From 071b501172cdbfb7df05f4741a4303cd2a1758bf Mon Sep 17 00:00:00 2001
From: Malte Schokolowski <baw8441@uni-hamburg.de>
Date: Wed, 12 Jan 2022 01:41:53 +0100
Subject: [PATCH] fixed bug in verarbeitung.update_graph
---
.../construct_new_graph/initialize_graph.py | 6 +-
verarbeitung/test/input_test.py | 2 +-
verarbeitung/test/update_graph_unittest.py | 174 +++++++++++++++---
.../update_graph/connect_new_input.py | 121 ++++++++----
verarbeitung/update_graph/update_graph.py | 35 +++-
5 files changed, 272 insertions(+), 66 deletions(-)
diff --git a/verarbeitung/construct_new_graph/initialize_graph.py b/verarbeitung/construct_new_graph/initialize_graph.py
index 3807842..1368f6e 100644
--- a/verarbeitung/construct_new_graph/initialize_graph.py
+++ b/verarbeitung/construct_new_graph/initialize_graph.py
@@ -190,7 +190,5 @@ def init_graph_construction(doi_input_list, search_depth, search_height, test_va
del edges
gc.collect()
- if test_var:
- return(new_nodes, new_edges)
- else:
- return(new_nodes, new_edges, error_doi_list)
+
+ return(new_nodes, new_edges, error_doi_list)
diff --git a/verarbeitung/test/input_test.py b/verarbeitung/test/input_test.py
index a701f7e..352f983 100644
--- a/verarbeitung/test/input_test.py
+++ b/verarbeitung/test/input_test.py
@@ -100,5 +100,5 @@ list_of_arrays = [beispiel1, beispiel2, beispiel3, zyklus1, zyklus2, inner_ed
right_height01, right_height02, right_height1, right_height2, right_height3, right_depth01, right_depth02, right_depth1, right_depth2, right_depth3,
large_graph_1_h21, large_graph_1_h22, large_graph_1_h23, large_graph_1_h11, large_graph_1_h12, large_graph_1_i, large_graph_1_d11, large_graph_1_d12,
large_graph_1_d21, large_graph_1_d22, large_graph_1_d23, large_graph_2_h21, large_graph_2_h22, large_graph_2_h23, large_graph_2_h24, large_graph_2_h11, large_graph_2_h12,
- large_graph_2_i, large_graph_2_d11, large_graph_2_d12, large_graph_2_d21, large_graph_2_d22, large_graph_2_d23, crossed_graph_h21, crossed_graph_h22, crossed_graph_h11,
+ large_graph_2_i, large_graph_2_d11, large_graph_2_d12, large_graph_2_d21, large_graph_2_d22, large_graph_2_d23, large_graph_2_d24, crossed_graph_h21, crossed_graph_h22, crossed_graph_h11,
crossed_graph_i, crossed_graph_d11, crossed_graph_d12, crossed_graph_d21, crossed_graph_d22]
diff --git a/verarbeitung/test/update_graph_unittest.py b/verarbeitung/test/update_graph_unittest.py
index 765fcbf..77746ff 100644
--- a/verarbeitung/test/update_graph_unittest.py
+++ b/verarbeitung/test/update_graph_unittest.py
@@ -14,40 +14,74 @@ from verarbeitung.update_graph.update_depth import reduce_max_height_depth_test,
from verarbeitung.update_graph.update_edges import back_to_valid_edges
from verarbeitung.update_graph.delete_nodes_edges import search_ref_cit_graph_rec_test
from verarbeitung.update_graph.compare_old_and_new_node_lists import compare_old_and_new_node_lists
-from verarbeitung.update_graph.connect_new_input import find_furthermost_citations_test
+from verarbeitung.update_graph.connect_new_input import find_furthermost_citations_test, complete_changed_group_nodes_test
