added height/depth adjustment

verarbeitung/.gitignore

@@ -54,4 +54,8 @@ coverage.xml
 docs/_build/
 
 # PyBuilder
-target/
\ No newline at end of file
+target/
+
+#CodeCounter
+
+.VSCodeCounter/
\ No newline at end of file

verarbeitung/construct_new_graph/add_citations_rec.py

@@ -184,4 +184,4 @@ def add_citations(input_nodes, input_edges, citations_pub_obj_list, search_depth
     edges = input_edges
 
     process_citations_rec(citations_pub_obj_list, search_depth, search_depth_max, cit_type, test_var)
-    return(nodes, edges)
\ No newline at end of file
+    #return(nodes, edges)
\ No newline at end of file

verarbeitung/construct_new_graph/export_to_json.py

@@ -60,7 +60,7 @@ def format_edges(edges):
     return list_of_edge_dicts
    
 
-def output_to_json(nodes, edges, test_var):
+def output_to_json(nodes, edges, json_file = 'json_text.json', test_var = False):
     '''
         :param nodes:       list of publications to export to json
         :type nodes:        List[Publication]
@@ -78,9 +78,9 @@ def output_to_json(nodes, edges, test_var):
     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):
+    if (test_var and json_file == 'json_text.json'):
         with open('test_output.json','w') as outfile:
             json.dump(dict_of_all, outfile)
     else:
-        with open('json_text.json','w') as outfile:
+        with open(json_file,'w') as outfile:
             json.dump(dict_of_all, outfile)

verarbeitung/construct_new_graph/initialize_graph.py

@@ -142,6 +142,6 @@ def init_graph_construction(doi_input_list, search_height, search_depth, test_va
     complete_inner_edges()
 
     # calls a skript to save nodes and edges of graph in .json file
-    output_to_json(nodes, edges, test_var)
+    #output_to_json(nodes, edges, test_var)
 
     return(nodes,edges)

verarbeitung/dev_files/print_graph_test.py

@@ -20,7 +20,7 @@ import sys
 sys.path.append("../../")
 from verarbeitung.construct_new_graph.initialize_graph import init_graph_construction
 from verarbeitung.update_graph.import_from_json import input_from_json
-from verarbeitung.update_graph.update_graph import check_graph_updates
+from verarbeitung.update_graph.update_graph import update_graph
 
 # a function to print nodes and edges from a graph
 def print_graph(nodes, edges):
@@ -93,7 +93,7 @@ def try_delete_nodes():
     # list_of_nodes_py, list_of_edges_py = input_from_json('json_text.json')
     # doi_list = []
     # doi_list.append('https://pubs.acs.org/doi/10.1021/acs.jcim.9b00249')
-    # valid_nodes, valid_edges = check_graph_updates(doi_list, list_of_nodes_py, list_of_edges_py)
+    # valid_nodes, valid_edges = update_graph(doi_list, list_of_nodes_py, list_of_edges_py)
     # print_simple(valid_nodes, valid_edges)
 
 def try_import():

verarbeitung/new_height.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_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_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}], "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_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_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_1_h12", "target": "doi_lg_1_d12"}, {"source": "doi_lg_1_h11", "target": "doi_lg_1_h12"}]}
\ No newline at end of file

verarbeitung/process_main.py

+# -*- coding: utf-8 -*-
+"""
+main function to call to generate a graph representing citations between multiple ACS/Nature journals
 
+"""
 
