Document format: inline formatting and formulas

#Inline formatting

Inline markup is recognised inside any text block (headings, paragraphs, blockquotes, list items). In a heading it follows headings.inlineMarks, on by default: the title's italic, bold, scripts, small capitals and links print as in a paragraph, and an italic run in an italic heading comes out upright. With it off the markers are dropped and the title prints in the heading's own style; configurations stored by postext 1.4 or earlier whose headings carry marks are read that way (see Configuration › Headings). A heading design prints {titleText} as plain text either way.

Markup	Syntax	Notes
Bold	bold or bold	Rendered with bodyText.boldFontWeight. An optional bodyText.boldColor overrides the default body color for bold spans.
Italic	italic or italic	Rendered with the italic variant of the current font family. An optional bodyText.italicColor overrides the default body color for italic spans. An underscore marks emphasis only at a word boundary, as in CommonMark: one between two letters or digits — snake_case_name, a URL's SR_AIR_EN.pdf — is text (also for bold); inside a word, use asterisks (unbelievable). Chinese, Japanese and Korean letters do not count here, since those scripts have no spaces: 中文_斜体_中文 and 中文__粗体__中文 are italic and bold. A __ that does not make bold never turns into italics: foo__bar__baz stays text.
Bold italic	both or both	Both flags combine.
Superscript	^text^	Set at 58% of the text size and raised a third of it — an exponent (10^-8^), an ion charge (Na^+^). The marked text starts and ends with a non-space character; a lone caret in prose stays literal, and so do the faces ^_^ and ^o^ (n.^o^ and 1^o^ are still superscripts).
Subscript	~text~	Same size, lowered 0.15 of the text size, so it stays within the descenders — a chemical index (H~2~O, pK~a~). A tilde with a digit on each side is a range and stays literal: 3~5 days, 需要3~5天. So is one between two Chinese words, which opens no subscript: 周一~周五, 北京~上海; a subscript still closes before a Chinese character (F~合~等于). Combines with bold and italic (H~2~O). A subscript and a superscript written together, with nothing between them, are stacked as in a formula: T~0~^2^ sets the 2 over the 0, whichever comes first, the subscript lowered 0.25 of the text size to clear the superscript, and the pair takes the width of the wider one. A space or a letter between them sets them one after the other; a word joiner (U+2060) does too, with nothing visible between them. A stacked pair never parts at a line break: a word too wide for the line breaks before the pair.
Small caps	:smallcaps[text]	Lowercase letters set as capitals at 70% of the text size — a character's name in a play, an acronym in running text. Takes other marks inside and around it; see Small capitals.
Orientation in vertical text	:tcy[12], :upright[GDP], :sideways[12]	In vertical text: one upright cell (tate-chu-yoko), each character upright in a cell of its own, or the whole run turned. No effect in horizontal text; see Orientation in vertical text.
Chinese marks	:dots[不可], :name[賈寶玉], :book[石頭記]	Emphasis dots (着重号), the proper-name line (专名号) and the book-title mark (书名号: 《》 or a wavy line, per cjk.bookTitleMark). The text stays as written; see Chinese marks, ruby and warichu.
Ruby	:ruby[紅樓]{rt="hóng lóu"} or {紅樓|hóng|lóu}	Pinyin or zhuyin readings over (or beside) the base characters. The compact form needs a Han, kana or bopomofo base, so {x|x>0} stays text.
Warichu	:warichu[note]{open="〔" close="〕"}	A note set in two half-size rows inside the line (双行夹注), broken across lines and pages.
Inline code	code	Backticks are stripped; the span is rendered as plain text, exactly as written: a , a chip, maths, a link or an emphasis marker inside it prints literally, which is how a text shows the syntax. It is set in the text's face, or, with codeStyle.inline, in the code face as one unit (see Configuration › Code listings). A listing goes in a code block.
Escape	, , , , ,	A backslash sets the marker character itself — the asterisk of a table footnote (), a literal caret, a dollar sign that opens no formula — instead of opening a span. Works in the body, in chips and in captions, cells and notes. (Up to postext 1.4 captions, cells, notes and chips printed the backslash of .)
No-break space	the character itself: U+00A0, U+202F, U+2007	Glues the words on either side, so a line never breaks between them: a number and its unit (37 °C), a page reference (p. 12), a group of thousands (225 000). The no-break space (U+00A0), the narrow no-break space (U+202F) and the figure space (U+2007) all glue — in the body, captions, cells, boxes and running heads — and each keeps its own width: justification stretches the word spaces only. Many faces have no glyph for the narrow no-break space or the figure space; they are then set as browsers set them, in the canvas, the HTML and the PDF alike: half a word space and the width of a digit. Practically every face has U+00A0, so it is the safe choice. A group glued this way that is wider than the whole line breaks at its last no-break space rather than inside a word. The word joiner (U+2060) glues without taking room. Type the character itself: an HTML entity such as prints as it is written. See Where a line never breaks.
Soft hyphen	the character itself: U+00AD	A place where the word may be divided, with a hyphen added when a line ends there: between two letters or digits only. A word that carries one gets no other break from the dictionary, so the soft hyphen is its one division; typed before the first letter of a word it divides nothing and keeps that occurrence whole. Next to a bracket, a quotation mark or punctuation it does nothing. See Where a line never breaks.
Link	text	The visible text is kept in the flow, set exactly as without the link. The URL makes the words a live link in the HTML and PDF output; see Links.
Image		Inline image markdown is removed from the text. Images must be declared on PostextContent.resources so the layout engine can place them according to resourcePlacement rules.
Chip	:chip[text]	A boxed run of text that wraps as one unit — a word bank, a key, a tag. Styled by chipStyles; see Inline chips.
Tab	:tab, :tab{at=120mm align=end leader="."}	Sends the words after it to the next tab stop: one of the stops of the paragraph's style, or the one its attributes set. A tab character typed in the text is a tab only where the style sets stops. See Tabs and tab stops.
Inline math	$…$	A LaTeX formula that flows with the surrounding text, e.g. $e^+1=0$. Typeset by MathJax and rendered as vector paths on every backend. Use \$ for a literal dollar sign. Scaling, the display ($$ … $$) form, and error handling are covered under Mathematical formulas.

