Implement String#split (#301)

std/assembly/index.d.ts

@@ -491,6 +491,7 @@ declare class String {
   padStart(targetLength: i32, padString?: string): string;
   padEnd(targetLength: i32, padString?: string): string;
   repeat(count?: i32): string;
+  split(separator?: string, limit?: i32): string[];
   toString(): string;
   static fromUTF8(ptr: usize, len: usize): string;
   toUTF8(): usize;

std/assembly/string.ts

@@ -10,6 +10,10 @@ import {
   parse
 } from "./internal/string";
 
+import {
+  storeUnsafe as storeUnsafeArray
+} from "./internal/arraybuffer";
+
 @sealed
 export class String {
 
@@ -200,6 +204,7 @@ export class String {
     return compareUnsafe(left, 0, right, 0, length) <= 0;
   }
 
+  @inline
   includes(searchString: String, position: i32 = 0): bool {
     return this.indexOf(searchString, position) != -1;
   }
@@ -400,14 +405,69 @@ export class String {
       throw new RangeError("Invalid count value");
     }
 
-    if (count === 0 || !length) return changetype<String>("");
+    if (count == 0 || !length) return changetype<String>("");
-    if (count === 1) return this;
+    if (count == 1) return this;
 
     var result = allocateUnsafe(length * count);
     repeatUnsafe(result, 0, this, count);
     return result;
   }
 
+  split(separator: String = null, limit: i32 = i32.MAX_VALUE): String[] {
+    assert(this !== null);
+    if (!limit) return new Array<String>();
+    if (separator === null) return <String[]>[this];
+    var length: isize = this.length;
+    var sepLen: isize = separator.length;
+    if (limit < 0) limit = i32.MAX_VALUE;
+    if (!sepLen) {
+      if (!length) return new Array<String>();
+      // split by chars
+      length = min<isize>(length, <isize>limit);
+      let result = new Array<String>(length);
+      let buffer = <ArrayBuffer>result.buffer_;
+      for (let i: isize = 0; i < length; ++i) {
+        let char = allocateUnsafe(1);
+        store<u16>(
+          changetype<usize>(char),
+          load<u16>(
+            changetype<usize>(this) + (<usize>i << 1),
+            HEADER_SIZE
+          ),
+          HEADER_SIZE
+        );
+        storeUnsafeArray<String,String>(buffer, i, char);
+      }
+      return result;
+    } else if (!length) {
+      return <String[]>[changetype<String>("")];
+    }
+    var result = new Array<String>();
+    var end = 0, start = 0, i = 0;
+    while ((end = this.indexOf(separator, start)) != -1) {
+      let len = end - start;
+      if (len > 0) {
+        let out = allocateUnsafe(len);
+        copyUnsafe(out, 0, this, start, len);
+        result.push(out);
+      } else {
+        result.push(changetype<String>(""));
+      }
+      if (++i == limit) return result;
+      start = end + sepLen;
+    }
+    if (!start) return <String[]>[this];
+    var len = length - start;
+    if (len > 0) {
+      let out = allocateUnsafe(len);
+      copyUnsafe(out, 0, this, start, len);
+      result.push(out);
+    } else {
+      result.push(changetype<String>(""));
+    }
+    return result;
+  }
+
   toString(): String {
     return this;
   }

std/portable/index.d.ts

@@ -310,6 +310,7 @@ declare class String {
   padEnd(targetLength: i32, padString?: string): string;
   replace(search: string, replacement: string): string;
   repeat(count?: i32): string;
+  split(separator?: string, limit?: i32): string[];
   toString(): string;
 }
 

tests/compiler/std/array-access.optimized.wat

@@ -132,7 +132,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 16)
-     (i32.const 239)
+     (i32.const 244)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/array-access.untouched.wat

@@ -225,7 +225,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 16)
-     (i32.const 239)
+     (i32.const 244)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/array.optimized.wat

@@ -7930,7 +7930,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 1616)
-     (i32.const 106)
+     (i32.const 110)
      (i32.const 4)
     )
     (unreachable)
@@ -8104,7 +8104,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 1616)
-     (i32.const 264)
+     (i32.const 269)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/array.untouched.wat

@@ -12670,7 +12670,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 1616)
-     (i32.const 54)
+     (i32.const 58)
      (i32.const 4)
     )
     (unreachable)
@@ -12764,7 +12764,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 1616)
-     (i32.const 106)
+     (i32.const 110)
      (i32.const 4)
     )
     (unreachable)
@@ -13051,7 +13051,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 1616)
-     (i32.const 264)
+     (i32.const 269)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/hash.untouched.wat

