Redesing the base function is_conference, is_institute, is_thesis.

modules/invenio_tools/base.py

@@ -31,64 +31,81 @@ REG_YEAR = re.compile(r"(\d{4})")
 THESIS_DIR = "dir."
 
 
-def is_conference(record):
+def is_conference(recjson):
     """True when the record describes a publication related to a conference.
 
     Args:
-        record (Record): MARC12 record associated to a publication
-            or to and institute.
+        recjson (dict):
+            record associated to a publication or to and institute.
 
     Return:
-        bool: ``True`` when the MARC record describes a publication related
+        bool:
+            ``True`` when the record describes a publication related
             to a conference.
 
     """
-    if "111" in record:
+    # ConferencePaper in collection
+    # find proceeding in both stores
+    if "collection" in recjson:
+        data = recjson["collection"]
+        data = (data if isinstance(data, list) else [data])
+
+        value = "ConferencePaper"
+        li = [di for di in data if "primary" in di and di["primary"] == value]
+
+        if len(li) > 0:
+            return True
+
+    # try to identify talk in cds
+    # look for a conference key or for a subject equal to Talk
+    found = \
+        ("aleph_linking_page" in recjson) or \
+        ("subject" in recjson
+         and "term" in recjson["subject"]
+         and recjson["subject"]["term"] == u"Talk")
+
+    if found:
         return True
 
-    # try with the conference key
-    # the location of this values depends on the store
-    # cds.cern.ch (962, n) and inspirehep.net (773,w).
-    if record.host().startswith("cds"):
-        field, subfield = "962", "n"
-    else:
-        field, subfield = "773", "w"
+    # 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 len(record._get(field, subfield)) > 0
+    return False
 
 
-def is_institute(record):
+def is_institute(recjson):
     """True when the record describes an institute.
 
     Args:
-        record (Record): MARC12 record associated to a publication
-            or to and institute.
+        recjson (dict):
+            record associated to a publication or to and institute.
 
     Return:
-        bool: ``True`` when the MARC record describes an institute.
+        bool:
+            ``True`` when the record describes an institute.
 
     """
-    # "980": [
-    #    {"b": ["CK90", "HEP200", "PDGLIST", "PPF", "TOP500", "WEB"]},
-    #    {"a": "INSTITUTION"},
-    #    {"a": "CORE"}
-    # ]
-    if "980" in record:
-
-        if isinstance(record["980"], list):
-            for di in record["980"]:
-                for k, v in di.items():
-                    if k == "a" and v == "INSTITUTION":
-                        return True
-
-        elif isinstance(record["980"], dict) and "a" in record["980"] and \
-                record["980"]["a"] == "INSTITUTION":
-                return True
+    # 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
 
 
-def is_thesis(record):
+def is_thesis(recjson):
     """True when the record describes a thesis.
 
     Args:
@@ -99,6 +116,13 @@ def is_thesis(record):
         bool: ``True`` when the MARC record describes a thesis.
 
     """
-    li = record._get("980", "a", force_list=True)
-    val = ", ".join(li)
-    return "THESIS" in val
+    # THESIS in collection
+    if "collection" in recjson:
+        data = recjson["collection"]
+        data = (data if isinstance(data, list) else [data])
+
+        value = "THESIS"
+        li = [di for di in data if "primary" in di and di["primary"] == value]
+
+        if len(li) > 0:
+            return True

tests/basis/test_02_Record.py

+# -*- coding: utf-8 -*-
+"""test_instantiate_Record
+
+    * Test functions to introspect the type of record.
+    * Test functions to upcast the Record from the JSON object.
+
+"""
+from invenio_tools.base import is_conference, is_institute, is_thesis
+from invenio_tools.factory import build_record
+from invenio_tools.inveniostore import InvenioStore
+from invenio_tools.recordconf import RecordConf
+from invenio_tools.recordinst import RecordInst
+from invenio_tools.recordpubli import RecordPubli
+from invenio_tools.recordthesis import RecordThesis
+
+
+def test_conference_cds():
+
+    store = InvenioStore("cds.cern.ch")
+    recjson = store.get_record(1411352)
+
+    assert is_conference(recjson)
+    assert not is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordConf)
+
+
+def test_conference_inspirehep():
+
+    store = InvenioStore("inspirehep.net")
+    recjson = store.get_record(1276938)
+
+    assert is_conference(recjson)
+    assert not is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordConf)
+
+
+def test_institute():
+    """CPPM"""
+    store = InvenioStore("inspirehep.net")
+    recjson = store.get_record(902989)
+
+    assert not is_conference(recjson)
+    assert is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordInst)
+
+
+def test_publi_cds():
+    """Precision luminosity measurements at LHCb"""
+
+    store = InvenioStore("cds.cern.ch")
+    recjson = store.get_record(1951625)
+
+    assert not is_conference(recjson)
+    assert not is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordPubli)
+
+
+def test_publi_inspirehep():
+    """Precision luminosity measurements at LHCb"""
+
+    store = InvenioStore("inspirehep.net")
+    recjson = store.get_record(1319638)
+
+    assert not is_conference(recjson)
+    assert not is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordPubli)
+
+
+def test_talk_cds():
+
+    store = InvenioStore("cds.cern.ch")
+    recjson = store.get_record(2239092)
+
+    assert is_conference(recjson)
+    assert not is_institute(recjson)
+    assert not is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordConf)
+
+def test_thesis_cds():
+
+    store = InvenioStore("cds.cern.ch")
+    recjson = store.get_record(1632177)
+
+    assert not is_conference(recjson)
+    assert not is_institute(recjson)
+    assert is_thesis(recjson)
+
+    record = build_record(recjson)
+    assert isinstance(record, RecordThesis)

tests/invenio_tools/Marc12/test_marc.py

-"""test_marc
-
-"""
-from invenio_tools import (is_conference,
-                           is_institute,
-                           is_thesis,
-                           load_record,
-                           Record,
-                           RecordConf,
-                           RecordInst,
-                           RecordPubli,
-                           RecordThesis)
-
-
-def test_conference_cds():
-    record = load_record("cds.cern.ch", 1411352)
-    assert is_conference(record) == True
-    assert is_institute(record) == False
-    assert is_thesis(record) == False
-    assert isinstance(record, RecordConf)
-
-def test_conference_inspirehep():
-    record = load_record("inspirehep.net", 1276938)
-    assert is_conference(record) == True
-    assert is_institute(record) == False
-    assert is_thesis(record) == False
-    assert isinstance(record, RecordConf)
-
-def test_institute():
-    """CPPM"""
-    record = load_record("inspirehep.net", 902989)
-    assert is_conference(record) == False
-    assert is_institute(record) == True
-    assert is_thesis(record) == False
-    assert isinstance(record, RecordInst)
-
-
-def test_publi_cds():
-    """Precision luminosity measurements at LHCb"""
-    record = load_record("cds.cern.ch", 1951625)
-    assert is_conference(record) == False
-    assert is_institute(record) == False
-    assert is_thesis(record) == False
-    assert isinstance(record, RecordPubli)
-
-
-def test_publi_inspirehep():
-    """Precision luminosity measurements at LHCb"""
-    record = load_record("inspirehep.net", 1319638)
-    assert is_conference(record) == False
-    assert is_institute(record) == False
-    assert is_thesis(record) == False
-    assert isinstance(record, RecordPubli)
-
-
-def test_thesis_cds():
-    record = load_record("cds.cern.ch", 1632177)
-    assert is_conference(record) == False
-    assert is_institute(record) == False
-    assert is_thesis(record) == True
-    assert isinstance(record, RecordThesis)