#Links

A Markdown link, [text](url), keeps its text in the flow exactly as the text would be set without it: the link never changes line breaks, spacing or colour. It keeps its URL for the outputs that can follow it:

• HTML (renderToHtml): the linked words are wrapped in <a href="…" rel="noopener noreferrer">, which takes the text colour and has no underline. There is one anchor per run of linked words on a line.
• PDF (postext-pdf): each run of linked words on a line becomes a clickable URI link annotation. In a tagged PDF it is a Link element whose text is the linked words.
• Canvas: painted as plain text, since a canvas has no click surface.

Read the [configuration guide](https://postext.dev/en/docs/configuration "Configuration")
or write to [the team](mailto:team@example.com).

• The destination may carry a title, which is ignored ([text](https://example.com "Title")). It may also be wrapped in angle brackets, where spaces become %20 ([text](<https://example.com/a b>)).
• Only safe targets are kept: http:, https:, mailto:, tel:, ftp: and relative URLs (../guide, #top). Any other scheme (javascript:, data:, file:…) sets the text without a link. The PDF links absolute URLs only.
• The destination may contain parentheses as long as they balance, as in CommonMark: [photo](https://commons.wikimedia.org/wiki/File:Bike_(Unsplash).jpg) links the whole URL. A backslash escapes a character (\(, \_). When the parentheses do not balance, the destination ends at its first ) and the text is set without a link; encode a lone parenthesis as %28 or %29. A space also ends the destination unless it is wrapped in angle brackets.
• A word glued to the link text shares the link. In see [the site](https://example.com). the full stop is part of the clickable area.
• The link text may carry emphasis ([**bold** words](…)) and may sit inside it (**[words](…)**).
• Links work in paragraphs, list items, blockquotes, callouts, captions, notes and table cells, and in headings while headings.inlineMarks is on (the default). The running heads, the outline and the contents rows made from a heading keep the text only, and so does the text of a :chip[…].
• Reference-style links ([text][id]) and autolinks (<https://…>) are not recognised (see What is NOT supported).

