Migrate RecordInst.

modules/invenio_tools/factory.py

@@ -2,6 +2,8 @@
 """ invenio_tools.factory
 
 """
+import requests
+
 from base import (is_conference,
                   is_institute,
                   is_thesis,
@@ -20,6 +22,44 @@ from recordpubli import RecordPubli
 from recordthesis import RecordThesis
 
 
+def add_affiliation_keys(recjson):
+    """A the affiliation keys to the record describing an institute:
+
+        * The XML record contains the affiliation keys used by inspirehep.net.
+          They are located in the field 110__u and 110__t (future).
+
+        * The JSON record does not contains this information.
+
+        * This tool add the affiliation keys to the JSON record.
+          They are located:
+
+             +----------------+------------------------------------+
+             | field (limbra) | subfield                           |
+             +----------------+------------------------------------+
+             | corporate_note | identifier, futur_identifier, name |
+             +----------------+------------------------------------+
+
+    Args
+        recjson (dict): record data (MarcJSON)
+
+    """
+    url = "http://inspirehep.net/record/%i" % recjson[u"recid"]
+    rep = requests.get(url, params={"ot": "110", "of": "txt"})
+
+    # decode the string: '000recid 110__ $$aXXX$$bYYY$$tZZZ\n'
+    txt = rep.content.replace("\n", "")
+    li = txt[txt.find("$"):].split("$$")
+
+    di = {}
+    for el in li:
+        if len(el) == 0:
+            continue
+        di[el[0:1]] = el[1:]
+
+    recjson[u"corporate_note"] = {u"identifier": di["u"],
+                                  u"future_identifier": di["t"]}
+
+
 def add_conference_data(recjson):
     """Add the conference data to the recjson.
 
@@ -83,6 +123,7 @@ def add_conference_data(recjson):
 
     # get the data
     if conf_id is not None:
+        conf_id = (conf_id if isinstance(conf_id, int) else int(conf_id))
         confjson = get_conference_data(host, conf_id=conf_id)
 
     else:
@@ -117,6 +158,7 @@ def build_record(recjson):
         upcast_record = RecordConf(recjson)
 
     elif is_institute(recjson):
+        add_affiliation_keys(recjson)
         upcast_record = RecordInst(recjson)
 
     elif is_thesis(recjson):
@@ -135,7 +177,7 @@ def get_conference_data(host, conf_id=None, key=None):
         host (unicode):
             possible values are ``cds.cern.ch`` or ``inspirehep.net``.
 
-        conf_id (unicode):
+        conf_id (int):
             the conference identifier in the store.
             This is the preferred way.
 
@@ -159,7 +201,7 @@ def get_conference_data(host, conf_id=None, key=None):
     #
     if conf_id is not None:
         recjson = cds.get_record(conf_id)
-        if str(recjson["recid"]) != conf_id:
+        if recjson["recid"] != conf_id:
             raise CdsException(MSG_INV_CONF)
         return recjson
 

modules/invenio_tools/recordinst.py

 """ invenio_tools.recordinst
 
 """
+<<<<<<< HEAD
 from .base import is_institute
 from .exception import RecordException
 from .record import Record
+=======
+from base import is_institute, REG_OAI
+from exception import RecordException
+from record import Record
+>>>>>>> Migrate RecordInst.
 
 
 MSG_INVALID_ARG = "Invalid argument record"
@@ -12,70 +18,99 @@ MSG_INVALID_RECORD = "Invalid record, it is not describing an institute"
 
 
 class RecordInst(Record):
-    """The MARC record describing an institute.
-    The relation between methods and MARC fields are the following::
-
-        ------------------------+-------------+
-        |                       |  INSPIREHEP |
-        ------------------------+-------------+
-        | institute identifier  | 110 u       |
-        | future institute id   | 110 t       |
-        | name                  | 110 b       |
-        | type of record        | 980 a       |
-        ------------------------+-------------+
+    """The record describing an institute.
+    Fields are::
+
+         +-----------------------------+----------------------------------+
+         | field (inspirehep)          | subfield                         |
+         +-----------------------------+----------------------------------+
+         | FIXME_OAI                   | id, set                          |
+         | administrative_history      |                                  |
+         | authority_institution       | institution                      |
+         | cataloguer_info             | creation_date, modification_date |
+         | collection                  | primary, secondary               |
+         | corporate_name              | name, subordinate_unit           |
+         | creation_date               |                                  |
+         | files                       |                                  |
+         | filetypes                   |                                  |
+         | number_of_citations         |                                  |
+         | number_of_comments          |                                  |
+         | number_of_reviews           |                                  |
+         | persistent_identifiers_keys |                                  |
+         | recid                       |                                  |
+         | source_of_description       | note                             |
+         | system_control_number       | institute, value                 |
+         | url                         |                                  |
+         | version_id                  |                                  |
+         +-----------------------------+----------------------------------+
+
+    One field is added by limbra:
+         +-----------------------------+----------------------------------+
+         | field (limbra)              | subfield                         |
+         +-----------------------------+----------------------------------+
+         | corporate_note              | identifier, futur_identifier,    |
+         |                             | name                             |
+         +-----------------------------+----------------------------------+
 
     Args:
-        record (Record):
+        recjson (dict):
+            institute data (MarcJSON)
 
     """
-    def __init__(self, record):
+    def __init__(self, recjson):
 
-        if not isinstance(record, Record):
-            raise RecordException(MSG_INVALID_ARG)
+        host = REG_OAI.match(recjson[u"FIXME_OAI"][u"id"]).group(1)
+        if host != "inspirehep.net":
+            raise RecordException(MSG_INVALID_HOST)
 
-        if not is_institute(record):
+        if not is_institute(recjson):
             raise RecordException(MSG_INVALID_RECORD)
 
-        if record.host() != "inspirehep.net":
-            raise RecordException(MSG_INVALID_HOST)
-
-        Record.__init__(self, record)
+        Record.__init__(self, recjson)
 
     def future_identifier(self):
         """Future identifier of the institute.
 
