Decode seemingly malformed utf8 representation of Unicode pointer from Facebook data export

Question

Similar to in this post, is there a way to decode some of the seemingly malformed UTF-8 data returned from downloading a copy of my Facebook data?

Looking at a particular example, in one of my chats I have a sent message containing only the emoji . Opening the message_1.json file with vim and looking at the appropriate entry shows the text "\u00f0\u009f\u0092\u008e". However this differs from the view from my terminal (Mac OSX)

$ jq '.messages[0].content' message_1.json
"ð"  # stackoverflow seems to be truncating this string, there are 3 extra chars which show as spaces
$ jq '.messages[0].content' message_1.json > utf
$ cat utf
"ð"
$ od -h utf
0000000      c322    c2b0    c29f    c292    228e    000a
0000013
$ wc utf
       1       1      11 utf

This also differs the output from directly pasting the emoji into a file

$ echo '' > gem.txt
$ cat gem.txt

$ od -h gem.txt
0000000      9ff0    8e92    000a
0000005
$ wc gem.txt
       1       1       5 gem.txt

And I get seemingly different information when reading in these two files with python3

$ python3
Python 3.7.3 (default, Dec 13 2019, 19:58:14)
[Clang 11.0.0 (clang-1100.0.33.17)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> with open('gem.txt', 'r') as f:
...   gem = f.read()
...
>>> gem
'\n'
>>> len(gem)
2
>>> ord(gem[0])
128142
>>>
>>>
>>> with open('utf', 'r') as f:
...   utf = f.read()
...
>>> utf
'"ð\x9f\x92\x8e"\n'
>>> len(utf)
7
>>> for char in utf:
...   print(ord(char))
...
34
240
159
146
142
34
10
>>>

I have a few questions based on this behavior:

1. Is the data returned by Facebook encoded incorrectly? This page shows the proper Unicode pointer for the gem emoji to be U+1F48E, and the corresponding UTF-8 0xF0 0x9F 0x92 0x8E representation matches with the byte output from od
2. Is there a way for me to parse the returned string from Facebook? It seems that the previous question recommends a regular expression to transform the text before doing so, is this required?
3. The gem.txt had a length of 5 bytes, and subtracting the newline, 4 bytes to represent the emoji. This makes sense to me as its UTF-8 representation requires 4 bytes. Why does the utf document list 11 bytes (presumably 10 without the newline)?

Answer 1

It looks like the content of your JSON file indeed got mojibaked, ie. misinterpreted with the wrong encoding.

>>> import json
>>> # That's how it should look:
>>> print(json.dumps(''))
"\ud83d\udc8e"
>>> # That's what you got:
>>> mojibaked = ''.encode('utf8').decode('latin1')
>>> print(json.dumps(mojibaked))
"\u00f0\u009f\u0092\u008e"

Check if you can fix how the JSON file is created. Latin-1 is the default in some tools/protocols. The convenient thing is that you can always decode any stream of bytes as Latin-1 without exceptions. It might corrupt your input though, as happens here.

If you can't fix the source, you might be able to recover by doing the encoding round-trip in reverse:

>>> mojibaked = json.loads('"\\u00f0\\u009f\\u0092\\u008e"')
>>> mojibaked
'ð\x9f\x92\x8e'
>>> mojibaked.encode('latin1').decode('utf8')
''