data from sirano

bsoclinicaltrials/server/main/enrich_ct.py

@@ -2,7 +2,7 @@
 
 from bsoclinicaltrials.server.main.strings import normalize
 from bsoclinicaltrials.server.main.utils import chunks, get_dois_info
-
+from bsoclinicaltrials.server.main.sirano import get_sirano
 
 def tag_sponsor(x):
     x_normalized = normalize(x)
@@ -16,8 +16,9 @@ def tag_sponsor(x):
 def enrich(all_ct):
     res = []
     dois_to_get = []
+    sirano_dict = get_sirano()
     for ct in all_ct:
-        enriched = enrich_ct(ct)
+        enriched = enrich_ct(ct, sirano_dict)
         references = enriched.get('references', [])
         for r in references:
             if r.get('doi') and r.get('ReferenceType') in ['result', 'derived']:
@@ -76,7 +77,7 @@ def enrich(all_ct):
     return res
 
 
-def enrich_ct(ct):
+def enrich_ct(ct, sirano_dict):
     ct['study_start_year'] = None
     if isinstance(ct.get('study_start_date'), str):
         ct['study_start_year'] = int(ct['study_start_date'][0:4])
@@ -125,4 +126,6 @@ def enrich_ct(ct):
         status_simplified = 'Ongoing'
     ct['status_simplified'] = status_simplified
     ct['bso_country'] = ['fr']
+    if isinstance(ct.get('NCTId'), str) and ct['NCTId'] in sirano_dict:
+        ct.update(sirano_dict[ct['NCTId']])
     return ct

bsoclinicaltrials/server/main/sirano.py

@@ -0,0 +1,20 @@
+import pandas as pd
+from bsoclinicaltrials.server.main.logger import get_logger
+
+logger = get_logger(__name__)
+
+url = "https://www.data.gouv.fr/fr/datasets/r/c156ce7f-1ec8-4381-b8b9-fe5d2f933168"
+
+def get_sirano():
+    df = pd.read_csv(url, sep=';', encoding='iso-8859-1')
+    sirano_dict = {}
+    for ix, row in df.iterrows():
+        if isinstance(row.numero_registre_essais, str):
+            nct = row.numero_registre_essais
+            if nct not in sirano_dict:
+                sirano_dict[nct] = {'financement_total': 0.0, 'financements':[]}
+            sirano_dict[nct]['financements'].append(row.to_dict())
+            if isinstance(row.financement_total, float) or isinstance(row.financement_total, int):
+                sirano_dict[nct]['financement_total'] += row.financement_total
+    logger.debug(f'{len(sirano_dict)} essais dans SIRANo')
+    return sirano_dict

bsoclinicaltrials/server/main/tasks.py

@@ -5,7 +5,9 @@
 from bsoclinicaltrials.server.main.enrich_ct import enrich
 from bsoclinicaltrials.server.main.euctr import harvest_parse_euctr
 from bsoclinicaltrials.server.main.merge_sources import merge_all
+from bsoclinicaltrials.server.main.logger import get_logger
 
+logger = get_logger(__name__)
 
 def create_task_harvest(args: dict) -> dict:
     source = args.get('source', '').lower()