Preserve existing nav markup when converting from EPUB 3 to EPUB 3

src/calibre/ebooks/conversion/plugins/epub_input.py

@@ -289,7 +289,7 @@ class EPUBInput(InputFormatPlugin):
 
         epub3_nav = opf.epub3_nav
         if epub3_nav is not None:
-            self.convert_epub3_nav(epub3_nav, opf, log)
+            self.convert_epub3_nav(epub3_nav, opf, log, options)
 
         if len(parts) > 1 and parts[0]:
             delta = '/'.join(parts[:-1])+'/'
@@ -346,11 +346,11 @@ class EPUBInput(InputFormatPlugin):
 
         return os.path.abspath(u'content.opf')
 
-    def convert_epub3_nav(self, nav_path, opf, log):
+    def convert_epub3_nav(self, nav_path, opf, log, opts):
         from lxml import etree
         from calibre.ebooks.chardet import xml_to_unicode
         from calibre.ebooks.oeb.polish.parsing import parse
-        from calibre.ebooks.oeb.base import EPUB_NS, XHTML, NCX_MIME, NCX
+        from calibre.ebooks.oeb.base import EPUB_NS, XHTML, NCX_MIME, NCX, urlnormalize
         from calibre.ebooks.oeb.polish.toc import first_child
         from tempfile import NamedTemporaryFile
         with lopen(nav_path, 'rb') as f:
@@ -401,6 +401,9 @@ class EPUBInput(InputFormatPlugin):
         ncx_id = opf.add_path_to_manifest(f.name, NCX_MIME)
         for spine in opf.root.xpath('//*[local-name()="spine"]'):
             spine.set('toc', ncx_id)
+        href = os.path.relpath(nav_path).replace(os.sep, '/')
+        opts.epub3_nav_href = urlnormalize(href)
+        opts.epub3_nav_parsed = root
 
     def postprocess_book(self, oeb, opts, log):
         rc = getattr(self, 'removed_cover', None)

src/calibre/ebooks/conversion/plugins/epub_output.py

@@ -295,7 +295,8 @@ class EPUBOutput(OutputFormatPlugin):
         from calibre.ebooks.oeb.polish.container import EpubContainer
         container = EpubContainer(tdir, self.log)
         from calibre.ebooks.oeb.polish.upgrade import epub_2_to_3
-        epub_2_to_3(container, self.log.info)
+        existing_nav = getattr(self.opts, 'epub3_nav_parsed', None)
+        epub_2_to_3(container, self.log.info, previous_nav=existing_nav)
         container.commit()
         os.remove(f.name)
         try:

src/calibre/ebooks/oeb/polish/toc.py

@@ -655,14 +655,17 @@ def ensure_single_nav_of_type(root, ntype='toc'):
     return nav
 
 
-def commit_nav_toc(container, toc, lang=None, landmarks=None):
+def commit_nav_toc(container, toc, lang=None, landmarks=None, previous_nav=None):
     from calibre.ebooks.oeb.polish.pretty import pretty_xml_tree
     tocname = find_existing_nav_toc(container)
     if tocname is None:
         item = container.generate_item('nav.xhtml', id_prefix='nav')
         item.set('properties', 'nav')
         tocname = container.href_to_name(item.get('href'), base=container.opf_name)
-        root = container.parse_xhtml(P('templates/new_nav.html', data=True).decode('utf-8'))
+        if previous_nav is not None:
+            root = previous_nav
+        else:
+            root = container.parse_xhtml(P('templates/new_nav.html', data=True).decode('utf-8'))
         container.replace(tocname, root)
     else:
         root = container.parsed(tocname)

src/calibre/ebooks/oeb/polish/upgrade.py

@@ -100,7 +100,7 @@ guide_epubtype_map = {
 }
 
 
-def create_nav(container, toc, landmarks):
+def create_nav(container, toc, landmarks, previous_nav=None):
     lang = get_book_language(container)
     if lang == 'und':
         lang = None
@@ -109,10 +109,10 @@ def create_nav(container, toc, landmarks):
             entry['type'] = guide_epubtype_map.get(entry['type'].lower())
             if entry['type'] == 'cover' and container.mime_map.get(entry['dest'], '').lower() in OEB_DOCS:
                 container.apply_unique_properties(entry['dest'], 'calibre:title-page')
-    commit_nav_toc(container, toc, lang=lang, landmarks=landmarks)
+    commit_nav_toc(container, toc, lang=lang, landmarks=landmarks, previous_nav=previous_nav)
 
