def assert_pdf(pdf_part, raise_exc=True, extra=None):
if not pdf_part:
LOG.error('cannot check %r', pdf_part)
return False
pdf_part_s = repr(pdf_part)[:100]
start = None
if hasattr(pdf_part, 'seek'):
p = pdf_part.tell()
start = pdf_part.read(1024)
pdf_part.seek(p, 0)
elif isinstance(pdf_part, (basestring, buffer)):
start = pdf_part[:1024]
else:
LOG.warn('cannot check %s (%s %s)', pdf_part_s, type(pdf_part), extra)
return None
if not start:
LOG.error('cannot check %r', start)
if start.startswith('%PDF-'):
LOG.debug('pdf_check ok on %r (%s)', pdf_part_s, extra)
return True
if 'application/pdf' == guess_data_mime(start):
LOG.debug('pdf_check ok on %r (%s) by Magic!', pdf_part_s, extra)
return True
if raise_exc:
raise AssertionError('NOT PDF: %s (%r)\n%r'
% (extra, pdf_part_s, start))
else:
return False
def fix_mimetype(stream, typ):
if 'application/x-zip-compressed' == typ:
typ = 'application/zip'
already_pdf = assert_pdf(stream, raise_exc=False)
if already_pdf:
LOG.info('mimetype is %r, but seems to be a pdf already!', typ)
return 'application/pdf'
elif typ == 'application/pdf':
LOG.error('got %r as pdf, but it is not!', repr(stream)[:100])
typn = guess_data_mime(stream)
if typn and typn != typ:
LOG.info('typ=%s -> %s by magic!', typ, typn)
return typn
return typ