Update store_tools/base.py to add is_conference and friends to new inspirehep API

modules/store_tools/base.py

@@ -43,6 +43,10 @@ THESIS_DIR = "dir."
 def is_conference(recjson):
     """True when the record describes a publication related to a conference.
 
+    Note:
+        This tools works for JSON record coming from cds.cern.ch,
+        old.inspirehep.net as well as inspirehep.net
+
     Args:
         recjson (dict):
             record associated to a publication or to and institute.
@@ -53,8 +57,13 @@ def is_conference(recjson):
             to a conference.
 
     """
-    # ConferencePaper in collection
-    # find proceeding in both stores
+    # proceeding in inspirehep
+    if recjson.get("$schema", "").endswith("hep.json"):
+        for elt in recjson.get("document_type", []):
+            if elt == "conference paper":
+                return True
+
+    # proceeding in cds
     if "collection" in recjson:
         data = recjson["collection"]
         data = (data if isinstance(data, list) else [data])
@@ -65,7 +74,7 @@ def is_conference(recjson):
         if len(li) > 0:
             return True
 
-    # try to identify talk in cds
+    # talk in cds
     # look for a conference key or for a subject equal to Talk
     found = \
         ("aleph_linking_page" in recjson) or \
@@ -76,21 +85,16 @@ def is_conference(recjson):
     if found:
         return True
 
-    # try to identify talk in inspirehep
-    # look for a conference key
-    found = \
-        ("publication_info" in recjson
-         and "cnum" in recjson["publication_info"])
-
-    if found:
-        return True
-
     return False
 
 
 def is_institute(recjson):
     """True when the record describes an institute.
 
+    Note:
+        This tools works for JSON record coming from cds.cern.ch,
+        old.inspirehep.net as well as inspirehep.net
+
     Args:
         recjson (dict):
             record associated to a publication or to and institute.
@@ -100,23 +104,16 @@ def is_institute(recjson):
             ``True`` when the record describes an institute.
 
     """
-    # INSTITUTION in collection
-    if "collection" in recjson:
-        data = recjson["collection"]
-        data = (data if isinstance(data, list) else [data])
-
-        value = "INSTITUTION"
-        li = [di for di in data if "primary" in di and di["primary"] == value]
-
-        if len(li) > 0:
-            return True
-
-    return False
+    return recjson.get("$schema", "").endswith("institutions.json")
 
 
 def is_thesis(recjson):
     """True when the record describes a thesis.
 
+    Note:
+        This tools works for JSON record coming from cds.cern.ch,
+        old.inspirehep.net as well as inspirehep.net
+
     Args:
         recjson (dict):
             record associated to a publication or to and institute.
@@ -125,7 +122,13 @@ def is_thesis(recjson):
         bool: ``True`` when the record describes a thesis.
 
     """
-    # THESIS in collection
+    # Thesis in inspirehep
+    if recjson.get("$schema", "").endswith("hep.json"):
+        for elt in recjson.get("document_type", []):
+            if elt == "thesis":
+                return True
+
+    # Thesis in cds
     if "collection" in recjson:
         data = recjson["collection"]
         data = (data if isinstance(data, list) else [data])
@@ -135,3 +138,5 @@ def is_thesis(recjson):
 
         if len(li) > 0:
             return True
+
+    return False