The bz2 implementation in python 2.7 is unable to decompress multi-stream bzip-files. This is especially problematic because it does not actually output an error or a warning and just acts as if the file was parsed completely [0].

When using the debug flag you will see something like this:

…
2019-03-28 12:54:57,119: [DEBUG] Invalid line detected (line did not match): xxx.xxx.xxx.xxx - - [03/Feb/2019:08:42:34 +0100] "GET /resources/some.pdf HTTP/2.0" 200 811576 "-" "Mozilla/5.0 (iPhone; CPU iPhone OS 12_1_2 like Mac OS X)
…

After this, the parsing for the file ends. This happens because the content of the uncompressed log-file is not separated into streams be line (but by size), so you will most likely have the first and only stream the bz2-module actually reads end in the middle of a long line.
In the files I tried the first stream ended after exactly 900.000 bytes, so that could be an indication.

To check your bz2-file you might want to try running bzip2 on console (this is a multi-stream file):

~$ bzip2 -tvvv ../logs/access.log-20190203.bz2 
  ../logs/access.log-20190203.bz2: 
    [1: huff+mtf rt+rld {0x0af23f00, 0x0af23f00}]
    [2: huff+mtf rt+rld {0xb291b802, 0xb291b802}]
    combined CRCs: stored = 0xa775c602, computed = 0xa775c602
    [1: huff+mtf rt+rld {0x12f00a0d, 0x12f00a0d}]
    [2: huff+mtf rt+rld {0xcbad012a, 0xcbad012a}]
…
    combined CRCs: stored = 0xfcecf793, computed = 0xfcecf793
    [1: huff+mtf rt+rld {0x84243689, 0x84243689}]
    combined CRCs: stored = 0x84243689, computed = 0x84243689
    ok

Possible solutions

• check the file-pointer after reading the compressed logfile, if it's exactly 900.000 bytes, output a warning or an error message
• Use an external dependency like bz2file [1] to read the compressed data

[0] https://docs.python.org/2/library/bz2.html#bz2.BZ2File
[1] https://pypi.org/project/bz2file/