In the VDT the target is on each linked segment, as VDTLineSegment.href. The parser keeps a span's links as ranges of its text, in InlineSpan.links, and never splits a span for a link, which is why a link cannot change the layout.

#Inline chips

:chip[text] sets text in a box — a rounded, tinted "chip" — that flows with the line: the words of a word bank or a classify exercise, keyboard keys, tags. :chip[text]{style="key"} picks a named style from chipStyles (see the configuration reference); without style, or with an id no style declares, the chip takes the first style (a built-in chip style when the configuration has none; the sandbox warns about an unknown id).

Classify: :chip[battery] :chip[cable] :chip[switch] :chip[bulb]
 
Press :chip[Ctrl]{style="key"} + :chip[C]{style="key"} to copy.

• One unit. A chip is never broken or hyphenated inside; the line breaks between chips, at the word spaces around them. A chip wider than the whole line overflows it rather than splitting.
• Width. Its advance is the text plus the horizontal padding and the outline on both sides. Justification stretches the word spaces only, never the inside of a chip. The style's gap is the least room kept between the box and a neighbouring word or chip across a space — a narrower space is topped up (not at a line edge, and not against glued punctuation such as :chip[a],).
• Height. The box is a band around the baseline, 0.8 em above and 0.25 em below at the chip's size, grown by the vertical padding and the outline. The vertical padding paints outside the line box and never changes the line height, so the baseline grid holds; a box taller than the line pitch can run into a chip on the line above or below, and the sandbox flags two chips on different lines that overlap ("Chips touch the next line") so the padding, outline or size can be reduced.
• Text. The chip text takes its own inline marks (:chip[**bold** word], :chip[x^2^]) and the emphasis around it (**:chip[a]**); the style may set its family, size, colour, bold and italic. Write \] for a literal bracket inside. References, swatches and math inside a chip stay literal, and \$ prints a dollar sign (up to postext 1.4 it printed the backslash too).
• Empty. A chip of spaces alone (:chip[ ], no-break spaces included) is an empty box — an answer blank, an end mark: as wide as its style's horizontal padding and outline, as tall as a chip with words. Size the blank with a style's paddingX (:chip[ ]{style="blank"}). With nothing at all between the brackets, :chip[] stays literal text.
• Where. Paragraphs, list items, blockquotes, callouts, table cells, captions and notes. Headings keep :chip[…] as literal text.
• Output. Canvas, HTML and PDF paint the box and set the words as real text: selectable in the HTML, extractable and in reading order in the PDF (in a tagged PDF the box is a layout artifact and the words belong to the paragraph).

#Small capitals

:smallcaps[text] sets text in small capitals — the names of the characters in a play, an acronym in running text, the first words of a chapter. Lowercase letters are set as capitals at 70% of the text size, while capitals, digits and punctuation keep the full size, so :smallcaps[Hamlet] prints a full-size H followed by a small-capital AMLET.

Enter :smallcaps[Hamlet] and :smallcaps[Horatio], reading.
 
The :smallcaps[unesco] report and the :smallcaps[who] guidelines.

