clean up trie construction

gen.js

@@ -17,8 +17,6 @@ function writeTrie(filename, varname, trie) {
 #include <stdint.h>
 __attribute__((used))
 uint32_t ${varname}[] = {${buffer.join(', ')}};
-__attribute__((used))
-uint32_t ${varname}_len = ${buffer.length};
 `);
 }
 

src/grapheme-break.ts

@@ -1,13 +1,13 @@
 // All code based on foliojs/grapheme-breaker at time of writing
-import UnicodeTrie from './unicode-trie.js';
+import {createTrie} from './unicode-trie.js';
 import wasm from './wasm.js';
 
-const heapu32 = new Uint32Array(wasm.instance.exports.memory.buffer);
-const len = heapu32[wasm.instance.exports.grapheme_break_trie_len.value >> 2];
 // I don't know why the pointer value is stored directly in the .value here.
 // It must be an emscripten weirdness, so watch out in the future
-const ptr = wasm.instance.exports.grapheme_break_trie.value >> 2;
-const trie = new UnicodeTrie(heapu32.subarray(ptr, ptr + len));
+const trie = createTrie(
+  wasm.instance.exports.memory.buffer,
+  wasm.instance.exports.grapheme_break_trie.value
+);
 
 // Gets a code point from a UTF-16 string
 // handling surrogate pairs appropriately

src/itemize.ts

@@ -1,12 +1,12 @@
-import UnicodeTrie from './unicode-trie.js';
+import {createTrie} from './unicode-trie.js';
 import wasm from './wasm.js';
 
-const heapu32 = new Uint32Array(wasm.instance.exports.memory.buffer);
-const len = heapu32[wasm.instance.exports.emoji_trie_len.value >> 2];
 // I don't know why the pointer value is stored directly in the .value here.
 // It must be an emscripten weirdness, so watch out in the future
-const ptr = wasm.instance.exports.emoji_trie.value >> 2;
-const emojiTrie = new UnicodeTrie(heapu32.subarray(ptr, ptr + len));
+const emojiTrie = createTrie(
+  wasm.instance.exports.memory.buffer,
+  wasm.instance.exports.emoji_trie.value
+);
 
 const {
   // SheenBidi

src/line-break.ts

@@ -1,13 +1,13 @@
 // All code based on foliojs/linebreak at time of writing
-import UnicodeTrie from './unicode-trie.js';
+import {createTrie} from './unicode-trie.js';
 import wasm from './wasm.js';
 
-const heapu32 = new Uint32Array(wasm.instance.exports.memory.buffer);
-const len = heapu32[wasm.instance.exports.line_break_trie_len.value >> 2];
 // I don't know why the pointer value is stored directly in the .value here.
 // It must be an emscripten weirdness, so watch out in the future
-const ptr = wasm.instance.exports.line_break_trie.value >> 2;
-const trie = new UnicodeTrie(heapu32.subarray(ptr, ptr + len));
+const trie = createTrie(
+  wasm.instance.exports.memory.buffer,
+  wasm.instance.exports.line_break_trie.value
+);
 
 const DI_BRK = 0; // Direct break opportunity
 const IN_BRK = 1; // Indirect break opportunity

src/unicode-trie-builder.ts

@@ -816,9 +816,10 @@ export default class UnicodeTrieBuilder {
 
     // calculate the sizes of, and allocate, the index and data arrays
     const indexLength = allIndexesLength + this.dataLength;
-    const data = new Uint32Array(indexLength + 2);
+    const data = new Uint32Array(indexLength + 3);
     let destIdx = 0;
 
+    data[destIdx++] = indexLength + 3;
     data[destIdx++] = this.highStart;
     data[destIdx++] = this.errorValue;
 

src/unicode-trie.ts

@@ -106,9 +106,9 @@ export default class UnicodeTrie {
   data: Uint32Array;
 
   constructor(data: Uint32Array) {
-    this.highStart = data[0];
-    this.errorValue = data[1];
-    this.data = data.subarray(2);
+    this.highStart = data[1];
+    this.errorValue = data[2];
+    this.data = data.subarray(3);
   }
 
   get(codePoint: number) {
@@ -138,3 +138,9 @@ export default class UnicodeTrie {
   }
 }
 
+export function createTrie(memory: ArrayBuffer, ptr: number) {
+  const ptr32 = ptr >> 2;
+  const heapu32 = new Uint32Array(memory);
+  const len = heapu32[ptr32];
+  return new UnicodeTrie(heapu32.subarray(ptr32, ptr32 + len));
+}