127 lines
3.3 KiB
Python
127 lines
3.3 KiB
Python
import json
|
|
from pprint import pprint
|
|
import sys
|
|
|
|
nodes = {}
|
|
edges = {}
|
|
# we need a separate counter, because some, once created, edges
|
|
# can get removed, if their target node does not materialize
|
|
edge_count = 0
|
|
|
|
wanted_node_types = {
|
|
"xyzri:XYZInstrument": 'instrument',
|
|
"xyzri:XYZOrganization": 'organization',
|
|
"xyzri:XYZPerson": 'person',
|
|
"xyzri:XYZProject": 'project',
|
|
"xyzri:XYZPublication": 'publication',
|
|
"xyzri:XYZTopic": 'topic',
|
|
"xyzri:XYZObjective": 'objective',
|
|
"xyzri:XYZDataset": 'dataset',
|
|
}
|
|
|
|
wanted_edge_types = {
|
|
'associated_with': 'associated_with',
|
|
'attributed_to': 'attributed_to',
|
|
'generated_by': 'generated_by',
|
|
'delegated_by': 'delegated_by',
|
|
'influenced_by': 'influenced_by',
|
|
'part_of': 'part_of',
|
|
'about': 'about',
|
|
}
|
|
|
|
|
|
def get_node_label(rec: dict) -> str | None:
|
|
for src in (
|
|
'xyzrins:annotation-tags/graphviz-label',
|
|
'short_name',
|
|
'name',
|
|
'title',
|
|
'family_name',
|
|
'display_label',
|
|
):
|
|
# consider annotations, and then a direct property
|
|
label = rec.get('annotations', {}).get(src)
|
|
if isinstance(label, dict):
|
|
label = label.get('annotation_value')
|
|
if label is None:
|
|
label = rec.get(src)
|
|
if label:
|
|
if src == 'family_name':
|
|
return f'{rec["given_name"][0]}.{label}'
|
|
else:
|
|
return label
|
|
return None
|
|
|
|
|
|
def get_node_url(rec: dict) -> str | None:
|
|
url_override = rec.get('annotations', {}).get('xyzrins:annotation-tags/psyinf-ns-id')
|
|
if url_override:
|
|
if isinstance(url_override, dict):
|
|
url_override = url_override['annotation_value']
|
|
return f'/{url_override}'
|
|
url = rec['pid']
|
|
www_root_prefix = 'xyzrins:'
|
|
if not url.startswith(www_root_prefix):
|
|
return None
|
|
url = f'/{url[len(www_root_prefix):]}'
|
|
return url
|
|
|
|
|
|
def add_edge(src: str, target: str, kind: str) -> None:
|
|
global edge_count
|
|
# use the edge properties as ID to get auto-deduplication
|
|
edge_id = (src, target)
|
|
edges[edge_id] = {
|
|
'id': f'e{edge_count}',
|
|
'source': src,
|
|
'target': target,
|
|
}
|
|
edge_count += 1
|
|
|
|
|
|
for line in sys.stdin:
|
|
rec = json.loads(line)
|
|
schema_type = rec.get('schema_type')
|
|
if schema_type not in wanted_node_types:
|
|
continue
|
|
|
|
pid = rec['pid']
|
|
if pid in nodes:
|
|
print(f'Ignoring duplicate node {pid}', file=sys.stderr)
|
|
continue
|
|
|
|
for prop, edge_kind in wanted_edge_types.items():
|
|
for rel in rec.get(prop, []):
|
|
obj = rel['object'] if isinstance(rel, dict) else rel
|
|
add_edge(pid, obj, edge_kind)
|
|
|
|
nodes[pid] = {
|
|
'id': pid,
|
|
'label': get_node_label(rec),
|
|
'type': wanted_node_types[schema_type],
|
|
'size': 1,
|
|
'url': get_node_url(rec),
|
|
}
|
|
|
|
kept_edges = []
|
|
nodes_missing = set()
|
|
for edge in edges.values():
|
|
err = False
|
|
for prop in ('source', 'target'):
|
|
if edge[prop] not in nodes:
|
|
nodes_missing.add(edge[prop])
|
|
err = True
|
|
if err:
|
|
continue
|
|
nodes[edge['target']]['size'] += 1
|
|
kept_edges.append(edge)
|
|
|
|
out = {'nodes': list(nodes.values()), 'edges': kept_edges}
|
|
|
|
json.dump(out, sys.stdout)
|
|
|
|
if nodes_missing:
|
|
pprint(
|
|
f"Missing nodes: {nodes_missing!r}",
|
|
stream=sys.stderr,
|
|
)
|