relev_user_view_builder(dico_param, relevant_modules):
     new_dico["subworkflows"] = []
     for i in range(len(new_nodes)):
         new_nodes[i].sort()
-        new_name = get_name_new_node(get_names_tab(dico, new_nodes[i]), relevant_modules)
+        new_name = get_name_new_node(get_names_tab(dico, new_nodes[i]), relevant_modules, tag, alias_2_tools)
         name_printed = new_name.split(tag)[0]
         shape = "ellipse"
         if(name_printed==""):