Docs Reference

Rules

The engine has 33 rules, 22 on by default (the rest are off until named in enable). typographic runs them in the order below, each on the output of the one before. An id can be switched off with disable (see api.md); a group name switches off every rule that starts with it (space is every space.* rule, dash every dash.*, symbol every symbol.*).

This file is generated by bun run docs:rules from RULES in src/core/rules.ts and from the examples in scripts/gen-rules-doc.ts. Do not edit it by hand: src/docs.test.ts fails when it is out of date.

Each example runs the engine with every rule except the one shown switched off, so it shows what that rule does and nothing else. In a full run the other rules also act (an English " is turned into a quote by quote before anything else sees it). The text of each example is plain; a no-break space is shown as ~ (the character U+00A0 in the real output; an ordinary space stays a space). Text in a code span, a web address and an email are never changed by any rule.

Order

#	Rule	Category	Default
1	`quote`	tradition	on
2	`apostrophe`	tradition	on
3	`ellipsis`	tradition	on
4	`space.before-punctuation`	tradition	off
5	`space.after-comma`	tradition	off
6	`punctuation.double`	tradition	off
7	`arrow`	symbols	on
8	`symbol.implies`	math	off
9	`symbol.legal`	symbols	on
10	`dash.dialogue`	tradition	on
11	`phone.bind`	data	off
12	`number.group`	data	off
13	`dash.range`	data	on
14	`dash.sentence`	tradition	on
15	`symbol.plusminus`	math	off
16	`symbol.minus`	math	on
17	`symbol.multiply`	math	on
18	`symbol.fraction`	math	off
19	`space.collapse`	tradition	on
20	`space.title`	tradition	on
21	`space.initial`	tradition	on
22	`space.abbreviation`	tradition	on
23	`space.number`	data	on
24	`space.percent`	data	on
25	`symbol.degree`	data	off
26	`space.section`	symbols	on
27	`space.association`	tradition	on
28	`space.copyright`	symbols	on
29	`space.operator`	symbols	on
30	`space.short-word`	tradition	off
31	`space.orphan`	tradition	on
32	`space.arrow`	symbols	on
33	`space.trailing`	tradition	off

`quote`

Quotation marks by language, register and nesting level, from any source style

Category: tradition.

Why: sources.md, disputed.

Characters it can insert, in any language: « » „ “ ‚ ‘ ’ ” › ‹ ~. A font that lacks one of them is handled by avoid (see api.md).

Before: He said "stone" and left.
After: He said «stone» and left.

Display register, the default.

Before: He said "we said 'no' twice" and left.
After: He said «we said „no“ twice» and left.

Nesting level two. Source marks may be straight, curly or guillemets.

en, literary

Before: He said "we said 'no' twice" and left.
After: He said ‘we said “no” twice’ and left.

British English in the literary register.

Before: Er sagte "Sie rief 'Haus' laut" und ging.
After: Er sagte »Sie rief ›Haus‹ laut« und ging.

`apostrophe`

A straight ' in a word or before a year becomes ’

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: ’. A font that lacks one of them is handled by avoid (see api.md).

Before: It's Maria's turn, don't wait.
After: It’s Maria’s turn, don’t wait.

Inside a word.

Before: The '90s were loud.
After: The ’90s were loud.

Before a year.

`ellipsis`

… becomes …

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: …. A font that lacks one of them is handled by avoid (see api.md).

Before: Wait... what?
After: Wait… what?

en, avoid …

Before: Wait... what?
After: Wait... what?

With the ellipsis in avoid (a font that cannot draw it), the text stays as typed.

`space.before-punctuation`

Removes a space before , . ; : ! ? ) (French keeps the space before : ; ! ?)

Category: tradition.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Before: Hello , world . Is it ready ? (yes )
After: Hello, world. Is it ready? (yes)

Off by default.

Before: Oui : non , bon !
After: Oui : non, bon !

French keeps its space before : ; ! ?, the rest is tidied.

