I used Flask-Caching to cache the response of a Flask view in Redis. Getting the cached data directly from Redis returns some bytes. How can I parse this in Python to examine the cached value?
b'!\x80\x03cflask.wrappers\nResponse\nq\x00)\x81q\x01}q\x02(X\x07\x00\x00\x00headersq\x03cwerkzeug.datastructures\nHeaders\nq\x04)\x81q\x05}q\x06X\x05\x00\x00\x00_listq\x07]q\x08X\x0c\x00\x00\x00Content-Typeq\tX\x10\x00\x00\x00application/jsonq\n\x86q\x0basbX\x0c\x00\x00\x00_status_codeq\x0cK\xc8X\x07\x00\x00\x00_statusq\rX\x06\x00\x00\x00200 OKq\x0eX\x12\x00\x00\x00direct_passthroughq\x0f\x89X\t\x00\x00\x00_on_closeq\x10]q\x11X\x08\x00\x00\x00responseq\x12Xn\x00\x00\x00[\n {\n "desc": "pronoun object", \n "tag": "CLO"\n }, \n {\n "desc": "pronoun", \n "tag": "CLS"\n }\n]q\x13X\x01\x00\x00\x00\nq\x14\x86q\x15ub.'
Flask-Cache uses the RedisCache backend provided by Werkzeug, which serializes values using pickle.dumps. It also prepends a ! for help when deserializing. You should typically not mess with these values directly, and let Flask-Caching handle it.
You can use pickletools.dis to safely examine the representation, then pickle.loads to deserialize it. Security note: pickle.loads can execute arbitrary code, so make sure you understand untrusted data first using pickletools.dis.
>>> data = b'!\x80\x03cflask.wrappers\nResponse\nq\x00)\x81q\x01}q\x02(X\x07\x00\x00\x00headersq\x03cwerkzeug.datastructures\nHeaders\nq\x04)\x81q\x05}q\x06X\x05\x00\x00\x00_listq\x07]q\x08X\x0c\x00\x00\x00Content-Typeq\tX\x10\x00\x00\x00application/jsonq\n\x86q\x0basbX\x0c\x00\x00\x00_status_codeq\x0cK\xc8X\x07\x00\x00\x00_statusq\rX\x06\x00\x00\x00200 OKq\x0eX\x12\x00\x00\x00direct_passthroughq\x0f\x89X\t\x00\x00\x00_on_closeq\x10]q\x11X\x08\x00\x00\x00responseq\x12Xn\x00\x00\x00[\n {\n "desc": "pronoun object", \n "tag": "CLO"\n }, \n {\n "desc": "pronoun", \n "tag": "CLS"\n }\n]q\x13X\x01\x00\x00\x00\nq\x14\x86q\x15ub.'
>>> value = value[1:] # strip leading !
>>> import pickletools
>>> pickletools.dis(data)
0: \x80 PROTO 3
2: c GLOBAL 'flask.wrappers Response'
27: q BINPUT 0
29: ) EMPTY_TUPLE
30: \x81 NEWOBJ
31: q BINPUT 1
33: } EMPTY_DICT
34: q BINPUT 2
36: ( MARK
37: X BINUNICODE 'headers'
49: q BINPUT 3
51: c GLOBAL 'werkzeug.datastructures Headers'
84: q BINPUT 4
86: ) EMPTY_TUPLE
87: \x81 NEWOBJ
88: q BINPUT 5
90: } EMPTY_DICT
91: q BINPUT 6
93: X BINUNICODE '_list'
103: q BINPUT 7
105: ] EMPTY_LIST
106: q BINPUT 8
108: X BINUNICODE 'Content-Type'
125: q BINPUT 9
127: X BINUNICODE 'application/json'
148: q BINPUT 10
150: \x86 TUPLE2
151: q BINPUT 11
153: a APPEND
154: s SETITEM
155: b BUILD
156: X BINUNICODE '_status_code'
173: q BINPUT 12
175: K BININT1 200
177: X BINUNICODE '_status'
189: q BINPUT 13
191: X BINUNICODE '200 OK'
202: q BINPUT 14
204: X BINUNICODE 'direct_passthrough'
227: q BINPUT 15
229: \x89 NEWFALSE
230: X BINUNICODE '_on_close'
244: q BINPUT 16
246: ] EMPTY_LIST
247: q BINPUT 17
249: X BINUNICODE 'response'
262: q BINPUT 18
264: X BINUNICODE '[\n {\n "desc": "pronoun object", \n "tag": "CLO"\n }, \n {\n "desc": "pronoun", \n "tag": "CLS"\n }\n]'
379: q BINPUT 19
381: X BINUNICODE '\n'
387: q BINPUT 20
389: \x86 TUPLE2
390: q BINPUT 21
392: u SETITEMS (MARK at 36)
393: b BUILD
394: . STOP
highest protocol among opcodes = 2
>>> import pickle
>>> response = pickle.loads(data)
>>> response.content_type
'application/json'
>>> response.json
[{'desc': 'pronoun object', 'tag': 'CLO'}, {'desc': 'pronoun', 'tag': 'CLS'}]