new branch and finished connect_new_input

verarbeitung/construct_new_graph/initialize_graph.py

@@ -96,7 +96,7 @@ 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, input_nodes = [], input_edges = []):
     '''
         :param doi_input_list:  input list of doi from UI
         :type doi_input_list:   List[String]
@@ -128,8 +128,8 @@ def init_graph_construction(doi_input_list, search_depth, search_height, test_va
     
     # creates empty lists to save nodes and edges
     global nodes, edges
-    nodes = [] 
-    edges = [] 
+    nodes = input_nodes
+    edges = input_edges 
 
     # 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(doi_input_list,search_depth, search_height, test_var)

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("../")
 
@@ -42,13 +42,13 @@ 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)
-            
-    except IOError:
-        nodes, edges = init_graph_construction(url_list, search_depth, search_height)
+    json_file = Path(json_file)
     
+    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
     output_to_json(nodes, edges, json_file)
     
\ No newline at end of file

verarbeitung/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, 2, 2, 'test.json')

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 connect_old_and_new_input(json_file, new_doi_list, search_depth, search_height, test_var = False):
+def find_furthermost_citations(node, old_search_depth, cit_type):
     '''
-        :param json_file:       json file with old graph
-        :type json_file:        json file
+        :param node:                node which is known but not from input group
+        :type node:                 Publication
 
-        :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 cit_type:            determines whether the function call is for a reference or citation
+        :type cit_type:             String
 
-        :param search_height:   height to search for citations
-        :type search_height:    int
+        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_citations.append(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 inserted_nodes:      list of nodes which are inserted to new input array  
+        :type inserted_nodes:       List[String]
 
-        :param test_var:        variable to differenciate between test and url call
-        :type test_var:         boolean
+        :param old_search_depth:    depth to search for references from old construction call
+        :type old_search_depth:     int
 
-        connetcs the old graph and the new input dois to a complete new graph
-    '''
-    global nodes, edges
-    nodes = []
-    edges = []
+        :param old_search_height:   height to search for citations from old construction call
+        :type old_search_height:    int
 
-    nodes, edges = input_from_json(json_file)
+        :param new_search_depth:    depth to search for references from new construction call
+        :type new_search_depth:     int
 
-    complete_changed_group_nodes(new_doi_list, search_depth, search_height, test_var)
-    
-    # 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)
+        :param new_search_height:   height to search for citations from new construction call
+        :type new_search_height:    int
 
-    # 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)
+        :param test_var:            variable to differenciate between test and url call
+        :type test_var:             boolean
 
-    # adds edges between reference group and citation group of known publications
-    complete_inner_edges(test_var)
+        completes the references and citations for nodes which were known in non input group
+    '''
+    #changed_group_node_citations = []
+    #changed_group_node_references = []
 
-    # calls a skript to save nodes and edges of graph in .json file
-    output_to_json(nodes,edges, test_var)
+    # saves which nodes were handled because they were known before
+    handled_inserted_nodes = []
+    for node in nodes:
 
-    return(nodes, edges)
+        # 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)
+            
+        # 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)
 
+            node.group = 0
+            handled_inserted_nodes.append(node)
 
-def complete_changed_group_nodes(new_doi_list, search_depth_max, search_height_max, test_var):
+    return(handled_inserted_nodes)
+
+
+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):
     '''
-    work in progress
+        :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]]
+
+        :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)
 
-        elif (node.group > 0) and (node.doi in new_doi_list):
-            node.group = "input"
+    # 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]
+    
+    # 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, input_nodes = nodes, input_edges = edges)
 

verarbeitung/update_graph/update_depth.py

@@ -133,6 +133,8 @@ def update_depth(obj_input_list, input_edges, new_depth, new_height, test_var):
     # 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
 
@@ -20,6 +20,7 @@ sys.path.append("../../")
 
 from input.publication import Publication
 from verarbeitung.get_pub_from_input import get_pub
+from verarbeitung.construct_new_graph.initialize_graph import init_graph_construction
 from .Knoten_Vergleich import doi_listen_vergleichen
 from .delete_nodes_edges import delete_nodes_and_edges
 from .connect_new_input import connect_old_and_new_input
@@ -85,7 +86,6 @@ 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
@@ -104,9 +104,9 @@ def update_graph(new_doi_input_list, json_file, search_depth, search_height, tes
     if (len(deleted_nodes) > 0):
         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)
+    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)