Merge branch 'complete_graph'

verarbeitung/start_script.py → start_script.py

@@ -8,4 +8,5 @@ doi_list = []
 doi_list.append('https://pubs.acs.org/doi/10.1021/acs.jcim.9b00249')
 #doi_list.append('https://doi.org/10.1021/acs.jcim.9b00249')
 doi_list.append('https://pubs.acs.org/doi/10.1021/acs.jcim.1c00203')
-Processing(doi_list, 3, 3, 'test.json')
\ No newline at end of file
+doi_list.append('https://doi.org/10.1021/acs.jmedchem.0c01332')
+Processing(doi_list, 3, 2, 'test.json')

verarbeitung/construct_new_graph/add_citations_rec.py

@@ -32,10 +32,12 @@ def get_cit_type_list(pub, cit_type):
 
         function to create nodes and edges and call create_graph_structure_citations
     '''
-    if (cit_type == "Citation"):
+    if cit_type == "Citation":
         return(pub.citations)
-    else:
+    elif cit_type == "Reference":
         return(pub.references)
+    else:
+        return(ValueError)
 
 def create_global_lists_cit(input_nodes, input_edges, pub, search_depth, search_depth_max, cit_type, test_var):
     '''

verarbeitung/construct_new_graph/export_to_json.py

@@ -78,7 +78,7 @@ def output_to_json(nodes, edges, json_file = 'json_text.json', test_var = False)
     list_of_edge_dicts = format_edges(edges)
     dict_of_all["nodes"] = list_of_node_dicts
     dict_of_all["links"] = list_of_edge_dicts
-    if (test_var and json_file == 'json_text.json'):
+    if (test_var):
         with open('test_output.json','w') as outfile:
             json.dump(dict_of_all, outfile)
     else:

verarbeitung/construct_new_graph/initialize_graph.py

@@ -15,6 +15,7 @@ __status__ = "Production"
 
 
 import sys  
+import gc
 from pathlib import Path
 from os import error
 sys.path.append("../")
@@ -24,6 +25,43 @@ from verarbeitung.get_pub_from_input import get_pub
 from .export_to_json import output_to_json
 from .add_citations_rec import add_citations, create_global_lists_cit
 
+def initialize_nodes_list_test(doi_input_list, search_depth_max, search_height_max, test_var):
+    '''
+        :param doi_input_list:      input list of doi from UI
+        :type doi_input_list:       List[String]
+
+        :param search_depth_max:    maximum depth to search for references
+        :type search_depth_max:     int
+
+        :param search_height_max:   maximum height to search for citations
+        :type search_height_max:    int
+
+        :param test_var:            variable to differenciate between test and url call
+        :type test_var:             boolean
+
+        for unit test purposes only
+    '''
+    global nodes, edges
+    nodes = []
+    edges = []
+    return(initialize_nodes_list(doi_input_list, search_depth_max, search_height_max, test_var))
+
+def complete_inner_edges_test(test_nodes, test_edges):
+    '''
+        :param test_nodes:      list of publications from unit test
+        :type test_nodes:       List[Publication]
+
+        :param test_nodes:      list of links from unit test
+        :type test_nodes:       List[List[String,String]]
+
+        for unit test purposes only
+        '''
+        
+    global nodes, edges
+    nodes = test_nodes
+    edges = test_edges
+    complete_inner_edges()
+    return(nodes, edges)
 
 def initialize_nodes_list(doi_input_list, search_depth_max, search_height_max, test_var):
     '''
@@ -54,6 +92,7 @@ def initialize_nodes_list(doi_input_list, search_depth_max, search_height_max, t
 
         # checks if publication already exists in nodes
         not_in_nodes = True #boolean value to check if a node already exists in the set of nodes
+
         for node in nodes: #iterates over every node in the set of nodes
             if (pub.doi_url == node.doi_url): #determines that a node with this doi already is in the set
                 not_in_nodes = False #false --> node will not be created
@@ -96,7 +135,7 @@ def complete_inner_edges():
                         edges.append([node.doi_url,reference.doi_url])
 
 
-def init_graph_construction(doi_input_list, search_depth, search_height, test_var = False):
+def init_graph_construction(doi_input_list, search_depth, search_height, test_var = False, update_var = False, input_nodes = [], input_edges = []):
     '''
         :param doi_input_list:  input list of doi from UI
         :type doi_input_list:   List[String]
@@ -110,6 +149,15 @@ def init_graph_construction(doi_input_list, search_depth, search_height, test_va
         :param test_var:        variable to differenciate between test and url call
         :type test_var:         boolean
 
+        :param test_var:        variable to check if call is from update_graph with known nodes and edges or fresh construction
+        :type test_var:         boolean
+
+        :param input_nodes:     list of publications from update_graph
+        :type input_nodes:      List[Publication]
+
+        :param input_nodes:     list of links from update_graph
+        :type input_nodes:      List[List[String,String]]
+
         main function to start graph generation
     '''
 