from verarbeitung.get_pub_from_input import input_test_func
class UpdatingTest(unittest.TestCase):
maxDiff = None
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_old_single, edges_old_single, err_list = init_graph_construction(['doi_lg_1_i'],2,2,True)
+ nodes_old_both, edges_old_both, err_list = 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)
+ nodes_new_single, edges_new_single, err_list = 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)
+ nodes_old_single, edges_old_single, err_list = init_graph_construction(['doi_cg_i'],3,3,True)
+ nodes_old_two, edges_old_two, err_list = init_graph_construction(['doi_lg_1_i','doi_cg_i'],3,3,True)
+ nodes_old_three, edges_old_three, err_list = 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)
+ nodes_height_0, edges_height_0, err_list = init_graph_construction(['doi_lg_1_i'],2,0,True)
+ nodes_height_1, edges_height_1, err_list = init_graph_construction(['doi_lg_1_i'],2,1,True)
+ nodes_height_2, edges_height_2, err_list = 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)
+ nodes_new_height_1, edges_new_height_1, err_list = 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)
+ nodes_height_2, edges_height_2, err_list = 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)
+ nodes_new_height_0, edges_new_height_0, err_list = 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 test_ref_to_input(self):
+ nodes, edges, err_list = init_graph_construction(['doi_cg_i'], 2, 2, True)
+ nodes_2, edges_2, err_list_2 = init_graph_construction(['doi_cg_d11'], 2, 2, True)
+ output_to_json(nodes, edges, 'ref_to_input.json')
+ new_nodes, new_edges, new_err_list = update_graph(['doi_cg_d11'], 'ref_to_input.json', 2, 2, True)
+ self.assertCountEqual(new_nodes, nodes_2)
+ self.assertCountEqual(new_edges, edges_2)
+
+ nodes, edges, err_list = init_graph_construction(['doi_cg_i','doi_lg_2_i'], 2, 2, True)
+ nodes_2, edges_2, err_list_2 = init_graph_construction(['doi_cg_d11','doi_lg_2_i'], 2, 2, True)
+
+ output_to_json(nodes, edges, 'ref_to_input.json')
+ new_nodes, new_edges, new_err_list = update_graph(['doi_cg_d11','doi_lg_2_i'], 'ref_to_input.json', 2, 2, True)
+ self.assertCountEqual(new_nodes, nodes_2)
+ self.assertCountEqual(new_edges, edges_2)
+
+ output_to_json(nodes_2, edges_2, 'ref_to_input.json')
+ new_nodes, new_edges, new_err_list = update_graph(['doi_cg_d11','doi_lg_2_i'], 'ref_to_input.json', 2, 2, True)
+ self.assertCountEqual(new_nodes, nodes_2)
+ self.assertCountEqual(new_edges, edges_2)
+
+ nodes, edges, err_list = init_graph_construction(['doi_cg_i','doi_lg_2_i'], 2, 2, True)
+ nodes_2, edges_2, err_list_2 = init_graph_construction(['doi_cg_i','doi_lg_2_h11','doi_lg_1_i'], 3, 3, True)
+
+ output_to_json(nodes_2, edges_2, 'ref_to_input.json')
+ new_nodes, new_edges, new_err_list = update_graph(['doi_cg_i','doi_lg_2_i'], 'ref_to_input.json', 2, 2, True)
+ self.assertCountEqual(new_nodes, nodes)
+ self.assertCountEqual(new_edges, edges)
+
+
+
+
+
+
