def pubfetch(
config_path: str,
only_update_authorships: bool,
delay: int,
max_authorships: int
) -> None:
global pubmed_delay
pubmed_delay = delay
cfg = config.load(config_path)
pubmed_init(email=cfg.get("pubmed_email"),
api_key=cfg.get("pubmed_api_token"))
sup_conn: psql_connection
sup_conn = psycopg2.connect(host=cfg.get("sup_host"),
dbname=cfg.get("sup_database"),
user=cfg.get("sup_username"),
password=cfg.get("sup_password"),
port=cfg.get("sup_port"))
with sup_conn:
with sup_conn.cursor() as cursor:
update_authorships(cursor, max_authorships)
if not only_update_authorships:
fetch_publications(cursor)
sup_conn.close()
def serve(config_path: str):
global conn
global picture_path
global file_storage_alias
cfg = config.load(config_path)
if not cfg:
print('Error: Check config file')
sys.exit(-1)
try:
conn = psycopg2.connect(host=cfg.get("sup_host"),
dbname=cfg.get("sup_database"),
user=cfg.get("sup_username"),
password=cfg.get("sup_password"),
port=cfg.get("sup_port"))
except Exception:
print('Cannot connect to the database')
sys.exit(-1)
picture_path = cfg.get('picturepath', picture_path)
file_storage_alias = cfg.get('file_storage_alias', file_storage_alias)
secret_key = cfg.get('secret', os.getenv('SECRET_KEY', ''))
assert secret_key, (
"You must set a secret key for sessions in Flask\n"
"\thttps://flask.palletsprojects.com/en/1.1.x/quickstart/#sessions")
template_folder = cfg.get('forms', 'templates')
url_prefix = os.getenv('APPLICATION_ROOT', '')
server = Flask(__name__, template_folder=template_folder)
server.register_blueprint(app, url_prefix=url_prefix)
server.secret_key = secret_key
server.run()
def prefill(config_path: str):
cfg = config.load(config_path)
mwb_client = mwb.Client(cfg.get("mwb_host"), cfg.get("mwb_port"))
sup_conn: db.Connection = psycopg2.connect(
host=cfg.get("sup_host"),
dbname=cfg.get("sup_database"),
user=cfg.get("sup_username"),
password=cfg.get("sup_password"),
port=cfg.get("sup_port"))
embargoed_path = cfg.get("embargoed", "")
embargoed: List[str] = []
if embargoed_path:
with open(embargoed_path) as f:
embargoed = [line.strip() for line in f if line]
with sup_conn:
with sup_conn.cursor() as sup_cur:
process_projects_and_studies(mwb_client, sup_cur, embargoed)
add_developers(sup_cur)
sup_conn.close()
def generate(config_path: str, old_path: str):
timestamp = datetime.now()
path = os.path.join("data_out", timestamp.strftime("%Y"),
timestamp.strftime("%m"),
timestamp.strftime("%Y_%m_%d"))
os.makedirs(path, exist_ok=True)
org_file = os.path.join(path, 'orgs.nt')
people_file = os.path.join(path, 'people.nt')
project_file = os.path.join(path, 'projects.nt')
study_file = os.path.join(path, 'studies.nt')
dataset_file = os.path.join(path, 'datasets.nt')
tools_file = os.path.join(path, 'tools.nt')
photos_file = os.path.join(path, 'photos.nt')
pubs_file = os.path.join(path, 'pubs.nt')
add_file = os.path.join(path, 'add.nt')
sub_file = os.path.join(path, 'sub.nt')
cfg = config.load(config_path)
if not cfg.namespace.endswith('/'):
print(f"WARNING! Namespace doesn't end with '/': {cfg.namespace}")
mwb_conn: psycopg2.extensions.connection = psycopg2.connect(
host=cfg.get('mwb_host'),
dbname=cfg.get('mwb_database'),
user=cfg.get('mwb_username'),
password=cfg.get('mwb_password'),
port=cfg.get('mwb_port'))
sup_conn: psycopg2.extensions.connection = psycopg2.connect(
host=cfg.get('sup_host'),
dbname=cfg.get('sup_database'),
user=cfg.get('sup_username'),
password=cfg.get('sup_password'),
port=cfg.get('sup_port'))
with mwb_conn, sup_conn:
with mwb_conn.cursor() as mwb_cur, sup_conn.cursor() as sup_cur:
# Organizations
orgs = get_organizations(sup_cur)
org_triples = make_organizations(cfg.namespace, orgs)
print_to_file(org_triples, org_file)
# People
all_people = get_people(sup_cur)
people = {k: v for k, v in all_people.items() if not v.withheld}
withheld_people = {
k: v
for k, v in all_people.items() if v.withheld
}
people_triples = make_people(cfg.namespace, people)
people_triples.extend(
link_people_to_org(cfg.namespace, sup_cur, people, orgs))
print_to_file(people_triples, people_file)
# Photos
photos = get_photos(cfg.get("picturepath", "."), people)
photos_triples = make_photos(cfg.namespace, photos)
print_to_file(photos_triples, photos_file)
# Tools
print("Gathering Tools")
yaml_tools = get_yaml_tools(cfg)
csv_tools = list(fetch_mtw_tools(sup_cur))
all_tools = yaml_tools + csv_tools
tools_triples = make_tools(cfg.namespace, all_tools, people,
withheld_people, mwb_cur, sup_cur)
print_to_file(tools_triples, tools_file)
# Publications
pubs = get_publications(sup_cur)
pubs_triples = make_publications(cfg.namespace, pubs)
print_to_file(pubs_triples, pubs_file)
# Projects
projects = get_projects(mwb_cur, sup_cur, people, orgs)
project_data = make_projects(cfg.namespace, projects)
project_triples, project_summaries = project_data
all_proj_triples = project_triples + project_summaries
print_to_file(all_proj_triples, project_file)
# Studies
# Study file printed after datasets
embargoed_path = cfg.get('embargoed', '')
embargoed: List[str] = []
if embargoed_path:
with open(embargoed_path) as f:
embargoed = [line.strip() for line in f if line]
studies = get_studies(mwb_cur, sup_cur, people, orgs, embargoed)
study_triples, study_summaries = \
make_studies(cfg.namespace, studies, projects)
# Datasets
datasets = get_datasets(mwb_cur)
dataset_triples, study_sup_triples = \
make_datasets(cfg.namespace, datasets, studies)
print_to_file(dataset_triples, dataset_file)
all_study_triples = study_triples + study_summaries \
+ study_sup_triples
print_to_file(all_study_triples, study_file)
if old_path:
add, sub = diff(old_path, path)
with open(add_file, 'w') as f:
f.writelines(add)
with open(sub_file, 'w') as f:
f.writelines(sub)
sup_conn.close()
mwb_conn.close()
