From e7b96bb072d74dbb27de2b99cbd34eebe8f4141b Mon Sep 17 00:00:00 2001
From: Malte Schokolowski <baw8441@uni-hamburg.de>
Date: Mon, 13 Dec 2021 17:23:42 +0100
Subject: [PATCH] changed update_graph_delete function
---
verarbeitung/test/Processing_unittest.py | 2 +-
.../update_graph/connect_new_input.py | 2 +-
verarbeitung/update_graph/import_from_json.py | 2 +-
verarbeitung/update_graph/update_graph.py | 2 +-
verarbeitung/update_graph/update_graph_del.py | 104 ++++++------------
5 files changed, 39 insertions(+), 73 deletions(-)
diff --git a/verarbeitung/test/Processing_unittest.py b/verarbeitung/test/Processing_unittest.py
index b17f5e4..35cade7 100644
--- a/verarbeitung/test/Processing_unittest.py
+++ b/verarbeitung/test/Processing_unittest.py
@@ -120,4 +120,4 @@ def keep_only_dois(nodes):
if __name__ == "__main__":
- unittest.main()
\ No newline at end of file
+ unittest.main()
\ No newline at end of file
diff --git a/verarbeitung/update_graph/connect_new_input.py b/verarbeitung/update_graph/connect_new_input.py
index 4f6b91f..32af1fc 100644
--- a/verarbeitung/update_graph/connect_new_input.py
+++ b/verarbeitung/update_graph/connect_new_input.py
@@ -18,7 +18,7 @@ from pathlib import Path
from os import error
sys.path.append("../")
-from import_from_json import input_from_json
+from .import_from_json import input_from_json
from verarbeitung.construct_new_graph.Processing 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
diff --git a/verarbeitung/update_graph/import_from_json.py b/verarbeitung/update_graph/import_from_json.py
index 01a3886..f79e42b 100644
--- a/verarbeitung/update_graph/import_from_json.py
+++ b/verarbeitung/update_graph/import_from_json.py
@@ -16,7 +16,7 @@ __status__ = "Production"
import json
import sys
-sys.path.append("../../")
+sys.path.append("../")
from input.publication import Publication, Citation
diff --git a/verarbeitung/update_graph/update_graph.py b/verarbeitung/update_graph/update_graph.py
index 0571cd0..068304a 100644
--- a/verarbeitung/update_graph/update_graph.py
+++ b/verarbeitung/update_graph/update_graph.py
@@ -100,7 +100,7 @@ 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, deleted_nodes, old_doi_input_list, old_edges_list)
+ processed_input_list_del, valid_edges_del = delete_nodes_and_edges(processed_input_list, common_nodes, old_edges_list)
if (len(inserted_nodes) > 0):
connect_old_and_new_input(json_file, inserted_nodes, search_depth, search_height, test_var)
diff --git a/verarbeitung/update_graph/update_graph_del.py b/verarbeitung/update_graph/update_graph_del.py
index eaa25d6..5879d96 100644
--- a/verarbeitung/update_graph/update_graph_del.py
+++ b/verarbeitung/update_graph/update_graph_del.py
@@ -20,17 +20,18 @@ sys.path.append("../../")
from .Kanten_Vergleich import back_to_valid_edges
-def delete_ref_nodes_rec(pub):
+def search_ref_graph_rec(pub):
'''
- :param pub: pub go get deleted after recursive call
+ :param pub: pub go get appended to usable_nodes
:type pub: Publication
- function that removes nodes of group "height", if they aren't reachable from input nodes
+ function that appends nodes of group "reference" to list usable_nodes, if they are reachable from input nodes
'''
for reference in pub.references:
- for ref_pub in processed_list:
- if (ref_pub.doi_url == reference.doi_url):
-
+ for ref_pub in input_obj_list:
+ if ((reference.doi_url == ref_pub.doi_url) and (ref_pub not in usable_nodes)):
+ usable_nodes.append(ref_pub)
+
# to find a cyclus and avoid recursion error
not_in_citations = True
for citation in pub.citations:
@@ -38,23 +39,20 @@ def delete_ref_nodes_rec(pub):
not_in_citations = False
break
if (not_in_citations):
- delete_ref_nodes_rec(ref_pub)
-
- # removes publication from list after recursion and if it's not of group input
- if (pub.group != 0):
