snake_case Converter

Spaces, tabs, line breaks and hyphens all become word boundaries, and so does the point inside a camelCase or PascalCase name where a small letter meets a capital. Every word is then made lowercase and the words are joined with a single underscore. Runs of separators collapse, and any at the start or end are dropped, so you never get __ or a leading underscore.

Note the hyphen in E-mail: it is a separator like any other, so e and mail end up as two words.

The usual job is turning headers like Order Date or Customer E-mail Address into database column or dataframe names. Paste the headers one at a time. If you paste a column of headers, the line breaks are treated as separators and the whole list becomes one name: first name and last name on two lines give first_name_last_name.

A faster route for many headers is to put them on one line separated by a character the tool keeps, such as a comma, convert them together, and split the result back apart. Order Date,Ship Date gives order_date,ship_date.

Only separators are replaced. Brackets, symbols, apostrophes and dots stay in the output, and most of them aren't valid in an unquoted SQL column or a Python variable:

• order total ($) gives order_total_($)
• Is Active? gives is_active?
• user.email stays user.email

A name that starts with a digit is also a problem in both SQL and Python. 10 Tips for SEO becomes 10_tips_for_seo, which needs quoting in SQL and is a syntax error as a Python name. Strip punctuation from the headers first, and add a word in front of leading numbers, such as tip_10.

Splitting at capitals only works for the unaccented letters A to Z, and only where a lowercase letter comes first:

• iPhone case gives i_phone_case, which is probably not what you want.
• XMLHttpRequest gives xmlhttp_request, because a run of capitals is read as part of the next word.
• caféMenu gives cafémenu, because the é before the M isn't recognised as a lowercase letter.

Accented letters themselves are kept, so Über Cool Größe becomes über_cool_größe. Many databases accept that only when the name is quoted, so for column names it is safer to replace accented letters first.