diff --git a/src/calibre/customize/profiles.py b/src/calibre/customize/profiles.py
index 2b5eb5011e..ba0cd187e4 100644
--- a/src/calibre/customize/profiles.py
+++ b/src/calibre/customize/profiles.py
@@ -61,6 +61,7 @@ class SonyReaderInput(InputProfile):
     dpi                       = 168.451
     fbase                     = 12
     fsizes                    = [7.5, 9, 10, 12, 15.5, 20, 22, 24]
+    #unsupported_unicode_chars = [\u2018, \u2019, \u201a, \u201b, \u201c, \u201d, \u201e, \u201f]
 
 class SonyReader300Input(SonyReaderInput):
 
@@ -250,6 +251,9 @@ class OutputProfile(Plugin):
 
     #: The character used to represent a star in ratings
     ratings_char = u'*'
+    
+    #: Unsupported unicode characters to be replaced during preprocessing
+    unsupported_unicode_chars = []
 
     @classmethod
     def tags_to_string(cls, tags):
diff --git a/src/calibre/ebooks/conversion/preprocess.py b/src/calibre/ebooks/conversion/preprocess.py
index 0b981cf6f7..b4815cb35e 100644
--- a/src/calibre/ebooks/conversion/preprocess.py
+++ b/src/calibre/ebooks/conversion/preprocess.py
@@ -182,8 +182,10 @@ class Dehyphenator(object):
         lookupword = self.removesuffixes.sub('', dehyphenated)
         if self.prefixes.match(firsthalf) is None:
            lookupword = self.removeprefix.sub('', lookupword)
-        booklookup = re.compile(u'%s' % lookupword, re.IGNORECASE)
+        # escape any meta-characters which may be in the lookup word
+        lookupword = re.sub(r'(?P<meta>[\[\]\\\^\$\.\|\?\*\+\(\)])', r'\\\g<meta>', lookupword)
         #print "lookup word is: "+str(lookupword)+", orig is: " + str(hyphenated)
+        booklookup = re.compile(u'%s' % lookupword, re.IGNORECASE)
         if self.format == 'html_cleanup':
            match = booklookup.search(self.html)
            hyphenmatch = re.search(u'%s' % hyphenated, self.html)