@@ -128,6 +176,10 @@ def init_graph_construction(doi_input_list, search_depth, search_height, test_va
     
     # creates empty lists to save nodes and edges
     global nodes, edges
+    if update_var:
+        nodes = input_nodes
+        edges = input_edges 
+    else:
         nodes = []
         edges = []
 
@@ -141,7 +193,11 @@ def init_graph_construction(doi_input_list, search_depth, search_height, test_va
     # adds edges between reference group and citation group of known publications
     complete_inner_edges()
 
-    # calls a skript to save nodes and edges of graph in .json file
-    #output_to_json(nodes, edges, test_var)
+    # garbage collection to delete nodes and edges lists. Needed because python keeps lists after function end till next function call
+    new_nodes = nodes.copy()
+    new_edges = edges.copy()
+    del nodes
+    del edges
+    gc.collect()
 
-    return(nodes,edges)
+    return(new_nodes,new_edges)

verarbeitung/process_main.py

@@ -16,7 +16,7 @@ __status__ = "Production"
 
 import sys  
 from pathlib import Path
-from os import error
+from os import error, path
 
 sys.path.append("../")
 
@@ -42,11 +42,11 @@ def Processing(url_list, search_depth, search_height, json_file = 'json_text.jso
     '''
 
     # updates graph if json file is known in directory otherwise starts new graph construction
-    try:
-        with open(json_file) as f:
-            nodes, edges = update_graph(url_list, json_file, search_depth, search_height)
+    json_file = Path(json_file)
     
-    except IOError:
+    if json_file.is_file():
+        nodes, edges = update_graph(url_list, json_file, search_depth, search_height)
+    else:
         nodes, edges = init_graph_construction(url_list, search_depth, search_height)
 
     # exports graph to given json file name

verarbeitung/test/construct_graph_unittest.py

 import unittest
 
 import sys
-from pathlib import Path
+  
+
 sys.path.append("../")
 
-from verarbeitung.construct_new_graph.initialize_graph import init_graph_construction
+from verarbeitung.construct_new_graph.initialize_graph import init_graph_construction, initialize_nodes_list_test, complete_inner_edges_test
+from verarbeitung.construct_new_graph.add_citations_rec import get_cit_type_list
+from verarbeitung.test.input_test import input_test_func
 
 class ConstructionTest(unittest.TestCase):
      maxDiff = None
 
 
      def testCycle(self):
-         nodes, edges = init_graph_construction(['doiz1'],1,1,True)
+         nodes, edges = init_graph_construction(['doiz1'],1,1,True,False)
          doi_nodes = keep_only_dois(nodes)
          self.assertCountEqual(doi_nodes, ['doiz1', 'doiz2'])
          self.assertCountEqual(edges, [['doiz1', 'doiz2'], ['doiz2', 'doiz1']])
 
-         nodes, edges = init_graph_construction(['doiz1'],2,2,True)
+         nodes, edges = init_graph_construction(['doiz1'],2,2,True,False)
          doi_nodes = keep_only_dois(nodes)
          self.assertCountEqual(doi_nodes, ['doiz1', 'doiz2'])
          self.assertCountEqual(edges, [['doiz2', 'doiz1'], ['doiz1', 'doiz2']])
@@ -28,61 +31,114 @@ class ConstructionTest(unittest.TestCase):
     #def testEmptyDepth(self):
 
      def testEmptyDepthHeight(self):
-         nodes, edges = init_graph_construction(['doi1'],0,0,True)
+         nodes, edges = init_graph_construction(['doi1'],0,0,True,False)
          doi_nodes = keep_only_dois(nodes)
          self.assertCountEqual(doi_nodes,['doi1'])
          self.assertCountEqual(edges, [])
 
-         nodes, edges = init_graph_construction(['doi1', 'doi2'],0,0,True)
+         nodes, edges = init_graph_construction(['doi1', 'doi2'],0,0,True,False)
          doi_nodes = keep_only_dois(nodes)
          self.assertCountEqual(doi_nodes, ['doi1','doi2'])
          self.assertCountEqual(edges, [['doi1', 'doi2']])
 
-         nodes, edges = init_graph_construction(['doi1', 'doi2', 'doi3'],0,0,True)
+         nodes, edges = init_graph_construction(['doi1', 'doi2', 'doi3'],0,0,True,False)
          doi_nodes = keep_only_dois(nodes)
          self.assertCountEqual(doi_nodes, ['doi1','doi2', 'doi3'])
          self.assertCountEqual(edges, [['doi3', 'doi1'], ['doi1', 'doi2']])
 
 
      def testInnerEdges(self):
-        nodes, edges = init_graph_construction(['doi_ie1'],1,1,True)
+        nodes, edges = init_graph_construction(['doi_ie1'],1,1,True,False)
         doi_nodes = keep_only_dois(nodes)
         self.assertCountEqual(doi_nodes,['doi_ie1','doi_ie2','doi_ie3'])
         self.assertCountEqual(edges,[['doi_ie1','doi_ie2'],['doi_ie3','doi_ie1'],['doi_ie3','doi_ie2']])
      
      def testRightHeight(self):
-          nodes, edges = init_graph_construction(['doi_h01'],1,0,True)
+          nodes, edges = init_graph_construction(['doi_h01'],0,1,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_h01'])
           self.assertCountEqual(edges, [])
 
