Skip to content
This repository was archived by the owner on Mar 1, 2025. It is now read-only.

Commit 49bf2cf

Browse files
luxassluxass
committedFeb 16, 2025··
chore: dump
1 parent 4c21308 commit 49bf2cf

File tree

5 files changed

+559
-4
lines changed

5 files changed

+559
-4
lines changed
 

‎src/adapter/base.ts

+1-1
Original file line numberDiff line numberDiff line change
@@ -119,7 +119,7 @@ export default defineMojiAdapter({
119119
emojis: notImplemented("emojis"),
120120
variations: notImplemented("variations"),
121121
unicodeNames: async (ctx) => {
122-
return fetchCache(`https://unicode.org/Public/${ctx.emojiVersion}.0/ucd/UnicodeData.txt`, {
122+
return fetchCache(`https://unicode.org/Public/${ctx.emojiVersion === "13.1" ? "13.0" : ctx.emojiVersion}.0/ucd/UnicodeData.txt`, {
123123
cacheKey: `v${ctx.emojiVersion}/unicode-names.json`,
124124
parser(data) {
125125
const lines = data.split("\n");

‎src/adapter/v13.ts

+186-1
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,193 @@
1-
import { defineMojiAdapter } from ".";
1+
import type { Emoji, EmojiData, EmojiSequence, EmojiVariation, Property } from "../types";
2+
import { defineMojiAdapter } from "../adapter";
3+
import { FEMALE_SIGN, MALE_SIGN } from "../constants";
4+
import { extractEmojiVersion, extractUnicodeVersion } from "../utils";
5+
import { fetchCache } from "../utils/cache";
6+
import { expandHexRange } from "../utils/hexcode";
27

38
export default defineMojiAdapter({
49
name: "v13",
510
description: "adapter for version 13 & 13.1",
611
range: ">=13.0.0 <14.0.0",
712
extend: "base",
13+
sequences: async (ctx) => {
14+
const [sequences, zwj] = await Promise.all([
15+
{
16+
cacheKey: `v${ctx.emojiVersion}/sequences.json`,
17+
url: `https://unicode.org/Public/emoji/${ctx.emojiVersion}/emoji-sequences.txt`,
18+
},
19+
{
20+
cacheKey: `v${ctx.emojiVersion}/zwj-sequences.json`,
21+
url: `https://unicode.org/Public/emoji/${ctx.emojiVersion}/emoji-zwj-sequences.txt`,
22+
},
23+
].map(async ({ cacheKey, url }) => {
24+
return await fetchCache(url, {
25+
cacheKey,
26+
parser(data) {
27+
const lines = data.split("\n");
28+
29+
const sequences: EmojiSequence[] = [];
30+
31+
for (let line of lines) {
32+
// skip empty line & comments
33+
if (line.trim() === "" || line.startsWith("#")) {
34+
continue;
35+
}
36+
37+
// remove line comment
38+
const commentIndex = line.indexOf("#");
39+
if (commentIndex !== -1) {
40+
line = line.slice(0, commentIndex).trim();
41+
}
42+
43+
const [hex, property, description] = line.split(";").map((col) => col.trim()).slice(0, 4);
44+
45+
if (hex == null || property == null || description == null) {
46+
throw new Error(`invalid line: ${line}`);
47+
}
48+
49+
const expandedHex = expandHexRange(hex);
50+
51+
for (const hex of expandedHex) {
52+
sequences.push({
53+
hex: hex.replace(/\s+/g, "-"),
54+
property,
55+
description,
56+
gender: hex.includes(FEMALE_SIGN) ? "female" : hex.includes(MALE_SIGN) ? "male" : null,
57+
});
58+
}
59+
}
60+
61+
return sequences;
62+
},
63+
bypassCache: ctx.force,
64+
});
65+
}));
66+
67+
return {
68+
sequences: sequences || [],
69+
zwj: zwj || [],
70+
};
71+
},
72+
async emojis(ctx) {
73+
const unicodeNames = await this.unicodeNames!(ctx);
74+
// const { sequences, zwj } = await this.sequences!(ctx);
75+
// const metadata = await this.metadata!(ctx);
76+
// const variations = await this.variations!(ctx);