## Ab hier die Tests für die einzelnen Funktionen ##
# update_graph.py:
@@ -137,7 +171,7 @@ class UpdatingTest(unittest.TestCase):
# import_from_json.py:
def test_input_from_json(self):
- nodes_old, edges_old = init_graph_construction(['doi_lg_1_i'],2,2,True)
+ nodes_old, edges_old, err_list = 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)
@@ -223,11 +257,11 @@ class UpdatingTest(unittest.TestCase):
pub_lg_2_d22.group = -2
pubs = [pub_lg_2_i, pub_lg_2_h11, pub_lg_2_h12, pub_lg_2_d11, pub_lg_2_d12, pub_lg_2_h21, pub_lg_2_h22, pub_lg_2_d21, pub_lg_2_d22]
- self.assertCountEqual(find_furthermost_citations_test(pubs, pub_lg_2_h11, 2, "Citation"),[pub_lg_2_h21, pub_lg_2_h22])
- self.assertCountEqual(find_furthermost_citations_test(pubs, pub_lg_2_h11, 1, "Citation"),[pub_lg_2_h11])
+ self.assertCountEqual(find_furthermost_citations_test(pubs, [], pub_lg_2_h11, 2, "Citation"),[pub_lg_2_h21, pub_lg_2_h22])
+ self.assertCountEqual(find_furthermost_citations_test(pubs, [], pub_lg_2_h11, 1, "Citation"),[pub_lg_2_h11])
- self.assertCountEqual(find_furthermost_citations_test(pubs, pub_lg_2_d11, 2, "Reference"),[pub_lg_2_d21, pub_lg_2_i])
- self.assertCountEqual(find_furthermost_citations_test(pubs, pub_lg_2_d11, 1, "Reference"),[pub_lg_2_d11])
+ self.assertCountEqual(find_furthermost_citations_test(pubs, [], pub_lg_2_d11, 2, "Reference"),[pub_lg_2_d21, pub_lg_2_i])
+ self.assertCountEqual(find_furthermost_citations_test(pubs, [], pub_lg_2_d11, 1, "Reference"),[pub_lg_2_d11])
def test_complete_changed_group_nodes(self):
@@ -239,11 +273,109 @@ class UpdatingTest(unittest.TestCase):
pub_cg_h21.group = 2
pub_cg_h22 = input_test_func('doi_cg_h22')
pub_cg_h22.group = 2
- pubs = []
+ pub_cg_d11 = input_test_func('doi_cg_d11')
+ pub_cg_d11.group = -1
+ pub_cg_d12 = input_test_func('doi_cg_d12')
+ pub_cg_d12.group = -1
+ pub_cg_d21 = input_test_func('doi_cg_d21')
+ pub_cg_d21.group = -2
+ pub_cg_d22 = input_test_func('doi_cg_d22')
+ pub_cg_d22.group = -2
+ pub_lg_1_h23 = input_test_func('doi_lg_1_h23')
+ pub_lg_1_h23.group = 2
+ pub_lg_1_d23 = input_test_func('doi_lg_1_d23')
+ pub_lg_1_d23.group = -2
-
-
+ pub_lg_2_i = input_test_func('doi_lg_2_i')
+ pub_lg_2_i.group = 0
+ pub_lg_2_h11 = input_test_func('doi_lg_2_h11')
+ pub_lg_2_h11.group = 1
+ pub_lg_2_h21 = input_test_func('doi_lg_2_h21')
+ pub_lg_2_h21.group = 2
+ pub_lg_2_h22 = input_test_func('doi_lg_2_h22')
+ pub_lg_2_h22.group = 2
+ pub_lg_2_d11 = input_test_func('doi_lg_2_d11')
+ pub_lg_2_d11.group = -1
+ pub_lg_2_d12 = input_test_func('doi_lg_2_d12')
+ pub_lg_2_d12.group = -1
+ pub_lg_2_d21 = input_test_func('doi_lg_2_d21')
+ pub_lg_2_d21.group = -2
+ pub_lg_2_d22 = input_test_func('doi_lg_2_d22')
+ pub_lg_2_d22.group = -2
+ pub_lg_2_d23 = input_test_func('doi_lg_2_d23')
+ pub_lg_2_d23.group = -2
+ pub_lg_2_d24 = input_test_func('doi_lg_2_d24')
+ pub_lg_2_d24.group = -2
+
+
+ moved_1_pub_cg_i = input_test_func('doi_cg_i')
+ moved_1_pub_cg_i.group = 1
+ moved_1_pub_cg_h11 = input_test_func('doi_cg_h11')
+ moved_1_pub_cg_h11.group = 2
+ moved_1_pub_cg_h21 = input_test_func('doi_cg_h21')
+ moved_1_pub_cg_h21.group = 3
