Emoji Detection in Text Strings: Algorithms and Libraries

Emoji Detection in Text Strings

Detecting whether a string contains emojis—and extracting them accurately—is harder than it looks. The Unicode standard has grown to include over 3,700 emoji characters spread across multiple code point ranges, with new ones added every year. A naive approach using a fixed range check will miss most of them.

This guide covers the algorithms, Unicode properties, and production-ready libraries you need to detect emojis reliably.

Why Simple Range Checks Fail

A common first attempt is checking whether a code point falls in the range U+1F600–U+1F64F (Emoticons block). This catches classics like 😀, 😂, and 😎, but misses:

Basic emoji: ©️ (U+00A9), ® (U+00AE), ™️ (U+2122) — in the Latin Extended range
Dingbats: ✅ (U+2705), ❌ (U+274C) — in the Dingbats block
Enclosed alphanumerics: 🅰️, 🅱️
ZWJ零宽连接符（ZWJ） (ZWJ)
一个不可见的Unicode字符（U+200D），用于将多个表情符号合并为单一复合表情符号，例如将人物与物体组合成职业表情符号。 sequences: 👨‍💻 (man technologist) — multiple code points joined by U+200D
Keycap sequences: 1️⃣ — digit + variation selector + combining enclosing keycap
Flag sequences: 🇺🇸 — pairs of Regional Indicator letters

The only reliable approach is to use the official Unicode emoji property data.

The Unicode Emoji Properties Approach

Unicode defines several properties relevant to emoji detection, published in emoji-data.txt from the Unicode Character Database (UCD):

Property	Meaning
`Emoji`	The code point is an emoji
`Emoji_Presentation`	Displayed as emoji by default (not text)
`Emoji_Modifier`	A skin tone modifier (🏻–🏿)
`Emoji_Modifier_Base`	Can be modified by a skin tone modifier
`Emoji_Component`	Used in emoji sequences (ZWJ, keycap, etc.)
`Extended_Pictographic`	Broader set including reserved ranges

For most detection tasks you want Extended_Pictographic, which includes current emoji plus code points reserved for future emoji assignments.

Detection in Python

Using the `emoji` Library

The emoji library maintains an up-to-date Unicode dataset:

import emoji

text = "Hello 👋 world! Check this out 🚀"

# Check if string contains any emoji
has_emoji = emoji.emoji_count(text) > 0
print(has_emoji)  # True

# Count emojis
count = emoji.emoji_count(text)
print(count)  # 2

# Extract emoji with positions
for item in emoji.emoji_list(text):
    print(item)
# {'match_start': 6, 'match_end': 7, 'emoji': '👋'}
# {'match_start': 26, 'match_end': 27, 'emoji': '🚀'}

# Replace emojis
clean = emoji.replace_emoji(text, replace="")
print(clean)  # "Hello  world! Check this out "

Using the `regex` Module with Unicode Properties

The third-party regex module (not the built-in re) supports Unicode properties:

import regex

# Match any Extended_Pictographic character or emoji sequence
EMOJI_PATTERN = regex.compile(
    r'\p{Extended_Pictographic}'
    r'(?:\uFE0F)?'           # optional variation selector-16
    r'(?:\u20E3)?'           # optional combining enclosing keycap
    r'(?:\uFE0F\u20E3)?'     # keycap sequence
    r'(?:\u200D\p{Extended_Pictographic}(?:\uFE0F)?)*'  # ZWJ sequences
    r'(?:[\U0001F1E0-\U0001F1FF]{2})?',  # flag sequences
    regex.UNICODE
)

text = "Deploying 🚀 to production 👨‍💻 — fingers crossed 🤞🏽"
matches = EMOJI_PATTERN.findall(text)
print(matches)  # ['🚀', '👨‍💻', '🤞🏽']

Pure stdlib with unicodedata

For simpler cases without extra dependencies, check the unicodedata category:

import unicodedata

def contains_emoji_simple(text: str) -> bool:
    for char in text:
        cat = unicodedata.category(char)
        # So (Symbol, other) covers many but not all emoji
        if cat == "So":
            return True
    return False

This is fast but incomplete — it misses many emoji that fall in other categories.