-def Processing(url):
-    print(url)
\ No newline at end of file
+__authors__ = "Donna Löding, Alina Molkentin, Xinyi Tang, Judith Große, Malte Schokolowski"
+__email__ = "cis-project2021@zbh.uni-hamburg.de"
+__status__ = "Production"
+#__copyright__ = ""
+#__credits__ = ["", "", "", ""]
+#__license__ = ""
+#__version__ = ""
+#__maintainer__ = ""
+
+
+import sys  
+from pathlib import Path
+from os import error
+
+sys.path.append("../")
+
+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.update_graph.update_graph import update_graph
+
+def Processing(url_list, search_depth, search_height, json_file = 'json_text.json'):
+    '''
+        :param url_list:        list of urls to construct publication graph for
+        :type url_list:         List[String]
+        
+        :param search_depth:    maximum depth to search for references
+        :type search_depth:     int
+
+        :param search_height:   maximum height to search for citations
+        :type search_height:    int
+
+        :param json_file:       file to export graph to
+        :type json_file:        String
+
+        main function to construct new or updated publication graphs
+    '''
+
+    # 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)
+            
+    except IOError:
+        nodes, edges = init_graph_construction(url_list, search_depth, search_height)
+    
+    # exports graph to given json file name
+    output_to_json(nodes, edges, json_file)
+    
\ No newline at end of file

verarbeitung/start_script.py

 import sys
 from pathlib import Path
+from verarbeitung.process_main import Processing
 from verarbeitung.dev_files.print_graph_test import try_known_publications, try_delete_nodes
 
-try_delete_nodes()
\ No newline at end of file
+
+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, 2, 2, 'test.json')
\ No newline at end of file

verarbeitung/test/update_graph_unittest.py

@@ -2,17 +2,20 @@ 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.export_to_json import output_to_json
 from verarbeitung.update_graph.import_from_json import input_from_json
-from verarbeitung.update_graph.update_graph import check_graph_updates
+from verarbeitung.update_graph.update_graph import update_graph
 
 class UpdatingTest(unittest.TestCase):
      maxDiff = None
 
      def test_import_from_json(self):
           nodes_old, edges_old = init_graph_construction(['doi_lg_1_i'],2,2,True)
+          output_to_json(nodes_old, edges_old, test_var = True)
           nodes_new, edges_new = input_from_json('test_output.json')
           self.assertCountEqual(nodes_old,nodes_new)
           self.assertCountEqual(edges_old, edges_new)
@@ -20,8 +23,8 @@ class UpdatingTest(unittest.TestCase):
      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)
-          nodes_new_both, edges_new_both = input_from_json('test_output.json')
-          nodes_new_single, edges_new_single = check_graph_updates(['doi_lg_1_i'], nodes_old_both, edges_old_both, 'test_output.json', 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)
 
@@ -29,6 +32,29 @@ class UpdatingTest(unittest.TestCase):
           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)
+          nodes_height_1, edges_height_1 = init_graph_construction(['doi_lg_1_i'],2,1,True)
+          nodes_height_2, edges_height_2 = init_graph_construction(['doi_lg_1_i'],2,2,True)
+
+          output_to_json(nodes_height_2, edges_height_2, 'new_height.json', True)
+          nodes_new_height_1, edges_new_height_1 = update_graph(['doi_lg_1_i'], 'new_height.json', 2, 1, True)
+          self.assertCountEqual(nodes_height_1, nodes_new_height_1)
+          self.assertCountEqual(edges_height_1, edges_new_height_1)
+
+          nodes_height_2, edges_height_2 = init_graph_construction(['doi_lg_1_i'],2,2,True)
+          output_to_json(nodes_height_2, edges_height_2, 'new_height.json', True)
+          nodes_new_height_0, edges_new_height_0 = update_graph(['doi_lg_1_i'], 'new_height.json', 2, 0, True)
+          self.assertCountEqual(nodes_height_0, nodes_new_height_0)
+          self.assertCountEqual(edges_height_0, edges_new_height_0)
+
+
+
+
+
+          
+
+
 
 def keep_only_dois(nodes):
      '''
@@ -36,7 +62,7 @@ def keep_only_dois(nodes):
           :type nodes:   List[Publication]
 
           gets nodes of type pub and return only their doi
-    '''
+     '''
      doi_list = []
      for node in nodes:
           doi_list.append(node.doi_url)

verarbeitung/update_graph/update_graph_del.py → verarbeitung/update_graph/delete_nodes_edges.py