+ moved_1_pub_cg_h22 = input_test_func('doi_cg_h22')
+ moved_1_pub_cg_h22.group = 3
+ moved_1_pub_cg_d11 = input_test_func('doi_cg_d11')
+ moved_1_pub_cg_d11.group = 0
+
+ moved_1_pub_cg_d21 = input_test_func('doi_cg_d21')
+ moved_1_pub_cg_d21.group = -1
+
+ moved_1_pub_lg_1_h23 = input_test_func('doi_lg_1_h23')
+ moved_1_pub_lg_1_h23.group = 2
+ moved_1_pub_lg_1_d23 = input_test_func('doi_lg_1_d23')
+ moved_1_pub_lg_1_d23.group = -1
+
+ moved_1_pub_lg_2_h11 = input_test_func('doi_lg_2_h11')
+ moved_1_pub_lg_2_h11.group = 1
+
+
+ moved_2_pub_cg_i = input_test_func('doi_cg_i')
+ moved_2_pub_cg_i.group = -1
+ moved_2_pub_cg_d11 = input_test_func('doi_cg_d11')
+ moved_2_pub_cg_d11.group = -2
+ moved_2_pub_cg_d12 = input_test_func('doi_cg_d12')
+ moved_2_pub_cg_d12.group = -2
+ moved_2_pub_cg_d21 = input_test_func('doi_cg_d21')
+ moved_2_pub_cg_d21.group = -3
+ moved_2_pub_cg_d22 = input_test_func('doi_cg_d22')
+ moved_2_pub_cg_d22.group = -3
+
+ moved_2_pub_lg_1_d23 = input_test_func('doi_lg_1_d23')
+ moved_2_pub_lg_1_d23.group = -3
+
+ moved_2_pub_lg_2_h21 = input_test_func('doi_lg_2_h21')
+ moved_2_pub_lg_2_h21.group = 1
+ moved_2_pub_lg_2_h22 = input_test_func('doi_lg_2_h22')
+ moved_2_pub_lg_2_h22.group = 1
+ moved_2_pub_lg_2_h11 = input_test_func('doi_lg_2_h11')
+ moved_2_pub_lg_2_h11.group = 0
+ moved_2_pub_lg_2_i = input_test_func('doi_lg_2_i')
+ moved_2_pub_lg_2_i.group = -2
+ moved_2_pub_lg_2_d11 = input_test_func('doi_lg_2_d11')
+ moved_2_pub_lg_2_d11.group = -2
+ moved_2_pub_lg_2_d12 = input_test_func('doi_lg_2_d12')
+ moved_2_pub_lg_2_d12.group = -2
+ moved_2_pub_lg_2_d21 = input_test_func('doi_lg_2_d21')
+ moved_2_pub_lg_2_d21.group = -3
+ moved_2_pub_lg_2_d22 = input_test_func('doi_lg_2_d22')
+ moved_2_pub_lg_2_d22.group = -3
+ moved_2_pub_lg_2_d23 = input_test_func('doi_lg_2_d23')
+ moved_2_pub_lg_2_d23.group = -3
+ moved_2_pub_lg_2_d24 = input_test_func('doi_lg_2_d24')
+ moved_2_pub_lg_2_d24.group = -3
+
+ pubs = [pub_cg_i, pub_cg_h11, pub_cg_h21, pub_cg_h22, pub_cg_d11, pub_cg_d12, pub_cg_d21, pub_cg_d22, pub_lg_1_h23, pub_lg_1_d23, pub_lg_2_h21, pub_lg_2_h22, pub_lg_2_h11, pub_lg_2_i, pub_lg_2_d11, pub_lg_2_d12, pub_lg_2_d21, pub_lg_2_d22, pub_lg_2_d23, pub_lg_2_d24]
+ edges = []
+ nodes, edges, handled_nodes = complete_changed_group_nodes_test(pubs, edges, 'doi_cg_d11', 2, 2, 2, 2)
+ self.assertCountEqual(nodes, [moved_1_pub_cg_d11, moved_1_pub_cg_d21, moved_1_pub_lg_1_d23, moved_1_pub_cg_i, moved_1_pub_lg_1_h23, moved_1_pub_cg_h11, moved_1_pub_lg_2_h11])
+ self.assertCountEqual(edges, [['doi_cg_d11','doi_lg_1_d23'],['doi_cg_d11','doi_cg_d21'],['doi_cg_i','doi_cg_d11'],['doi_lg_1_h23','doi_cg_i'],['doi_cg_h11','doi_cg_i'],['doi_lg_2_h11','doi_cg_i']])
+
+ nodes, edges, handled_nodes = complete_changed_group_nodes_test(pubs, edges, 'doi_lg_2_h11', 2, 2, 3, 3)
+ self.assertCountEqual(nodes, [moved_2_pub_cg_i, moved_2_pub_cg_d11, moved_2_pub_lg_1_d23, moved_2_pub_cg_d21, moved_2_pub_cg_d12, moved_2_pub_cg_d22, moved_2_pub_lg_2_h21, moved_2_pub_lg_2_h22, moved_2_pub_lg_2_h11, moved_2_pub_lg_2_i, moved_2_pub_lg_2_d11, moved_2_pub_lg_2_d21, moved_2_pub_lg_2_d12, moved_2_pub_lg_2_d22, moved_2_pub_lg_2_d23, moved_2_pub_lg_2_d24])
