Clean up the handling of anchors and page breaks.

src/calibre/ebooks/mobi/mobiml.py

@@ -44,6 +44,7 @@ class BlockState(object):
         self.vmargin = 0.
         self.pbreak = False
         self.istate = None
+        self.content = False
 
 class FormatState(object):
     def __init__(self):
@@ -120,17 +121,20 @@ class MobiMLizer(object):
         return result
     
     def mobimlize_content(self, tag, text, bstate, istates):
+        bstate.content = True
         istate = istates[-1]
-        if istate.ids:
-            body = bstate.body
-            index = max((0, len(body) - 1))
-            for id in istate.ids:
-                body.insert(index, etree.Element('a', attrib={'id': id}))
-            istate.ids.clear()
         para = bstate.para
         if tag in SPECIAL_TAGS and not text:
             para = para if para is not None else bstate.body
         elif para is None:
+            body = bstate.body
+            if bstate.pbreak:
+                etree.SubElement(body, MBP('pagebreak'))
+                bstate.pbreak = False
+            if istate.ids:
+                for id in istate.ids:
+                    etree.SubElement(body, 'a', attrib={'id': id})
+                istate.ids.clear()
             bstate.istate = None
             parent = bstate.nested[-1] if bstate.nested else bstate.body
             indent = istate.indent
@@ -140,9 +144,6 @@ class MobiMLizer(object):
                 indent = 0
             elif indent != 0 and abs(indent) < self.profile.fbase:
                 indent = (indent / abs(indent)) * self.profile.fbase
-            if bstate.pbreak:
-                etree.SubElement(parent, MBP('pagebreak'))
-                bstate.pbreak = False
             if tag in NESTABLE_TAGS:
                 para = wrapper = etree.SubElement(parent, tag)
                 bstate.nested.append(para)
@@ -270,7 +271,7 @@ class MobiMLizer(object):
                 else:
                     last = elem[-1]
                     last.text = (last.text or '') + (u'\xa0' * spaces)
-        if style['page-break-before'] in PAGE_BREAKS:
+        if bstate.content and style['page-break-before'] in PAGE_BREAKS:
             bstate.pbreak = True
         istate.fsize = self.mobimlize_font(style['font-size'])
         istate.italic = True if style['font-style'] == 'italic' else False
@@ -340,7 +341,7 @@ class MobiMLizer(object):
                     tail = COLLAPSE.sub(' ', child.tail)
             if tail:
                 self.mobimlize_content(tag, tail, bstate, istates)
-        if style['page-break-after'] in PAGE_BREAKS:
+        if bstate.content and style['page-break-after'] in PAGE_BREAKS:
             bstate.pbreak = True
         if isblock:
             para = bstate.para

src/calibre/ebooks/mobi/writer.py

@@ -103,7 +103,7 @@ class Serializer(object):
             buffer.write('<reference title="%s" type="%s" '
                          % (ref.title, ref.type))
             self.serialize_href(ref.href)
-            buffer.write(' />')
+            buffer.write('/>')
         buffer.write('</guide>')
 
     def serialize_href(self, href, base=None):
@@ -118,8 +118,7 @@ class Serializer(object):
         if item and item.spine_position is None:
             return False
         id =  item.id if item else base.id
-        frag = frag if frag else 'calibre_top'
+        href = '#'.join((id, frag)) if frag else id
-        href = '#'.join((id, frag))
         buffer.write('filepos=')
         self.href_offsets[href].append(buffer.tell())
         buffer.write('0000000000')
@@ -138,10 +137,10 @@ class Serializer(object):
         buffer = self.buffer
         if not item.linear:
             self.breaks.append(buffer.tell() - 1)
-        self.id_offsets[item.id + '#calibre_top'] = buffer.tell()
+        self.id_offsets[item.id] = buffer.tell()
         for elem in item.data.find(XHTML('body')):
             self.serialize_elem(elem, item)
-        buffer.write(' <mbp:pagebreak/>')
+        buffer.write('<mbp:pagebreak/>')
 
     def serialize_elem(self, elem, item, nsrmap=NSRMAP):
         buffer = self.buffer
@@ -155,6 +154,9 @@ class Serializer(object):
                 id = '#'.join((item.id, elem.attrib[attr]))
                 self.id_offsets[id] = buffer.tell()
                 del elem.attrib[attr]
+        if tag == 'a' and not elem.attrib \
+           and not len(elem) and not elem.text:
+            return
         buffer.write('<')
         buffer.write(tag)
         if elem.attrib:

src/calibre/ebooks/oeb/base.py

@@ -249,10 +249,10 @@ class Metadata(object):
         self.oeb = oeb
         self.items = defaultdict(list)
 
-    def add(self, term, value, attrib={}, index=-1, **kwargs):
+    def add(self, term, value, attrib={}, **kwargs):
         item = self.Item(term, value, attrib, **kwargs)
         items = self.items[barename(item.term)]
-        items.insert(index, item)
+        items.append(item)
         return item
 
     def iterkeys(self):