 
-def epub_2_to_3(container, report):
+def epub_2_to_3(container, report, previous_nav=None):
     upgrade_metadata(container.opf)
     collect_properties(container)
     toc = get_toc(container)
@@ -123,7 +123,7 @@ def epub_2_to_3(container, report):
     landmarks = get_landmarks(container)
     for guide in container.opf_xpath('./opf:guide'):
         guide.getparent().remove(guide)
-    create_nav(container, toc, landmarks)
+    create_nav(container, toc, landmarks, previous_nav)
     container.opf.set('version', '3.0')
     fix_font_mime_types(container)
     container.dirty(container.opf_name)

src/calibre/ebooks/oeb/transforms/flatcss.py

@@ -179,6 +179,11 @@ class CSSFlattener(object):
             titlepage = titlepage.item
             if titlepage is not None and titlepage not in self.items:
                 self.items.append(titlepage)
+        epub3_nav = None
+        if getattr(self.opts, 'epub3_nav_href', None):
+            epub3_nav = self.oeb.manifest.hrefs.get(self.opts.epub3_nav_href)
+            if epub3_nav is not None and epub3_nav not in self.items:
+                self.items.append(epub3_nav)
 
         self.filter_css = frozenset()
         if self.opts.filter_css:
@@ -210,6 +215,8 @@ class CSSFlattener(object):
         self.sbase = self.baseline_spine() if self.fbase else None
         self.fmap = FontMapper(self.sbase, self.fbase, self.fkey)
         self.flatten_spine()
+        if epub3_nav is not None:
+            self.opts.epub3_nav_parsed = epub3_nav.data
 
     def get_embed_font_info(self, family, failure_critical=True):
         efi = []
@@ -436,12 +443,11 @@ class CSSFlattener(object):
             cssdict['font-weight'] = 'normal'  # ADE chokes on font-weight medium
 
         fsize = font_size
-        is_drop_cap = (cssdict.get('float', None) == 'left' and 'font-size' in
+        is_drop_cap = (cssdict.get('float', None) == 'left' and 'font-size' in cssdict and len(node) == 0 and node.text and (
-                       cssdict and len(node) == 0 and node.text and
+            len(node.text) == 1 or (len(node.text) == 2 and 0x2000 <= ord(node.text[0]) <= 0x206f)))
-                       (len(node.text) == 1 or (len(node.text) == 2 and 0x2000 <= ord(node.text[0]) <= 0x206f)))
         # Detect drop caps generated by the docx input plugin
-        if (node.tag and node.tag.endswith('}p') and len(node) == 0 and node.text and len(node.text.strip()) == 1 and
+        if node.tag and node.tag.endswith('}p') and len(node) == 0 and node.text and len(node.text.strip()) == 1 and \
-            not node.tail and 'line-height' in cssdict and 'font-size' in cssdict):
+                not node.tail and 'line-height' in cssdict and 'font-size' in cssdict:
             dp = node.getparent()
             if dp.tag and dp.tag.endswith('}div') and len(dp) == 1 and not dp.text:
                 if stylizer.style(dp).cssdict().get('float', None) == 'left':
@@ -473,8 +479,8 @@ class CSSFlattener(object):
         if cssdict:
             for x in self.filter_css:
                 popval = cssdict.pop(x, None)
-                if (self.body_font_family and popval and x == 'font-family' and
+                if self.body_font_family and popval and x == 'font-family' \
-                    popval.partition(',')[0][1:-1] == self.body_font_family.partition(',')[0][1:-1]):
+                    and popval.partition(',')[0][1:-1] == self.body_font_family.partition(',')[0][1:-1]:
                     cssdict[x] = popval
 
         if cssdict:
@@ -499,8 +505,7 @@ class CSSFlattener(object):
             lineh = self.lineh / psize
             cssdict['line-height'] = "%0.5fem" % lineh
 
-        if (self.context.remove_paragraph_spacing or
+        if (self.context.remove_paragraph_spacing or self.context.insert_blank_line) and tag in ('p', 'div'):
-                self.context.insert_blank_line) and tag in ('p', 'div'):
             if item_id != 'calibre_jacket' or self.context.output_profile.name == 'Kindle':
                 for prop in ('margin', 'padding', 'border'):
                     for edge in ('top', 'bottom'):
@@ -510,8 +515,7 @@ class CSSFlattener(object):
                     '%fem'%self.context.insert_blank_line_size
             indent_size = self.context.remove_paragraph_spacing_indent_size
             keep_indents = indent_size < 0.0
-            if (self.context.remove_paragraph_spacing and not keep_indents and
+            if (self.context.remove_paragraph_spacing and not keep_indents and cssdict.get('text-align', None) not in ('center', 'right')):
-                cssdict.get('text-align', None) not in ('center', 'right')):
                 cssdict['text-indent'] =  "%1.1fem" % indent_size
 
         pseudo_classes = style.pseudo_classes(self.filter_css)
@@ -618,8 +622,7 @@ class CSSFlattener(object):
             items = sorted(stylizer.page_rule.items())
             css = ';\n'.join("%s: %s" % (key, val) for key, val in items)
             css = ('@page {\n%s\n}\n'%css) if items else ''
-            rules = [r.cssText for r in stylizer.font_face_rules +
+            rules = [r.cssText for r in stylizer.font_face_rules + self.embed_font_rules]
-                    self.embed_font_rules]
             raw = '\n\n'.join(rules)
             css += '\n\n' + raw
             global_css[css].append(item)