Fix bug #2914: Ignore invalid unicode code points in eReader.

src/calibre/ebooks/pdb/ereader/reader132.py

@@ -52,6 +52,8 @@ class Reader132(FormatReader):
     def __init__(self, header, stream, log, encoding=None):
         self.log = log
         self.encoding = encoding
+        
+        self.log.debug('132 byte header version found.')
 
         self.sections = []
         for i in range(header.num_sections):

src/calibre/ebooks/pdb/ereader/reader202.py

@@ -38,6 +38,8 @@ class Reader202(FormatReader):
         self.log = log
         self.encoding = encoding
 
+        self.log.debug('202 byte header version found.')
+
         self.sections = []
         for i in range(header.num_sections):
             self.sections.append(header.section_data(i))

src/calibre/ebooks/pml/pmlconverter.py

@@ -10,6 +10,7 @@ __docformat__ = 'restructuredtext en'
 
 import re
 
+from calibre import my_unichr
 from calibre.ebooks.pdb.ereader import image_name
 
 PML_HTML_RULES = [
@@ -35,7 +36,7 @@ PML_HTML_RULES = [
     (re.compile(r'\\Sb(?P<text>.+?)\\Sb', re.DOTALL), lambda match: '<sub>%s</sub>' % match.group('text')),
     (re.compile(r'\\k(?P<text>.+?)\\k', re.DOTALL), lambda match: '<small>%s</small>' % match.group('text')),
     (re.compile(r'\\a(?P<num>\d\d\d)'), lambda match: '&#%s;' % match.group('num')),
-    (re.compile(r'\\U(?P<num>\d+)'), lambda match: '%s' % unichr(int(match.group('num'), 16))),
+    (re.compile(r'\\U(?P<num>\d+)'), lambda match: '%s' % my_unichr(int(match.group('num'), 16))),
     (re.compile(r'\\m="(?P<name>.+?)"'), lambda match: '<img src="images/%s" />' % image_name(match.group('name')).strip('\x00')),
     (re.compile(r'\\q="(?P<target>#.+?)"(?P<text>.+?)\\q', re.DOTALL), lambda match: '<a href="%s">%s</a>' % (match.group('target'), match.group('text'))),
     (re.compile(r'\\Q="(?P<target>.+?)"'), lambda match: '<span id="%s"></span>' % match.group('target')),