Sentence Boundary
Matches sentence boundaries (punctuation followed by whitespace and a capital letter).
Try it in RegexPro →Available in
JavaScript / ECMAScript code
const re = new RegExp("[.!?]\\s+(?=[A-Z])", "g");
const input = "Hello world. How are you? I'm fine!";
const matches = [...input.matchAll(re)];
console.log(matches.map(m => m[0]));Uses `String.prototype.matchAll` for global iteration (Node 12+ / all modern browsers).
Python (re) code
import re
pattern = re.compile(r"[.!?]\s+(?=[A-Z])")
input_text = "Hello world. How are you? I'm fine!"
for m in pattern.finditer(input_text):
print(m.group(0))Stdlib `re` module — no third-party dependency. Works on Python 3.6+.
Why it doesn't work in GO
Go's RE2 engine doesn't support lookarounds (`(?=...)`, `(?<=...)`, etc.) — they break the linear-time matching guarantee.
Approach
Restructure to capture the surrounding context as a group instead, or use JS / Python where lookarounds are available.
Read the full guide →Workaround code in Go (RE2)
package main
import (
"fmt"
"regexp"
"unicode"
)
// RE2 doesn't support lookarounds. The fix: match a BROADER candidate
// without the lookaround, then verify the condition in Go code.
//
// Example: instead of `(?=.*\d)[A-Za-z\d]{8,}` (require a digit),
// match the candidate then check `HasDigit(s)` separately.
func HasDigit(s string) bool {
for _, r := range s {
if unicode.IsDigit(r) {
return true
}
}
return false
}
func main() {
re := regexp.MustCompile(`[A-Za-z\d]{8,}`) // simplified, no lookaround
input := "Password1 weakpass StrongerOne9"
for _, candidate := range re.FindAllString(input, -1) {
if HasDigit(candidate) { // the lookaround condition, in code
fmt.Println(candidate)
}
}
}Match a broader candidate without the lookaround, then verify the constraint in Go. Keeps the linear-time guarantee.
Pattern
[.!?]\s+(?=[A-Z]) (flags: g)Raw source: [.!?]\s+(?=[A-Z])
How it works
Examples
Input
Hello world. How are you? I'm fine!Matches
.?
Common use cases
- •Sentence splitting
- •NLP preprocessing
- •Reading-level analysis
Related patterns
Word Boundary (Whole Word Match)
Text ProcessingMatch the whole word `cat` without matching it inside other words like `concatenate` or `category`.
Hashtag
Text ProcessingMatch hashtags (# followed by word characters) in social media posts, including accented Latin characters.
JSON Boolean / Null Literal
Text ProcessingMatch JSON `true`, `false`, and `null` literal values, with word boundaries to avoid partial matches.
Positive Lookahead (Password)
Text ProcessingUse positive lookahead `(?=...)` to require the password contain at least one digit and one uppercase letter.