-          nodes, edges = init_graph_construction(['doi_h02'],1,0,True)
+          nodes, edges = init_graph_construction(['doi_h02'],0,1,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_h02','doi_h1'])
           self.assertCountEqual(edges, [['doi_h1','doi_h02']])
 
-          nodes, edges = init_graph_construction(['doi_h02'],2,0,True)
+          nodes, edges = init_graph_construction(['doi_h02'],0,2,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_h02','doi_h1','doi_h2'])
           self.assertCountEqual(edges, [['doi_h1','doi_h02'], ['doi_h2','doi_h1']])
 
      def testRightDepth(self):
-          nodes, edges = init_graph_construction(['doi_d01'],0,1,True)
+          nodes, edges = init_graph_construction(['doi_d01'],1,0,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_d01'])
           self.assertCountEqual(edges, [])
 
-          nodes, edges = init_graph_construction(['doi_d02'],0,1,True)
+          nodes, edges = init_graph_construction(['doi_d02'],1,0,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_d02','doi_d1'])
           self.assertCountEqual(edges, [['doi_d02','doi_d1']])
 
-          nodes, edges = init_graph_construction(['doi_d02'],0,2,True)
+          nodes, edges = init_graph_construction(['doi_d02'],2,0,True,False)
           doi_nodes = keep_only_dois(nodes)
           self.assertCountEqual(doi_nodes,['doi_d02','doi_d1','doi_d2'])
           self.assertCountEqual(edges, [['doi_d02','doi_d1'], ['doi_d1','doi_d2']])
 
 
+     ## Ab hier die Tests für die einzelnen Funktionen ##
+
+     # initialize_graph.py:
+
+     def test_initialize_nodes_list(self):
+          references_pub_obj_list, citations_pub_obj_list = initialize_nodes_list_test(['doi_lg_1_i','doi_lg_2_i'], 0, 0, True)
+          doi_references = keep_only_dois(references_pub_obj_list)
+          doi_citations = keep_only_dois(citations_pub_obj_list)
+          self.assertCountEqual(doi_references, [])
+          self.assertCountEqual(doi_citations, [])
+
+          references_pub_obj_list, citations_pub_obj_list = initialize_nodes_list_test(['doi_lg_1_i','doi_lg_2_i'], 1, 1, True)
+          doi_references = keep_only_dois(references_pub_obj_list)
+          doi_citations = keep_only_dois(citations_pub_obj_list)
+          self.assertCountEqual(doi_references, ['doi_lg_1_d11','doi_lg_1_d12','doi_lg_2_d11','doi_lg_2_d12'])
+          self.assertCountEqual(doi_citations, ['doi_lg_1_h11','doi_lg_1_h12','doi_cg_i','doi_lg_2_h11','doi_lg_2_h12'])
+
+     def test_complete_inner_edges(self):
+          pub_lg_1_i = input_test_func('doi_lg_1_i')
+          pub_lg_1_i.group = 0
+          pub_lg_1_h_12 = input_test_func('doi_lg_1_h12')
+          pub_lg_1_h_12.group = 1
+          pub_lg_1_d_12 = input_test_func('doi_lg_1_d12')
+          pub_lg_1_d_12.group = -1
+          nodes = [pub_lg_1_i, pub_lg_1_h_12, pub_lg_1_d_12]
+          edges = [['doi_lg_1_i','doi_lg_1_d12'],['doi_lg_1_h12','doi_lg_1_i']]
+          processed_nodes, processed_edges = complete_inner_edges_test(nodes, edges)
+          self.assertCountEqual(processed_nodes, [pub_lg_1_i, pub_lg_1_h_12, pub_lg_1_d_12])
+          self.assertCountEqual(processed_edges, [['doi_lg_1_i','doi_lg_1_d12'],['doi_lg_1_h12','doi_lg_1_i'],['doi_lg_1_h12','doi_lg_1_d12']])
+
+     #    add_citations_rec.py:
+
+     def test_get_type_list(self):
+          pub_lg_1_i = input_test_func('doi_lg_1_i')
+          pub_lg_1_i.group = 0
+          self.assertEqual(get_cit_type_list(pub_lg_1_i, "Hallo"), ValueError)
+          
+          pub_lg_1_h_12 = input_test_func('doi_lg_1_h12')
+          pub_lg_1_h_12.group = 1
+          pub_lg_1_h_12_refs = get_cit_type_list(pub_lg_1_h_12, "Reference")
+          pub_lg_1_h_12_cits = get_cit_type_list(pub_lg_1_h_12, "Citation")
+          self.assertCountEqual(keep_only_dois(pub_lg_1_h_12_refs), keep_only_dois(pub_lg_1_h_12.references))
+          self.assertCountEqual(keep_only_dois(pub_lg_1_h_12_cits), keep_only_dois(pub_lg_1_h_12.citations))
+          
+          pub_lg_1_d_12 = input_test_func('doi_lg_1_d12')
+          pub_lg_1_d_12.group = -1
+          pub_lg_1_d_12_refs = get_cit_type_list(pub_lg_1_d_12, "Reference")
+          pub_lg_1_d_12_cits = get_cit_type_list(pub_lg_1_d_12, "Citation")
+          self.assertCountEqual(keep_only_dois(pub_lg_1_d_12_refs), keep_only_dois(pub_lg_1_d_12.references))
+          self.assertCountEqual(keep_only_dois(pub_lg_1_d_12_cits), keep_only_dois(pub_lg_1_d_12.citations))
+
+     def test_create_graph_structure_citations(self):
+          print("Hallo")
 