• Synthesised. The small capitals are drawn from the face's own capitals, the same way on the canvas, in the HTML and in the PDF, so the text measures and breaks identically everywhere. A face's real small caps (the OpenType smcp feature) are not used; for them, set the text in a small-caps family (a family whose name ends in "SC", for example) through a paragraph style.
• Marks. The text takes its own inline marks (:smallcaps[**Ophelia**]) and the emphasis around it (*:smallcaps[Act I]*); write \] for a literal bracket inside. A reference or a chip inside it is set in small capitals too, and a reference stays one link in the HTML and the PDF; a formula is not.
• Breaking. Words wrap and hyphenate as usual; hyphenation reads each word in its own case.
• Where. Paragraphs, list items, blockquotes, callouts, table cells, captions and notes, and headings while headings.inlineMarks is on (the default); with it off, the markup is removed from a heading and the text prints as written.
• Whole paragraphs. A paragraph style or a callout body with smallCaps: true sets all its text this way (see Paragraph styles).
• Text. The canvas, the HTML and the PDF carry the letters as painted: copying :smallcaps[Hamlet] gives "HAMLET".

#Tabs and tab stops

A tab sends the words after it to a stop on the line: the price of a dish flush with the end of the measure behind a row of dots, the actor after the part in a cast list, the marks of an exam question at the margin. Write :tab where the tab goes; the spaces around it are taken into it. The stops are set in the configuration: bodyText.tabStops for every paragraph, list item and quotation, a paragraph style's tabStops for the paragraphs of a :::paragraphs group, and a callout style's body.tabStops inside its boxes (see Configuration › Tab stops). A menu whose prices end at the end of the measure, behind dots:

paragraphStyles: [
  { id: 'menu', firstLineIndent: { value: 0, unit: 'em' }, tabStops: [{ position: 'end', align: 'end', leader: '. ' }] },
],

:::paragraphs{style="menu"}
Onion soup :tab 8.50
 
Grilled sea bream with fennel and lemon :tab 21.00
:::

:tab{…} sets a stop for that one tab, with no style: a form's blank to the end of the line, an amount on its decimal point.

Name and surname :tab{at=end leader=rule}
 
Total due :tab{at=80% align=decimal} 1,234.50

• Where a tab goes. To the first stop past the text before it, among the stops after the one the line's previous tab took, so the second tab of a line usually takes the second stop. Past the last stop, tabInterval sets more stops every interval from the start of the measure; without it the tab is a word space.
• Tab characters. A tab typed in the text (U+0009) is a tab only in a paragraph whose style sets tabStops or tabInterval. Everywhere else it is the word space it always was, so a document written with tabs for spaces keeps its layout. Tab characters in a row go on that many stops. A tab at the start of a source line is trimmed with the line's other leading blanks: write :tab to start a line at a stop.
• What stays text. A :tab followed by a letter (3:table), and a :tab in inline code, in a formula, in a link's address or in a directive's attributes.
• Where. Paragraphs, list items, quotations, boxes and :::paragraphs groups. Headings, captions, table cells and the text of designs (openers, running heads) have no tab stops.
• Text past the stop. When the text before a tab has already passed every stop left on its line, an end, center or decimal stop takes the last word before the tab down to the next line with the text at the stop, and the text before it is set narrower, as a contents row wraps its title to keep the page number. A start stop breaks the line before the tab, and the text after it starts the next line at its stop. A tab that opens its line and still finds no stop takes no room.
• Leaders. A leader fills the room before its stop and ends flush with the end of that room, so the dots of several lines line up whatever the text before them. gap keeps it clear of the text on both sides; there is no gap before it when the tab opens the line, and none after it when nothing follows. A leader is set with three repeats or more; a room for one or two is left empty, as in the contents.
• Line breaking. A paragraph that holds a tab is set line by line, never with Knuth–Plass. In justified text only the word spaces after a line's last tab take the slack, and the paragraph's last line is not stretched (see Justification › Tabbed lines). A centred or right-aligned paragraph sets its tabbed lines from the start side, since the stops are positions on the measure. Right to left, the stops are measured from the right and an end stop sits at the left (see Arabic layout). In vertical text a tab is a word space, and a :tab there raises the content warning tabInVerticalText.
• Text and outputs. The plain text has a tab character (\t) at each tab; the leaders are not text, and copies, searches and the source map leave them out. The canvas, the HTML viewer and the fixed-layout EPUB paint them (hidden from screen readers in the HTML); a tagged PDF marks them as artifacts, so its text reads Onion soup 8.50. The reflowable EPUB cannot keep the stops exactly: each line of a tabbed paragraph becomes a row of the parts between its tabs, where an end, center or decimal stop, or one with a leader, pushes the next part to the end of the row (its leader a dotted or solid border), and a start stop keeps the part before it at its printed width. A Word export writes :tab as a Word tab, and the Word import reads a tab in body text back as :tab with the stop Word set for it (see Sandbox › Word documents).

