Edit Book: Allow direct importing of HTML or DOCX files as new books without doing a full conversion. Use File->Import HTML or DOCX to import a file as a new EPUB book with minimal changes from the original.

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

@@ -0,0 +1,64 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+from __future__ import (unicode_literals, division, absolute_import,
+                        print_function)
+
+__license__ = 'GPL v3'
+__copyright__ = '2014, Kovid Goyal <kovid at kovidgoyal.net>'
+
+import os, sys
+
+from calibre.ptempfile import TemporaryDirectory
+from calibre.ebooks.conversion.plumber import Plumber
+from calibre.ebooks.oeb.polish.container import Container, OEB_DOCS, OEB_STYLES
+from calibre.ebooks.epub import initialize_container
+
+from calibre.utils.logging import default_log
+
+def auto_fill_manifest(container):
+    manifest_id_map = container.manifest_id_map
+    manifest_name_map = {v:k for k, v in manifest_id_map.iteritems()}
+
+    for name, mt in container.mime_map.iteritems():
+        if name not in manifest_name_map and not container.ok_to_be_unmanifested(name):
+            mitem = container.generate_item(name)
+            gname = container.href_to_name(mitem.get('href'), container.opf_name)
+            if gname != name:
+                raise ValueError('This should never happen (gname=%r, name=%r)' % (gname, name))
+            manifest_name_map[name] = mitem.get('id')
+            manifest_id_map[mitem.get('id')] = name
+
+def import_book_as_epub(srcpath, destpath, log=default_log):
+    if not destpath.lower().endswith('.epub'):
+        raise ValueError('Can only import books into the EPUB format, not %s' % (os.path.basename(destpath)))
+    with TemporaryDirectory('eei') as tdir:
+        plumber = Plumber(srcpath, tdir, log)
+        plumber.setup_options()
+        if srcpath.lower().endswith('.opf'):
+            plumber.opts.dont_package = True
+        if hasattr(plumber.opts, 'no_process'):
+            plumber.opts.no_process = True
+        plumber.input_plugin.for_viewer = True
+        with plumber.input_plugin, open(plumber.input, 'rb') as inf:
+            pathtoopf = plumber.input_plugin(inf, plumber.opts, plumber.input_fmt, log, {}, tdir)
+        if hasattr(pathtoopf, 'manifest'):
+            from calibre.ebooks.oeb.iterator.book import write_oebbook
+            pathtoopf = write_oebbook(pathtoopf, tdir)
+
+        c = Container(tdir, pathtoopf, log)
+        auto_fill_manifest(c)
+        # Auto fix all HTML/CSS
+        for name, mt in c.mime_map.iteritems():
+            if mt in set(OEB_DOCS) | set(OEB_STYLES):
+                c.parsed(name)
+                c.dirty(name)
+        c.commit()
+
+        zf = initialize_container(destpath, opf_name=c.opf_name)
+        with zf:
+            for name in c.name_path_map:
+                zf.writestr(name, c.raw_data(name, decode=False))
+
+if __name__ == '__main__':
+    import_book_as_epub(sys.argv[-2], sys.argv[-1])
+

src/calibre/gui2/tweak_book/boss.py

@@ -37,7 +37,7 @@ from calibre.gui2.tweak_book.toc import TOCEditor
 from calibre.gui2.tweak_book.editor import editor_from_syntax, syntax_from_mime
 from calibre.gui2.tweak_book.editor.insert_resource import get_resource_data, NewBook
 from calibre.gui2.tweak_book.preferences import Preferences
-from calibre.gui2.tweak_book.widgets import RationalizeFolders, MultiSplit
+from calibre.gui2.tweak_book.widgets import RationalizeFolders, MultiSplit, ImportForeign
 
 _diff_dialogs = []
 
@@ -167,6 +167,19 @@ class Boss(QObject):
                 create_book(d.mi, path, fmt=fmt)
                 self.open_book(path=path)
 
+    def import_book(self):
+        if not self._check_before_open():
+            return
+        d = ImportForeign(self.gui)
+        if d.exec_() == d.Accepted:
+            from calibre.ebooks.oeb.polish.import_book import import_book_as_epub
+            src, dest = d.data
+            self._clear_notify_data = True
+            def func(src, dest, tdir):
+                import_book_as_epub(src, dest)
+                return get_container(dest, tdir=tdir)
+            self.gui.blocking_job('import_book', _('Importing book, please wait...'), self.book_opened, func, src, dest, tdir=self.mkdtemp())
+
     def open_book(self, path=None, edit_file=None, clear_notify_data=True):
         if not self._check_before_open():
             return

src/calibre/gui2/tweak_book/ui.py

@@ -299,6 +299,8 @@ class Main(MainWindow):
         self.action_quit = reg('quit.png', _('&Quit'), self.boss.quit, 'quit', 'Ctrl+Q', _('Quit'))
         self.action_preferences = reg('config.png', _('&Preferences'), self.boss.preferences, 'preferences', 'Ctrl+P', _('Preferences'))
         self.action_new_book = reg('book.png', _('Create &new, empty book'), self.boss.new_book, 'new-book', (), _('Create a new, empty book'))
+        self.action_import_book = reg('book.png', _('&Import an HTML or DOCX file as a new book'),
+                                      self.boss.import_book, 'import-book', (), _('Import an HTML or DOCX file as a new book'))
 
