Add option to detect chapters based on tagname and attributes

src/calibre/ebooks/lrf/__init__.py

@@ -158,12 +158,13 @@ def option_parser(usage, gui_mode=False):
     chapter = parser.add_option_group('CHAPTER OPTIONS')
     chapter.add_option('--disable-chapter-detection', action='store_true', 
                       default=False, dest='disable_chapter_detection', 
-                      help=_('''Prevent the automatic insertion of page breaks'''
-                      ''' before detected chapters.'''))
+                      help=_('''Prevent the automatic detection chapters.'''))
     chapter.add_option('--chapter-regex', dest='chapter_regex', 
                       default='chapter|book|appendix',
                       help=_('''The regular expression used to detect chapter titles.'''
                       ''' It is searched for in heading tags (h1-h6). Defaults to %default'''))
+    chapter.add_option('--chapter-attr', default='$,,$', 
+                       help=_('Detect a chapter beginning at an element having the specified attribute. The format for this option is tagname regexp,attribute name,attribute value regexp. For example to match all heading tags that have the attribute class="chapter" you would use "h\d,class,chapter". Default is %default'''))
     chapter.add_option('--page-break-before-tag', dest='page_break', default='h[12]',
                       help=_('''If html2lrf does not find any page breaks in the '''
                       '''html file and cannot detect chapter headings, it will '''

src/calibre/ebooks/lrf/html/convert_from.py

@@ -1427,6 +1427,18 @@ class HTMLConverter(object, LoggingInterface):
                 return
         except KeyError:
             pass
+        if not self.disable_chapter_detection and \
+           (self.chapter_attr[0].match(tagname) and \
+           tag.has_key(self.chapter_attr[1]) and \
+           self.chapter_attr[2].match(tag[self.chapter_attr[1]])):
+                self.log_debug('Detected chapter %s', tagname)
+                self.end_page() 
+                self.page_break_found = True
+
+                if self.options.add_chapters_to_toc:
+                    self.extra_toc_entries.append((self.get_text(tag,
+                        limit=1000), self.current_block))
+
         end_page = self.process_page_breaks(tag, tagname, tag_css)
         try:
             if tagname in ["title", "script", "meta", 'del', 'frameset']:            
@@ -1850,6 +1862,9 @@ def process_file(path, options, logger=None):
          re.compile('$')
     fpb = re.compile(options.force_page_break, re.IGNORECASE) if options.force_page_break else \
          re.compile('$')
+    cq = options.chapter_attr.split(',')
+    options.chapter_attr = [re.compile(cq[0], re.IGNORECASE), cq[1], 
+                            re.compile(cq[2], re.IGNORECASE)]
     options.force_page_break = fpb
     options.link_exclude = le
     options.page_break = pb

src/calibre/gui2/dialogs/lrf_single.py

@@ -1,8 +1,8 @@
 __license__   = 'GPL v3'
 __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
-import os, cPickle, codecs
+import os, codecs
 
-from PyQt4.QtCore import QObject, SIGNAL, Qt, QVariant, QByteArray
+from PyQt4.QtCore import QObject, SIGNAL, Qt
 from PyQt4.QtGui import QAbstractSpinBox, QLineEdit, QCheckBox, QDialog, \
                         QPixmap, QTextEdit, QListWidgetItem, QIcon
 
@@ -65,7 +65,7 @@ class LRFSingleDialog(QDialog, Ui_LRFSingleDialog):
         QObject.connect(self.categoryList, SIGNAL('itemEntered(QListWidgetItem *)'),
                         self.show_category_help)
         QObject.connect(self.cover_button, SIGNAL("clicked(bool)"), self.select_cover)
-        self.categoryList.leaveEvent = self.reset_help
+        #self.categoryList.leaveEvent = self.reset_help
         self.reset_help()
         self.selected_format = None
         self.initialize_common()
@@ -277,9 +277,9 @@ class LRFSingleDialog(QDialog, Ui_LRFSingleDialog):
                 obj.setWhatsThis(help)
                 self.option_map[guiname] = opt
                 obj.__class__.enterEvent = show_item_help
-                obj.leaveEvent = self.reset_help
+                #obj.leaveEvent = self.reset_help
         self.preprocess.__class__.enterEvent = show_item_help
-        self.preprocess.leaveEvent = self.reset_help
+        #self.preprocess.leaveEvent = self.reset_help
             
     
     def show_category_help(self, item):
@@ -293,7 +293,8 @@ class LRFSingleDialog(QDialog, Ui_LRFSingleDialog):
         self.set_help(help[text])
         
     def set_help(self, msg):
-        self.help_view.setHtml('<html><body>%s</body></html>'%(msg,))
+        if msg and getattr(msg, 'strip', lambda:True)():
+            self.help_view.setHtml('<html><body>%s</body></html>'%(msg,))
     
     def reset_help(self, *args):
         self.set_help(_('<font color="gray">No help available</font>'))

src/calibre/gui2/dialogs/lrf_single.ui

@@ -115,7 +115,7 @@
           <item row="0" column="0" >
            <widget class="QStackedWidget" name="stack" >
             <property name="currentIndex" >
-             <number>2</number>
+             <number>0</number>
             </property>
             <widget class="QWidget" name="metadata_page" >
              <property name="geometry" >
@@ -951,6 +951,19 @@
                  <item row="2" column="1" >
                   <widget class="QLineEdit" name="gui_force_page_break_before_attr" />
                  </item>
+                 <item row="3" column="0" >
+                  <widget class="QLabel" name="label_28" >
+                   <property name="text" >
+                    <string>Detect chapter &amp;at tag:</string>
+                   </property>
+                   <property name="buddy" >
+                    <cstring>gui_chapter_attr</cstring>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="3" column="1" >
+                  <widget class="QLineEdit" name="gui_chapter_attr" />
+                 </item>
                 </layout>
                </widget>
               </item>