I am working on a migration project to upgrade a layer of web server from python 2.7.8 to python 3.6.3 and I have hit a roadblock for some special cases.
When a request is received from a client, payload is transmitted locally using pyzmq which now interacts in bytes in python3 instead of str (as it is in python2).
Now, the payload which I am receiving is encoded using iso-8859-1 (latin-1) scheme and I can easily convert it into string as payload.decode('latin-1') and pass it to next service (svc-save-entity) which expects string argument.
However, the subsequent service 'svc-save-entity' expects non-latin-1 chars (if present) to be represented in ASCII Character Reference (such as é for é) rather than in Hex (such as \xe9 for é).
I am struggling to find an efficient way to achieve this conversion. Can any python expert guide me here? Essentially I need the definition of a function say decode_tostring():
payload = b'Banco Santander (M\xe9xico)' #payload is in bytes
payload_str = decode_tostring(payload) #function to convert into string
payload_str == 'Banco Santander (México)' #payload_str is a string in ASCII Character Reference
Definition of decode_tostring() please. :)
The encode() and decode() methods accept a parameter called errors which allows you to specify how characters which are not representable in the specified encoding should be handled. The one you're looking for is XML numeric character reference replacement, which is fortunately one of the standard handlers provided in the codecs module.
Now, it's a little complex to actually do the replacement the way you want it, because the operation of replacing non-ASCII characters with their corresponding XML numeric character references happens during encoding, not decoding. After all, encoding is the process that takes in characters and emits bytes, so it's only during encoding that you can tell whether you have a character that is not part of ASCII. The cleanest way I can think of at the moment to get the transformation you want is to decode, re-encode, and re-decode, applying the XML entity reference replacement during the encoding step.
def decode_tostring(payload):
return payload.decode('latin-1').encode('ascii', errors='xmlcharrefreplace').decode('ascii')
I wouldn't be surprised if there is a method somewhere out there that will replace all non-ASCII characters in a string with their XML numeric character refs and give you back a string, and if so, you could use it to replace the encoding and the second decoding. But I don't know of one. The closest I found at the moment was xml.sax.saxutils.escape(), but that only acts on certain specific characters.
This isn't really relevant to your main question, but I did want to clarify one thing: the numeric entities like é are a feature of SGML, HTML, and XML, which are markup languages - a way to represent structured data as text. They have nothing to do with ASCII. A character encoding like ASCII is nothing more than a table of some characters and some byte sequences such that each character in the table is mapped to one byte sequence in the table and vice versa, with a few constraints to make the mapping unambiguous.
If you have a string with characters that are not in a particular encoding's table, you can't encode the string using that encoding. But what you can do is convert the string into a new string by replacing the characters which aren't in the table with sequences of characters that are in the table, and then encode the new string. There are many ways to do the replacement, of which XML numeric entity references are one example. Some of the other error handlers in Python's codecs module represent other approaches to this replacement.