         # Editor actions
         group = _('Editor actions')
@@ -429,6 +431,7 @@ class Main(MainWindow):
         f.addAction(self.action_import_files)
         f.addAction(self.action_open_book)
         f.addAction(self.action_new_book)
+        f.addAction(self.action_import_book)
         self.recent_books_menu = f.addMenu(_('&Recently opened books'))
         self.update_recent_books()
         f.addSeparator()

src/calibre/gui2/tweak_book/widgets.py

@@ -7,9 +7,10 @@ __license__ = 'GPL v3'
 __copyright__ = '2014, Kovid Goyal <kovid at kovidgoyal.net>'
 
 from PyQt4.Qt import (
-    QDialog, QDialogButtonBox, QGridLayout, QLabel, QLineEdit, QVBoxLayout)
+    QDialog, QDialogButtonBox, QGridLayout, QLabel, QLineEdit, QVBoxLayout,
+    QFormLayout, QHBoxLayout, QToolButton, QIcon, QApplication, Qt)
 
-from calibre.gui2 import error_dialog
+from calibre.gui2 import error_dialog, choose_files, choose_save_file
 from calibre.gui2.tweak_book import tprefs
 
 class Dialog(QDialog):
@@ -140,3 +141,86 @@ class MultiSplit(Dialog):  # {{{
         return self._xpath.xpath
 
 # }}}
+
+class ImportForeign(Dialog):  # {{{
+
+    def __init__(self, parent=None):
+        Dialog.__init__(self, _('Choose file to import'), 'import-foreign')
+
+    def sizeHint(self):
+        ans = Dialog.sizeHint(self)
+        ans.setWidth(ans.width() + 200)
+        return ans
+
+    def setup_ui(self):
+        self.l = l = QFormLayout(self)
+        self.setLayout(l)
+
+        la = self.la = QLabel(_(
+            'You can import an HTML or DOCX file directly as an EPUB and edit it. The EPUB'
+            ' will be generated with minimal changes from the source, unlike doing a full'
+            ' conversion in calibre.'))
+        la.setWordWrap(True)
+        l.addRow(la)
+
+        self.h1 = h1 = QHBoxLayout()
+        self.src = src = QLineEdit(self)
+        src.setPlaceholderText(_('Choose the file to import'))
+        h1.addWidget(src)
+        self.b1 = b = QToolButton(self)
+        b.setIcon(QIcon(I('document_open.png')))
+        b.setText(_('Choose file'))
+        h1.addWidget(b)
+        l.addRow(_('Source file'), h1)
+        b.clicked.connect(self.choose_source)
+        b.setFocus(Qt.OtherFocusReason)
+
+        self.h2 = h1 = QHBoxLayout()
+        self.dest = src = QLineEdit(self)
+        src.setPlaceholderText(_('Choose the location for the newly created EPUB'))
+        h1.addWidget(src)
+        self.b2 = b = QToolButton(self)
+        b.setIcon(QIcon(I('document_open.png')))
+        b.setText(_('Choose file'))
+        h1.addWidget(b)
+        l.addRow(_('Destination file'), h1)
+        b.clicked.connect(self.choose_destination)
+
+        l.addRow(self.bb)
+
+    def choose_source(self):
+        path = choose_files(self, 'edit-book-choose-file-to-import', _('Choose file'), filters=[
+            (_('HTML files'), ['htm', 'html', 'xhtml', 'xhtm']),
+            (_('DOCX files'), ['docx'])], select_only_single_file=True)
+        if path:
+            self.src.setText(path[0])
+            self.dest.setText(self.data[1])
+
+    def choose_destination(self):
+        path = choose_save_file(self, 'edit-book-destination-for-generated-epub', _('Choose destination'), filters=[
+            (_('EPUB files'), ['epub'])], all_files=False)
+        if path:
+            if not path.lower().endswith('.epub'):
+                path += '.epub'
+            self.dest.setText(path)
+
+    def accept(self):
+        if not unicode(self.src.text()):
+            return error_dialog(self, _('Need document'), _(
+                'You must specify the source file that will be imported.'), show=True)
+        Dialog.accept(self)
+
+    @property
+    def data(self):
+        src = unicode(self.src.text()).strip()
+        dest = unicode(self.dest.text()).strip()
+        if not dest:
+            dest = src.rpartition('.')[0] + '.epub'
+        return src, dest
+# }}}
+
+if __name__ == '__main__':
+    app = QApplication([])
+    d = ImportForeign()
+    d.exec_()
+    print (d.data)