CHM Input: Do not choke on CHM files with non ascii internal filenames on windows. Fixes #917696 (Several problems in handling CHM file)

src/calibre/ebooks/chm/input.py

@@ -11,6 +11,7 @@ from calibre.customize.conversion import InputFormatPlugin
 from calibre.ptempfile import TemporaryDirectory
 from calibre.utils.localization import get_lang
 from calibre.utils.filenames import ascii_filename
+from calibre.constants import filesystem_encoding
 
 class CHMInput(InputFormatPlugin):
 
@@ -36,6 +37,8 @@ class CHMInput(InputFormatPlugin):
 
         log.debug('Processing CHM...')
         with TemporaryDirectory('_chm2oeb') as tdir:
+            if not isinstance(tdir, unicode):
+                tdir = tdir.decode(filesystem_encoding)
             html_input = plugin_for_input_format('html')
             for opt in html_input.options:
                 setattr(options, opt.option.name, opt.recommended_value)

src/calibre/ebooks/chm/reader.py

@@ -4,7 +4,7 @@ __license__ = 'GPL v3'
 __copyright__  = '2008, Kovid Goyal <kovid at kovidgoyal.net>,' \
                  ' and Alex Bramley <a.bramley at gmail.com>.'
 
-import os, re
+import os, re, codecs
 
 from calibre import guess_type as guess_mimetype
 from calibre.ebooks.BeautifulSoup import BeautifulSoup, NavigableString
@@ -99,8 +99,17 @@ class CHMReader(CHMFile):
 
     def ExtractFiles(self, output_dir=os.getcwdu(), debug_dump=False):
         html_files = set([])
+        try:
+            x = self.GetEncoding()
+            codecs.lookup(x)
+            enc = x
+        except:
+            enc = None
         for path in self.Contents():
-            lpath = os.path.join(output_dir, path)
+            fpath = path
+            if not isinstance(path, unicode) and enc:
+                fpath = path.decode(enc)
+            lpath = os.path.join(output_dir, fpath)
             self._ensure_dir(lpath)
             try:
                 data = self.GetFile(path)
@@ -123,6 +132,7 @@ class CHMReader(CHMFile):
                     self.log.warn('%r filename too long, skipping'%path)
                     continue
                 raise
+
         if debug_dump:
             import shutil
             shutil.copytree(output_dir, os.path.join(debug_dump, 'debug_dump'))