Update RecordThesis in order to use author iterators.

modules/invenio_tools/recordthesis.py

@@ -4,9 +4,15 @@
 """
 from base import REG_AUTHOR, THESIS_DIR
 from filters import CLEAN_THESIS_DEFENSE
+from iterauthors import author_name, iter_author_fields
+from itertools import ifilter, ifilterfalse, imap
 from recordpubli import RecordPubli
 
 
+# the argument x is a dict corresponding to the field 100 or 700 for one author
+is_thesis_dir = lambda x: "e" in x and x["e"] == THESIS_DIR
+
+
 class RecordThesis(RecordPubli):
     """The MARC record describing a thesis.
     The relation between methods and MARC fields are the following::
@@ -28,50 +34,11 @@ class RecordThesis(RecordPubli):
             list: the list is empty when authors are not defined.
 
         """
-        authors = []
-        first_author = self.first_author()
-
-        # NOTE
-        # the content of the 700 field depend on the record type.
-        # For thesis it also contains the name of the director
-        if u"700" in self and isinstance(self[u"700"], dict):
-            if not ("e" in self[u"700"] and self[u"700"]["e"] == THESIS_DIR):
-                if "a" in self[u"700"]:
-                    authors.append(self[u"700"]["a"])
-
-        elif u"700" in self and isinstance(self[u"700"], list):
-            for di in self[u"700"]:
-                if "e" in di and di["e"] == THESIS_DIR:
-                    continue
-
-                if "a" in di:
-                    author = di["a"]
-
-                    # PROTECTION
-                    # in most of the case the author is a string
-                    # but it can be a list, e.g inspirehep.net/138663:
-                    # [u'Zuniga, J.', u'(the A.N.T.ARES. Collaboration)']
-                    if isinstance(author, unicode):
-                        authors.append(di["a"])
-
-                    elif isinstance(author, list):
-                        for elt in author:
-                            if REG_AUTHOR.match(elt):
-                                authors.append(elt)
-                                break
-
-        # the first author is defined not the other one
-        elif u"100" in self:
-            authors.append(first_author)
-
-        # sometime the first author is missing
-        if len(authors) == 0:
-            authors = [first_author]
-
-        elif first_author != authors[0]:
-            authors.insert(0, first_author)
-
-        return authors
+        # for a thesis, the author field 700 can contains names of director.
+        # they have to be removed.
+
+        iter_filter = ifilterfalse(is_thesis_dir, iter_author_fields(self))
+        return list(imap(author_name, iter_filter))
 
     def these_defense(self):
         """The defence date for a master/phd thesis.
@@ -109,21 +76,11 @@ class RecordThesis(RecordPubli):
                 * Empty string when it is not defined.
 
         """
-        # NOTE director are stored in the author field
-        # the first author is not a director !
-        li = []
-        if u"700" in self and isinstance(self[u"700"], dict):
-            if "e" in self[u"700"] and self[u"700"]["e"] == THESIS_DIR:
-                if "a" in self[u"700"]:
-                    li.append(self[u"700"]["a"])
-
-        elif u"700" in self and isinstance(self[u"700"], list):
-            for di in self[u"700"]:
-                if "e" in di and di["e"] == THESIS_DIR:
-                    if "a" in di:
-                        li.append(di["a"])
-
-        return ', '.join(li)
+        # for a thesis, the author field 700 field can contains
+        # names of the director.
+
+        iter_filter = ifilter(is_thesis_dir, iter_author_fields(self))
+        return u", ".join(list(imap(author_name, iter_filter)))
 
     def these_town(self):
         """The town where the thesis took place.