Elysia Tools
Navigation
Workflow Playbook
Find hidden or confusable Unicode characters, inspect emoji and domain representations, normalize only when justified, and verify the cleaned copy before it reaches search, URLs, forms, or parsers.
Hubs
Copy the exact value into a preserved reference and a separate working copy. Record where it came from and what failed: visual equality, copy and paste, search matching, a URL, a form rule, or a parser. Also note whether emoji, joiners, variation selectors, width forms, and non-Latin scripts are valid for that field.
Use hex-unicode-converter to expose a code-point-oriented representation. Use emoji-extractor to list emoji and positions, special-character-validator for punctuation and symbols, and text-fake-detector for heuristic clues about invisible characters, unusual spaces, and common homoglyphs. These are diagnostic signals. They do not establish that a multilingual character is wrong.
If the value is a domain, compare the human-readable label with the ASCII result from punycode-encoder, and use punycode-decoder to inspect an existing ASCII label. If the issue is width, choose exactly the direction required by the consumer. Neither branch is a general security or application-compatibility proof.
Apply zero-width-remover or text-emoji-remover only to the working copy after deciding what may be deleted. Preserve whitespace when it carries layout meaning and be careful with joiners that form emoji sequences. Never turn a detector's suggestion into an unreviewed replacement.
Repeat the diagnostic tools and compare before-and-after representations, counts, and intended emoji. Then test the candidate in the real search box, URL parser, form, or downstream parser. Accept it only when the observed behavior is corrected and every remaining character difference is explained.
Workflow playbook
Keep the raw string unchanged. Convert it to visible hex or Unicode escape tokens, extract emoji and their positions, list ordinary special characters, and review suspicious invisible, confusable, or unusual-space clues. Treat every report as evidence to investigate, not as permission to replace a character.
For a domain only, decode Punycode labels to inspect what a reader sees, then encode the intended international label back to ASCII and compare label by label. Keep this branch separate from ordinary text cleanup; it confirms representation, not ownership, DNS resolution, certificate coverage, or phishing safety.
Choose the width direction that matches the receiving system and run it on a copy. Compare the converted text with the original using the saved code-point evidence. Do not call a width conversion general Unicode normalization, and do not normalize a meaningful CJK or display value merely to make it look simpler.
Start with Hex/Unicode Converter, Emoji Extractor, Special Character Validator, and Check If Text Is Fake. They expose representations, positions, emoji, and heuristic clues without changing the source. Use a remover only after the suspicious characters and the intended policy are recorded.
Use Punycode Decoder to inspect an ASCII IDN label and Punycode Encoder to produce the ASCII representation of an international label. Do not use those tools as domain registration, DNS, certificate, or phishing-safety validation.
Choose Fullwidth to Halfwidth for a consumer that expects narrow ASCII-oriented forms, or Halfwidth to Fullwidth for a defined CJK presentation rule. Do not run both directions without a stated target because a reversible-looking conversion can still change matching or display behavior.
Zero-width Character Remover and Text Emoji Remover alter the text. Preserve the original, select options deliberately, and manually review whether a joiner, emoji, control character, whitespace, or symbol carries meaning before accepting the result.
Only after diagnosis, use Zero Width Character Remover for confirmed unwanted hidden or control characters, or Text Emoji Remover when the field policy explicitly excludes emoji. Preserve whitespace when layout matters, avoid broad symbol removal unless approved, and record what disappeared. Never use a character-spoofing generator as a repair step.
Run the diagnostic views again on the candidate, compare the before-and-after tokens and emoji list, and recheck confusable clues. For a domain, verify the intended Unicode-to-ASCII round trip. Finally paste, search, parse, or submit the candidate through the actual consumer and reject it if an unexplained difference remains.