quote-punctuation
The opt-in quote-punctuation rule moves a comma or full stop from after a closing quotation mark to before it, for the locale whose convention puts it there, and declines wherever the placement depends on meaning.
quote-punctuation moves a comma or a full stop from after a closing quotation mark to before it: “hello”, becomes “hello,”. It applies only to locales whose cited convention puts these two marks inside the quotation mark. In the shipped locale data that is en-US (the Chicago Manual of Style); every other locale is set to none, and the rule does nothing there.
This rule is off by default. It changes where a character the author typed sits, not just which glyph it is. Turn it on with rules: { "quote-punctuation": true }.
Why it is opt-in #
Where a comma or full stop goes relative to a closing quotation mark is a convention, and in most locales it depends on meaning: a mark that belongs to the quoted words goes inside, one that belongs to the surrounding sentence goes outside. "Stop." and "most popular", look alike on the surface, and a rule that scans code points cannot tell them apart. So the rule is defined only where the convention is positional, a comma or full stop goes inside whatever the sense, and the locale says none everywhere else.
Even for en-US the move is not always right. In technical text a quoted literal absorbs the punctuation: set it to "strict". becomes set it to “strict.”. Code spans and code blocks are skipped in the html and markdown modes, so a literal written as code is safe. That is why the rule is an opt-in.
How it works #
The rule looks at each closing primary quotation mark of the locale (for en-US, U+201D) and moves the comma or full stop that follows it to before it, when all of these hold:
- A letter comes before the quotation mark. A digit, another punctuation mark, a bracket or another quotation mark does not qualify, so
“Route 66”.,“Stop!”.,“e.g.”,and the nested“He said ‘hi’”.are left alone. - The mark is a comma or a full stop. Semicolons, colons, question marks and exclamation marks are never moved.
- A space, a line break or the end of the text comes next. A letter, a bracket, another dot or a quotation mark does not qualify, so
“file”.txt,(“hi”.)and“wait”…are left alone.
Nothing is inserted or deleted, so a call keeps its length, and the output is a fixed point: run again, nothing more moves.
Modes #
A boundary between the quotation mark and the punctuation, or immediately before or after the pair, blocks the move. In html and markdown that means an element or an emphasis delimiter: <em>“hello”</em>, and *“hello”*, keep the comma where it was. In text mode a range given to protect counts as a boundary in the same way. The end of a document is not a boundary.
Locale data #
The rule reads quotes.punctuationPlacement, one of inside or none.
| Value | Locales |
|---|---|
inside | en-US |
none | every other locale |
none means the rule does nothing. For en-GB and de-DE the cited convention places these marks by meaning (the University of Oxford Style Guide (2014) and the Amtliches Regelwerk of 2024), so there is nothing a scan can apply. For the others no convention has been verified.
Options #
Enable the rule explicitly:
transform(text, { locale: 'en-US', rules: { 'quote-punctuation': true } })false is the default. There is no other setting.
Edge cases #
- Only the closing primary quotation mark is read. The secondary close, which in
en-USis also the apostrophe, is never a trigger:the dogs’, bowlsis untouched. - A duplicate is not removed:
“Stop!”.keeps both marks. - An unpaired closing quotation mark that the author typed in curly form moves its comma too. The rule trusts the glyph, not the pairing.
- A quotation after a number (
“Route 66”.) and a nested quotation keep the punctuation outside.
Related #
- quotes: the rule that produces the quotation marks this rule reads.
- Rule order
- Options
Normative text: spec/rules/quote-punctuation.md.