 
 def keep_only_dois(nodes):

verarbeitung/test/update_graph_unittest.py

@@ -20,17 +20,17 @@ class UpdatingTest(unittest.TestCase):
      #      self.assertCountEqual(nodes_old,nodes_new)
      #      self.assertCountEqual(edges_old, edges_new)
 
-     # def test_deleted_input_dois(self):
-     #      nodes_old_single, edges_old_single = init_graph_construction(['doi_lg_1_i'],2,2,True)
-     #      nodes_old_both, edges_old_both = init_graph_construction(['doi_lg_1_i','doi_lg_2_i'],2,2,True)
-     #      output_to_json(nodes_old_both, edges_old_both, test_var=True)
-     #      nodes_new_single, edges_new_single = update_graph(['doi_lg_1_i'], 'test_output.json', 2, 2, True)
-     #      self.assertCountEqual(nodes_old_single,nodes_new_single)
-     #      self.assertCountEqual(edges_old_single, edges_new_single)
-
-     #      nodes_old_single, edges_old_single = init_graph_construction(['doi_cg_i'],3,3,True)
-     #      nodes_old_two, edges_old_two = init_graph_construction(['doi_lg_1_i','doi_cg_i'],3,3,True)
-     #      nodes_old_three, edges_old_three = init_graph_construction(['doi_lg_1_i','doi_lg_2_i','doi_cg_i'],3,3,True)
+     def test_deleted_input_dois(self):
+          nodes_old_single, edges_old_single = init_graph_construction(['doi_lg_1_i'],2,2,True)
+          nodes_old_both, edges_old_both = init_graph_construction(['doi_lg_1_i','doi_lg_2_i'],2,2,True)
+          output_to_json(nodes_old_both, edges_old_both, test_var=True)
+          nodes_new_single, edges_new_single = update_graph(['doi_lg_1_i'], 'test_output.json', 2, 2, True)
+          self.assertCountEqual(nodes_old_single,nodes_new_single)
+          self.assertCountEqual(edges_old_single, edges_new_single)
+
+          nodes_old_single, edges_old_single = init_graph_construction(['doi_cg_i'],3,3,True)
+          nodes_old_two, edges_old_two = init_graph_construction(['doi_lg_1_i','doi_cg_i'],3,3,True)
+          nodes_old_three, edges_old_three = init_graph_construction(['doi_lg_1_i','doi_lg_2_i','doi_cg_i'],3,3,True)
 
      def test_new_height(self):
           nodes_height_0, edges_height_0 = init_graph_construction(['doi_lg_1_i'],2,0,True)