#Orientation in vertical text

In vertical text (layout.writingMode: 'vertical-rl') Chinese characters stand upright, Latin words and long numbers are turned sideways, and a number of at most two digits stands upright in one cell, unless it sits in a Latin sentence, whose words it follows (see Configuration › Numbers in vertical text). Three marks set a run apart by hand:

第:tcy[120]回，:upright[GDP]增長:sideways[12]倍。

• :tcy[…] (tate-chu-yoko, 縱中橫) sets its text side by side in one upright cell of one em, squeezed across when it is wider: :tcy[120], :tcy[3.0], :tcy[A+]. Up to about four characters read well.
• :upright[…] stands each character upright in a cell of its own, Latin letters centred in it: an acronym read letter by letter down the column. A line never breaks inside it.
• :sideways[…] turns the whole run with the line, Chinese characters too: a two-digit number the author wants on its side.
• The text stays in the plain text; the marks take other inline marks inside (:tcy[**12**]), a link (:sideways[[iPhone](https://…)]) and each other, the innermost winning (:tcy[:upright[AB]] stands A and B upright), and write \] for a literal bracket. In horizontal text they change nothing.
• A reference, a note marker, a formula, a chip or a swatch inside a mark keeps its own setting: :sideways[iPhone[^1]] turns iPhone and leaves the note number as every note number is set; a reference's number stands in one cell by cjk.uprightDigits like any short number.
• A design text element with inlineMarks: true takes them too when it is set vertically (see Configuration › Vertical text elements).
• For breaking and justifying, a :tcy cell and each :upright character count as Chinese characters, and no Han–Latin space is set next to them.

#Chinese marks, ruby and warichu

Chinese and Japanese editions mark text between the lines rather than with italics or underlining, gloss characters with their readings, and set commentary inside the line. Seven directives do it. Chinese layout and Japanese layout explain the conventions behind them. They keep their text: the characters between the brackets stay in the paragraph (search, the contents, index anchors and copied text read them as written), and the directives take the other inline marks inside and around them, nested ones included. Write \] for a literal bracket.

此事:dots[不可]輕忽。:name[賈寶玉]與:name[林黛玉]讀:book[西廂記]。
 
{滿紙|mǎn|zhǐ}荒唐言，:ruby[一把]{rt="yì bǎ"}辛酸淚！:ruby[都]{rt="ㄉㄡ"}云作者痴。
 
寶玉:warichu[甲戌側批：此是第一首標題詩。]{open="〔" close="〕"}道：……
 
{東京|とう|きょう}の:ruby[紫陽花]{rt="あじさい"}は*とんと*:sideline[見事]だ。:book[こころ]を読む。
 
:kunten[學]{okuri="ビテ"}而:kunten[時]{okuri="ニ"}:kunten[習]{kaeri="二" okuri="フ"}:kunten[之]{kaeri="一" okuri="ヲ"}。

