Turns fancy and lookalike characters back into plain a-z: 𝖘𝖚𝖒, 𝕤𝓾𝓂 and ⓢⓤⓜ all become sum. A Cyrillic а like the one in “pаypal.com” is swapped for a Latin a without a trace, so the output is safe to search, sort or paste into code, but it looks exactly like what you read and won't tell you a swap happened. Curly quotes, long dashes and invisible zero-width spaces come out plain too.
Spam and phishing mail fakes text for two reasons. Letters from another alphabet slip past filters that look for “PayPal” or “password”, and a lookalike domain passes a quick glance. Paste a subject written in 𝐟𝐚𝐧𝐜𝐲 letters here and it comes out as plain words you can read and search; a letter that drops out had no Latin twin at all.
A lookalike domain is the case this tool can't show you: the fake а becomes a real a, and the output reads “paypal.com” like the original. Paste the domain into the Punycode Converter instead. A domain with a foreign letter in it comes out as xn--, and “pаypal.com” becomes xn--pypal-4ve.com.
It won't repair mojibake. Text like “Café” was garbled by the wrong encoding, not faked, and the fixer only makes it worse: “CafA”. Open the file again as UTF-8 instead.
Most fake text is built from a handful of Unicode blocks:
They exist because each script needs its own letters even where the shapes match, which is what a homoglyph is: two characters that look the same and mean different things. Every block is listed at Unicode Explorer.
Text garbled by being read in the wrong encoding: "Café" for "Café", or "it’s" for "it's". The Japanese name means roughly "character transformation". It isn't fake text, so this fixer can't undo it and only makes it worse. The cure is to open the file again as UTF-8, or to re-import it with the right encoding.
A trick that swaps a letter for one that looks the same but isn't: a Cyrillic а (U+0430) in "pаypal.com" instead of the Latin a (U+0061). Your eye reads PayPal, your browser goes somewhere else. This fixer won't catch it: it turns the fake а into a plain a, so the output looks the same as the original. Paste the domain into a Punycode converter instead, where a foreign letter shows up as xn--: "pаypal.com" becomes xn--pypal-4ve.com.
Read fancy-letter spam as plain words, and check a lookalike domain separately.
Garbled text like "Café" comes from reading a file in the wrong encoding, so the fix is to read it again correctly.
Usually an encoding mismatch: the message was written in UTF-8 and read as an older Western encoding, or a header saying which encoding it is got lost on the way. Accented letters, curly quotes and emoji suffer first, because they take more than one byte. Mail that looks deliberately odd, 𝐟𝐫𝐞𝐞 or ⓦⓘⓝ, is a different thing: that's fake text, and this tool reads it.
Find fancy-text letters and characters that have no plain equivalent.
UTF-8 can write every Unicode character, using one to four bytes each. Latin-1 (ISO-8859-1) has 256 characters, one byte each, enough for Western European languages. Read UTF-8 as Latin-1 and every two-byte letter turns into two wrong ones: "café" becomes "café". That's mojibake, and the fix is to read the file with the right encoding.
Either the font has no picture for those characters, which gives boxes (□), or something on the way couldn't store them and wrote ? instead, which is a database or a file saved in an old encoding. Boxes are a display problem and the text is still there. Question marks usually mean the characters are gone and have to be typed again.
Strip invisible characters that break search, comparisons and copy-paste.