EPUB Input: Handle EPUB files with components encoded in an encoding other than UTF-8 correctly, though why anyone would do that is a mystery. Fixes #7196 (Non UTF-8 in ePub)

src/calibre/customize/conversion.py

@@ -120,6 +120,11 @@ class InputFormatPlugin(Plugin):
     #: to make its output suitable for viewing
     for_viewer = False
 
+    #: The encoding that this input plugin creates files in. A value of
+    #: None means that the encoding is undefined and must be
+    #: detected individually
+    output_encoding = 'utf-8'
+
     #: Options shared by all Input format plugins. Do not override
     #: in sub-classes. Use :attr:`options` instead. Every option must be an
     #: instance of :class:`OptionRecommendation`.

src/calibre/ebooks/conversion/plumber.py

@@ -838,7 +838,8 @@ OptionRecommendation(name='timestamp',
                 self.opts_to_mi(self.user_metadata)
             if not hasattr(self.oeb, 'manifest'):
                 self.oeb = create_oebbook(self.log, self.oeb, self.opts,
-                        self.input_plugin)
+                        self.input_plugin,
+                        encoding=self.input_plugin.output_encoding)
             self.input_plugin.postprocess_book(self.oeb, self.opts, self.log)
             self.opts.is_image_collection = self.input_plugin.is_image_collection
             pr = CompositeProgressReporter(0.34, 0.67, self.ui_reporter)

src/calibre/ebooks/epub/input.py

@@ -16,6 +16,7 @@ class EPUBInput(InputFormatPlugin):
     author      = 'Kovid Goyal'
     description = 'Convert EPUB files (.epub) to HTML'
     file_types  = set(['epub'])
+    output_encoding = None
 
     recommendations = set([('page_breaks_before', '/', OptionRecommendation.MED)])