- processed_list.remove(pub)
+ search_ref_graph_rec(ref_pub)
-def delete_cit_nodes_rec(pub):
+def search_cit_graph_rec(pub):
'''
- :param pub: publication to be removed after recursive call
+ :param pub: pub go get appended to usable_nodes
:type pub: Publication
- function that removes publications of group "depth" from input_list, if they aren't reachable from input nodes
- '''
+ function that appends nodes of group "citation" to list usable_nodes, if they are reachable from input nodes
+ '''
for citation in pub.citations:
- for cit_pub in processed_list:
- if (cit_pub.doi_url == citation.doi_url):
+ for cit_pub in input_obj_list:
+ if ((citation.doi_url == cit_pub.doi_url) and (cit_pub not in usable_nodes)):
+ usable_nodes.append(cit_pub)
# to find a cyclus and avoid recursion error
not_in_references = True
@@ -63,68 +61,36 @@ def delete_cit_nodes_rec(pub):
not_in_references = False
break
if (not_in_references):
- delete_cit_nodes_rec(cit_pub)
+ search_cit_graph_rec(cit_pub)
- # removes publication from list after recursion and if it's not of group input
- if (pub.group != 0):
- processed_list.remove(pub)
-def delete_nodes_and_edges(input_list, deleted_nodes, old_doi_input_list, old_edges_list):
+def delete_nodes_and_edges(input_list, common_nodes, old_edges_list):
'''
:param input_list: list of publications to get reduced
:type input_list: List[Publication]
- :param deleted_nodes: list of input dois which are not in new call
- :type deleted_nodes: List[String]
-
- :param old_doi_input_list: list of input dois from old call
- :type old_doi_input_list: List[String]
+ :param common_nodes: list of input dois which are in old and new input call
+ :type common_nodes: List[String]
:param old_edges_list: list of links between publications from old call
:type old_edges_list: List[List[String,String]]
- function to start recursive node removal for references and citations and to return edge list to valid state
+ function to start recursive node removal for references and citations and to change edge list to valid state
'''
- # global list to save the process of removing unneeded publications
- global processed_list
- processed_list = input_list.copy()
-
- for del_node in deleted_nodes:
- for pub in processed_list:
- if (del_node == pub.doi_url):
-
- # checks for every reference if it is citet more than once. If not it calls deletion function
- for reference in pub.references:
- only_reference = True
- for ref_cit in processed_list:
- if (reference == ref_cit.doi_url):
- for citation in ref_cit.citations:
- if ((citation in old_doi_input_list) and (citation != del_node)):
- only_reference = False
- break
- if (only_reference == False):
- break
- if (only_reference):
- delete_ref_nodes_rec(pub)
-
- # checks for every citation if it cites more than once. If not it calls deletion function
- for citation in pub.citations:
- only_reference = True
- for cit_ref in processed_list:
- if (citation == cit_ref.doi_url):
- for reference in cit_ref.references:
- if ((reference in old_doi_input_list) and (reference != del_node)):
- only_reference = False
- break
- if (only_reference == False):
- break
- if (only_reference):
- delete_cit_nodes_rec(pub)
-
- # removes publication of type input after start of recursive call to both directions
- processed_list.remove(pub)
-
- valid_edges = back_to_valid_edges(old_edges_list, processed_list)
- return(processed_list, valid_edges)
\ No newline at end of file
+ global usable_nodes, input_obj_list
+ usable_nodes = []
+ input_obj_list = input_list.copy()
+
+ # starts for every common input node a tree-search and adds found nodes to usable_nodes
+ for common in common_nodes:
+ for pub in input_obj_list:
+ if (common == pub.doi_url):
+ usable_nodes.append(pub)
+ 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
--
GitLab