verarbeitung/test_output.json

-{"nodes": [{"doi": "doi_lg_1_i", "name": "title_lg_1_i", "author": ["contributor_lg_1_i"], "year": "date_lg_1_i", "journal": "journal_lg_1_i", "group": "Input", "depth": 0, "citations": 2}, {"doi": "doi_lg_1_d11", "name": "title_lg_1_d11", "author": ["contributor_lg_1_d11"], "year": "date_lg_1_d11", "journal": "journal_lg_1_d11", "group": "Reference", "depth": -1, "citations": 1}, {"doi": "doi_lg_1_d12", "name": "title_lg_1_d12", "author": ["contributor_lg_1_d12"], "year": "date_lg_1_d12", "journal": "journal_lg_1_d12", "group": "Reference", "depth": -1, "citations": 2}, {"doi": "doi_lg_1_h11", "name": "title_lg_1_h11", "author": ["contributor_lg_1_h11"], "year": "date_lg_1_h11", "journal": "journal_lg_1_h11", "group": "Citedby", "depth": 1, "citations": 2}, {"doi": "doi_lg_1_h12", "name": "title_lg_1_h12", "author": ["contributor_lg_1_h12"], "year": "date_lg_1_h12", "journal": "journal_lg_1_h12", "group": "Citedby", "depth": 1, "citations": 2}, {"doi": "doi_lg_2_i", "name": "title_lg_2_i", "author": ["contributor_lg_2_i"], "year": "date_lg_2_i", "journal": "journal_lg_2_i", "group": "Input", "depth": 0, "citations": 4}, {"doi": "doi_lg_2_d11", "name": "title_lg_2_d11", "author": ["contributor_lg_2_d11"], "year": "date_lg_2_d11", "journal": "journal_lg_2_d11", "group": "Reference", "depth": -1, "citations": 1}, {"doi": "doi_lg_2_d12", "name": "title_lg_2_d12", "author": ["contributor_lg_2_d12"], "year": "date_lg_2_d12", "journal": "journal_lg_2_d12", "group": "Reference", "depth": -1, "citations": 2}, {"doi": "doi_lg_2_h11", "name": "title_lg_2_h11", "author": ["contributor_lg_2_h11"], "year": "date_lg_2_h11", "journal": "journal_lg_2_h11", "group": "Citedby", "depth": 1, "citations": 2}, {"doi": "doi_lg_2_h12", "name": "title_lg_2_h12", "author": ["contributor_lg_2_h12"], "year": "date_lg_2_h12", "journal": "journal_lg_2_h12", "group": "Citedby", "depth": 1, "citations": 2}, {"doi": "doi_cg_i", "name": "title_cg_i", "author": ["contributor_cg_i"], "year": "date_cg_i", "journal": "journal_cg_i", "group": "Citedby", "depth": 1, "citations": 3}, {"doi": "doi_lg_1_h21", "name": "title_lg_1_h21", "author": ["contributor_lg_1_h21"], "year": "date_lg_1_h21", "journal": "journal_lg_1_h21", "group": "Citedby", "depth": 2, "citations": 0}, {"doi": "doi_lg_1_h22", "name": "title_lg_1_h22", "author": ["contributor_lg_1_h22"], "year": "date_lg_1_h22", "journal": "journal_lg_1_h22", "group": "Citedby", "depth": 2, "citations": 0}, {"doi": "doi_lg_1_h23", "name": "title_lg_1_h23", "author": ["contributor_lg_1_h23"], "year": "date_lg_1_h23", "journal": "journal_lg_1_h23", "group": "Citedby", "depth": 2, "citations": 0}, {"doi": "doi_lg_2_h21", "name": "title_lg_2_h21", "author": ["contributor_lg_2_h21"], "year": "date_lg_2_h21", "journal": "journal_lg_2_h21", "group": "Citedby", "depth": 2, "citations": 0}, {"doi": "doi_lg_2_h22", "name": "title_lg_2_h22", "author": ["contributor_lg_2_h22"], "year": "date_lg_2_h22", "journal": "journal_lg_2_h22", "group": "Citedby", "depth": 2, "citations": 0}, {"doi": "doi_lg_2_h23", "name": "title_lg_2_h23", "author": ["contributor_lg_2_h23"], "year": "date_lg_2_h23", "journal": "journal_lg_2_h23", "group": "Citedby", "depth": 2, "citations": 1}, {"doi": "doi_lg_2_h24", "name": "title_lg_2_h24", "author": ["contributor_lg_2_h24"], "year": "date_lg_2_h24", "journal": "journal_lg_2_h24", "group": "Citedby", "depth": 2, "citations": 1}, {"doi": "doi_cg_h11", "name": "title_cg_h11", "author": ["contributor_cg_h11"], "year": "date_cg_h11", "journal": "journal_cg_h11", "group": "Citedby", "depth": 2, "citations": 