77+
78+
const emojis: Record<string, Record<string, Record<string, Emoji>>> = {};
79+
80+
const emojiData = await fetchCache(`https://unicode.org/Public/${ctx.emojiVersion === "13.1" ? "13.0" : ctx.emojiVersion}.0/ucd/emoji/emoji-data.txt`, {
81+
cacheKey: `v${ctx.emojiVersion}/emoji-data.json`,
82+
parser(data) {
83+
const lines = data.split("\n");
84+
85+
const emojiData: Record<string, EmojiData> = {};
86+
87+
for (const line of lines) {
88+
// skip empty line & comments
89+
if (line.trim() === "" || line.startsWith("#")) {
90+
continue;
91+
}
92+
93+
const lineCommentIndex = line.indexOf("#");
94+
const lineComment = lineCommentIndex !== -1 ? line.slice(lineCommentIndex + 1).trim() : "";
95+
96+
let [hex, property] = line.split(";").map((col) => col.trim()).slice(0, 4);
97+
98+
if (hex == null || property == null) {
99+
throw new Error(`invalid line: ${line}`);
100+
}
101+
102+
// remove line comment from property
103+
const propertyCommentIndex = property.indexOf("#");
104+
if (propertyCommentIndex !== -1) {
105+
property = property.slice(0, propertyCommentIndex).trim();
106+
}
107+
108+
if (property === "Extended_Pictographic") {
109+
continue;
110+
}
111+
112+
const expandedHex = expandHexRange(hex);
113+
const emojiVersion = extractEmojiVersion(lineComment);
114+
115+
const emoji: EmojiData = {
116+
description: lineComment,
117+
hexcode: "",
118+
gender: null,
119+
properties: [(property as Property) || "Emoji"],
120+
unicodeVersion: extractUnicodeVersion(emojiVersion, ctx.unicodeVersion),
121+
emojiVersion,
122+
name: unicodeNames[hex] || "",
123+
};
124+
125+
for (const hex of expandedHex) {
126+
if (emojiData[hex] != null) {
127+
emojiData[hex].properties = [...new Set([...emojiData[hex].properties, ...emoji.properties])];
128+
} else {
129+
emojiData[hex] = {
130+
...emoji,
131+
hexcode: hex.replace(/\s+/g, "-"),
132+
};
133+
}
134+
}
135+
}
136+
137+
return emojiData;
138+
},
139+
bypassCache: ctx.force,
140+
});
141+
142+
return {
143+
emojiData,
144+
emojis,
145+
};
146+
},
147+
variations: async (ctx) => {
148+
return fetchCache(`https://unicode.org/Public/${ctx.emojiVersion === "13.1" ? "13.0" : ctx.emojiVersion}.0/ucd/emoji/emoji-variation-sequences.txt`, {
149+
cacheKey: `v${ctx.emojiVersion}/variations.json`,
150+
parser(data) {
151+
const lines = data.split("\n");
152+
153+
const variations: EmojiVariation[] = [];
154+
155+
for (let line of lines) {
156+
// skip empty line & comments
157+
if (line.trim() === "" || line.startsWith("#")) {
158+
continue;
159+
}
160+
161+
// remove line comment
162+
const commentIndex = line.indexOf("#");
163+
if (commentIndex !== -1) {
164+
line = line.slice(0, commentIndex).trim();
165+
}
166+
167+
const [hex, style] = line.split(";").map((col) => col.trim()).slice(0, 4);
168+
169+
if (hex == null || style == null) {
170+
throw new Error(`invalid line: ${line}`);
171+
}
172+
173+
const hexcode = hex.replace(/\s+/g, "-");
174+
175+
const type = style.replace("style", "").trim();
176+
177+
if (type !== "text" && type !== "emoji") {
178+
throw new Error(`invalid style: ${style}`);
179+
}
180+
181+
variations.push({
182+
emoji: type === "emoji" ? hexcode : null,
183+
text: type === "text" ? hexcode : null,
184+
property: ["Emoji"],
185+
});
186+
}
187+
188+
return variations;
189+
},
190+
bypassCache: ctx.force,
191+
});
192+
},
8193
});

