日本語English

Unicode Inspector — Character & Invisible Text Analyzer

Paste text to break it into characters with U+ codes, byte sequences and categories. Hidden zero-width spaces, BOMs, bidi controls and look-alike letters are detected and can be removed in one click.

On-device・No upload

Your files and input are processed only inside your browser on this device and are never uploaded to our servers.

Text to inspect
Characters (visible): 0
Enter text to see the analysis here

How to use

  1. Paste characters or text into “Text to inspect” (or try the sample).

  2. In “Characters”, read each character’s code point (U+XXXX), UTF-8 / UTF-16 bytes, general category, script, block and HTML reference. Emoji sequences are grouped as one visible character.

  3. Open “Invisible” to see zero-width spaces, bidi controls and other hidden characters, highlighted in place; tick the kinds to fix and press the fix button.

  4. Use “Normalization” to see exactly what NFC, NFD, NFKC and NFKD would change, and apply a form if you need it.

  5. Check “Look-alikes” and “Japanese / Shift_JIS” for spoofing letters and characters that break legacy Japanese systems. Every fix can be undone.

What it does

Per-character detail table

Code point, UTF-8 and UTF-16 bytes, general category (all 30), script, Unicode block and HTML reference. Emoji ZWJ sequences, flags and combined characters are grouped with Intl.Segmenter and their code points listed.

Invisible character detection & cleanup

15 kinds by severity — zero-width characters, BOM, unusual spaces, soft hyphens, bidi controls, tag and control characters and more. Legitimate uses such as ZWJ inside emoji or ZWNJ in Persian are skipped, and odd spaces become normal spaces instead of vanishing.

Trojan Source warning

Counts lines with unterminated bidirectional overrides or isolates (RLO, LRI…) — the technique behind CVE-2021-42574, where code displays differently from how it executes.

Normalization side by side

Lists what NFC, NFD, NFKC and NFKD change (e.g. “é” composed vs decomposed, “①” → “1”, full-width “Ａ” → “A”) with the code points before and after.

Homoglyph / mixed-script detection

Flags words that mix Latin with Cyrillic, Greek or Armenian look-alikes, or are written entirely in look-alike letters, and can replace them with Latin letters.

Japanese legacy-encoding check

Using the browser’s Shift_JIS table, finds Windows-only characters (①, ㈱, ㍻) and characters Shift_JIS cannot store (emoji, rare kanji, wave dash), with safe replacements.

Use cases

  • Find out why a copied password, ID or search term does not match (zero-width space, NBSP)
  • Check a suspicious link for Cyrillic letters disguised as Latin ones
  • Audit source code or commands for Trojan Source bidirectional characters
  • See how many code points an emoji really uses and why lengths differ between systems
  • Clean text before importing it into a database or CSV

Capability matrix

FeatureSupportNotes
Code point, UTF-8 / UTF-16, category, scriptSupported—
Unicode block namesPartialAll BMP blocks and the main supplementary ones (emoji, CJK extensions, kana, math…). Some historic-script blocks show as “not in table”
Unicode character namesPartialOnly for invisible, space and other noteworthy characters; full name data is not bundled
Invisible character detection & removalSupported—
NFC / NFD / NFKC / NFKD comparisonSupported—
Homoglyph detectionPartialLatin vs Cyrillic / Greek / Armenian only, not the full Unicode confusables list
Japanese platform-dependent & non-Shift_JIS charactersSupportedBased on Windows-31J
Meanings or pronunciations of charactersNot supported—
Offline useSupportedWorks without a connection once loaded

Limits & notes

  • Input up to 1 million characters. The table shows 300 characters at a time; the visualized view covers the first 20,000.
  • Categories and scripts follow the Unicode version your browser supports; very new emoji may show as unassigned in older browsers.
  • The Japanese check uses Windows-31J (CP932); Apple-specific extensions are not evaluated.
  • Replacement suggestions exist for common characters only.

Privacy & safety

Text is analyzed only inside your browser and never sent to or stored on a server, so you can inspect passwords or internal documents safely. Closing the page discards everything.

FAQ

What is a zero-width space and why is it a problem?
U+200B is a space with no width, so you cannot see it. It sneaks in when copying from web pages or PDFs and makes otherwise identical passwords, IDs or search terms fail to match. The “Invisible” tab finds and removes it.
How can I tell if a link uses look-alike letters?
Paste it here and open “Look-alikes”. Words that mix Latin with Cyrillic, Greek or Armenian letters are flagged, with the disguised letters highlighted and the Latin equivalent shown.
What is Trojan Source?
An attack that hides bidirectional control characters (such as U+202E RIGHT-TO-LEFT OVERRIDE) in source code so it displays differently from how the compiler reads it. The tool warns when such controls are left unterminated on a line.
What is the difference between NFC and NFKC?
NFC only unifies characters that are canonically identical (such as “é” written as one or two code points). NFKC also folds compatibility characters like ① → 1 or full-width Ａ → A. Use NFKC for search keys and IDs, NFC for storing text.
Why does one emoji count as several characters elsewhere?
Many emoji are sequences: 👨‍👩‍👧 is three people joined by zero-width joiners — five code points and eight UTF-16 units. This tool shows it as one visible character and lists its parts.
Is my text uploaded?
No. All analysis and fixes run in your browser.

Related tools

Date & Time Studio — Unix Timestamp & Time Zone Converter

Paste a Unix timestamp or almost any date string — the unit and format are detected automatically and shown in 15 formats and multiple time zones. DST gaps, month ends and Japanese era boundaries are handled precisely.

Encode / Decode Studio — Base64, URL, JWT & more

Paste %E3%81%82, an eyJ… token or any encoded string — the format is detected and decoded instantly. Errors say exactly which character is wrong and why, and nothing leaves your browser.

Markdown Studio — Live Editor & Preview

A split-view Markdown editor with instant GFM preview, a table formatter that aligns pipes, an automatic table of contents and HTML export. Output is always sanitized and external images load only when you allow them.

Text Studio — Character Counter & Text Cleaner

Paste text to see characters, words, lines and bytes instantly, then clean it up in one click — change case, remove duplicate or blank lines, sort naturally, convert line breaks. Every action reports what changed and can be undone.

IT Calculator (Storage, Transfer, Uptime, RAID)

Convert storage units (SI GB vs IEC GiB), estimate transfer time from file size and line speed, turn an uptime percentage into allowed downtime, and size RAID arrays, all with the formula shown. Type values with units, like “1.5GB” or “100Mbps”.

MIME & Content-Type Studio

Search extensions and MIME types both ways, get problems in a Content-Type header pointed out, and compare a dropped file’s extension, browser file.type and real content. Server config drafts are generated too.

Last updated: