[CALCITE-6388] PsTableFunction throws NumberFormatException when the …

[asolimando]

…'user' column has spaces

[sonarcloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[snuyanzin]

I guess the problem here that other columns also could contain space separated values like e.g. comm

I tried this ./sqlsh select distinct \`user\` from ps for the branch for this PR
and the output is

6.9
13.3
0.0
0.1
1.0
0.2
1.1
0.3
1.2
69.7
39.5
0.4
0.5
0.6
0.7
108
4.4
0.8
1.8
35.8
0.9

while for main same query is

message+
serg
avahi
lp
kernoops
root
rtkit
syslog
systemd+
colord

after some checks I noticed that command could also contain space separated values and probably this is the reason...

[caicancai]

Can we add some tests to illustrate the jira case?

[asolimando]

@snuyanzin, you are right, both user and comm can contain spaces, I have looked up if any of the other fields could be used in an unambiguous manner as an anchor to understand when user is over, unfortunately it doesn't seem to be any.

What I came up with is to move user as first field, split by whitespace as before, and stop parsing user whenever a purely numeric token is found. For comm I am just composing all the tokens counting on the fact that the number of fields is known.

So at the moment we support spaces, but not when one of the tokens is purely numeric like root 123. I don't have better ideas, it seems already an improvement over the current implementation which doesn't properly support spaces neither in user nor in comm.

@caicancai I have added a test, I had to unpack a bit the current implementation to that aim, thanks for the reminder.

Eager to hear your thoughts, thanks for your comments!

[sonarcloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
86.7% Coverage on New Code
9.4% Duplication on New Code

See analysis details on SonarCloud