2}, {"doi": "doi_lg_1_d21", "name": "title_lg_1_d21", "author": ["contributor_lg_1_d21"], "year": "date_lg_1_d21", "journal": "journal_lg_1_d21", "group": "Reference", "depth": -2, "citations": 2}, {"doi": "doi_lg_1_d22", "name": "title_lg_1_d22", "author": ["contributor_lg_1_d22"], "year": "date_lg_1_d22", "journal": "journal_lg_1_d22", "group": "Reference", "depth": -2, "citations": 2}, {"doi": "doi_lg_1_d23", "name": "title_lg_1_d23", "author": ["contributor_lg_1_d23"], "year": "date_lg_1_d23", "journal": "journal_lg_1_d23", "group": "Reference", "depth": -2, "citations": 2}, {"doi": "doi_lg_2_d21", "name": "title_lg_2_d21", "author": ["contributor_lg_2_d21"], "year": "date_lg_2_d21", "journal": "journal_lg_2_d21", "group": "Reference", "depth": -2, "citations": 1}, {"doi": "doi_lg_2_d22", "name": "title_lg_2_d22", "author": ["contributor_lg_2_d22"], "year": "date_lg_2_d22", "journal": "journal_lg_2_d22", "group": "Reference", "depth": -2, "citations": 1}, {"doi": "doi_lg_2_d23", "name": "title_lg_2_d23", "author": ["contributor_lg_2_d23"], "year": "date_lg_2_d23", "journal": "journal_lg_2_d23", "group": "Reference", "depth": -2, "citations": 1}], "links": [{"source": "doi_lg_1_i", "target": "doi_lg_1_d11"}, {"source": "doi_lg_1_i", "target": "doi_lg_1_d12"}, {"source": "doi_lg_1_h11", "target": "doi_lg_1_i"}, {"source": "doi_lg_1_h12", "target": "doi_lg_1_i"}, {"source": "doi_lg_2_i", "target": "doi_lg_2_d11"}, {"source": "doi_lg_2_i", "target": "doi_lg_2_d12"}, {"source": "doi_lg_2_h11", "target": "doi_lg_2_i"}, {"source": "doi_lg_2_h12", "target": "doi_lg_2_i"}, {"source": "doi_cg_i", "target": "doi_lg_2_i"}, {"source": "doi_lg_1_h21", "target": "doi_lg_1_h11"}, {"source": "doi_lg_1_h22", "target": "doi_lg_1_h11"}, {"source": "doi_lg_1_h22", "target": "doi_lg_1_h12"}, {"source": "doi_lg_1_h23", "target": "doi_lg_1_h12"}, {"source": "doi_lg_2_h21", "target": "doi_lg_2_h11"}, {"source": "doi_lg_2_h22", "target": "doi_lg_2_h11"}, {"source": "doi_lg_2_h23", "target": "doi_lg_2_h12"}, {"source": "doi_lg_2_h24", "target": "doi_lg_2_h12"}, {"source": "doi_lg_2_h24", "target": "doi_lg_2_h23"}, {"source": "doi_lg_2_h23", "target": "doi_lg_2_h24"}, {"source": "doi_lg_1_h23", "target": "doi_cg_i"}, {"source": "doi_cg_h11", "target": "doi_cg_i"}, {"source": "doi_lg_2_h11", "target": "doi_cg_i"}, {"source": "doi_lg_1_d11", "target": "doi_lg_1_d21"}, {"source": "doi_lg_1_d11", "target": "doi_lg_1_d22"}, {"source": "doi_lg_1_d21", "target": "doi_lg_1_d22"}, {"source": "doi_lg_1_d22", "target": "doi_lg_1_d21"}, {"source": "doi_lg_1_d12", "target": "doi_lg_1_d23"}, {"source": "doi_lg_2_d11", "target": "doi_lg_2_i"}, {"source": "doi_lg_2_d11", "target": "doi_lg_2_d21"}, {"source": "doi_lg_2_d12", "target": "doi_lg_2_d22"}, {"source": "doi_lg_2_d12", "target": "doi_lg_2_d23"}, {"source": "doi_lg_1_h12", "target": "doi_lg_1_d12"}, {"source": "doi_lg_1_h11", "target": "doi_lg_1_h12"}, {"source": "doi_lg_2_h24", "target": "doi_lg_2_d12"}]}
\ No newline at end of file
+{"nodes": [{"doi": "doi_h02", "name": "title_h02", "author": ["contributor_h02"], "year": "date_h02", "journal": "journal_h02", "group": "Input", "depth": 0, "citations": 1}, {"doi": "doi_h1", "name": "title_h1", "author": ["contributor_h1"], "year": "date_h1", "journal": "journal_h1", "group": "Citedby", "depth": 1, "citations": 1}, {"doi": "doi_h2", "name": "title_h2", "author": ["contributor_h2"], "year": "date_h2", "journal": "journal_h2", "group": "Citedby", "depth": 2, "citations": 1}], "links": [{"source": "doi_h1", "target": "doi_h02"}, {"source": "doi_h2", "target": "doi_h1"}]}
\ No newline at end of file

