ftp-statemodel-learner

LearnLib harness to infer every FTP state model via Active Learning.

Setting the configuration file

Before running the learner, you need to edit the config.properties. Here you find a typical setup:

• ip: 127.0.0.1

  [ip target]

• port: 2200

  [server port]

• commands: USER ubuntu,NOOP,PWD,TYPE,PORT,CDUP,CWD,RETR,ABOR,DELE,PASS ubuntu,REST,SIZE,MKD,RMD,STOR,SYST,APPE,RNFR,RNTO,OPTS,AUTH,PBSZ,PROT,SITE [list of commands, separated by a comma] NB:You don't have to add the QUIT command

• debug: true

  [the default value is false]

• timing_info: true

  [if enabled, the learner will add the cost -- in terms of seconds -- for every transition]

Running the learner

• Via jar file:

  1. java -jar ftpLearner.jar

• via Eclipse:

  1. File -> Open Project from File System
  2. Select the project
  3. Run it

Output

If you enable the debug mode, you can see every request and response:

The inferred state model is saved in a .dot file:

Light FTP state model

Cleaning the state model

If you want to merge all the clouds into a single edge labelled with "others/{list of response codes}", you can run the script simplify_model.py

python3 simplify_model.py <input_file> <output_file>

to obtain something like this:

Light FTP state model

Case studies

Implementation	Inferred	Number of states
Light FTP	Yes	5
bftpd	Yes	8
pure-ftp	Yes	5
ProFTPd	Yes	7

Todo