Improve isSpace util, refactor fromCharCode and fix String#trim{End} (#…

…692)
AssemblyScript · Jun 27, 2019 · 316b2d4 · 316b2d4
1 parent cf0f4f5
commit 316b2d4
Show file tree

Hide file tree

Showing 7 changed files with 2,962 additions and 2,137 deletions.
diff --git a/std/assembly/string.ts b/std/assembly/string.ts
@@ -1,7 +1,7 @@
 /// <reference path="./rt/index.d.ts" />
 
 import { BLOCK, BLOCK_OVERHEAD, BLOCK_MAXSIZE } from "./rt/common";
-import { compareImpl, strtol, strtod, isWhiteSpaceOrLineTerminator } from "./util/string";
+import { compareImpl, strtol, strtod, isSpace } from "./util/string";
 import { E_INVALIDLENGTH } from "./util/error";
 import { ArrayBufferView } from "./arraybuffer";
 import { idof } from "./builtins";
@@ -11,23 +11,18 @@ import { idof } from "./builtins";
   @lazy static readonly MAX_LENGTH: i32 = BLOCK_MAXSIZE >>> alignof<u16>();
 
   static fromCharCode(unit: i32, surr: i32 = -1): string {
-    var out: usize;
-    if (~surr) {
-      out = __alloc(4, idof<string>());
-      store<u16>(out, <u16>unit);
-      store<u16>(out, <u16>surr, 2);
-    } else {
-      out = __alloc(2, idof<string>());
-      store<u16>(out, <u16>unit);
-    }
+    var hasSur = surr > 0;
+    var out = __alloc(2 << i32(hasSur), idof<string>());
+    store<u16>(out, <u16>unit);
+    if (hasSur) store<u16>(out, <u16>surr, 2);
     return changetype<string>(out); // retains
   }
 
   static fromCodePoint(code: i32): string {
     assert(<u32>code <= 0x10FFFF);
-    var sur = code > 0xFFFF;
-    var out = __alloc((i32(sur) + 1) << 1, idof<string>());
-    if (!sur) {
+    var hasSur = code > 0xFFFF;
+    var out = __alloc(2 << i32(hasSur), idof<string>());
+    if (!hasSur) {
       store<u16>(out, <u16>code);
     } else {
       code -= 0x10000;
@@ -205,25 +200,15 @@ import { idof } from "./builtins";
   trim(): String {
     var length = this.length;
     var size: usize = length << 1;
-    while (
-      size &&
-      isWhiteSpaceOrLineTerminator(
-        load<u16>(changetype<usize>(this) + size)
-      )
-    ) {
+    while (size && isSpace(load<u16>(changetype<usize>(this) + size - 2))) {
       size -= 2;
     }
     var offset: usize = 0;
-    while (
-      offset < size &&
-      isWhiteSpaceOrLineTerminator(
-        load<u16>(changetype<usize>(this) + offset)
-      )
-    ) {
+    while (offset < size && isSpace(load<u16>(changetype<usize>(this) + offset))) {
       offset += 2; size -= 2;
     }
     if (!size) return changetype<String>("");
-    if (!start && size == length << 1) return this;
+    if (!offset && size == length << 1) return this;
     var out = __alloc(size, idof<String>());
     memory.copy(out, changetype<usize>(this) + offset, size);
     return changetype<String>(out); // retains
@@ -242,12 +227,7 @@ import { idof } from "./builtins";
   trimStart(): String {
     var size = <usize>this.length << 1;
     var offset: usize = 0;
-    while (
-      offset < size &&
-      isWhiteSpaceOrLineTerminator(
-        load<u16>(changetype<usize>(this) + offset)
-      )
-    ) {
+    while (offset < size && isSpace(load<u16>(changetype<usize>(this) + offset))) {
       offset += 2;
     }
     if (!offset) return this;
@@ -261,12 +241,7 @@ import { idof } from "./builtins";
   trimEnd(): String {
     var originalSize = <usize>this.length << 1;
     var size = originalSize;
-    while (
-      size &&
-      isWhiteSpaceOrLineTerminator(
-        load<u16>(changetype<usize>(this) + size)
-      )
-    ) {
+    while (size && isSpace(load<u16>(changetype<usize>(this) + size - 2))) {
       size -= 2;
     }
     if (!size) return changetype<String>("");