verarbeitung/update_graph/connect_new_input.py

@@ -16,69 +16,138 @@ __status__ = "Production"
 import sys  
 from pathlib import Path
 from os import error
+
 sys.path.append("../")
 
-from .import_from_json import input_from_json
-from verarbeitung.construct_new_graph.initialize_graph import initialize_nodes_list, complete_inner_edges
-from verarbeitung.construct_new_graph.add_citations_rec import add_citations
-from verarbeitung.construct_new_graph.export_to_json import output_to_json
+from verarbeitung.construct_new_graph.initialize_graph import init_graph_construction
+from verarbeitung.construct_new_graph.add_citations_rec import add_citations, get_cit_type_list, create_global_lists_cit
+
+
+def find_furthermost_citations(node, old_search_depth, cit_type):
+    '''
+        :param node:                node which is known but not from input group
+        :type node:                 Publication
 
-def connect_old_and_new_input(json_file, new_doi_list, search_depth, search_height, test_var = False):
+        :param old_search_depth:    depth to search for references from old construction call
+        :type old_search_depth:     int
+
+        :param cit_type:            determines whether the function call is for a reference or citation
+        :type cit_type:             String
+
+        function to find the furthermost citation/reference for given node which is from the same group
+    '''
+    citations_saved = get_cit_type_list(node, cit_type)
+
+    # group of node and old search depth/height determines how often the loop needs to be repeated
+    for depth in range(old_search_depth - abs(node.group)):
+        new_citations = []
+        for citation in citations_saved:
+            for cit_node in nodes:
+                if citation.doi_url == cit_node.doi_url:
+                    for new_cit in get_cit_type_list(cit_node, cit_type):
+                        for new_cit_node in nodes:
+                            if new_cit.doi_url == new_cit_node.doi_url:
+                                new_cit_node.group -= node.group
+                                new_citations.append(new_cit_node)
+        citations_saved = new_citations
+
+    # returns the references/citations which needs to be processed to complete contruction
+    return(citations_saved)
+
+def complete_changed_group_nodes(inserted_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, test_var):
     '''
-        :param json_file:       json file with old graph
-        :type json_file:        json file
+        :param inserted_nodes:      list of nodes which are inserted to new input array  
+        :type inserted_nodes:       List[String]
 
-        :param new_doi_list:    additional dois which has to be connected to the old graph
-        :type new_doi_list:     list of strings
+        :param old_search_depth:    depth to search for references from old construction call
+        :type old_search_depth:     int
 
-        :param search_depth:    depth to search for references
-        :type search_depth:     int
+        :param old_search_height:   height to search for citations from old construction call
+        :type old_search_height:    int
 
-        :param search_height:   height to search for citations
-        :type search_height:    int
+        :param new_search_depth:    depth to search for references from new construction call
+        :type new_search_depth:     int
+
+        :param new_search_height:   height to search for citations from new construction call
+        :type new_search_height:    int
 
         :param test_var:            variable to differenciate between test and url call
         :type test_var:             boolean
 
-        connetcs the old graph and the new input dois to a complete new graph
+        completes the references and citations for nodes which were known in non input group
     '''
-    global nodes, edges
-    nodes = []
-    edges = []
+    #changed_group_node_citations = []
+    #changed_group_node_references = []
 
-    nodes, edges = input_from_json(json_file)
+    # saves which nodes were handled because they were known before
+    handled_inserted_nodes = []
+    for node in nodes:
 
-    complete_changed_group_nodes(new_doi_list, search_depth, search_height, test_var)
+        # moves known reference node to input and completes citations and references for this node
+        if (node.group < 0) and (node.doi_url in inserted_nodes):
+            node_citations = create_global_lists_cit(nodes, edges, node, 1, new_search_height, "Citation", test_var)
+            add_citations(nodes, edges, node_citations, 1, new_search_height, "Citation", test_var)
+            old_max_references = find_furthermost_citations(node, old_search_depth, "Reference")
+            add_citations(nodes, edges, old_max_references, old_search_depth, new_search_depth, "Reference", test_var)
+            node.group = 0
+            handled_inserted_nodes.append(node)
             
-    # initializes nodes/edges from input and gets a list with publication objects for citations and references returned
-    references_obj_list, citations_obj_list = initialize_nodes_list(new_doi_list,search_depth, search_height, test_var)
+        # moves known citation node to input and completes citations and references for this node
+        elif (node.group > 0) and (node.doi_url in inserted_nodes):
+            node_references = create_global_lists_cit(nodes, edges, node, 1, new_search_depth, "Reference", test_var)
+            add_citations(nodes, edges, node_references, 1, new_search_depth, "Reference", test_var)
+            old_max_citations = find_furthermost_citations(node, old_search_height, "Citation")
+            add_citations(nodes, edges, old_max_citations, old_search_height, new_search_height, "Citation", test_var)
 
-    # function calls to begin recursive processing up to max depth/height
-    add_citations(nodes, edges, citations_obj_list, 1, search_height, "Citation", test_var)
-    add_citations(nodes, edges, references_obj_list, 1, search_depth, "Reference", test_var)
+            node.group = 0
+            handled_inserted_nodes.append(node)
 
-    # adds edges between reference group and citation group of known publications
-    complete_inner_edges(test_var)
+    return(handled_inserted_nodes)
 
-    # calls a skript to save nodes and edges of graph in .json file
-    output_to_json(nodes,edges, test_var)
 
-    return(nodes, edges)
+def connect_old_and_new_input(input_nodes_list, input_edges_list, inserted_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, test_var = False):
+    '''
+        :param input_nodes_list:    list of nodes which are processed for new construction call
+        :type input_nodes_list:     List[Publication]
 
+        :param input_edges_list:    list of links between nodes from input_nodes_list.
+        :type input_edges_list:     List[List[String,String]]
 
-def complete_changed_group_nodes(new_doi_list, search_depth_max, search_height_max, test_var):
-    '''
-    work in progress
+        :param inserted_nodes:      list of nodes which are inserted to new input array  
+        :type inserted_nodes:       List[String]
+
+        :param old_search_depth:    depth to search for references from old construction call
+        :type old_search_depth:     int
+
+        :param old_search_height:   height to search for citations from old construction call
+        :type old_search_height:    int
+
+        :param new_search_depth:    depth to search for references from new construction call
+        :type new_search_depth:     int
+
+        :param new_search_height:   height to search for citations from new construction call
+        :type new_search_height:    int
+
+        :param test_var:            variable to differenciate between test and url call
+        :type test_var:             boolean
+
+        completes the references and citations for nodes which were known in non input group
     '''
-    changed_group_node_citations = []
-    changed_group_node_references = []
+    global nodes, edges
+    nodes = input_nodes_list
+    edges = input_edges_list
 