@@ -78,10 +78,9 @@ def delete_nodes_and_edges(input_list, common_nodes, old_edges_list):
     
     function to start recursive node removal for references and citations and to change edge list to valid state
     '''
-
     global usable_nodes, input_obj_list
     usable_nodes = []
-    input_obj_list = input_list.copy()
+    input_obj_list = input_list
 
     # starts for every common input node a tree-search and adds found nodes to usable_nodes
     for common in common_nodes:

verarbeitung/update_graph/import_from_json.py

@@ -34,7 +34,8 @@ def create_pubs_from_json(input_dict):
     for node in input_dict["nodes"]:         
         #creates for the nodes the objects class Publication
         
-        pub = Publication(node["doi"], node["name"], node["author"], node["journal"], node["year"], node["depth"] )
+        pub = Publication(node["doi"], node["name"], node["author"], node["journal"], node["year"], [])
+        pub.group = node["depth"]
         #appends the objects to a list
         list_of_nodes_py.append(pub) 
 

verarbeitung/update_graph/update_depth.py

+# -*- coding: utf-8 -*-
+"""
+Functions to update the citation depth of recursive graph construction
+
+"""
+
+__authors__ = "Donna Löding, Alina Molkentin, Xinyi Tang, Judith Große, Malte Schokolowski"
+__email__ = "cis-project2021@zbh.uni-hamburg.de"
+__status__ = "Production"
+#__copyright__ = ""
+#__credits__ = ["", "", "", ""]
+#__license__ = ""
+#__version__ = ""
+#__maintainer__ = ""
+
+import sys  
+sys.path.append("../../")
+
+from verarbeitung.construct_new_graph.add_citations_rec import add_citations
+from verarbeitung.construct_new_graph.initialize_graph import complete_inner_edges
+from .Kanten_Vergleich import back_to_valid_edges
+
+
+def reduce_max_depth_height(max_depth_height):
+    '''
+        :param max_depth_height:        new maximum depth/height to reduce publications in publication list to
+        :type max_depth_height:         int
+
+        function to remove all publications which are not in new maximum depth/height threshold
+    '''
+    for pub in processed_input_list:
+        if (abs(pub.group) > max_depth_height):
+            processed_input_list.remove(pub)
+
+def get_old_height_depth():
+    '''
+        function to get old max height and max depth from previous construction call
+    '''
+    max_height = 0 
+    max_depth = 0
+    for pub in processed_input_list:
+        if (pub.group < 0):
+            max_depth = max(max_depth, abs(pub.group))
+        if (pub.group > 0):
+            max_height = max(max_height, pub.group)
+    return(max_height, max_depth)
+
+def get_old_max_references(old_depth):
+    '''
+        :param old_depth:       old maximum depth to search for citations
+        :type old_depth:        int
+
+        function to get references for new recursive levels
+    '''
+    old_max_references = []
+    for pub in processed_input_list:
+        if (abs(pub.group) == old_depth):
+            old_max_references.append(pub.references)
+    return(old_max_references)
+
+def get_old_max_citations(old_height):
+    '''
+        :param old_height:      old maximum height to search for citations
+        :type old_height:       int
+
+        function to get citations for new recursive levels
+    '''
+    old_max_citations = []
+    for pub in processed_input_list:
+        if (abs(pub.group) == old_height):
+            old_max_citations.append(pub.citations)
+    return(old_max_citations)
+
+def update_depth(obj_input_list, input_edges, new_depth, new_height, test_var):
+    '''
+        :param obj_input_list:  input list of publications of type Publication from update_graph
+        :type obj_input_list:   List[Publication]
+
+        :param input_edges:     list of publications from update_graph
+        :type input_edges:      List[Publication]
+
+        :param new_depth:       new maximum depth to search for references
+        :type new_depth:        int
+
+        :param new_height:      new maximum height to search for citations
+        :type new_height:       int
+
+        :param test_var:        variable to differenciate between test and url call
+        :type test_var:         boolean
+
+        function to adjust old publication search depth to update call
+    '''
+
+    global processed_input_list, valid_edges
+    processed_input_list = obj_input_list
+    valid_edges = input_edges
+
+    old_height, old_depth = get_old_height_depth()
+
+    # removes publications and links from recursion levels which aren't needed anymore
+    if (old_depth > new_depth):
+        reduce_max_depth_height(new_depth)
+        valid_edges = back_to_valid_edges(processed_input_list, input_edges)
+    elif (old_height > new_height):
+        reduce_max_depth_height(new_height)
+        valid_edges = back_to_valid_edges(processed_input_list, valid_edges)
+    
+    # adds publications and links for new recursion levels
+    elif (old_depth < new_depth):
+        old_max_references = get_old_max_references()
+        add_citations(processed_input_list, valid_edges, old_max_references, old_depth+1, new_depth, "Reference", test_var)
+    elif (old_height < new_height):
+        old_max_citations = get_old_max_citations()
+        add_citations(processed_input_list, valid_edges, old_max_citations, old_height+1, new_height, "Citation", test_var)
+
+    # adds edges between reference group and citation group of known publications
+    complete_inner_edges()
+
+
+    
+

