Implement fixing html to create well formed markup

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

@@ -0,0 +1,22 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+from __future__ import (unicode_literals, division, absolute_import,
+                        print_function)
+
+__license__ = 'GPL v3'
+__copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
+
+
+from calibre.ebooks.oeb.polish.parsing import parse
+from calibre.ebooks.oeb.base import serialize, OEB_DOCS
+
+def fix_html(raw):
+    root = parse(raw)
+    return serialize(root, 'text/html').decode('utf-8')
+
+def fix_all_html(container):
+    for name, mt in container.mime_map.iteritems():
+        if mt in OEB_DOCS:
+            container.parsed(name)
+            container.dirty(name)
+

src/calibre/gui2/tweak_book/boss.py

@@ -19,6 +19,7 @@ from calibre.ptempfile import PersistentTemporaryDirectory
 from calibre.ebooks.oeb.base import urlnormalize
 from calibre.ebooks.oeb.polish.main import SUPPORTED, tweak_polish
 from calibre.ebooks.oeb.polish.container import get_container as _gc, clone_container, guess_type
+from calibre.ebooks.oeb.polish.pretty import fix_all_html
 from calibre.ebooks.oeb.polish.replace import rename_files
 from calibre.ebooks.oeb.polish.split import split, merge, AbortError
 from calibre.gui2 import error_dialog, choose_files, question_dialog, info_dialog
@@ -306,6 +307,18 @@ class Boss(QObject):
             self.update_global_history_actions()
     # }}}
 
+    def fix_html(self, current):
+        if current:
+            ed = self.gui.central.current_editor
+            if hasattr(ed, 'fix_html'):
+                ed.fix_html()
+        else:
+            if not self.check_dirtied():
+                return
+            self.add_savepoint(_('Fix HTML'))
+            fix_all_html(current_container())
+            self.update_editors_from_container()
+
     def mark_selected_text(self):
         ed = self.gui.central.current_editor
         if ed is not None:
@@ -657,6 +670,7 @@ class Boss(QObject):
             actions['editor-cut'].setEnabled(ed.copy_available)
             actions['editor-copy'].setEnabled(ed.cut_available)
             actions['go-to-line-number'].setEnabled(ed.has_line_numbers)
+            actions['fix-html-current'].setEnabled(ed.syntax == 'html')
             self.gui.keyboard.set_mode(ed.syntax)
             name = None
             for n, x in editors.iteritems():

src/calibre/gui2/tweak_book/editor/text.py

@@ -164,6 +164,10 @@ class TextEdit(QPlainTextEdit):
             sel.append(self.current_search_mark)
         self.setExtraSelections(sel)
 
+    def fix_html(self):
+        from calibre.ebooks.oeb.polish.pretty import fix_html
+        self.replace_text(fix_html(unicode(self.toPlainText())).decode('utf-8'))
+
     # Search and replace {{{
     def mark_selected_text(self):
         sel = QTextEdit.ExtraSelection()

src/calibre/gui2/tweak_book/editor/widget.py

@@ -130,6 +130,9 @@ class Editor(QMainWindow):
                 b.addAction(actions['editor-%s' % x])
             except KeyError:
                 pass
+        self.tools_bar = b = self.addToolBar(_('Editor tools'))
+        if self.syntax == 'html':
+            b.addAction(actions['fix-html-current'])
 
     def break_cycles(self):
         self.modification_state_changed.disconnect()
@@ -178,6 +181,10 @@ class Editor(QMainWindow):
     def contextMenuEvent(self, ev):
         ev.ignore()
 
+    def fix_html(self):
+        if self.syntax == 'html':
+            self.editor.fix_html()
+
 def launch_editor(path_to_edit, path_is_raw=False, syntax='html'):
     if path_is_raw:
         raw = path_to_edit

src/calibre/gui2/tweak_book/ui.py

@@ -210,6 +210,10 @@ class Main(MainWindow):
         # Tool actions
         group = _('Tools')
         self.action_toc = reg('toc.png', _('&Edit Table of Contents'), self.boss.edit_toc, 'edit-toc', (), _('Edit Table of Contents'))
+        self.action_fix_html_current = reg('html-fix.png', _('&Fix HTML'), partial(self.boss.fix_html, True), 'fix-html-current', (),
+                                           _('Fix HTML in the current file'))
+        self.action_fix_html_all = reg('html-fix.png', _('&Fix HTML - all files'), partial(self.boss.fix_html, False), 'fix-html-all', (),
+                                       _('Fix HTML in all files'))
 
         # Polish actions
         group = _('Polish Book')
@@ -286,6 +290,7 @@ class Main(MainWindow):
         e.addAction(self.action_embed_fonts)
         e.addAction(self.action_subset_fonts)
         e.addAction(self.action_smarten_punctuation)
+        e.addAction(self.action_fix_html_all)
 
         e = b.addMenu(_('&View'))
         t = e.addMenu(_('Tool&bars'))