feat: extend MojiAdapter to support emoji variations and dataset generation

Author: luxass

src/adapter/index.ts

@@ -1,4 +1,4 @@
-import type { EmojiGroup, EmojiSequence } from "../types";
+import type { EmojiGroup, EmojiSequence, EmojiVariation } from "../types";
 import semver from "semver";
 
 export interface MojiAdapter {
@@ -31,6 +31,16 @@ export interface MojiAdapter {
    * A function to generate the emoji sequences for the specified version
    */
   sequences?: SequenceFn;
+
+  /**
+   * A function to generate the emoji dataset for the specified version.
+   */
+  emojis?: EmojiFn;
+
+  /**
+   * A function to generate emoji variations for the specified version.
+   */
+  variations?: EmojiVariationFn;
 }
 
 export interface BaseAdapterContext {
@@ -40,6 +50,8 @@ export interface BaseAdapterContext {
 
 export type GroupFn = (ctx: BaseAdapterContext) => Promise<EmojiGroup[]>;
 export type SequenceFn = (ctx: BaseAdapterContext) => Promise<{ zwj: EmojiSequence[]; sequences: EmojiSequence[] }>;
+export type EmojiFn = (ctx: BaseAdapterContext) => Promise<any>;
+export type EmojiVariationFn = (ctx: BaseAdapterContext) => Promise<EmojiVariation[]>;
 
 export const ADAPTERS = new Map<string, MojiAdapter>();
 

src/adapter/v16.ts

@@ -1,4 +1,4 @@
-import type { EmojiSequence } from "../types";
+import type { EmojiSequence, EmojiVariation } from "../types";
 import { defineMojiAdapter } from "../adapter";
 import { FEMALE_SIGN, MALE_SIGN } from "../constants";
 import { fetchCache } from "../utils/cache";
@@ -68,4 +68,52 @@ export default defineMojiAdapter({
       zwj: zwj || [],
     };
   },
+  async emojis({ version, force }) {
+  },
+  variations: async (ctx) => {
+    return fetchCache(`https://unicode.org/Public/${ctx.version}.0/ucd/emoji/emoji-variation-sequences.txt`, {
+      cacheKey: `v${ctx.version}/variations.json`,
+      parser(data) {
+        const lines = data.split("\n");
+
+        const variations: EmojiVariation[] = [];
+
+        for (let line of lines) {
+          // skip empty line & comments
+          if (line.trim() === "" || line.startsWith("#")) {
+            continue;
+          }
+
+          // remove line comment
+          const commentIndex = line.indexOf("#");
+          if (commentIndex !== -1) {
+            line = line.slice(0, commentIndex).trim();
+          }
+
+          const [hex, style] = line.split(";").map((col) => col.trim()).slice(0, 4);
+
+          if (hex == null || style == null) {
+            throw new Error(`invalid line: ${line}`);
+          }
+
+          const hexcode = hex.replace(/\s+/g, "-");
+
+          const type = style.replace("style", "").trim();
+
+          if (type !== "text" && type !== "emoji") {
+            throw new Error(`invalid style: ${style}`);
+          }
+
+          variations.push({
+            emoji: type === "emoji" ? hexcode : null,
+            text: type === "text" ? hexcode : null,
+            property: ["Emoji"],
+          });
+        }
+
+        return variations;
+      },
+      bypassCache: ctx.force,
+    });
+  },
 });

src/types.ts

@@ -44,3 +44,47 @@ export interface EmojiSequence {
   description: string;
   gender: string | null;
 }
+
+export type Property =
+  // An emoji character.
+  | "Basic_Emoji"
+  | "Emoji"
+  // The presentation in which to display the emoji character. Either emoji or text.
+  | "Emoji_Presentation"
+  // An emoji or unicode character for modifying complex sequences (hair style, skin tone, etc),
+  // and should never be used as a stand-alone emoji.
+  | "Emoji_Component"
+  // An emoji character that modifies a preceding emoji base character.
+  | "Emoji_Modifier"
+  // An emoji character. Can be modified with a subsequent emoji modifier.
+  | "Emoji_Modifier_Base"
+  // A sequence of a base and modifier ("Emoji_Modifier_Base" + "Emoji_Modifier").
+  | "Emoji_Modifier_Sequence"
+  // A sequence of unicode characters representing the available keys on a
+  // phone dial: 0-9, *, # ("Key" + "FE0F" + "20E3").
+  | "Emoji_Keycap_Sequence"
+  // A sequence of 2 regional indicators representing a region flag (nation).
+  // "Regional_Indicator" + "Regional_Indicator".
+  | "Emoji_Flag_Sequence"
+  // A sequence of characters that are not ZWJ or flag sequences.
+  // Currently used for representing sub-region/division flags (country).
+  | "Emoji_Tag_Sequence"
+  // A sequence of multiple emoji characters joined with a zero-width-joiner (200D).
+  | "Emoji_ZWJ_Sequence"
+  // Either an "Emoji", "Emoji" + "FE0F", or "Emoji_Keycap_Sequence".
+  | "Emoji_Combining_Sequence"
+  // An emoji slot that is reserved for future allocations and releases.
+  | "Extended_Pictographic"
+  // A unicode character representing one of the 26 letters of the alphabet, A-Z.
+  | "Regional_Indicator"
+  // v13+ renamed properties
+  | "RGI_Emoji_Flag_Sequence"
+  | "RGI_Emoji_Modifier_Sequence"
+  | "RGI_Emoji_Tag_Sequence"
+  | "RGI_Emoji_ZWJ_Sequence";
+
+export interface EmojiVariation {
+  text: string | null;
+  emoji: string | null;
+  property?: Property[];
+}