probably-china 1.1.0 → 1.1.1

package/README.md

@@ -5,10 +5,14 @@
 
 This is a simple package for the browser environment that makes a quick guess
 to see if the current webpage is running in China, or if the user is from China,
-based on the available browser info. It currently uses two strategies:
-
-1. Check if the current browser favors simplified Chinese or related languages.
-2. Check if the current timezone is one of the Chinese timezones.
+based on the available browser info. It currently uses the following strategies:
+
+1. Check if the current timezone is one of the Chinese timezones.
+2. Check if the current browser favors simplified Chinese or related languages
+   (unless a Taiwanese language tag comes first).
+3. Check if the user agent is an in-app browser or a browser that is mostly used in China
+   (such as WeChat, QQ, Alipay, Douyin, or HarmonyOS).
+4. Check if the user comes from a Chinese website (such as Baidu or Zhihu).
 
 The use case for this package is not to avoid showing information that are sensitive in China.
 

package/dist/index.js

@@ -1,37 +1,94 @@
-
-const detectChinaLanguage = ["zh-CN", "zh-CHS", "zh-Hans", "zh-HK", "zh-MO"];
-const detectChinaTimeZone = ["Asia/Shanghai", "Asia/Urumqi", "Asia/Hong_Kong", "Asia/Macau"];
-
-function checkTimeZone() {
-	if(typeof Intl !== "undefined") {
-		const timeZone = new Intl.DateTimeFormat().resolvedOptions().timeZone;
-		return detectChinaTimeZone.indexOf(timeZone) >= 0;
-	} else {
-		// legacy browser
-		const str = new Date().toString();
-		return str.indexOf("\u4e2d") > 0 || str.indexOf("\u9999") > 0;
-	}
-}
-
-function checkChinaLanguages() {
-	if(navigator.languages) { // https://caniuse.com/mdn-api_navigator_languages
-		for(let i = 0; i < navigator.languages.length; i++) {
-			const l = navigator.languages[i];
-			// Some Taiwanese users might also have "zh-CN" in their language list.
-			// We use the following criteria to eliminate such false-positive,
-			// so that it returns false if "zh-TW" comes first in the language list.
-			if(l === "zh-TW") return false;
-			if(detectChinaLanguage.indexOf(l) >= 0) return true;
-		}
-		return false;
-	}
-	// legacy browser
-	return detectChinaLanguage.indexOf(navigator.language) >= 0;
-}
-
-/**
- * A lucky guess that we are probably in China or facing a Chinese user.
- */
-export const probablyChina = checkTimeZone() || checkChinaLanguages();
-
-export default probablyChina;
+const detectChinaRegion = ["cn", "hk", "mo"];
+const detectChinaTimeZone = [
+	"Asia/Shanghai", "Asia/Urumqi", "Asia/Hong_Kong", "Asia/Macau",
+	// Legacy aliases, which may be returned as-is in some environments
+	"PRC", "Asia/Chongqing", "Asia/Chungking", "Asia/Harbin", "Asia/Kashgar", "Asia/Macao", "Hongkong",
+];
+
+// In-app browsers or browsers that are mostly used in China only.
+// Browsers with a large user base outside China (such as UC, Mi, HeyTap, vivo or Huawei) are intentionally excluded.
+const detectChinaUserAgent = new RegExp([
+	"MicroMessenger", "MQQBrowser", "QQ/", "Quark", "baiduboxapp", "BIDUBrowser", "AlipayClient", "DingTalk",
+	"Weibo", "aweme", "NewsArticle", "Feishu", "SogouMobileBrowser", "MetaSr", "360SE", "360EE",
+	"2345Explorer", "LBBROWSER", "OpenHarmony", "ArkWeb",
+].join("|"));
+
+const detectChinaReferrer = /^https?:\/\/([^/]+\.)?(baidu\.com|so\.com|sogou\.com|sm\.cn|cn\.bing\.com|weixin\.qq\.com|weixin110\.qq\.com|zhihu\.com|weibo\.com|weibo\.cn|xiaohongshu\.com|douyin\.com|bilibili\.com)(:\d+)?(\/|$)/i;