@@ -13,6 +13,8 @@
  (global $~lib/internal/allocator/AL_SIZE i32 (i32.const 8))
  (global $~lib/internal/allocator/AL_MASK i32 (i32.const 7))
  (global $~lib/internal/allocator/MAX_SIZE_32 i32 (i32.const 1073741824))
+ (global $~lib/internal/arraybuffer/HEADER_SIZE i32 (i32.const 8))
+ (global $~lib/internal/arraybuffer/MAX_BLENGTH i32 (i32.const 1073741816))
  (global $~lib/internal/string/HEADER_SIZE i32 (i32.const 4))
  (global $~lib/internal/string/MAX_LENGTH i32 (i32.const 536870910))
  (global $~lib/internal/hash/FNV_OFFSET i32 (i32.const -2128831035))

tests/compiler/std/string-utf8.optimized.wat

@@ -2365,7 +2365,7 @@
            (call $~lib/env/abort
             (i32.const 0)
             (i32.const 72)
-            (i32.const 450)
+            (i32.const 510)
             (i32.const 8)
            )
            (unreachable)
@@ -2440,7 +2440,7 @@
              (call $~lib/env/abort
               (i32.const 0)
               (i32.const 72)
-              (i32.const 454)
+              (i32.const 514)
               (i32.const 8)
              )
              (unreachable)
@@ -2559,7 +2559,7 @@
              (call $~lib/env/abort
               (i32.const 0)
               (i32.const 72)
-              (i32.const 466)
+              (i32.const 526)
               (i32.const 8)
              )
              (unreachable)
@@ -2645,7 +2645,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 72)
-     (i32.const 475)
+     (i32.const 535)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/string-utf8.untouched.wat

@@ -2911,7 +2911,7 @@
             (call $~lib/env/abort
              (i32.const 0)
              (i32.const 72)
-             (i32.const 450)
+             (i32.const 510)
              (i32.const 8)
             )
             (unreachable)
@@ -2988,7 +2988,7 @@
              (call $~lib/env/abort
               (i32.const 0)
               (i32.const 72)
-              (i32.const 454)
+              (i32.const 514)
               (i32.const 8)
              )
              (unreachable)
@@ -3132,7 +3132,7 @@
              (call $~lib/env/abort
               (i32.const 0)
               (i32.const 72)
-              (i32.const 466)
+              (i32.const 526)
               (i32.const 8)
              )
              (unreachable)
@@ -3225,7 +3225,7 @@
     (call $~lib/env/abort
      (i32.const 0)
      (i32.const 72)
-     (i32.const 475)
+     (i32.const 535)
      (i32.const 4)
     )
     (unreachable)

tests/compiler/std/string.ts

@@ -51,6 +51,8 @@ assert("abc".padEnd(8, "abc") == "abcabcab");
 
 assert("".indexOf("") == 0);
 assert("".indexOf("hi") == -1);
+assert("a".indexOf("a") == 0);
+assert(str.indexOf(str) == 0);
 assert(str.indexOf("") == 0);
 assert(str.indexOf(",") == 2);
 assert(str.indexOf("x") == -1);
@@ -131,6 +133,41 @@ assert("a".repeat(5) == "aaaaa");
 assert("a".repeat(6) == "aaaaaa");
 assert("a".repeat(7) == "aaaaaaa");
 
+var sa: string[];
+
+sa = "".split();
+assert(sa.length == 1 && sa[0] == "");
+sa = "".split("");
+assert(sa.length == 0);
+sa = "".split(",");
+assert(sa.length == 1 && sa[0] == "");
+sa = "a,b,c".split(".");
+assert(sa.length == 1 && sa[0] == "a,b,c");
+sa = "a,b,c".split(",");
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+sa = "a, b, c".split(", ");
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+sa = "a,b,,c".split(",");
+assert(sa.length == 4 && sa[0] == "a" && sa[1] == "b" && sa[2] == "" && sa[3] == "c");
+sa = ",a,b,c".split(",");
+assert(sa.length == 4 && sa[0] == "" && sa[1] == "a" && sa[2] == "b" && sa[3] == "c");
+sa = "a,b,c,".split(",");
+assert(sa.length == 4 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c" && sa[3] == "");
+sa = "abc".split("");
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+sa = "abc".split("", 0);
+assert(sa.length == 0);
+sa = "abc".split("", 1);
+assert(sa.length == 1 && sa[0] == "a");
+sa = "a,b,c".split(",", 1);
+assert(sa.length == 1 && sa[0] == "a");
+sa = "abc".split("", 4);
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+sa = "abc".split("", -1);
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+sa = "a,b,c".split(",", -1);
+assert(sa.length == 3 && sa[0] == "a" && sa[1] == "b" && sa[2] == "c");
+
 assert(itoa32(0) == "0");
 assert(itoa32(1) == "1");
 assert(itoa32(8) == "8");