Detection in JavaScript

Using the `emoji-regex` Package

import emojiRegex from 'emoji-regex';

const regex = emojiRegex();
const text = "Meeting at 3pm 📅 — bring your laptop 💻";

// Test for presence
console.log(regex.test(text)); // true

// Extract all emojis
const matches = [...text.matchAll(regex)];
matches.forEach(m => {
  console.log(`Found: ${m[0]} at index ${m.index}`);
});
// Found: 📅 at index 15
// Found: 💻 at index 38

// Count
const count = [...text.matchAll(regex)].length;
console.log(count); // 2

Note that emoji-regex is generated directly from Unicode data, so it stays accurate across emoji versions.

Native Unicode Property Escapes (ES2018+)

Modern JavaScript engines support \p{} in regex with the u flag:

// Requires Node.js 10+ or modern browsers
const emojiRx = /\p{Emoji}/u;
const extPictoRx = /\p{Extended_Pictographic}/u;

console.log(emojiRx.test("Hello 🌍")); // true
console.log(extPictoRx.test("No emoji here")); // false

// Extract using matchAll
const text = "Status: ✅ Build passed, 🔴 Tests failed";
const allEmoji = [...text.matchAll(/\p{Extended_Pictographic}/gu)];
console.log(allEmoji.map(m => m[0])); // ['✅', '🔴']

Detection in Go

package main

import (
    "fmt"
    "unicode"
    "golang.org/x/text/unicode/rangetable"
)

// Basic check using unicode.Is
func containsEmoji(s string) bool {
    for _, r := range s {
        if unicode.Is(unicode.So, r) || // Symbol, other
           (r >= 0x1F600 && r <= 0x1FFFF) || // Supplemental symbols
           (r >= 0x2600 && r <= 0x27BF) {    // Misc symbols
            return true
        }
    }
    return false
}

func main() {
    texts := []string{
        "Hello world",
        "Rocket 🚀 launched",
        "©️ Copyright symbol",
    }
    for _, t := range texts {
        fmt.Printf("%q → %v\n", t, containsEmoji(t))
    }
}

For production Go code, consider the github.com/rivo/uniseg package, which handles grapheme cluster segmentation correctly and can identify emoji clusters.

Handling Edge Cases

Variation Selectors

Many emoji have both a text (VS15, U+FE0E) and emoji (VS16, U+FE0F) presentation. The digit ☎ can appear as ☎︎ (text) or ☎️ (emoji). Your detection should account for the variation selector:

phone_text = "\u260E\uFE0E"   # ☎︎  text presentation
phone_emoji = "\u260E\uFE0F"  # ☎️  emoji presentation

import emoji
print(emoji.emoji_count(phone_text))   # 0
print(emoji.emoji_count(phone_emoji))  # 1

ZWJ Sequences

👨‍💻 is a single grapheme cluster composed of 👨 + ZWJ (U+200D) + 💻. When counting or extracting emoji, treat ZWJ sequences as one unit. Libraries like emoji (Python) and emoji-regex (JS) handle this automatically.

Regional Indicator Flags

Country flags like 🇩🇪 consist of two Regional Indicator letters (U+1F1E6–U+1F1FF). They are only valid in pairs. A single 🇩 without a following 🇪 is not a flag.

Performance Considerations

For high-throughput text processing:

Pre-compile your regex — do it once at module load, not per call
Short-circuit on ASCII — if all bytes are < 128, there are no emoji (they are all non-ASCII)
Use a library — regex-based approaches with proper Unicode support are faster than custom range tables you maintain yourself

def fast_has_emoji(text: str) -> bool:
    # Short-circuit: emoji require non-ASCII bytes in UTF-8UTF-8
每字符使用1至4个字节的可变长度Unicode编码方式，是互联网上最主流的编码，被98%以上的网站采用。
    if text.isascii():
        return False
    return emoji.emoji_count(text) > 0

Explore More on EmojiFYI

Analyze emoji sequences in detail: Sequence Analyzer
Browse the emoji glossary: Glossary
Integrate emoji data in your app: API Reference
Find specific emoji fast: Search

Emoji Detection in Text Strings: Algorithms and Libraries