+
+function checkTimeZone() {
+	if(typeof Intl !== "undefined") {
+		const timeZone = new Intl.DateTimeFormat().resolvedOptions().timeZone;
+		return detectChinaTimeZone.indexOf(timeZone) >= 0;
+	} else {
+		// legacy browser
+		const str = new Date().toString();
+		return str.indexOf("中") > 0 || str.indexOf("香") > 0;
+	}
+}
+
+/**
+ * Classify a language tag.
+ * @param {string} tag
+ * @returns {boolean | undefined} `true` for China, `false` for Taiwan, `undefined` if undecided.
+ */
+function classifyLanguage(tag) {
+	// Old browsers (such as Safari < 10) report tags in lower case.
+	const parts = tag.toLowerCase().split("-");
+	if(parts[0] !== "zh") return undefined;
+
+	// Legacy Microsoft tags
+	if(parts[1] === "chs") return true;
+	if(parts[1] === "cht") return false;
+
+	let script = "", region = "";
+	for(let i = 1; i < parts.length; i++) {
+		if(parts[i].length === 4) script = parts[i];
+		else if(parts[i].length === 2) region = parts[i];
+	}
+
+	// Region takes precedence, so for example "zh-Hans-SG" is not China.
+	if(region === "tw") return false;
+	if(detectChinaRegion.indexOf(region) >= 0) return true;
+	if(region) return undefined;
+
+	// No region; "zh" and "zh-Hans" maximize to "zh-Hans-CN", while "zh-Hant" maximizes to "zh-Hant-TW".
+	if(script === "hant") return false;
+	return true;
+}
+
+function checkChinaLanguages() {
+	const list = [];
+	if(navigator.languages) { // https://caniuse.com/mdn-api_navigator_languages
+		for(let i = 0; i < navigator.languages.length; i++) list.push(navigator.languages[i]);
+	} else if(navigator.language) {
+		// legacy browser
+		list.push(navigator.language);
+	}
+	// The browser UI locale, which in some browsers (such as Firefox) can differ from the content language list.
+	if(typeof Intl !== "undefined") list.push(new Intl.DateTimeFormat().resolvedOptions().locale);
+
+	for(let i = 0; i < list.length; i++) {
+		// Some Taiwanese users might also have "zh-CN" in their language list.
+		// The first decisive tag wins, so that it returns false if "zh-TW" comes first in the language list.
+		const result = classifyLanguage(list[i]);
+		if(result !== undefined) return result;
+	}
+	return false;
+}
+
+function checkUserAgent() {
+	// False positive: navigator.userAgent exists in early Opera versions, but the compat data only records it since Opera 10.5.
+	// eslint-disable-next-line compat/compat
+	return detectChinaUserAgent.test(navigator.userAgent || "");
+}
+
+function checkReferrer() {
+	return typeof document !== "undefined" && detectChinaReferrer.test(document.referrer);
+}
+
+/**
+ * A lucky guess that we are probably in China or facing a Chinese user.
+ */
+export const probablyChina = checkTimeZone() || checkChinaLanguages() || checkUserAgent() || checkReferrer();
+
+export default probablyChina;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "probably-china",
-  "version": "1.1.0",
+  "version": "1.1.1",
   "description": "A lucky guess that we are probably in China or facing a Chinese user.",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -8,7 +8,10 @@
     "China"
   ],
   "author": "Mu-Tsun Tsai <don.m.t.tsai@gmail.com> (https://github.com/MuTsunTsai)",
-  "repository": "github:MuTsunTsai/probably-china",
+  "repository": {
+    "type": "git",
+    "url": "github:MuTsunTsai/probably-china"
+  },
   "license": "MIT",
   "files": [
     "dist"