`space.after-comma`

A comma followed directly by a letter gets a space (a,b becomes a, b)

Category: tradition.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Before: apples,pears,plums
After: apples, pears, plums

Digits are left alone: 1,5 stays.

Before: Hallo,Welt und 3,14
After: Hallo, Welt und 3,14

`punctuation.double`

A run of ! or ? becomes one mark (Wow!!! becomes Wow!); ?! and !? stay

Category: tradition.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Before: Wow!!! Really??
After: Wow! Really?

Before: What?!! No!!!? Stop!!!
After: What?!! No!? Stop!

?! and !? stay as a pair; a run beside one collapses.

`arrow`

→ and ← become → and ←

Category: symbols.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: → ←. A font that lacks one of them is handled by avoid (see api.md).

Before: Read more ->
After: Read more →

Before: <- Back
After: ← Back

`symbol.implies`

=> becomes ⇒

Category: math.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ⇒. A font that lacks one of them is handled by avoid (see api.md).

Before: A => B
After: A ⇒ B

Before: if x=>y then z
After: if x⇒y then z

Not =>>, ==>, =>= or <=>.

`symbol.legal`

Category: symbols.

Why: sources.md, verified.

Characters it can insert, in any language: © ® ™. A font that lacks one of them is handled by avoid (see api.md).

Before: (c) 2026 Studio (R) (TM)
After: © 2026 Studio ® ™

`dash.dialogue`

A paragraph opening with a dash is a line of dialogue, in languages that write dialogue that way: their dash and a no-break space (Ukrainian — Привіт)

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: — ~ –. A font that lacks one of them is handled by avoid (see api.md).

Before: - Привіт, - сказав він.
After: — Привіт, - сказав він.

Only the dash that opens the paragraph; the second one belongs to dash.sentence. Both registers.

Before: - ¿Esperamos? - preguntó.
After: —¿Esperamos? - preguntó.

Spanish writes the raya attached to the turn.

Before: - Odotetaanko? kysyi hän.
After: – Odotetaanko? kysyi hän.

An en dash where the language uses one. A paragraph whose later line also opens with a dash is a list and is left alone.

`phone.bind`

An international phone number (+ or 00, at least seven digits) is kept whole: no-break spaces between its groups and non-breaking hyphens inside it, with the digits as typed

Category: data.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ~ ‑. A font that lacks one of them is handled by avoid (see api.md).

Before: Call +380 44 123 45 67 or +1 (555) 123-4567
After: Call +380 44 123 45 67 or +1 (555) 123‑4567

Off by default. The digits are not reformatted; the hyphen becomes U+2011, which does not break.

Before: Not 555-1234, not 2026-10-20, not DE89 3704 0044 0532 0130 00
After: Not 555-1234, not 2026-10-20, not DE89 3704 0044 0532 0130 00

A local number, a date and an IBAN are left alone: the number must open with + or 00 and hold seven to fifteen digits.

`number.group`

Digits of an integer of five or more are grouped in threes with the language separator (1,000,000; 1 000 000)

Category: data.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: , ~. A font that lacks one of them is handled by avoid (see api.md).

Before: We sold 1234567 copies in 2026.
After: We sold 1,234,567 copies in 2026.

A run of five or more digits; years are left.

Before: Wir verkauften 1234567 Exemplare, Pi ist 3,14159265.
After: Wir verkauften 1 234 567 Exemplare, Pi ist 3,14159265.

A no-break space; the fraction is left alone.

Before: Продано 12345 примірників, код v1000000.
After: Продано 12 345 примірників, код v1000000.

A name glued to digits is left.

`dash.range`

A hyphen between two numbers (21–26, 10:00–12:00) becomes an unspaced en dash

Category: data.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: –. A font that lacks one of them is handled by avoid (see api.md).

Before: Open 21-26 September, 10:00-12:00.
After: Open 21–26 September, 10:00–12:00.

Before: Pages 5 - 9
After: Pages 5–9

Spaced, too.

