lauhg / re1 Goto Github PK

What steps will reproduce the problem?
1.# ./re a 1a2

What is the expected output? What do you see instead?
expected: "thompson match (1,2)"
get: "thompson match (0,2)"

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?


Please provide any additional information below.

Segfault generated when the following is run:
./re "a|b" a

The following instructions are generated for the above regexp:
 0. split 3, 1
 1. any
 2. jmp 0
 3. save 0
 4. split 4, 7
 5. char a
 6. jmp 8
 7. char b
 8. save 1
 9. match

The full output:
Cat(NgStar(Dot), Paren(0, Alt(Lit(a), Lit(b))))
 0. split 3, 1
 1. any
 2. jmp 0
 3. save 0
 4. split 4, 7
 5. char a
 6. jmp 8
 7. char b
 8. save 1
 9. match
#2 a
Segmentation fault

I attached a patch that seems to fix the problem.

The recursive writing of addthread obscures the fact that the Split instruction 
is the single reason we need a recursive thread processing order.

The switch cases in addthread in both pike.c and thompson.c all end upon tail 
calls. This should be acknowledged and maybe coded with an explicit loop.