+ self.assertCountEqual(edges, [['doi_cg_d11','doi_lg_1_d23'],['doi_cg_d11','doi_cg_d21'],['doi_cg_i','doi_cg_d11'],['doi_cg_i','doi_cg_d12'],['doi_cg_d12','doi_cg_d22'],['doi_lg_2_h11','doi_cg_i'],['doi_cg_i','doi_lg_2_i'],['doi_lg_2_h21','doi_lg_2_h11'],['doi_lg_2_h22','doi_lg_2_h11'],['doi_lg_2_h11','doi_lg_2_i'],['doi_lg_2_i','doi_lg_2_d11'],['doi_lg_2_d11','doi_lg_2_i'],['doi_lg_2_d11','doi_lg_2_d21'],['doi_lg_2_i','doi_lg_2_d12'],['doi_lg_2_d12','doi_lg_2_d22'],['doi_lg_2_d12','doi_lg_2_d23'],['doi_lg_2_d12','doi_lg_2_d24']])
def keep_only_dois(nodes):
diff --git a/verarbeitung/update_graph/connect_new_input.py b/verarbeitung/update_graph/connect_new_input.py
index d9cbf82..df28ee6 100644
--- a/verarbeitung/update_graph/connect_new_input.py
+++ b/verarbeitung/update_graph/connect_new_input.py
@@ -23,22 +23,24 @@ from verarbeitung.construct_new_graph.initialize_graph import init_graph_constru
from verarbeitung.construct_new_graph.add_citations_rec import add_citations, get_cit_type_list, create_global_lists_cit
-def find_furthermost_citations_test(test_nodes, changed_node, old_search_depth, cit_type):
- global nodes
+def find_furthermost_citations_test(test_nodes, test_edges, changed_node, old_search_depth, cit_type):
+ global nodes, edges
nodes = test_nodes
+ edges = test_edges
- return(find_furthermost_citations(changed_node, old_search_depth, cit_type))
+ return(find_furthermost_citations(nodes, edges, changed_node, old_search_depth, cit_type))
-def complete_changed_group_nodes_test(test_nodes, inserted_test_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height):
- global nodes
+def complete_changed_group_nodes_test(test_nodes, test_edges, inserted_test_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height):
+ global nodes, edges
nodes = test_nodes
+ edges = test_edges
- handled_nodes = complete_changed_group_nodes(inserted_test_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, True)
- return(nodes, handled_nodes)
+ handled_nodes, new_nodes, new_edges = complete_changed_group_nodes(inserted_test_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, True)
+ return(new_nodes, new_edges, handled_nodes)
-def find_furthermost_citations(node, old_search_depth, cit_type):
+def find_furthermost_citations(new_nodes, new_edges, node, old_search_depth, cit_type):
'''
:param node: node which is known but not from input group
:type node: Publication
@@ -62,9 +64,43 @@ def find_furthermost_citations(node, old_search_depth, cit_type):
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)
+
+ if cit_type == "Citation":
+
+ # to find a cyclus and not change height
+ not_in_citations = True
+ for new_cit_node_citation in new_cit_node.citations:
+ if (cit_node.doi_url == new_cit_node_citation.doi_url):
+ not_in_citations = False
+ break
+ if (not_in_citations):
+ new_citations.append(new_cit_node)
+
+ # change height accordingly and add link to edge
+ new_cit_node.group = node.group + depth + 1
