Assertion error when trying to parse a legit CANoe generated BLF file

[z33kz33k]

I'm trying to parse a BLF using this approach:

import can
from pprint import pprint as pp

filename = "logfile.blf"
logging = can.BLFReader(filename)
logging = list(logging)
pp(logging)

and get

Traceback (most recent call last):
  File "D:\eclipse-workspace\blf2csv\pythoncan_blf2csv.py", line 7, in <module>
    logging = list(logging)
  File "C:\Users\my_user_name\AppData\Local\Python\Python36\lib\site-packages\can\io\blf.py", line 129, in __iter__
    assert header[0] == b"LOBJ", "Parse error"
AssertionError: Parse error

The file is legit (I can convert it to CSV using Vector CANoe Tool >> Log File Conversion dialog (so I can convert a single file but looked at python-can as a means to do a batch conversion of multiple log files)).

[felixdivo]

I do now know what that line should achieve, but as noted in #108 by @christiansandberg, finding documentation about that format is difficult ... Maybe he knows what's wrong with this.

[z33kz33k]

Additional info.
When trying to parse a smaller file (on the same setup) I get a different error:

Traceback (most recent call last):
  File "D:\eclipse-workspace\blf2csv\pythoncan_blf2csv.py", line 7, in <module>
    logging = list(logging)
  File "C:\Users\my_user_name\AppData\Local\Python\Python36\lib\site-packages\can\io\blf.py", line 122, in __iter__
    data = zlib.decompress(obj_data, 15, uncompressed_size)
MemoryError

This file was recorded over a couple of seconds and has ca. 3.5MB. Trying the same code on a slightly weighter file (ca. 6MB) recorded on the same setup gives the above assertion error.

The files where generated on Vector CANoe version 10.0.

[christiansandberg]

Try the master. I've done some changes since last release since the reference implementation I've based my implementation on has made some changes regarding uncompressed_size that might cause memory issues.

[christiansandberg]

Also try printing each message as they are read. Do you see any messages at all?

import can

filename = "logfile.blf"
logging = can.BLFReader(filename)
for msg in logging:
    print(msg)

[z33kz33k]

Trying to read it one message at a time didn't change anything. But I'll try with the master branch and report back, thx.

Edit: no good.

[felixdivo]

The master branch should be the same as the one on PyPi, please use the develop branch.

[z33kz33k]

Update: I tried develop branch - the same result.

[felixdivo]

Okay, then it is a legit problem ... Does anybody know why the header should start with "LOBJ"?

[christiansandberg]

It’s a signature that starts each object to make sure you are reading the file correctly. Since the format is somewhat secret, Vector might change it at their own will at anytime or there could be some details we are missing that did not show up on the files I’ve tested it with.

[christiansandberg]

It could be that the file header size has changed. Right now it's hardcoded to 144 bytes but it can be read from the file instead which would make it more robust to future changes. I'll try to fix that tonight.

[christiansandberg]

Or actually I believe it is the usage of a newer object header (v2). I'll fix that as well.

[dhruvildarji]

Hey @christiansandberg I am still getting this issue with Newest version 3.3.2
Should I checkout to some different commit to solve it?