The class CheckAndFix used the function search_synonym.

languages/fr-fr.py

@@ -480,7 +480,9 @@
 'Reject': 'Rejeter',
 'Reject article is not published': "Rejeté l'article n'est pas publié",
 'Reject collaboration is not well formed': 'Rejeté la collaboration est mal formatté',
+'Reject collaborations is not defined': "Rejeté la collaboration n'est pas définie",
 'Reject conference dates is not well formed': 'Rejecté les dates de la conférence dates sont mal formatté',
+'Reject countries is not defined': "Rejeté le pays n'est pas définie",
 'Reject editor is not well formed': "Rejeté l'éditeur est mal formatté",
 'Reject incomplete paper reference': 'Rejeté la référence du papier est incomplète',
 'Reject invalid country': 'Rejeté pays inconnu',
@@ -502,11 +504,15 @@
 'Reject preprint is a conference': 'Rejeté ce preprint est une  conférence',
 'Reject preprint is a published paper': 'Rejeté ce preprint est un article publié',
 'Reject preprint is a thesis': 'Rejeté ce preprint est une thèse',
+'Reject publishers is not defined': "Rejeté la revue n'est pas définie",
 'Reject submission date is not well formed': "Rejeté la date de soumission n'est pas correcte",
 'Reject the talk match a proceeding': 'Rejeté cette présentation correspond à un actes de conférence',
 'Reject to many first author': 'Rejeté trop de premier autheur',
 'Reject to many submit date': 'Rejeté plusieurs date de soumission',
 'Reject to many year': 'Rejeté plusieurs année',
+'Reject too many collaborations synonyms': 'Rejeté synonyme de collaboration défini plusieurs fois',
+'Reject too many countries synonyms': 'Rejeté synonyme de pays défini plusieurs fois',
+'Reject too many publishers synonyms': 'Rejeté synonyme de revue défini plusieurs fois',
 'Reject XML is not well formed': "Rejeté la chaine XML n'est pas correcte",
 'Rejected': 'Rejeté',
 "Rejeté la chaine XML n'est pas correcte": "Rejeté la chaine XML n'est pas correcte",

modules/harvest_tools/base.py

@@ -13,8 +13,8 @@ MSG_CRASH = "Crash: %s"
 MSG_FIX_ORIGIN = "Fixed the origin field"
 MSG_IN_DB = "Already in the database"
 MSG_LOAD = "Load in the database"
-MSG_NO_ENTRY = "Reject %s is not defined."
-MSG_TOOMANY_SYNONYM = "Reject too many %s synonyms."
+MSG_NO_ENTRY = "Reject %s is not defined"
+MSG_TOOMANY_SYNONYM = "Reject too many %s synonyms"
 
 
 def family_name_fr(full_name):

modules/harvest_tools/checkandfix.py

@@ -5,12 +5,10 @@
 import re
 import regex
 
+from base import search_synonym
 from exception import CheckException
 from invenio_tools import (load_record,
-                           MSG_NO_COUNTRY,
                            MSG_NO_CONF,
-                           MSG_NO_PUBLISHER,
-                           MSG_WELL_FORMED_COLLABORATION,
                            MSG_NO_THESIS,
                            OAI_URL,
                            RecordConf,
@@ -358,21 +356,16 @@ class CheckAndFix(object):
             record (RecordPubli): record describing a publication.
 
         Raises:
-            CheckException: when the collaboration value is not well formed
-                and not entered as a synonym.
+            ToolException: when the collaboration value is defined
+                nor entered as a synonym.
 
         """
         val = record.collaboration()
         if not val:
             return
 
-        if REG_COLLABORATION.match(val):
-            return
-
-        if self._is_synonym("collaborations", val):
-            return
-
-        raise CheckException(MSG_WELL_FORMED_COLLABORATION)
+        db = self.db
+        search_synonym(db.collaborations, "collaboration", val)
 
     def country(self, record):
         """Check conference country.
@@ -382,8 +375,8 @@ class CheckAndFix(object):
             record (RecordConf): record describing a talk or a proceeding.
 
         Raises:
-            CheckException: when the country is not defined
-                and not entered as a synonym.
+            ToolException: when the country is not defined
+                nor entered as a synonym.
 
         """
         if not isinstance(record, RecordConf):
@@ -391,15 +384,7 @@ class CheckAndFix(object):
 
         db = self.db
         val = record.conference_country()
-
-        id_country = get_id(db.countries, country=val)
-        if id_country:
-            return
-
-        if self._is_synonym("countries", val):
-            return
-
-        raise CheckException(MSG_NO_COUNTRY)
+        search_synonym(db.countries, "country", val)
 
     def conference_date(self, record):
         """Check conference date.
@@ -774,8 +759,8 @@ class CheckAndFix(object):
             record (RecordPubli): record describing a publication.
 
         Raises:
-            CheckException: when the publisher is not defined
-                and not entered as a synonym.
+            ToolException: when the publisher is not defined
+                nor entered as a synonym.
 
         """
         db = self.db
@@ -787,14 +772,7 @@ class CheckAndFix(object):
         if isinstance(val, list):
             val = val[0]
 
-        id_publisher = get_id(db.publishers, abbreviation=val)
-        if id_publisher:
-            return
-
-        if self._is_synonym("publishers", val):
-            return
-
-        raise CheckException(MSG_NO_PUBLISHER)
+        search_synonym(db.publishers, "abbreviation", val)
 
     def recover_oai(self, record, host):
         """Recover the OAI identifier when it is not defined

modules/harvest_tools/msg.py

@@ -5,12 +5,14 @@
 import json
 
 
+from base import MSG_NO_ENTRY, MSG_TOOMANY_SYNONYM
 from gluon import current
 from gluon.storage import Storage
-from invenio_tools import (MSG_NO_COUNTRY,
-                           MSG_NO_PUBLISHER,
-                           MSG_WELL_FORMED_COLLABORATION,
-                           OAI_URL)
+from invenio_tools import OAI_URL
+
+
+MSGS = (MSG_NO_ENTRY, MSG_TOOMANY_SYNONYM)
+TABLES = ("collaborations", "countries", "publishers")
 
 
 class Msg(Storage):
@@ -117,14 +119,15 @@ class Msg(Storage):
         self.action = 'reject'
 
         if record is not None:
-            if str(txt) == MSG_NO_COUNTRY:
-                self.synonym = record.country()
-
-            elif str(txt) == MSG_WELL_FORMED_COLLABORATION:
-                self.synonym = record.collaboration()
-
-            elif str(txt) == MSG_NO_PUBLISHER:
-                self.synonym = record.publisher()
+            for msg in MSGS:
+                for tablename in TABLES:
+                    if str(txt) == msg % tablename:
+                        if tablename == "collaborations":
+                            self.synonym = record.collaboration()
+                        elif tablename == "countries":
+                            self.synonym = record.country()
+                        elif tablename == "publishers":
+                            self.synonym = record.publisher()
 
         if year is None and record is not None:
             year = record.year()