+ if [cit_node.doi_url,cit_node.doi_url] not in new_edges:
+ new_edges.append([new_cit_node.doi_url,cit_node.doi_url])
+
+ elif cit_type == "Reference":
+
+ # to find a cyclus and not change depth
+ not_in_citations = True
+ for new_cit_node_reference in new_cit_node.references:
+ if (new_cit_node.doi_url == new_cit_node_reference.doi_url):
+ not_in_citations = False
+ break
+ if (not_in_citations):
+ new_citations.append(new_cit_node)
+
+ # change height accordingly and add link to edge
+ new_cit_node.group = node.group - depth - 1
+ if [cit_node.doi_url, new_cit_node.doi_url] not in new_edges:
+ new_edges.append([cit_node.doi_url, new_cit_node.doi_url])
+
citations_saved = new_citations
+ for new_citation in new_citations:
+ if new_citation not in new_nodes:
+ new_nodes.append(new_citation)
# returns the references/citations which needs to be processed to complete contruction
return(citations_saved)
@@ -97,28 +133,50 @@ def complete_changed_group_nodes(inserted_nodes, old_search_depth, old_search_he
# saves which nodes were handled because they were known before
handled_inserted_nodes = []
+ new_nodes = []
+ new_edges = []
for node in nodes:
-
+
# 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
+ new_max_citations = find_furthermost_citations(new_nodes, new_edges, node, old_search_height + abs(node.group), "Citation")
+ add_citations(new_nodes, new_edges, new_max_citations, old_search_height, new_search_height, "Citation", test_var)
+
+ new_nodes, new_edges, error_doi_list_ref = init_graph_construction([node.doi_url], new_search_height, 0, test_var, True, new_nodes, new_edges)
+
+ for err_node in error_doi_list_ref:
+ if err_node not in error_doi_list:
+ error_doi_list.append(err_node)
+
+
+ old_max_references = find_furthermost_citations(new_nodes, new_edges, node, old_search_depth, "Reference")
+ add_citations(new_nodes, new_edges, old_max_references, old_search_depth, new_search_depth, "Reference", test_var)
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
+ new_max_references = find_furthermost_citations(new_nodes, new_edges, node, old_search_depth + abs(node.group), "Reference")
+ add_citations(new_nodes, new_edges, new_max_references, old_search_depth, new_search_depth, "Reference", test_var)
+ #new_nodes.append(new_max_references)
+
+ new_nodes, new_edges, error_doi_list_ref = init_graph_construction([node.doi_url], new_search_depth, 0, test_var, True, new_nodes, new_edges)
+ for err_node in error_doi_list_ref:
+ if err_node not in error_doi_list:
+ error_doi_list.append(err_node)
+
+
+ old_max_citations = find_furthermost_citations(new_nodes, new_edges, node, old_search_height, "Citation")
+ add_citations(new_nodes, new_edges, old_max_citations, old_search_height, new_search_height, "Citation", test_var)
handled_inserted_nodes.append(node)
- return(handled_inserted_nodes)
+ for new_node in new_nodes:
+ for inserted_node in inserted_nodes:
+ if new_node.doi_url == inserted_node:
+ new_node.group = 0
+
+ return(handled_inserted_nodes, new_nodes, new_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):
