Case Converter

How to use the case converter

Paste text, choose one of ten styles, and the converted version appears immediately. What separates this from a find and replace is the first step it takes: before applying any style it splits the input into words, and it does that by reading the casing already present rather than looking only for spaces.

So XMLHttpRequest is understood as three words and becomes xml-http-request in kebab-case, and parse_json_file becomes ParseJsonFile in PascalCase. A naive converter treats each of those as a single unbroken token and returns xmlhttprequest, which is the single most common way a case converter quietly gets it wrong.

That split makes the programmer styles reversible. Run someVariableName through snake_case and you get some_variable_name; run that back through camelCase and you get someVariableName again, character for character.

The same round trip works through kebab-case, which is what makes the tool usable for renaming across languages that disagree on convention: Python and Ruby want snake_case, JavaScript wants camelCase, Go and C sharp want PascalCase, CSS and URLs want kebab-case, and environment variables want CONSTANT_CASE. Splitting once and rejoining differently is the only way that stays consistent, and it is why the same word list underlies all nine mechanical styles.

Why is Title Case the one that needs judgement?

Title Case is the one that needs judgement, and it is the reason this page is longer than a list of ten buttons. Capitalising every word is wrong: Of Mice And Men is not how the book is printed, and no house style writes it that way.

What the styles agree on is that short function words stay lowercase inside a title, and that the first and last word are always capitalised whatever they are. This converter holds a list of 22 such words, chosen as the intersection of Associated Press and Chicago Manual of Style practice, so the output is defensible under either. What Are You Waiting For keeps its capital on the closing For for exactly that reason.

The edges are worth knowing before you trust the output on anything published. Sentence case behaves differently again: it lowercases everything and then capitalises the first letter after each terminal mark, which is right for prose and wrong for anything containing a proper noun, so hello there. how are you? fine! becomes Hello there. How are you? Fine! with the names left flat. Read the output rather than assuming it, particularly on titles containing brand names, acronyms, or a colon.

The formula, worked line by line

Nine of the ten styles are the same two operations in different order: split the input into a list of lowercase words, then rejoin them with a chosen separator and a chosen capitalisation. All the intelligence is in the split, because the split is what lets an identifier that already carries casing be understood as several words rather than one.

The split runs three passes. It inserts a boundary between a lowercase letter or digit and a following capital, which is what breaks someVariableName apart. It inserts a second boundary inside a run of capitals followed by a capital and a lowercase letter, which is what stops XMLHttpRequest collapsing into one token and turns it into XML, Http, Request. Then it splits on whitespace, underscores, hyphens, and full stops, drops the empty pieces, and lowercases everything that survives.

words        = split on caseChange, whitespace, _ , - and . then lowercase
camelCase    = words[0] + capitalise(rest), joined with nothing
PascalCase   = capitalise(every word), joined with nothing
snake_case   = words joined with _        kebab-case = words joined with -
CONSTANT     = snake_case, uppercased
Title Case   = capitalise(word) unless it is a minor word away from either end

The minor-word list holds 22 entries: a, an, and, as, at, but, by, for, if, in, nor, of, on, or, per, so, the, to, up, via, vs, and yet. It is deliberately the intersection of the two dominant American style guides rather than either one in full.

The Chicago Manual of Style lowercases all articles, coordinating conjunctions, and prepositions regardless of length, so Chicago writes Gone with the Wind and Far from the Madding Crowd. Associated Press style capitalises any word of four letters or more, so AP writes Gone With the Wind and Far From the Madding Crowd.

Taking only the words both guides lowercase means the output is acceptable under either, at the cost of leaving longer prepositions capitalised. Verified outputs from this converter include the following.

• Of Mice and Men: and stays down, and the opening Of takes its capital as the first word.
• A Tale of Two Cities: of stays down in the middle of the title.
• To Kill a Mockingbird: to opens the title, so it is capitalised despite being on the list.
• The Old Man and the Sea: both and and the mid-title the stay down.
• East of Eden: one minor word between two capitalised ones.
• Up in the Air: the opening Up keeps its capital despite being on the list.

Two more limits are worth stating plainly. Full stops are treated as word separators, which is correct for file.name.parts and wrong for a sentence, so Title Case applied to prose eats the punctuation.

And the alternating style is the only one that ignores the word split entirely, walking the raw string character by character and flipping on index parity, which means spaces count towards the alternation and the pattern will look offset if you were expecting it to restart at each word. Use Title Case on titles, Sentence case on sentences, and the programmer styles on identifiers.

Questions people ask