switch
switch picks one of several branches by a value. It looks like C's, with four differences: cases don't fall through into each other, a case can be a range, you can switch on text, and an explicit fallthrough is there for the rare case you want it.
The basics
The value is compared with each case; the matching case runs. default runs when none matches. A case ends where the next one starts, so no break is needed.
void describe(i64 n) {
switch (n) {
case 0: printf("%d: zero\n", n);
case 1: printf("%d: one\n", n);
default: printf("%d: many\n", n);
}
}
void main() {
describe(0);
describe(1);
describe(7);
}0: zero
1: one
7: manyIn C, describe(0) would print all three lines. In ktm each case stands alone.
Without a default, a value that matches no case does nothing:
void main() {
for (i64 i = 0; i < 4; i++) {
switch (i) {
case 1: printf("one ");
case 3: printf("three ");
}
}
printf("\n");
}one three The parentheses are optional, as for if:
void main() {
i64 n = 2;
switch n {
case 2: printf("two\n");
default: printf("other\n");
}
}twobreak
break leaves the switch early, from the middle of a case:
void check(i64 n) {
switch (n) {
default:
if (n < 0) {
printf("%d: negative, stop\n", n);
break;
}
printf("%d: fine\n", n);
}
}
void main() {
check(-3);
check(4);
}-3: negative, stop
4: fineInside a loop, break in a switch leaves the switch, not the loop; continue goes to the loop's next pass.
void main() {
for (i64 i = 0; i < 5; i++) {
switch (i) {
case 1: continue;
case 3: break;
default: printf("%d ", i);
}
printf("| ");
}
printf("\n");
}0 | 2 | | 4 | Several values, one body
Labels stacked with nothing between them share the code after them.
u8 *dayType(i64 day) {
switch (day) {
case 0:
case 6:
return "weekend";
default:
return "weekday";
}
}
void main() {
printf("%s %s %s\n", dayType(0), dayType(3), dayType(6));
}weekend weekday weekendRanges
case lo ... hi: matches every value from lo to hi, both included.
void grade(i64 score) {
switch (score) {
case 90 ... 100: printf("%d: A\n", score);
case 80 ... 89: printf("%d: B\n", score);
case 0: printf("%d: no score\n", score);
default: printf("%d: C or below\n", score);
}
}
void main() {
grade(95);
grade(83);
grade(0);
grade(42);
}95: A
83: B
0: no score
42: C or belowRanges work on characters too, since a character is a number:
u8 *kind(u8 c) {
switch (c) {
case 'a' ... 'z': return "lower";
case 'A' ... 'Z': return "upper";
case '0' ... '9': return "digit";
default: return "other";
}
}
void main() {
printf("%s %s %s %s\n", kind('q'), kind('Q'), kind('7'), kind('!'));
}lower upper digit otherTwo cases may not cover the same value:
does not compilevoid main() {
i64 n = 5;
switch (n) {
case 1 ... 10: printf("low\n");
case 5: printf("five\n");
}
}
error:fallthrough
fallthrough; as the last statement of a case runs on into the next case.
void main() {
for (i64 i = 1; i <= 3; i++) {
printf("%d:", i);
switch (i) {
case 1:
printf(" one");
fallthrough;
case 2:
printf(" one-or-two");
case 3:
printf(" three");
}
printf("\n");
}
}1: one one-or-two
2: one-or-two
3: threefallthrough is an error anywhere else:
- outside a switch, or before other statements of its case;
- inside an
ifor a loop in the case; - in the last case, which has no next case.
does not compilevoid main() {
i64 n = 1;
switch (n) {
case 1:
fallthrough;
printf("never\n");
case 2:
printf("two\n");
}
}
error:Each case is a block
A case can declare its own variables without braces, and two cases can use the same name.
void main() {
for (i64 i = 0; i < 2; i++) {
switch (i) {
case 0:
i64 n = 10;
printf("case 0: n = %d\n", n);
case 1:
i64 n = 20;
printf("case 1: n = %d\n", n);
}
}
}case 0: n = 10
case 1: n = 20A defer in a case runs at the end of that case. It needs braces: case 1: { defer ...; }.
Switching on enums
The cases of an enum are its members.
enum Color { red, green, blue };
u8 *hex(Color c) {
switch (c) {
case Color.red: return "#f00";
case Color.green: return "#0f0";
case Color.blue: return "#00f";
}
return "?";
}
void main() {
printf("%s %s %s\n", hex(Color.red), hex(Color.green), hex(Color.blue));
}#f00 #0f0 #00fSwitching on text
A u8[] or a String can be switched on, with string literals as the cases. A case matches when the value has exactly its bytes: the same length, the same case, no trimming.
i64 run(u8[] cmd) {
switch (cmd) {
case "GET": return 1;
case "SET": return 2;
case "DEL":
case "REMOVE": return 3;
default: return 0;
}
}
void main() {
u8[] line = "SET name ktm";
printf("%d %d %d %d\n", run("GET"), run(line[0:3]), run("REMOVE"), run("get"));
}1 2 3 0- Every case is a string literal: no ranges and no integers.
- The same string twice is an error.
- The value is computed once, so
switch (readCommand())calls the function once. default, stacked labels andfallthroughwork as in a switch on numbers.
For a match that ignores case, switch on a lower-case copy:
import str;
i64 code(u8[] word) {
String low = str.from(word);
low.toLower();
i64 result = 0;
switch (low) {
case "yes": result = 1;
case "no": result = 2;
}
low.free();
return result;
}
void main() {
printf("%d %d %d\n", code("YES"), code("No"), code("maybe"));
}1 2 0Two extras
Implicit case values
A case: with no value is the previous case's value plus one, starting from 0.
void main() {
for (i64 i = 0; i < 4; i++) {
switch (i) {
case: printf("zero ");
case: printf("one ");
case: printf("two ");
default: printf("more");
}
}
printf("\n");
}zero one two moreswitch [x]: no range check
switch [x] leaves out the check that x is within the cases' range, for a jump table without the test. Use it only when x always matches a case: anything else is undefined. A switch on text always checks.
void main() {
for (i64 i = 0; i < 3; i++) {
switch [i] {
case 0: printf("a");
case 1: printf("b");
case 2: printf("c\n");
}
}
}abc