‎src/adapter/v14.ts

+186-1
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,193 @@
1-
import { defineMojiAdapter } from ".";
1+
import type { Emoji, EmojiData, EmojiSequence, EmojiVariation, Property } from "../types";
2+
import { defineMojiAdapter } from "../adapter";
3+
import { FEMALE_SIGN, MALE_SIGN } from "../constants";
4+
import { extractEmojiVersion, extractUnicodeVersion } from "../utils";
5+
import { fetchCache } from "../utils/cache";
6+
import { expandHexRange } from "../utils/hexcode";
27

38
export default defineMojiAdapter({
49
name: "v14",
510
description: "adapter for version 14",
611
range: ">=14.0.0 <15.0.0",
712
extend: "base",
13+
sequences: async (ctx) => {
14+
const [sequences, zwj] = await Promise.all([
15+
{
16+
cacheKey: `v${ctx.emojiVersion}/sequences.json`,
17+
url: `https://unicode.org/Public/emoji/${ctx.emojiVersion}/emoji-sequences.txt`,
18+
},
19+
{
20+
cacheKey: `v${ctx.emojiVersion}/zwj-sequences.json`,
21+
url: `https://unicode.org/Public/emoji/${ctx.emojiVersion}/emoji-zwj-sequences.txt`,
22+
},
23+
].map(async ({ cacheKey, url }) => {
24+
return await fetchCache(url, {
25+
cacheKey,
26+
parser(data) {
27+
const lines = data.split("\n");
28+
29+
const sequences: EmojiSequence[] = [];
30+
31+
for (let line of lines) {
32+
// skip empty line & comments
33+
if (line.trim() === "" || line.startsWith("#")) {
34+
continue;
35+
}
36+
37+
// remove line comment
38+
const commentIndex = line.indexOf("#");
39+
if (commentIndex !== -1) {
40+
line = line.slice(0, commentIndex).trim();
41+
}
42+
43+
const [hex, property, description] = line.split(";").map((col) => col.trim()).slice(0, 4);
44+
45+
if (hex == null || property == null || description == null) {
46+
throw new Error(`invalid line: ${line}`);
47+
}
48+
49+
const expandedHex = expandHexRange(hex);
50+
51+
for (const hex of expandedHex) {
52+
sequences.push({
53+
hex: hex.replace(/\s+/g, "-"),
54+
property,
55+
description,
56+
gender: hex.includes(FEMALE_SIGN) ? "female" : hex.includes(MALE_SIGN) ? "male" : null,
57+
});
58+
}
59+
}
60+
61+
return sequences;
62+
},
63+
bypassCache: ctx.force,
64+
});
65+
}));
66+
67+
return {
68+
sequences: sequences || [],
69+
zwj: zwj || [],
70+
};
71+
},
72+
async emojis(ctx) {
73+
const unicodeNames = await this.unicodeNames!(ctx);
74+
// const { sequences, zwj } = await this.sequences!(ctx);
75+
// const metadata = await this.metadata!(ctx);
76+
// const variations = await this.variations!(ctx);
77+
78+
const emojis: Record<string, Record<string, Record<string, Emoji>>> = {};
79+
80+
const emojiData = await fetchCache(`https://unicode.org/Public/${ctx.emojiVersion}.0/ucd/emoji/emoji-data.txt`, {
81+
cacheKey: `v${ctx.emojiVersion}/emoji-data.json`,
82+
parser(data) {
83+
const lines = data.split("\n");
84+
85+
const emojiData: Record<string, EmojiData> = {};
86+
87+
for (const line of lines) {
88+
// skip empty line & comments
89+
if (line.trim() === "" || line.startsWith("#")) {
90+
continue;
91+
}
92+
93+
const lineCommentIndex = line.indexOf("#");
94+
const lineComment = lineCommentIndex !== -1 ? line.slice(lineCommentIndex + 1).trim() : "";
95+
96+
let [hex, property] = line.split(";").map((col) => col.trim()).slice(0, 4);
97+
98+
if (hex == null || property == null) {
99+
throw new Error(`invalid line: ${line}`);
100+
}
101+
102+
// remove line comment from property
103+
const propertyCommentIndex = property.indexOf("#");
104+
if (propertyCommentIndex !== -1) {
105+
property = property.slice(0, propertyCommentIndex).trim();
106+
}
107+
108+
if (property === "Extended_Pictographic") {
109+
continue;
110+
}
111+
112+
const expandedHex = expandHexRange(hex);
113+
const emojiVersion = extractEmojiVersion(lineComment);
114+
115+
const emoji: EmojiData = {
116+
description: lineComment,
117+
hexcode: "",
118+
gender: null,
119+
properties: [(property as Property) || "Emoji"],
120+
unicodeVersion: extractUnicodeVersion(emojiVersion, ctx.unicodeVersion),
121+
emojiVersion,
122+
name: unicodeNames[hex] || "",
123+
};
124+
125+
for (const hex of expandedHex) {
126+
if (emojiData[hex] != null) {
127+
emojiData[hex].properties = [...new Set([...emojiData[hex].properties, ...emoji.properties])];
128+
} else {
129+
emojiData[hex] = {
130+
...emoji,
131+
hexcode: hex.replace(/\s+/g, "-"),
132+
};
133+
}
134+
}
135+
}
136+
137+
return emojiData;
138+
},
139+
bypassCache: ctx.force,
140+
});
141+
142+
return {
143+
emojiData,
144+
emojis,
145+
};
146+
},
147+
variations: async (ctx) => {
148+
return fetchCache(`https://unicode.org/Public/${ctx.emojiVersion}.0/ucd/emoji/emoji-variation-sequences.txt`, {
149+
cacheKey: `v${ctx.emojiVersion}/variations.json`,
150+
parser(data) {
151+
const lines = data.split("\n");
152+
153+
const variations: EmojiVariation[] = [];
154+
155+
for (let line of lines) {
156+
// skip empty line & comments
157+
if (line.trim() === "" || line.startsWith("#")) {
158+
continue;
159+
}
160+
161+
// remove line comment
162+
const commentIndex = line.indexOf("#");
163+
if (commentIndex !== -1) {
164+
line = line.slice(0, commentIndex).trim();
165+
}
166+
167+
const [hex, style] = line.split(";").map((col) => col.trim()).slice(0, 4);
168+
169+
if (hex == null || style == null) {
170+
throw new Error(`invalid line: ${line}`);
171+
}
172+
173+
const hexcode = hex.replace(/\s+/g, "-");
174+
175+
const type = style.replace("style", "").trim();
176+
177+
if (type !== "text" && type !== "emoji") {
178+
throw new Error(`invalid style: ${style}`);
179+
}
180+
181+
variations.push({
182+
emoji: type === "emoji" ? hexcode : null,
183+
text: type === "text" ? hexcode : null,
184+
property: ["Emoji"],
185+
});
186+
}
187+
188+
return variations;
189+
},
190+
bypassCache: ctx.force,
191+
});
192+
},
8193
});

0 commit comments

Comments
 (0)
This repository has been archived.