verarbeitung/update_graph/update_graph.py

@@ -14,16 +14,17 @@ __status__ = "Production"
 #__maintainer__ = ""
 
 
-import sys  
-from pathlib import Path
-from os import error
+import sys
+
 sys.path.append("../../")
 
 from input.publication import Publication
 from verarbeitung.get_pub_from_input import get_pub
 from .Knoten_Vergleich import doi_listen_vergleichen
-from .update_graph_del import delete_nodes_and_edges
+from .delete_nodes_edges import delete_nodes_and_edges
 from .connect_new_input import connect_old_and_new_input
+from .update_depth import update_depth
+from .import_from_json import input_from_json
 
 
 def get_old_input_dois(old_obj_input_list):
@@ -65,10 +66,10 @@ def get_new_input_dois(new_input, test_var):
     return(new_input_dois)
 
 
-def check_graph_updates(new_doi_input_list, old_obj_input_list, old_edges_list, json_file, search_depth, search_height, test_var = False):
+def update_graph(new_doi_input_list, json_file, search_depth, search_height, test_var = False):
     '''
         :param new_doi_input_list:  input list of doi from UI
-        :type new_doi_input_list:   list of strings
+        :type new_doi_input_list:   List[String]
 
         :param old_obj_input_list:  list of publications retrieved from old json file
         :type old_obj_input_list:   List[Publication]
@@ -82,14 +83,14 @@ def check_graph_updates(new_doi_input_list, old_obj_input_list, old_edges_list,
         function to compare old and new input, start node/edge removal and to return updated sets of nodes and edges
     '''
 
+    # gets information from previous cunstruction call
+    old_obj_input_list , old_edges_list = input_from_json(json_file)
+
     # one global list to save the process of removing unneeded publications and one to save valid edges
-    global processed_input_list, valid_edges
-    processed_input_list = old_obj_input_list.copy()
+    global processed_list, valid_edges
+    processed_list = old_obj_input_list
     valid_edges = []
 
-    # save the return values of global lists
-    processed_input_list_del = []
-    valid_edges_del = []
 
     # get dois from lists to compare for differences
     old_doi_input_list = get_old_input_dois(old_obj_input_list)
@@ -100,9 +101,11 @@ def check_graph_updates(new_doi_input_list, old_obj_input_list, old_edges_list,
 
     # deletes publications and edges from node_list if publications can no longer be reached
     if (len(deleted_nodes) > 0):
-        processed_input_list_del, valid_edges_del = delete_nodes_and_edges(processed_input_list, common_nodes, old_edges_list)
-
+        processed_list, valid_edges = delete_nodes_and_edges(processed_list, common_nodes, old_edges_list)
+    
+    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)
 
-    return(processed_input_list_del, valid_edges_del) 
+    return(processed_list, valid_edges)