Minor index fix, for entries with spaces

This commit is contained in:
Peter Garst 2014-05-31 15:54:32 -07:00
parent 02ce48b451
commit 0cf3dc0545

View File

@ -26,7 +26,12 @@ class Field(object):
if not raw:
return
if self.name is None:
self.name, raw = raw.strip().partition(' ')[0::2]
# There are cases where partial index entries end with
# a significant space, along the lines of
# <>Summary <> ... <>Hearing<>.
# No known examples of starting with a space yet.
# self.name, raw = raw.strip().partition(' ')[0::2]
self.name, raw = raw.lstrip().partition(' ')[0::2]
self.buf.append(raw)
def finalize(self):
@ -100,6 +105,7 @@ class Fields(object):
def __call__(self, doc, log):
all_ids = frozenset(XPath('//*/@w:id')(doc))
# import pydevd;pydevd.settrace()
c = 0
while self.index_bookmark_prefix in all_ids:
c += 1