@@ -149,20 +207,21 @@ def connect_old_and_new_input(input_nodes_list, input_edges_list, inserted_nodes
completes the references and citations for nodes which were known in non input group
'''
- global nodes, edges
- nodes = input_nodes_list
- edges = input_edges_list
+ global nodes, edges, error_doi_list
+ nodes = input_nodes_list.copy()
+ edges = input_edges_list.copy()
+ error_doi_list = []
- handled_inserted_nodes = complete_changed_group_nodes(inserted_nodes, old_search_depth, old_search_height, new_search_depth, new_search_height, test_var)
+ handled_inserted_nodes, new_nodes, new_edges = 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]
- # 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)
+ # function call to begin recursive processing up to max depth/height for unhandled nodes
+ nodes, edges, error_doi_list_new = init_graph_construction(not_handled_inserted_nodes, new_search_depth, new_search_height, test_var = test_var, update_var = True, input_nodes = new_nodes, input_edges = new_edges)
+ for err_node in error_doi_list_new:
+ if err_node not in error_doi_list:
+ error_doi_list.append(err_node)
- nodes, edges, error_doi_list = 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)
+ return(nodes, edges, error_doi_list)
\ No newline at end of file
diff --git a/verarbeitung/update_graph/update_graph.py b/verarbeitung/update_graph/update_graph.py
index b7b89ab..416be51 100644
--- a/verarbeitung/update_graph/update_graph.py
+++ b/verarbeitung/update_graph/update_graph.py
@@ -102,16 +102,33 @@ def update_graph(new_doi_input_list, json_file, search_depth, search_height, tes
common_nodes, inserted_nodes, deleted_nodes = compare_old_and_new_node_lists(old_doi_input_list, new_doi_input_list)
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(processed_list, valid_edges, inserted_nodes, old_search_depth, old_search_height, search_depth, search_height, test_var)
+
+ processed_list_copy = processed_list.copy()
+ valid_edges_copy = valid_edges.copy()
# 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, valid_edges)
-
-
- if test_var:
- return(processed_list, valid_edges)
- else:
- return(processed_list, valid_edges, error_doi_list)
+
+ if (len(inserted_nodes) > 0):
+ inserted_pub_nodes, inserted_edges, error_doi_list_new = connect_old_and_new_input(processed_list_copy, valid_edges_copy, inserted_nodes, old_search_depth, old_search_height, search_depth, search_height, test_var)
+ for err_node in error_doi_list_new:
+ if err_node not in error_doi_list:
+ error_doi_list.append(err_node)
+
+ for inserted_node in inserted_pub_nodes:
+ not_in_nodes = True
+ for node in processed_list:
+ if inserted_node.doi_url == node.doi_url:
+ processed_list.remove(node)
+ processed_list.append(inserted_node)
+ not_in_nodes = False
+ break
+ if not_in_nodes:
+ processed_list.append(inserted_node)
+
+ for inserted_edge in inserted_edges:
+ if inserted_edge not in valid_edges:
+ valid_edges.append(inserted_edge)
+
+ return(processed_list, valid_edges, error_doi_list)
--
GitLab