         Returns:
-            str: the future inspirehep identifier or an empty string
+            unicode:
+                the future inspirehep identifier or an empty string
                 if the identifier is not defined.
 
         """
-        return self._get("110", "t")
+        return self._get(u"corporate_note", u"future_identifier")
 
     def identifier(self):
         """Identifier of the institute.
 
         Returns:
-            str: the current inspirehep identifier (2015) or an empty
+            unicode:
+                the current inspirehep identifier (2015) or an empty
                 string if it is not defined.
 
         """
-        return self._get("110", "u")
+        return self._get(u"corporate_note", u"identifier")
 
     def name(self):
         """ Name of the institute.
 
         Returns:
-            str: the name of the institute or an empty string if
-                it is not defined.
+            unicode:
+                - the name of the institute.
+                - an empty string when it is not defined.
 
         """
-        return self._get("110", "a")
+        value = self._get(u"corporate_name", u"subordinate_unit")
+        if isinstance(value, list) and len(value) == 1:
+            return value[0]
+
+        return u""
 
     def rex(self):
         """ Regular expression to search authors affiliate to the institute.
 
         Returns:
-            str: the regular expression to search author affiliate
+            unicode:
+                the regular expression to search author affiliate
                 to the institute in the store ``cds.cern.ch`` or
                 ``inspirehep.net``.
 

tests/basis/test_02_record_factory.py

@@ -10,7 +10,8 @@ from invenio_tools.base import (is_conference,
                                 is_institute,
                                 is_thesis)
 
-from invenio_tools.factory import (add_conference_data,
+from invenio_tools.factory import (add_affiliation_keys,
+                                   add_conference_data,
                                    build_record,
                                    get_conference_data)
 
@@ -24,104 +25,7 @@ from invenio_tools.recordthesis import RecordThesis
 
 # ............................................................................
 #
-# Section to test introspection and instantiation
-#
-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)
-
-
-# ............................................................................
-#
-# Section to test tool to get and add conference data
+# Conference proceeding and talk
 #
 def test_get_conference_data():
     """ check the different approach to get the conference data
@@ -196,10 +100,13 @@ def test_add_conference_data():
 
     add_conference_data(recjson)
     assert "meeting_name" in recjson
-    assert "meeting" in recjson
+    assert "meeting_note" in recjson
     assert recjson["meeting_name"][0]["coference_code"] == "rome20101206"
-    assert recjson["meeting"]["url"] == "http://www.roma1.infn.it/discrete10"
-    assert recjson["meeting"]["recid"] == 1181092
+
+    assert recjson["meeting_note"]["url"] == \
+        "http://www.roma1.infn.it/discrete10"
+
+    assert recjson["meeting_note"]["recid"] == 1181092
 
     # ........................................................................
     #
@@ -210,7 +117,127 @@ def test_add_conference_data():
 
     add_conference_data(recjson)
     assert "meeting_name" in recjson
-    assert "meeting" in recjson
+    assert "meeting_note" in recjson
     assert recjson["meeting_name"][0]["coference_code"] == "C10-12-06"
-    assert recjson["meeting"]["url"] == "http://www.roma1.infn.it/discrete10"
-    assert recjson["meeting"]["recid"] == 980401
+
+    assert recjson["meeting_note"]["url"] == \
+        "http://www.roma1.infn.it/discrete10"
+
+    assert recjson["meeting_note"]["recid"] == 980401
+
+
+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_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)
+
+
+# ............................................................................
+#
+# Institute
+#
+def test_add_affiliation_keys():
+    """CPPM"""
+    store = InvenioStore("inspirehep.net")
+    recjson = store.get_record(902989)
+
+    add_affiliation_keys(recjson)
+
+    assert u"corporate_note" in recjson
+    assert recjson[u"corporate_note"][u"identifier"] == "Marseille, CPPM"
+    assert recjson[u"corporate_note"][u"futur_identifier"] == "CPPM, Marseille"
+
+
+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)
+
+
+# ............................................................................
+#
+# Article, ...
+#
+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)
+
+
+# ............................................................................
+#
+# Thesis
+#
+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/basis/test_07_RecordInst.py

+# -*- coding: utf-8 -*-
+"""test_06_RecordThesis
+
+Test specific methods of the RecordInst class for CPPM
+
+"""
+import pytest
+from invenio_tools import load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    return load_record("inspirehep.net", 902989)
+
+
+def test_future_identifer(record):
+    assert record.future_identifier() == u'CPPM, Marseille'
+
+
+def test_id(record):
+    assert record.id() == 902989
+
+
+def test_identifier(record):
+    assert record.identifier() == u'Marseille, CPPM'
+
+
+def test_name(record):
+    assert record.name() == \
+        u'Centre de Physique des Particules de Marseille (CPPM)'
+
+
+def test_rex(record):
+    assert record.rex() == \
+        r"Marseille, CPPM|CPPM, Marseille|" \
+        "Centre de Physique des Particules de Marseille (CPPM)"