FB2 Output: Experimental option to break book into sections based upon chapters.

.bzrignore

@@ -28,3 +28,4 @@ nbproject/
 *.userprefs
 .project
 .pydevproject
+.settings/

src/calibre/ebooks/fb2/fb2ml.py

@@ -86,10 +86,12 @@ class FB2MLizer(object):
         output.append(self.fb2_footer())
         output = ''.join(output).replace(u'ghji87yhjko0Caliblre-toc-placeholder-for-insertion-later8ujko0987yjk', self.get_toc())
         output = self.clean_text(output)
+        if self.opts.sectionize_chapters:
+            output = self.sectionize_chapters(output)
         return u'<?xml version="1.0" encoding="UTF-8"?>\n%s' % etree.tostring(etree.fromstring(output), encoding=unicode, pretty_print=True)
 
     def clean_text(self, text):
-        text = re.sub('<p>[ ]*</p>', '', text)
+        text = re.sub('<p>\s*</p>', '', text)
 
         return text
 
@@ -149,6 +151,11 @@ class FB2MLizer(object):
                     self.oeb.warn('Ignoring toc item: %s not found in document.' % item)
         return ''.join(toc)
 
+    def sectionize_chapters(self, text):
+        text = re.sub(r'(?imsu)(?P<anchor><a\s+id="calibre_link-\d+"\s*/>)\s*(?P<strong>(<p>)*\s*<strong>.+?</strong>\s*(</p>)*)', lambda mo: '</section><section>%s<title>%s</title>' % (mo.group('anchor'), mo.group('strong')), text)
+        text = re.sub(r'(?imsu)<p>\s*(?P<anchor><a\s+id="calibre_link-\d+"\s*/>)\s*</p>\s*(?P<strong>(<p>)*\s*<strong>.+?</strong>\s*(</p>)*)', lambda mo: '</section><section>%s<title>%s</title>' % (mo.group('anchor'), mo.group('strong')), text)
+        return text
+
     def get_text(self):
         text = []
         for item in self.oeb_book.spine:

src/calibre/ebooks/fb2/output.py

@@ -19,6 +19,12 @@ class FB2Output(OutputFormatPlugin):
         OptionRecommendation(name='inline_toc',
             recommended_value=False, level=OptionRecommendation.LOW,
             help=_('Add Table of Contents to beginning of the book.')),
+        OptionRecommendation(name='sectionize_chapters',
+            recommended_value=False, level=OptionRecommendation.LOW,
+            help=_('Try to turn chapters into individual sections. ' \
+                   'WARNING: ' \
+                   'This option is experimental. It can cause conversion ' \
+                   'to fail. It can also produce unexpected output.')),
     ])
 
     def convert(self, oeb_book, output_path, input_plugin, opts, log):

src/calibre/gui2/convert/fb2_output.py

@@ -16,6 +16,6 @@ class PluginWidget(Widget, Ui_Form):
     COMMIT_NAME = 'fb2_output'
 
     def __init__(self, parent, get_option, get_help, db=None, book_id=None):
-        Widget.__init__(self, parent, ['inline_toc'])
+        Widget.__init__(self, parent, ['inline_toc', 'sectionize_chapters'])
         self.db, self.book_id = db, book_id
         self.initialize_options(get_option, get_help, db, book_id)

src/calibre/gui2/convert/fb2_output.ui

@@ -14,7 +14,7 @@
    <string>Form</string>
   </property>
   <layout class="QGridLayout" name="gridLayout">
-   <item row="1" column="0">
+   <item row="2" column="0">
     <spacer name="verticalSpacer">
      <property name="orientation">
       <enum>Qt::Vertical</enum>
@@ -34,6 +34,13 @@
      </property>
     </widget>
    </item>
+   <item row="1" column="0">
+    <widget class="QCheckBox" name="opt_sectionize_chapters">
+     <property name="text">
+      <string>Sectionize Chapters (Use with care!)</string>
+     </property>
+    </widget>
+   </item>
   </layout>
  </widget>
  <resources/>