Move xml as PublicationTools constructor argument

controllers/harvest.py

@@ -224,11 +224,12 @@ def insert_marcxml():
                           selector.id_projects,
                           selector.controller,
                           selector.id_categories,
+                          xml=selector.xml,
                           year_start=selector.year_start,
                           year_end=selector.year_end,
                           dry_run=(selector.mode == DRY_RUN),
                           debug=False)
-        tool(selector.xml)
+        tool()
     
     except ToolException, e:
         return T(str(e))

modules/harvest_tools.py

@@ -408,7 +408,7 @@ class PublicationsTool(object):
     """
 
     def __init__(self, db, id_team, id_project, controller, id_category, host=None, collections=None, 
-                 year_start=None, year_end=None, dry_run=True, debug=False):
+                 xml=None, year_start=None, year_end=None, dry_run=True, debug=False):
         """
 
         @type db: gluon.dal.DAL
@@ -432,6 +432,9 @@ class PublicationsTool(object):
         @type collections: unicode
         @keyword collections: Request string to send to the host to get the publications
 
+        @type xml: unicode
+        @keyword xml: marc12 xml encoding of the publication record
+
         @type year_start: int
         @keyword year_start: Start year of search (i.e. '2014')
 
@@ -444,6 +447,8 @@ class PublicationsTool(object):
         @type debug: bool
         @param debug: activate the debug mode
 
+        The constructor expect that host and collections is provided or xml is provided instead.
+
         """
         self.collection_logs = []
         self.db = db
@@ -453,17 +458,31 @@ class PublicationsTool(object):
         self.id_category = id_category
         self.host = host
         self.collections = collections
+        self.xml = xml
         self.year_start = year_start
         self.year_end = year_end
         self.dbg = debug
-        self.harvester = None
         self.logs = []
 
         self.check = CheckAndFix()
         self.marc12 = Marc12()
+        
+        # check parameters
+        # protection team, project and/or category have to be defined    
+        if not self.id_team:
+            raise ToolException(MSG_NO_TEAM)
 
-        # Construct harvester Storage needed for the log
+        if not self.id_project:
+            raise ToolException(MSG_NO_PROJECT)
 
+        if not self.id_category:
+            raise ToolException(MSG_NO_CAT)
+
+        # xml or (host and collections) keyargs must be provided 
+        if not ((xml and not host and not collections) or (not xml and host and collections)):
+            raise ToolException(MSG_MISSING_PARAMETER)
+        
+        # Construct harvester Storage needed for the log        
         if host and collections: 
             self.harvester = Storage(id_teams=self.id_team,
                                      id_projects=self.id_project,
@@ -917,26 +936,15 @@ class PublicationsTool(object):
         if self.dbg:
             print "start processing", self.__class__.__name__
             print "decode request"
-
-        # protection team, project and/or category have to be defined
-        if not self.id_project:
-            raise ToolException(MSG_NO_PROJECT)
-
-        if not self.id_team:
-            raise ToolException(MSG_NO_TEAM)
-
+            
         if self.dbg:
             print "get harvest parameters"
 
         # process an XML request
-        if xml:
-            if not self.id_category:
-                raise ToolException(MSG_NO_CAT)
-            self.xml = xml 
-            
+        if self.xml:
             self.collection_logs.append(MsgCollection(found=1))
 
-            self.process_xml(xml)
+            self.process_xml(self.xml)
             return
 
         # retrieve the harvester parameter in the database