Affects PMD Version:
7.11.0
Rule:
NPathComplexity
Description:
This issue is actually more of a question of why PMD reports switch expressions with an incredibly high NPath complexity.
Code Sample demonstrating the issue:
The following code has an NPathComplexity of 1296:
public static byte toByte(final Register8 r) {
return switch (r) {
case AL, R8B -> (byte) 0x00;
case CL, R9B -> (byte) 0x01;
case DL, R10B -> (byte) 0x02;
case BL, R11B -> (byte) 0x03;
case AH, SPL, R12B -> (byte) 0x04;
case CH, BPL, R13B -> (byte) 0x05;
case DH, SIL, R14B -> (byte) 0x06;
case BH, DIL, R15B -> (byte) 0x07;
};
}
The following code (completely equivalent to the first one) has an NPathComplexity of 21:
public static byte toByte(final Register8 r) {
if (r == AL || r == R8B) {
return (byte) 0x00;
} else if (r == CL || r == R9B) {
return (byte) 0x01;
} else if (r == DL || r == R10B) {
return (byte) 0x02;
} else if (r == BL || r == R11B) {
return (byte) 0x03;
} else if (r == AH || r == SPL || r == R12B) {
return (byte) 0x04;
} else if (r == CH || r == BPL || r == R13B) {
return (byte) 0x05;
} else if (r == DH || r == SIL || r == R14B) {
return (byte) 0x06;
} else if (r == BH || r == DIL || r == R15B) {
return (byte) 0x07;
} else {
throw new NullPointerException();
}
}
Expected outcome:
I understand where the 1296 comes from: 4 cases with 2 values and 4 cases with 3 values produce 2 * 2 * 2 * 2 * 3 * 3 * 3 * 3 = 1296 possible execution paths. The Register8 enum has only those 20 values (as can be deduced from the switch missing the default case) so I would expect a complexity of 20 (maybe +1 for the hidden null case) like in the second example.
In short, I think PMD is doing the wrong calculation. The value of 1296 would make sense if the switch was a classic one without breaks like the one below. This would explain the use of multiplication in the computation of the NPath complexity but I think in a switch expression (in which each case has its own scope, AFAIK) that computation should be a simple addition of the complexity of each case.
switch(x) {
case 0, 1:
// some code...
// fallthrough
case 2, 3:
// some other code...
// fallthrough
default:
return ...;
}
Looking around in older issues I found this one which was apparently fixed in PMD 6.26.0, so this issue may be a duplicate.
Running PMD through: Gradle
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4