Pipeline file unreadable (by human)

cellprofiler

#1

Hi,

Since Cellprofiler 3.1.5 pipeline files (.cppipe) cannot be read by human. We use pipeline files to keep track of the changes we do in our pipelines but since the last release we can only see what seems to be hexadecimal sequences. See below

Extract metadata?:\xff\xfeY\x00e\x00s\x00
    Metadata data type:\xff\xfeT\x00e\x00x\x00t\x00
    Metadata types:\xff\xfe{\x00}\x00
    Extraction method count:\xff\xfe2\x00

Is there any way to revert that ?
Thank you
Daniel


#2

Hello,
I have the same problem. Especially the \x00 which usually is the NULL termination character of a String makes it hard to parse that by common parsers.
And you cannot simple skip everything starting with \x plus the next two characters because also some meaningful characters (e.g. “[”, “]”) are encoded this way.

Edit:
AFAIK its enough to just remove these three patterns:

  • \xff
  • \xfe
  • \x00

Edit II:
The Experiment.csv output file shows the same strange encoding.

Regards,
Patrick


#3

Hi guys,

That was related to a necessary change we needed to make in order to fix how CP handles Unicode stuff; we agree it’s super annoying and are going to try to get it fixed!