-    for node in nodes:
-        if (node.group < 0) and (node.doi in new_doi_list):
-            node.group = "input"
+    handled_inserted_nodes = complete_changed_group_nodes(inserted_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, test_var)
 
+    # copy all nodes from inserted_nodes to new node, if node is not in handled_inserted_nodes
+    not_handled_inserted_nodes = [node for node in inserted_nodes if node not in handled_inserted_nodes]
     
-        elif (node.group > 0) and (node.doi in new_doi_list):
-            node.group = "input"
+    # initializes nodes/edges from input and gets a list with publication objects for citations and references returned
+    #references_obj_list, citations_obj_list = initialize_nodes_list(not_handled_inserted_nodes, new_search_depth, new_search_height, test_var)
+
+    # function calls to begin recursive processing up to max depth/height
+    #add_citations(nodes, edges, citations_obj_list, 1, new_search_height, "Citation", test_var)
+    #add_citations(nodes, edges, references_obj_list, 1, new_search_depth, "Reference", test_var)
 
+    init_graph_construction(not_handled_inserted_nodes, new_search_depth, new_search_height, test_var = test_var, update_var = True, input_nodes = nodes, input_edges = edges)
 

verarbeitung/update_graph/delete_nodes_edges.py

@@ -90,6 +90,7 @@ def delete_nodes_and_edges(input_list, common_nodes, old_edges_list):
                 search_ref_graph_rec(pub)
                 search_cit_graph_rec(pub)
 
-    valid_edges = back_to_valid_edges(old_edges_list, usable_nodes)
-
-    return(usable_nodes, valid_edges)
\ No newline at end of file
+    back_to_valid_edges(old_edges_list, usable_nodes)
+    input_list = usable_nodes.copy()
+    print(input_list)
+    #return(usable_nodes, valid_edges)

verarbeitung/update_graph/update_depth.py

@@ -72,7 +72,10 @@ def get_old_max_references(old_depth):
     old_max_references = []
     for pub in processed_input_list:
         if (abs(pub.group) == old_depth):
-            old_max_references.append(pub.references)
+            for reference in pub.references:
+                for ref_pub in processed_input_list:
+                    if reference.doi_url == ref_pub.doi_url:
+                        old_max_references.append(ref_pub)
     return(old_max_references)
 
 def get_old_max_citations(old_height):
@@ -85,7 +88,10 @@ def get_old_max_citations(old_height):
     old_max_citations = []
     for pub in processed_input_list:
         if (abs(pub.group) == old_height):
-            old_max_citations.append(pub.citations)
+            for citation in pub.citations:
+                for cit_pub in processed_input_list:
+                    if citation.doi_url == cit_pub.doi_url:
+                        old_max_citations.append(cit_pub)
     return(old_max_citations)
 
 def update_depth(obj_input_list, input_edges, new_depth, new_height, test_var):
@@ -131,7 +137,8 @@ def update_depth(obj_input_list, input_edges, new_depth, new_height, test_var):
     back_to_valid_edges(valid_edges, processed_input_list)
 
     # adds edges between reference group and citation group of known publications
-    complete_inner_edges()
+
+    return(old_depth, old_height)
 
 
     

verarbeitung/update_graph/update_graph.py

-# -*- coding: utf-8 -*-
+1# -*- coding: utf-8 -*-
 """
 Functions to update a graph representing citations between multiple ACS/Nature journals
 
@@ -85,7 +85,6 @@ def update_graph(new_doi_input_list, json_file, search_depth, search_height, tes
 
     # gets information from previous cunstruction call
     old_obj_input_list , old_edges_list = input_from_json(json_file)
-    print(type(old_edges_list[1]))
 
     # one global list to save the process of removing unneeded publications and one to save valid edges
     global processed_list, valid_edges
@@ -102,11 +101,11 @@ def update_graph(new_doi_input_list, json_file, search_depth, search_height, tes
 
     # deletes publications and edges from node_list if publications can no longer be reached
     if (len(deleted_nodes) > 0):
-        processed_list, valid_edges = delete_nodes_and_edges(processed_list, common_nodes, old_edges_list)
+        delete_nodes_and_edges(processed_list, common_nodes, valid_edges)
     
-    update_depth(processed_list, valid_edges, search_depth, search_height, test_var)
+    old_search_depth, old_search_height = update_depth(processed_list, valid_edges, search_depth, search_height, test_var)
     
     if (len(inserted_nodes) > 0):
-        connect_old_and_new_input(json_file, inserted_nodes, search_depth, search_height, test_var)
+        connect_old_and_new_input(processed_list, valid_edges, inserted_nodes, old_search_depth, old_search_height, search_depth, search_height, test_var)
 
     return(processed_list, valid_edges)