Merge branch '66-no-conference-date' into 'master'

66 no conference date

* Fix `RecordConf.conference_dates` for record from inspirehep.net.
* Modify the logic to check and validate conference dates. It now depends on the record host.
* Close #66 

See merge request !66

controllers/harvest.py

@@ -216,7 +216,7 @@ def edit_insert():
 
             try:
                 check.country(record)
-                check.conference_date(record)
+                check.conference_date(record, selector.host)
 
             except CheckException:
                 pass

modules/harvest_tools/checkandfix.py

@@ -395,11 +395,13 @@ class CheckAndFix(object):
         except ToolException as e:
             raise CheckException(*e.args)
 
-    def conference_date(self, record):
+    def conference_date(self, record, host):
         """Check conference date.
 
         Args:
             record (RecordConf): record describing a talk or a proceeding.
+            host (str): possible values ares ``cds.cern.ch``
+                or ``inspirehep.net``
 
         Raises:
             CheckException: when dates are not found or not well formed.
@@ -409,11 +411,20 @@ class CheckAndFix(object):
         if not isinstance(record, RecordConf):
             return
 
-        # check and fix conference date
+        # inspirehep.net
+        if host == "inspirehep.net":
+            value = record.conference_dates()
+            if len(value) == 0:
+                raise CheckException(MSG_NO_CONF_DATE)
+
+            return
+
+        # cds.cern.ch
         if not (u"111" in record and "d" in record[u"111"]):
-            raise CheckException(MSG_NO_CONF_DATE)
+                raise CheckException(MSG_NO_CONF_DATE)
 
         value = record[u"111"]["d"]
+
         m = REG_CONF_DATES.match(value)
         if not m:
 

modules/harvest_tools/proceedings.py

@@ -35,7 +35,7 @@ class Proceedings(Automaton):
         try:
             self.check.is_conference(record)
             self.check.country(record)
-            self.check.conference_date(record)
+            self.check.conference_date(record, self.harvester.host)
 
             self.check.clean_erratum(record)
             self.check.submitted(record)

modules/harvest_tools/talks.py

@@ -35,7 +35,7 @@ class Talks(Automaton):
         try:
             self.check.is_conference(record)
             self.check.country(record)
-            self.check.conference_date(record)
+            self.check.conference_date(record, self.harvester.host)
 
             self.check.submitted(record)
             self.check.year(record)

modules/invenio_tools/record.py

@@ -6,7 +6,6 @@ import pprint
 
 
 from base import OAI, OAI_URL, REG_OAI
-from __builtin__ import isinstance
 
 
 class Record(dict):

modules/invenio_tools/recordconf.py

@@ -2,11 +2,18 @@
 """ invenio_tools.recordconf
 
 """
+import re
+
+
 from base import REG_YEAR
+from datetime import datetime
 from plugin_dbui import CLEAN_SPACES
 from recordpubli import RecordPubli
 
 
+REG_DATE = re.compile(r"(\d{4}-\d{2}-\d{2})")
+
+
 class RecordConf(RecordPubli):
     """The MARC record describing a conference talk or a proceeding.
     The relation between methods and MARC fields are the following::
@@ -14,8 +21,8 @@ class RecordConf(RecordPubli):
         +-----------------------+---------+----------+
         |                       |  CDS    | INSPIREP |
         +-----------------------+---------+----------+
-        | conference date       | 111 d   |          |
-        | conference end        | 111 z   | None     |
+        | conference date       | 111 d   | None     |
+        | conference end        | 111 z   | 111 y    |
         | conference key        | 111 g   |          |
         | conference location   | 111 c   |          |
         | conference title      | 111 a   |          |
@@ -38,7 +45,27 @@ class RecordConf(RecordPubli):
                   standardise.
 
         """
-        return self._get(u"111", "d")
+        if self.host().startswith("inspirehep"):
+
+            start = self._get(u"111", "x")
+            end = self._get(u"111", "y")
+
+            if not (REG_DATE.match(start) and REG_DATE.match(end)):
+                return u''
+
+            ds = datetime.strptime(start, "%Y-%m-%d")
+            de = datetime.strptime(end, "%Y-%m-%d")
+
+            if ds.month == de.month:
+                value = "%i-%i %s" % (ds.day, de.day, ds.strftime("%b %Y"))
+            else:
+                tpl = (ds.strftime("%d %b"), de.strftime("%d %b %Y"))
+                value = "%s - %s" % tpl
+
+        else:
+            value = self._get(u"111", "d")
+
+        return value
 
     def conference_country(self):
         """The country where the conference took place.