[BUGFIX] demanding more fields in istex-api query

parsing/FileParsers/ISTex.py

@@ -34,11 +34,13 @@ class ISTex(FileParser):
                 try:
                     # print(path," ==> ",len(json_doc[path]))
                     hyperdata[key] = json_doc[path]
-                except: pass
+                except:
+                    pass
 
             # print("|",hyperdata["language_iso3"])
 
-            if "doi" in hyperdata: hyperdata["doi"] = hyperdata["doi"][0]
+            if "doi" in hyperdata: 
+                hyperdata["doi"] = hyperdata["doi"][0]
             
             keywords = []
             if "keywords" in hyperdata:
@@ -76,7 +78,10 @@ class ISTex(FileParser):
                 if len(hyperdata["genre"])==0:
                     hyperdata.pop("genre")
             if "language_iso3" in hyperdata:
-                hyperdata["language_iso3"] = hyperdata["language_iso3"][0]
+                if len(hyperdata["language_iso3"])>0:
+                    hyperdata["language_iso3"] = hyperdata["language_iso3"][0]
+                else:
+                    hyperdata["language_iso3"] = "eng"
 
             RealDate = hyperdata["publication_date"]
             if "publication_date" in hyperdata: hyperdata.pop("publication_date")

scrappers/scrap_pubmed/views.py

@@ -66,7 +66,7 @@ def getGlobalStatsISTEXT(request ):
 		print ("LOG::TIME:_ "+datetime.datetime.now().isoformat()+" query =", query )
 		print ("LOG::TIME:_ "+datetime.datetime.now().isoformat()+" N =", N )
 		query_string = query.replace(" ","+")
-		url = "http://api.istex.fr/document/?q="+query_string
+		url = "http://api.istex.fr/document/?q="+query_string+"&output=id,title,abstract,pubdate,corpusName,authors,language"
 
 		tasks = MedlineFetcher()