diff --git a/std/assembly/util/string.ts b/std/assembly/util/string.ts
@@ -1,5 +1,5 @@
 export function compareImpl(str1: string, index1: usize, str2: string, index2: usize, len: usize): i32 {
-  var result: i32 = 0;
+  var result = 0;
   var ptr1 = changetype<usize>(str1) + (index1 << 1);
   var ptr2 = changetype<usize>(str2) + (index2 << 1);
   while (len && !(result = <i32>load<u16>(ptr1) - <i32>load<u16>(ptr2))) {
@@ -30,7 +30,7 @@ export const enum CharCode {
   N = 0x4E,
   O = 0x4F,
   X = 0x58,
-  Z = 0x5a,
+  Z = 0x5A,
   a = 0x61,
   b = 0x62,
   e = 0x65,
@@ -40,20 +40,30 @@ export const enum CharCode {
   z = 0x7A
 }
 
-export function isWhiteSpaceOrLineTerminator(c: i32): bool {
+export function isSpace(c: i32): bool {
+  if (c <= 0xFF) {
+    switch (c) {
+      case 0x09: // <TAB>
+      case 0x0A: // <LF>
+      case 0x0B: // <VT>
+      case 0x0C: // <FF>
+      case 0x0D: // <CR>
+      case 0x20: // <SP>
+      case 0xA0: return true; // <NBSP>
+    }
+    return false;
+  }
+  if (c >= 0x2000 && c <= 0x200A) return true;
   switch (c) {
-    case 9:    // <TAB>
-    case 10:   // <LF>
-    case 13:   // <CR>
-    case 11:   // <VT>
-    case 12:   // <FF>
-    case 32:   // <SP>
-    case 160:  // <NBSP>
-    case 8232: // <LS>
-    case 8233: // <PS>
-    case 65279: return true; // <ZWNBSP>
-    default: return false;
+    case 0x1680: // <LS> (1)
+    case 0x2028: // <LS> (2)
+    case 0x2029: // <PS>
+    case 0x202F: // <NNS>
+    case 0x205F: // <MMSP>
+    case 0x3000: // <IS>
+    case 0xFEFF: return true; // <ZWNBSP>
   }
+  return false;
 }
 
 /** Parses a string to an integer (usually), using the specified radix. */
@@ -69,7 +79,7 @@ export function strtol<T>(str: string, radix: i32 = 0): T {
   // @ts-ignore: cast
   var sign: T = 1;
   // trim white spaces
-  while (isWhiteSpaceOrLineTerminator(code)) {
+  while (isSpace(code)) {
     code = <i32>load<u16>(ptr += 2);
     --len;
   }
@@ -147,7 +157,7 @@ export function strtod(str: string): f64 {
   // determine sign
   var sign = 1.0;
   // trim white spaces
-  while (isWhiteSpaceOrLineTerminator(code)) {
+  while (isSpace(code)) {
     code = <i32>load<u16>(ptr += 2);
     --len;
   }

diff --git a/tests/compiler/std/string-encoding.optimized.wat b/tests/compiler/std/string-encoding.optimized.wat
@@ -2497,7 +2497,7 @@
    if
     i32.const 0
     i32.const 480
-    i32.const 592
+    i32.const 567
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -2520,7 +2520,7 @@
    if
     i32.const 0
     i32.const 480
-    i32.const 596
+    i32.const 571
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -2839,7 +2839,7 @@
   if
    i32.const 0
    i32.const 480
-   i32.const 610
+   i32.const 585
    i32.const 6
    call $~lib/builtins/abort
    unreachable

diff --git a/tests/compiler/std/string-encoding.untouched.wat b/tests/compiler/std/string-encoding.untouched.wat
@@ -4086,7 +4086,7 @@
    if
     i32.const 0
     i32.const 480
-    i32.const 592
+    i32.const 567
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -4110,7 +4110,7 @@
    if
     i32.const 0
     i32.const 480
-    i32.const 596
+    i32.const 571
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -4466,7 +4466,7 @@
   if
    i32.const 0
    i32.const 480
-   i32.const 610
+   i32.const 585
    i32.const 6
    call $~lib/builtins/abort
    unreachable