chore: refactor for unicode property of strings (#140)

Author: ota-meshi

scripts/update-unicode-properties.ts

@@ -54,6 +54,7 @@ const DATA_SOURCES = [
         binProperties: "#table-binary-unicode-properties",
         gcValues: getLatestUnicodeGeneralCategoryValues,
         scValues: getLatestUnicodeScriptValues,
+        binPropertiesOfStrings: "#table-binary-unicode-properties-of-strings",
     },
 ]
 const FILE_PATH = "src/unicode/properties.ts"
@@ -63,6 +64,7 @@ type Datum = {
     binProperties: string[]
     gcValues: string[]
     scValues: string[]
+    binPropertiesOfStrings: string[]
 }
 
 // Main
@@ -72,12 +74,14 @@ type Datum = {
         binProperties: new Set<string>(),
         gcValues: new Set<string>(),
         scValues: new Set<string>(),
+        binPropertiesOfStrings: new Set<string>(),
     }
 
     for (const {
         binProperties,
         gcValues,
         scValues,
+        binPropertiesOfStrings,
         url,
         version,
     } of DATA_SOURCES) {
@@ -86,6 +90,7 @@ type Datum = {
             binProperties: [],
             gcValues: [],
             scValues: [],
+            binPropertiesOfStrings: [],
         }
         data[version] = datum
 
@@ -120,6 +125,13 @@ type Datum = {
             scValues,
             existing.scValues,
         )
+        if (binPropertiesOfStrings) {
+            datum.binPropertiesOfStrings = await collectValues(
+                window,
+                binPropertiesOfStrings,
+                existing.binPropertiesOfStrings,
+            )
+        }
 
         logger.log("Done")
     }
@@ -140,6 +152,9 @@ const scValueSets = new DataSet(${Object.values(data)
 const binPropertySets = new DataSet(${Object.values(data)
         .map((d) => makeDataCode(d.binProperties))
         .join(",")})
+const binPropertyOfStringsSets = new DataSet(${Object.values(data)
+        .map((d) => makeDataCode(d.binPropertiesOfStrings))
+        .join(",")})
 
 export function isValidUnicodeProperty(version: number, name: string, value: string): boolean {
     if (gcNameSet.has(name)) {
@@ -169,6 +184,19 @@ export function isValidLoneUnicodeProperty(version: number, value: string): bool
         .filter(Boolean)
         .join(" || ")}
 }
+
+export function isValidLoneUnicodePropertyOfString(version: number, value: string): boolean {
+    return ${Object.entries(data)
+        .map(([version, { binPropertiesOfStrings }]) =>
+            makeVerificationCode(
+                version,
+                "binPropertyOfStringsSets",
+                binPropertiesOfStrings,
+            ),
+        )
+        .filter(Boolean)
+        .join(" || ")}
+}
 `
 
     logger.log("Formatting code...")

src/unicode/index.ts

@@ -2,6 +2,7 @@ export { isIdContinue, isIdStart } from "./ids"
 export {
     isValidLoneUnicodeProperty,
     isValidUnicodeProperty,
+    isValidLoneUnicodePropertyOfString,
 } from "./properties"
 
 export const NULL = 0x00

src/unicode/properties-of-strings.ts

@@ -119,6 +119,15 @@ const binPropertySets = new DataSet(
     "",
     "",
 )
+const binPropertyOfStringsSets = new DataSet(
+    "",
+    "",
+    "",
+    "",
+    "",
+    "",
+    "Basic_Emoji Emoji_Keycap_Sequence RGI_Emoji RGI_Emoji_Flag_Sequence RGI_Emoji_Modifier_Sequence RGI_Emoji_Tag_Sequence RGI_Emoji_ZWJ_Sequence",
+)
 
 export function isValidUnicodeProperty(
     version: number,
@@ -151,3 +160,10 @@ export function isValidLoneUnicodeProperty(
         (version >= 2021 && binPropertySets.es2021.has(value))
     )
 }
+
+export function isValidLoneUnicodePropertyOfString(
+    version: number,
+    value: string,
+): boolean {
+    return version >= 2024 && binPropertyOfStringsSets.es2024.has(value)
+}

src/unicode/properties.ts

@@ -73,6 +73,7 @@ import {
     isTrailSurrogate,
     isValidLoneUnicodeProperty,
     isValidUnicodeProperty,
+    isValidLoneUnicodePropertyOfString,
     isValidUnicode,
     AMPERSAND,
     NUMBER_SIGN,
@@ -83,7 +84,6 @@ import {
     TILDE,
     LATIN_SMALL_LETTER_Q,
 } from "./unicode"
-import { isValidLoneUnicodePropertyOfString } from "./unicode/properties-of-strings"
 
 // ^ $ \ . * + ? ( ) [ ] { } |
 const SYNTAX_CHARACTER = new Set([