この計算機の使い方
Building a pattern the way the engine sees it
A regular expression is a miniature program: the engine walks your text, and at every position it asks whether the pattern matches starting there. The tester makes that walk visible — every match is listed with its position, the exact text it matched, and the capture groups it produced, updated as you type. Start with the literal that must appear (request_id=, say), add the flexible parts around it (a w+ for the value), and check the result against real lines from your data before you trust it anywhere.
The engine here is the platform’s ECMAScript RegExp — the same one JavaScript, Node.js, Deno, and every modern browser use. That is the point of the tool’s fidelity promise: a pattern that matches here matches in your code, and one that fails here will fail there. Dialects like PCRE (PHP, grep, regex101’s default) have features ECMAScript lacks or spells differently — lookbehind syntax, named-group conventions, atomic groups — and the tester rejects them with the platform’s error message rather than pretending to support a dialect it does not run.
Using the flags deliberately
Flags are not decorations — they change what the pattern means. Without g, the engine stops at the first match, and the classic "why does my pattern match once?" confusion is usually a missing g. Without m, the anchors ^ and $ bind to the whole string, not to line boundaries, so a pattern that should match each log line silently matches nothing when the text has multiple lines. The s flag is the one that surprises everyone: . matches every character except line breaks unless s is set, so patterns that should span lines fail on the newline. And the difference between the unicode u and non-unicode modes changes how astral characters like emoji are treated — a point where the ECMAScript engine is stricter than older dialects.
The tester shows the flags you have set at a glance and re-runs the scan on every change, so each flag’s effect is visible the moment you toggle it. That is the fastest way to learn what a flag does: switch it, watch the match list change.
Reading matches, groups, and failures
Each result row shows the matched text, its start position, and its capture groups — the parts of the pattern inside parentheses, which are the values your code will actually consume. The classic use is extraction: with /request_id=(w+)/g against a log line, group 1 holds the id, and the tester shows both matches and their group values before you wire the pattern into a pipeline.
A pattern that matches nothing is a bug report, not a puzzle: the tool shows every match position so the failure is visible at a glance — the pattern matched a different place than you expected, or matched zero times because of a missing flag, a literal space, or an anchor bound to the wrong position. Malformed patterns return the platform’s own error message, which names the position of the problem, so fixing is a matter of reading the message rather than guessing.
Safety and the honest limits
Regular expressions can be pathological: certain patterns take exponential time on certain inputs — the classic "catastrophic backtracking" failure mode — and the tester caps match scanning so a hostile pattern cannot freeze the page. The cap is disclosed; extremely large texts or pathological patterns are truncated with a notice rather than run to completion.
Two things the tool deliberately does not do: it does not judge whether a pattern is efficient enough for production traffic (that is a load-testing question), and it does not claim your pattern is correct in the abstract — it shows what the engine does with your text, which is the only honest answer. And because patterns and pasted text often contain credentials and personal data, everything runs locally: nothing is uploaded, logged, or stored.
よくある質問
このツールが使う正規表現の方言は何ですか？
プラットフォームの ECMAScript エンジンです。JavaScript、Node.js、Deno、ブラウザが使うのと同じ方言です。PCRE（PHP、grep、regex101 のデフォルト）などの方言には、別の形式の後読みや別の構文の名前付きグループなど追加機能がありますが、それらは ECMAScript には含まれません。
パターンが何もマッチしないのはなぜですか？
よくある原因は、複数マッチを期待しているのに g フラグが付いていないこと、^ や $ のようなアンカーが m フラグなしでは行境界にマッチしないこと、意図せずリテラルの空白が入っていることです。このツールはすべてのマッチを位置付きで表示するため、失敗の原因はたいてい一目でわかります。
テキストはどこかに送信されますか？
いいえ。パターンとテキストはブラウザのタブ内だけで処理され、アップロード、ログ記録、保存は一切行われません。認証情報や個人データを含むことの多いログの抜粋や設定ファイルの扱いには重要です。
What does the u (unicode) flag actually change?
It switches the engine to strict Unicode mode: astral characters (emoji, rare scripts) are treated as single code points instead of surrogate halves, and some otherwise-lenient syntax becomes an error. Patterns that work without u can behave differently with it — the tester makes the difference visible as you toggle.
Can I use the same pattern in Python or Go?
Not safely — those languages use different engines (Python’s re, Go’s RE2). The fidelity promise here is specifically ECMAScript: what matches here matches in JavaScript, Node.js, Deno, and browsers. Porting a pattern to another language means re-testing it in that language’s tester.