`dash.sentence`

A spaced hyphen, dash or double hyphen becomes a spaced dash (em; en in German and literary British English), no-break before it

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: — ~ –. A font that lacks one of them is handled by avoid (see api.md).

Before: Wait - what happened -- really?
After: Wait — what happened — really?

An em dash in English.

Before: Warte - was ist passiert?
After: Warte – was ist passiert?

A spaced en dash in German.

en, literary

Before: Wait - what happened?
After: Wait – what happened?

A spaced en dash in British English, literary register.

`symbol.plusminus`

+- before a digit becomes ± (+-5 becomes ±5)

Category: math.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ±. A font that lacks one of them is handled by avoid (see api.md).

Before: Tolerance +-5 mm
After: Tolerance ±5 mm

Before: (+-0.2) and +-10%
After: (±0.2) and ±10%

Not a+-b or 1+-5.

`symbol.minus`

A hyphen before a digit (−5) becomes a minus sign (−5)

Category: math.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: −. A font that lacks one of them is handled by avoid (see api.md).

Before: It was -5 degrees (-3 inside).
After: It was −5 degrees (−3 inside).

`symbol.multiply`

The letter x between numbers (1920×1080, 3 × 4) becomes ×

Category: math.

Why: sources.md, verified.

Characters it can insert, in any language: ×. A font that lacks one of them is handled by avoid (see api.md).

Before: A 1920x1080 screen, a 3 x 4 grid.
After: A 1920×1080 screen, a 3 × 4 grid.

`symbol.fraction`

1/2, 1/4 and 3/4 become ½ ¼ ¾, as whole numbers only (not 11/2, 1/20 or a date)

Category: math.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ½ ¼ ¾. A font that lacks one of them is handled by avoid (see api.md).

Before: Add 1/2 cup and 3/4 of the milk, then 1/4.
After: Add ½ cup and ¾ of the milk, then ¼.

Only whole numbers: 11/2, 1/20, a1/2 and a date such as 1/2/2026 stay. A short word after the fraction gets the no-break space.

Before: Take 2 1/2 hours
After: Take 2 ½ hours

A mixed number keeps its integer: 2 ½.

`space.collapse`

A run of spaces becomes one

Category: tradition.

Why: sources.md, verified.

Before: Two spaces, four.
After: Two spaces, four.

`space.title`

No-break space after Prof., Dr., Mr. …

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Prof. Dr. Oksana opens.
After: Prof. Dr. Oksana opens.

`space.initial`

No-break space after an initial (K. Słoboda)

Category: tradition.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Written by K. Słoboda.
After: Written by K. Słoboda.

`space.abbreviation`

Parts of an abbreviation stay together (z. B.), and Nr., ul., S. stay with what follows

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: z. B. Nr. 5 auf S. 12
After: z. B. Nr. 5 auf S. 12

Before: вул. Хрещатик, т. д.
After: вул. Хрещатик, т. д.

Before: See Fig. 2 and p. 14.
After: See Fig. 2 and p. 14.

`space.number`

No-break space between a number and a short word or unit after it, up to nine letters (21 September, 5 %); money stays whole: the sign or code before the amount (USD 5 000, € 49.99) and the digit groups of an amount written in threes (1 000 грн)

Category: data.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: On 21 September, 15 people paid 5 %.
After: On 21 September, 15 people paid 5 %.

`space.percent`

German, French, Czech, Spanish, Italian, Portuguese, Irish, Maltese, Bulgarian, Croatian, Slovenian, Lithuanian, Latvian, Estonian, Danish, Finnish and Greek: a no-break space between a number and % (5 %); German, French, Czech, Spanish, Italian, Bulgarian, Croatian, Slovenian, Danish, Finnish and Greek: also before € after the number (49 €)

Category: data.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: 5% and 49€
After: 5 % and 49 €

The rule does nothing in other languages.

