fix some remaining spaces

modules/harvest_tools.py

@@ -175,50 +175,50 @@ def get_harvester_tool(controller):
     return Tool
 
 
-def build_harvester_tool(db, id_team, id_project, controller, id_category,  
+def build_harvester_tool(db, id_team, id_project, controller, id_category,
                  year_start=None, year_end=None, dry_run=True, debug=False):
     """
     Harvest tool factory function, returns non if no factory exist for controller.
-    
+
     @type db: gluon.dal.DAL
     @param db:
-    
+
     @type id_team: int
     @param id_team: Identifier of the team in the db
-    
+
     @type id_project: int
     @param id_project: Identifier of the project in the db
-    
+
     @type controller: unicode
     @param controller: Type of publication (i.e. 'article', 'proceedings', ...)
-    
+
     @type id_category: int
     @param id_category: Identifier of the category of publication (i.e. ACL, ACTI, ...)
-    
+
     @type year_start: int
     @keyword year_start: Start year of search (i.e. '2014')
-    
+
     @type year_end: int
     @keyword year_end: End year of search (i.e. '2015')
-    
+
     @type dry_run: boolean
     @keyword dry_run: True if no record is to be written to the db
-    
+
     @type debug: bool
     @param debug: activate the debug mode
-     
+
     """
     tool_class = get_harvester_tool(controller)
     if tool_class is None:
         return None
-    
-    return tool_class(db, id_team, id_project, controller, id_category,  
+
+    return tool_class(db, id_team, id_project, controller, id_category,
                  year_start, year_end, dry_run, debug)
 
 
-def learn_my_authors(db, authors=None, 
-                         id_project=None, 
-                         id_team=None, 
+def learn_my_authors(db, authors=None,
+                         id_project=None,
+                         id_team=None,
                          year=None):
     """Train the rescue list of the authors of my institute,
     stored in the database, using the list C{authors} provided in argument.
@@ -447,12 +447,12 @@ class PublicationsTool(object):
     are defined by the current request.
 
     """
-    
-    
-    def __init__(self, db, id_team, id_project, controller, id_category,  
+
+
+    def __init__(self, db, id_team, id_project, controller, id_category,
                  year_start=None, year_end=None, dry_run=True, debug=False):
         """
-        
+
         @note see C{build_harvester_tool} factory function building C{PublicationsTools}
 
         @type db: gluon.dal.DAL
@@ -469,19 +469,19 @@ class PublicationsTool(object):
 
         @type id_category: int
         @param id_category: Identifier of the category of publication (i.e. ACL, ACTI, ...)
-        
+
         @type year_start: int
         @keyword year_start: Start year of search (i.e. '2014')
 
         @type year_end: int
         @keyword year_end: End year of search (i.e. '2015')
-        
+
         @type dry_run: boolean
         @keyword dry_run: True if no record is to be written to the db
-        
+
         @type debug: bool
         @param debug: activate the debug mode
-                
+
         """
         self.collection_logs = []
         self.db = db
@@ -497,9 +497,9 @@ class PublicationsTool(object):
 
         self.check = CheckAndFix()
         self.marc12 = Marc12()
-        
+
         # check parameters
-        # protection team, project and/or category have to be defined    
+        # protection team, project and/or category have to be defined
         if not self.id_team:
             raise ToolException(MSG_NO_TEAM)
 
@@ -509,7 +509,7 @@ class PublicationsTool(object):
         if not self.id_category:
             raise ToolException(MSG_NO_CAT)
 
-        # Construct harvester Storage needed for the log        
+        # Construct harvester Storage needed for the log
         self.harvester = Storage(id_teams=self.id_team,
                                  id_projects=self.id_project,
                                  controller=self.controller,
@@ -535,7 +535,7 @@ class PublicationsTool(object):
         L{invenio_tools.InvenioStore.get_ids}.
 
         """
-        
+
         # INSPIREHEP store
         if collection.startswith('find'):
 
@@ -549,12 +549,12 @@ class PublicationsTool(object):
 
             elif self.year_start and self.year_end:
                 query += " and date > %s and date < %s " \
-                         % (self.year_start-1, self.year_end+1)
+                         % (self.year_start - 1, self.year_end + 1)
 
-            di = dict(p=query,              # query à la spires
-                      rg=1000,              # maximum number of records returned
-                      sf='year',            # sort by date
-                      so='d')               # descending order
+            di = dict(p=query,  # query à la spires
+                      rg=1000,  # maximum number of records returned
+                      sf='year',  # sort by date
+                      so='d')  # descending order
 
         # CERN INVENIO store
         else:
@@ -567,7 +567,7 @@ class PublicationsTool(object):
 
             elif self.year_start and self.year_end:
                 li = []
-                for year in range(self.year_start, self.year_end+1):
+                for year in range(self.year_start, self.year_end + 1):
                     li.append(str(year))
                 rex = '|'.join(li)
 
@@ -802,7 +802,7 @@ class PublicationsTool(object):
         """
         return 0
 
-        
+
     def process_url(self, host, collections):
         """Retrieve the MARC XML string and launch its decoding.
 
@@ -811,25 +811,25 @@ class PublicationsTool(object):
 
         @type host: unicode
         @keyword host: Web host name to query for publication
-        
+
         @type collections: unicode
         @keyword collections: Request string to send to the host to get the publications
-        
-        @rtype: int        
-        @return: one when the record is inserted / updated in the database 
+
+        @rtype: int
+        @return: one when the record is inserted / updated in the database
         zero otherwise.
-        
+
         """
         if self.dbg:
             print "process URL search"
-        
+
         self.host = host
         self.collections = collections
         marc12xmls = []
 
         # extend harvester for logs
         self.harvester.host = host
-        self.harvester.collections = collections                            
+        self.harvester.collections = collections
 
         store = InvenioStore(self.host)
 
@@ -952,7 +952,7 @@ class PublicationsTool(object):
 
         return dict(collection_logs=self.collection_logs,
                     controller=self.controller,
-                    logs=self.logs)    
+                    logs=self.logs)
 
 
     def __call__(self, xml):
@@ -970,12 +970,12 @@ class PublicationsTool(object):
 
         @type xml: unicode
         @keyword xml: marc12 xml encoding of the publication record
-                
+
         """
         if self.dbg:
             print "start processing", self.__class__.__name__
             print "decode request"
-            
+
         if self.dbg:
             print "get harvest parameters"