• :dots[text] sets an emphasis dot (着重号) under each character in horizontal text, right of it in vertical text, and none on punctuation or spaces. style="dot|circle|sesame" (default dot), fill="open" for an outline, pos="over|under" for the other side. Markdown emphasis *…* on Chinese or Japanese characters does the same under cjk.emphasis: 'dots', the default in a Chinese or Japanese document; Latin letters in the same emphasis stay italic. Attributes left out take cjk.emphasisMark: a dot under the text in Chinese, a sesame over it (right of it in vertical text) in Japanese.
• :name[text] draws the proper-name line (专名号) under the text (left of it in vertical text), and :book[text] the book-title mark (书名号): 《》 around the title (〈〉 for a title inside another), the wavy line of classical and Taiwan editions, or nothing, as cjk.bookTitleMark says (by default brackets on the mainland and in Japan, the wavy line in Taiwan and Hong Kong). In a Japanese document the brackets are 『』, and 「」 for a title inside another (cjk.bookTitleBrackets). One source serves a modern mainland edition and a classical one. Two names or titles side by side keep their lines apart.
• :ruby[base]{rt="…"} sets readings over the base (pinyin), or right of each character (zhuyin, the default for bopomofo readings). With as many readings as characters, split on spaces or |, each character gets its own and a line may break between them (mono ruby); group, or a count that does not match, sets one reading centred over the whole base, which never breaks. pos="over|under|right" picks the side. The compact form {紅樓|hóng|lóu} (one reading per |) or {紅樓|hónglóu} (one reading for the whole base) reads the same, only when the base holds a Han, kana or bopomofo character, outside maths, code and attributes, so {x|x>0} in Latin prose stays text. A backslash before the brace or the bar keeps a Chinese one as text: \{紅|hóng} and {紅\|hóng} print {紅|hóng}. A base that opens or closes a line aligns to that edge together with its reading. mode=mono|group|jukugo chooses how the readings go over the base (mode=group is the same as group), and align=center|jis|start how a reading shorter than its base is spaced. In a Japanese document the per-character form of a word of two or more characters ({東京|とう|きょう}, rt="とう|きょう") is jukugo ruby: each character keeps its reading, which may run onto the next character of the word, the line may break between them, and the readings are spaced and run onto kana as cjk.ruby.align and cjk.ruby.overhang say; one reading for the whole base ({紫陽花|あじさい}, Aozora's 《》) is group ruby; mode=mono keeps a reading per character with no running onto the neighbours.
• :sideline[text] draws a side line (傍線) beside the text, punctuation and spaces included: under it in horizontal text and right of it in vertical text. style="solid|double|wavy|dotted" (default solid), pos="over|under" (in vertical text over is right, under is left; position= works too). It works in any script, an underlined English phrase included.
• :warichu[note] sets a note (双行夹注) in two rows at half the text size inside the line, the upper row read first (the right one in vertical text). A long note fills what is left of the line and goes on with the next line, column or page, with at least two of its characters on each side of the break (cjk.warichu.minChars). open and close wrap it in brackets at the text size; cjk.warichu sets the size, colour and default brackets (（） in a Japanese document). A note in Latin script breaks between its words.
• :kunten[字]{kaeri="…" okuri="…" tate} sets the kanbun marks of the last character between the brackets: kaeri the return mark (レ, 一 二 三, 上 中 下, 甲 乙 丙, 天 地 人, and 一レ 上レ 甲レ 天レ; the code points ㆑ ㆒ … read the same), okuri the okurigana (Aozora's parentheses （ヲ） are dropped), and the flag tate the tatesen that joins it to the next character. The text is not reordered: the marks are set at half size beside it (cjk.kunten), and copied text reads 學ビテ而時ニ習フ之ヲ, with the okurigana and without the return marks. A ruby inside works for the characters read twice: :kunten[:ruby[未]{rt="ザル" pos=under}]{kaeri="レ" okuri="ダ"}.

The line pitch never changes: marks and readings live in the leading, and the build warns when a paragraph's line gap is too narrow for them (cjkMarksExceedLeading, rubyExceedsLeading, kuntenExceedsLeading), and when the marks under one line and the readings over the next do not fit in the gap they share, in one paragraph or across two. Give annotated paragraphs a paragraph style with more leading. Marks, readings and notes are drawn in paragraphs, headings, list items, blockquotes and boxes, and in captions, table cells and notes too, whose leading is checked the same way; in running heads and designs the text prints without them. Under cjk.bookTitleMark: 'brackets' the 《》 of a title are punctuation of the text: captions, cells, notes, contents rows, bookmarks and index entries keep them. See Configuration › Marks, ruby and warichu.

