Modify the signature of the method Article.get_field_by_origin.

modules/harvest_tools/articles.py

@@ -77,66 +77,6 @@ class Articles(Automaton):
 
         return True
 
-    def get_record_by_origin(self,
-                             id_publisher=None,
-                             my_authors=None,
-                             oai_url=None,
-                             pages=None,
-                             publication_url=None,
-                             title=None,
-                             volume=None,
-                             year=None):
-        """Get an existing record using the origin field and its value
-        defined in the ``oai_url`` keyword argument.
-        The other arguments are used to transform the corresponding preprint
-        into an article.
-
-        Keyword Args:
-            oai_url (unicode): the OAI identifier of the article.
-            id_publisher (int): identifier of the publisher in the database.
-            my_authors (unicode): authors of my institute separated by a comma.
-            pages (unicode):  the page reference.
-            publication_url (unicode): the URL of the publications
-            title (unicode): the title of the publication.
-            volume (unicode): the volume reference.
-            year (unicode): the year of publication.
-
-        Returns:
-            tuple: ``(id, status)`` which contains the ``id`` of the record.
-            It is equal to ``None`` when nothing is found.
-            The ``status`` is equal to one when the existing preprint was
-            modified into article, zero otherwise
-
-        """
-        if self.dbg:
-            print "check existing article by origin"
-
-        db = self.db
-
-        rec_id = get_id(db.publications, origin=oai_url)
-        if not rec_id:
-            return (None, 0)
-
-        # not a preprint ?
-        if db.publications[rec_id].id_categories != self.id_preprint:
-            self.logs[-1].idle(MSG_IN_DB, year)
-            return (rec_id, 0)
-
-        # transform a preprint into an article
-        self.logs[-1].modify(MSG_TRANSFORM_PREPRINT, year)
-        if not self.dry_run:
-            db.publications[rec_id] = dict(authors_institute=my_authors,
-                                           id_categories=self.id_category,
-                                           id_publishers=id_publisher,
-                                           id_status=UNDEF_ID,
-                                           pages=pages,
-                                           publication_url=publication_url,
-                                           title=title,
-                                           volume=volume,
-                                           year=year)
-
-        return (rec_id, 1)
-
     def get_record_by_fields(self,
                              oai_url,
                              year,
@@ -239,6 +179,85 @@ class Articles(Automaton):
 
         return (rec_id, 1)
 
+    def get_record_by_origin(self,
+                             primary_oai_url,
+                             year,
+                             id_publisher=None,
+                             my_authors=None,
+                             oai_url=None,
+                             pages=None,
+                             publication_url=None,
+                             title=None,
+                             volume=None):
+        """Get an existing record using the origin field and its value
+        defined in the *primary_oai_url* argument.
+
+        Note:
+            This method is required to transform a preprint into and article.
+            All the keyword arguments are needed by the transformation.
+
+        Args:
+            primary_oai_url (unicode): the *primary* OAI identifier of the
+                record. It is used by the search algorithm.
+
+            year (unicode): the year of publication which is used
+                by the logger.
+
+        Keyword Args:
+            id_publisher (int): identifier of the publisher in the database.
+            my_authors (unicode): authors of my institute separated by a comma.
+            oai_url (unicode): the full oai_url(s) of the article.
+            pages (unicode):  the page reference.
+            publication_url (unicode): the URL of the publications
+            title (unicode): the title of the publication.
+            volume (unicode): the volume reference.
+
+        Returns:
+            tuple: ``(id, status)`` which contains the ``id`` of the record.
+            It is equal to ``None`` when nothing is found.
+            The ``status`` is equal to one when the existing preprint was
+            modified into article, zero otherwise
+
+        """
+        if self.dbg:
+            print "check existing article by origin"
+
+        # alias
+        db = self.db
+        logs = self.logs
+        publications = db.publications
+
+        # search by origin
+        query = db.publications.origin.contains(primary_oai_url)
+        setrows = db(query)
+        if setrows.count() == 0:
+            return (None, 0)
+
+        # a record is found
+        rec_id = setrows.select(publications.id).first().id
+        publication = publications[rec_id]
+
+        # not a preprint ?
+        if publication.id_categories != self.id_preprint:
+            logs[-1].idle(MSG_IN_DB, year)
+            return (rec_id, 0)
+
+        # transform a preprint into an article
+        logs[-1].modify(MSG_TRANSFORM_PREPRINT, year)
+        if not self.dry_run:
+            db.publications[rec_id] = dict(authors_institute=my_authors,
+                                           id_categories=self.id_category,
+                                           id_publishers=id_publisher,
+                                           id_status=UNDEF_ID,
+                                           oai_url=oai_url,
+                                           pages=pages,
+                                           publication_url=publication_url,
+                                           title=title,
+                                           volume=volume,
+                                           year=year)
+
+        return (rec_id, 1)
+
     def insert_record(self, record):
         """Insert an article in the database.
 
@@ -286,10 +305,11 @@ class Articles(Automaton):
                       pages=pages,
                       publication_url=publication_url,
                       title=title,
-                      volume=volume,
-                      year=year)
+                      volume=volume)
 
-        rec_id, status = self.get_record_by_origin(**fields)
+        rec_id, status = self.get_record_by_origin(record.primary_oai_url(),
+                                                   year,
+                                                   **fields)
         if rec_id:
             return status