Check and convert text encoding

How to use

1. Choose the garbled CSV or text file.
2. Check the previews under “Possible encodings” and select the one that isn’t garbled.
3. Choose the encoding to save as (UTF-8 with BOM, UTF-8 or Shift_JIS) and the line breaks, then press “Convert and save”.

The original file is not changed. The saved file gets the encoding added to its name, for example “original_utf8-bom.csv”.

How the possible encodings are ordered

The order is a guess. The whole file is read in each encoding, and the readings are sorted by:

• the encoding indicated by a BOM at the start of the file
• fewer unreadable bytes (spots that become the replacement character �)
• fewer control characters and other characters that don’t appear in normal text
• fewer unusual sequences for that encoding, such as half-width katakana or runs of symbols and accented letters

Short files, or files that are mostly basic Latin letters and digits, may not be told apart. Always check the preview.

Shift_JIS and CP932

In browsers, “Shift_JIS” covers the same characters as CP932 (Windows-31J), the version used on Windows, as defined by the WHATWG Encoding Standard. NEC special characters (such as ① and ㈱) and IBM extended characters (such as kanji variants used in personal names) are read correctly.

Saving as Shift_JIS follows the same rules:

• If even one character has no Shift_JIS equivalent (emoji, rare kanji and so on), the file is not saved. Nothing is replaced with “?”; the positions are shown instead.
• ¥ is saved as 0x5C, ‾ as 0x7E, and − (minus sign) as － (full-width hyphen-minus). The number of such characters is shown before saving.

Line breaks

“Keep” leaves line breaks unchanged. CRLF or LF changes every line break to that type. In files ending in .csv or .tsv, line breaks inside quoted values (line breaks within a cell) are kept as they are.

What it can’t do

• Characters that have already become the replacement character (�) or “?” can’t be turned back into the original characters.
• It can’t always fix garbled text. If only part of a file was written in another encoding, every reading leaves some part unreadable.
• You can only save as UTF-8 (with or without BOM) or Shift_JIS.
• The limit is 20 MB.