#Text direction in the markup

A document runs in the direction its locale implies (direction in the configuration: right to left for Arabic, Persian, Urdu or Hebrew). Inside it, the markup sets the direction of a block or of a run of text, for an English quotation in an Arabic book or an Arabic one in an English book.

• A heading or a container: {dir=ltr} or {dir=rtl} in its attributes, # Introduction {dir=ltr}, :::paragraphs{dir=ltr}, :::callout{dir=rtl}. Every block inside a container takes the direction, down to a nested container or heading that sets another. A plain paragraph, a list or a quote has no attributes of its own: wrap it in :::paragraphs{dir=…}. Other values are ignored. A container's lang (:::paragraphs{dir=ltr lang=en}) names the language of the blocks inside it the same way: their ordered lists number in that language's digits.
• A run of text: :ltr[…] and :rtl[…] isolate their text (the LRI…PDI and RLI…PDI of the Unicode Bidirectional Algorithm): it is ordered inside, and the paragraph around it sees one neutral character. {lang=…} tags the run's language in the HTML and the PDF. A note marker, a :ref or a formula inside is part of the isolate, and isolates nest.

:::paragraphs{dir=ltr}
The opening of the *Nights* in Lane's translation.
:::
 
ترجمها :ltr[Edward William Lane]{lang=en} سنة ١٨٣٩.

A block set against the document's direction keeps its own start side: its indent, list markers and the flush end of its last line move to the side its text starts on. Use an isolate when a Latin title ends in a neutral character (a full stop, a bracket) that would otherwise join the Arabic around it. The Unicode controls themselves (U+2066–2069, U+202A–202E, U+200E, U+200F, U+061C) are honoured too. In the Sandbox editor each line runs in the direction of its first letter. See Arabic layout.

#Mathematical formulas

Math support is a first-class part of the document format. Postext parses $…$ for inline formulas and $$…$$ for display (block) formulas, and renders them via MathJax in SVG mode. The same vector paths drive all three backends, so the canvas preview, the HTML export, and the PDF output are pixel-for-pixel consistent — and the PDF stays fully vectorial regardless of the zoom level.

