[TRE-general] TRE coredump, regexp syntax, definition of i/d/s

[Stavros Macrakis]

Thanks very much for the TRE package!  Great stuff!

I'm afraid I have a few problems and questions which I'd appreciate your
help on. I am using TRE 0.7.5, gcc 3.4.4 under Cygwin/Windows XP.

-----

I have run into a problem where TRE coredumps, but I think that my input is
correct:

   echo foo | agrep
"((J{~1,1i+1d+1s<2})((a){~1,1i+1d+1s<2})|(~{~1,3d+10i+10s<4})J{~1,1i+1d+1s<2}o)"

==>

Program received signal SIGSEGV, Segmentation fault.
0x6ed0b2ea in tre_tnfa_run_approx (tnfa=0x673ad0, string=0x675988, len=9,
    type=STR_BYTE, match_tags=0x0, match=0x23c9b0, default_params=
      {cost_ins = 1, cost_del = 1, cost_subst = 1, max_cost = 0, max_ins =
2147483647, max_del = 2147483647, max_subst = 2147483647, max_err =
2147483647},
    eflags=0, match_end_ofs=0x23c8f8) at tre-match-approx.c:682
682           for (trans = reach[id].state; trans->state; trans++)

I tried taking out various parts of the regexp, and the problem goes away.

-----

I also seem not to understand the syntax of approximate matching settings.
 I thought the following would work:

  echo foo | agrep "g{1d+1i+1s<3}"
  echo foo | agrep -9 "g{1d+1i+1s<3}"

but I get "Invalid contents of {}" in both cases.  But the following works:

      echo foo | agrep "g{~9,1d+1i+1s<3}"

Am I misunderstanding something?

-----

I don't seem to understand the definition of
insertion/deletion/substitution.  For example, using the following test:

      echo XXX |  agrep -s  "(^PPP$){~8,Ii+Dd+Ss<MAX}"

              ==> Cost: …

I get the following results:

XXX

PPP

I

D

S

Max

Cost

abc

a

1

1

1

20

3

abc

a

1

1

20

20

4

aaa

a

1

1

1

20

2

aaa

a

1

1

20

20

2

aba

a

1

20

20

20

2

abc

a

1

20

20

20

No



I can't figure out how these costs are calculated – is there some sort of
"explain" mode?


Thanks,
               -s
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://laurikari.net/pipermail/tre-general/attachments/20080122/c27926bb/attachment.html