Before: 5% and 49€
After: 5 % and 49 €

`symbol.degree`

20 C, 20°C and 20 °C become 20 °C, with a no-break space before the unit

Category: data.

Why: sources.md, verified.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ° ~. A font that lacks one of them is handled by avoid (see api.md).

Before: It was 20°C, then 68 F and 20 C.
After: It was 20 °C, then 68 °F and 20 °C.

Every spelling becomes number, no-break space, °C or °F.

Before: Vitamin C 5 C and 5 °C
After: Vitamin C 5 C and 5 °C

A bare C or F after a label or a single letter is not a temperature.

`space.section`

No-break space after § and ¶, doubled or not, before a number

Category: symbols.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: See § 5 and ¶¶ 3, 4
After: See § 5 and ¶¶ 3, 4

`space.association`

No-break space before e.V.

Category: tradition.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Kulturverein e.V.
After: Kulturverein e.V.

`space.copyright`

Category: symbols.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

`space.operator`

No-break spaces around & + = and ’n’ standing alone between two words (Tom & Jerry, rock ’n’ roll)

Category: symbols.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Tom & Jerry, rock + roll
After: Tom & Jerry, rock + roll

Before: rock ’n’ roll
After: rock ’n’ roll

The ’n’ of rock ’n’ roll is an “and” with its letters dropped, so it is bound like one; quote sets the two marks curly before this rule runs.

`space.short-word`

English, Ukrainian, Polish, Czech and Slovak: a short word of two or three letters stays with the word after it (to~the shop), for titles and headings; never more than two words in a row

Category: tradition.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Welcome to the city of light
After: Welcome to the city of light

For titles and headings; off by default. Never more than two words in a row, so a chain cannot grow into a line.

Before: Zapraszamy do Gdańska
After: Zapraszamy do Gdańska

Before: Ласкаво просимо до міста
After: Ласкаво просимо до міста

`space.orphan`

No-break space after a one-letter word the language must not strand at a line end

Category: tradition.

Why: sources.md, verified.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: A cat and I
After: A cat and I

Before: Він і я в домі
After: Він і я в домі

Before: Idę w las i z nim
After: Idę w las i z nim

`space.arrow`

No-break space before → and after ←

Category: symbols.

Why: sources.md, general-knowledge.

Characters it can insert, in any language: ~. A font that lacks one of them is handled by avoid (see api.md).

Before: Read more →
After: Read more →

Before: ← Back
After: ← Back

`space.trailing`

Spaces and tabs at the end of a line are removed

Category: tradition.

Why: sources.md, general-knowledge.

Off by default: switch it on with enable (--enable on the command line).

Before: First line Second line Third
After: First line Second line Third

Spaces and tabs at the end of a line; the lines stay.

Before: One trailing space
After: One trailing space

Something wrong or unclear here? Report an issue.

Rules

Categories

Order

quote

apostrophe

ellipsis

space.before-punctuation

space.after-comma

punctuation.double

arrow

symbol.implies

symbol.legal

dash.dialogue

phone.bind

number.group

dash.range

dash.sentence

symbol.plusminus

symbol.minus

symbol.multiply

symbol.fraction

space.collapse

space.title

space.initial

space.abbreviation

space.number

space.percent

symbol.degree

space.section

space.association

space.copyright

space.operator

space.short-word

space.orphan

space.arrow

space.trailing

`quote`

`apostrophe`

`ellipsis`

`space.before-punctuation`

`space.after-comma`

`punctuation.double`

`arrow`

`symbol.implies`

`symbol.legal`

`dash.dialogue`

`phone.bind`

`number.group`

`dash.range`

`dash.sentence`

`symbol.plusminus`

`symbol.minus`

`symbol.multiply`

`symbol.fraction`

`space.collapse`

`space.title`

`space.initial`

`space.abbreviation`

`space.number`

`space.percent`

`symbol.degree`

`space.section`

`space.association`

`space.copyright`

`space.operator`

`space.short-word`

`space.orphan`

`space.arrow`

`space.trailing`