• Inline: $…$. Recognised inside any text block (paragraph, heading, blockquote, list item). Contributes a single atomic, non-breaking box to the line; Knuth-Plass treats it like a word that must not be split. If the formula's natural height would break the line box, it is scaled down uniformly so the baseline grid is preserved — very tall expressions belong in display mode.
• Display: $$…$$. Either on its own line ($$\int_0^1 x^2\,dx$$) or fenced across multiple lines with $$ markers on their own lines. Rendered centred on the column and snapped to the baseline grid with configurable top and bottom margins (math.marginTop, math.marginBottom) — exactly the same correction mechanism headings use, so the paragraph after the formula lands back on the grid.
• Text around a display: a display formula on its own line interrupts a paragraph even with no blank line above it. The text before it is a paragraph that leads into it; text written right under the closing $$, with no blank line, continues the interrupted paragraph and is set with no first-line indent, as TeX sets the "where …" after a display. A formula set off from the text above it by a blank line (or following a list, a quotation or a heading) interrupts nothing: the text after it is a new paragraph, indented as usual, blank line or not (math.indentAfterDisplay: false sets it flush too). Only a whole display interrupts a paragraph — one formula on the line, or a $$ fence closed before the next blank line; a line such as $$a$$ and $$b$$ stays text. math.keepWithLeadIn keeps a formula in the column of the line that introduces it. (Up to postext 1.4 a $$…$$ line with no blank line above it was read as part of the paragraph and printed as it was written.)
• Equation numbers: \label{eq:x} in a display formula numbers it, in reading order: (1), (2)… A formula with a number spans the column (or the callout's inner width), the equation centred and its number flush right on its line. In an align, gather, alignat, flalign or eqnarray each row that carries a \label gets its own number, unless the row says \nonumber or \notag. A formula without a \label is not numbered, as $$…$$ never was. \tag{…} still numbers a formula with the text you give it (\tag*{…} without the parentheses) and is not counted; a \label beside it names that text. The format, chapter or section numbers (2.3) and restarts are set in math.equationNumbering. Put the \label at the top level of the formula or of its row: not inside aligned, split or cases.
• References to equations: \eqref{eq:x} in the text prints the number as the formula does, (3), and \ref{eq:x} the bare number, 3; both are links, like every cross-reference. Eq.~\eqref{eq:x} keeps the ~ as LaTeX does: a no-break space. :ref{id="eq:x"} and @eq:x print (3) too. Inside a formula, \eqref and \ref print the number (not a link). A label nothing sets prints ? and is listed in the Checks panel.
• Escaping: \$ is a literal dollar sign. Unmatched $ or $$ delimiters produce an unclosedMath entry in the Sandbox's Checks panel with a click-to-focus source anchor.
• Captions, notes and table cells set inline maths too (since postext 1.19), in the line with the text and with the cell's alignment, in the snippet's size. There a $ opens a formula only by Pandoc's rule: no space after the opening $, no space before the closing one and no digit right after it, so the prices of $5 to $10 stay text; \$ prints a dollar sign, so one escaped string reads the same in the text and in a table. A chip's label takes no maths. Up to postext 1.18 a $ there always printed as it was.
• Errors: TeX source that MathJax rejects (undefined macros, syntax errors) surfaces as an invalidMath warning. The formula is replaced by a small red placeholder so the layout geometry stays valid.
• Configuration: the math section of the config exposes enabled, fontSizeScale (relative to the size of the surrounding text: at 1.0 one em of the formula is that size — the body size for display formulas and for inline maths in body text; since postext 1.5, which sets formulas about 13% smaller than 1.4 did, while bundles and Sandbox books saved by 1.4 keep their size — see Formula size), color (inherits the body colour when unset), and the display margins.
• The engine: MathJax loads on demand. The Sandbox and the layout worker start it themselves; in your own code, await initMathEngine() before buildDocument, or every formula is laid out as a grey placeholder box (see Starting the math engine).

The Euler identity $e^{i\pi}+1=0$ links the five fundamental constants.
 
$$
\int_0^{\infty} e^{-x^2}\,dx = \frac{\sqrt{\pi}}{2}
$$

A formula in the middle of a sentence, and the sentence going on after it with no indent:

For a pendulum of length $L$ the period is
$$T_0 = 2\pi\sqrt{L/g}$$
where $g$ is the acceleration of free fall.

Numbered equations and the references to them:

The energy is conserved,
$$
\begin{align}
E &= T + V \label{eq:energy} \\
  &= \tfrac12 m v^2 + m g h \nonumber
\end{align}
$$
and Eq.~\eqref{eq:energy} holds at every instant.

Theorems, lemmas and proofs. A callout style with numbering counts its boxes, as LaTeX's amsthm counts theorem environments: each box opens with its label and number in bold, "Theorem 2.", the fence's title after the number in parentheses, "Theorem 2 (Bradley–Terry).", and a reference to its {#id} prints "Theorem 2" (\ref{thm:x} or style=number the bare 2). Styles that name one counter share it, so a lemma after a theorem is Lemma 3. A proof style's endMark ends its boxes with ∎ or □, flush right on the last line. See Numbered statements and proofs for the styles.

:::callout{type="theorem" #thm:main title="Bradley–Terry"}
Every reward class holds one reparameterised reward.
:::
 
:::callout{type="proof"}
Apply the projection of \eqref{eq:projection} to any member of the class.
:::
 
By :ref{id="thm:main"}, the objective is unchanged.