@ai-matrx/print 0.6.2 → 0.7.1

package/CHANGELOG.md

@@ -1,5 +1,67 @@
 # Changelog
 
+## 0.7.1
+
+- **Fixed (PDF):** a heading followed by a table taller than a page vanished
+  together with the whole table — keep-with-next glued them into one
+  unbreakable block and pdfmake drops a block that cannot fit a page. A
+  heading now travels only with blocks that always fit (a short paragraph,
+  a figure, a short list); tables are never unbreakable (rows never split,
+  the header row repeats, a caption above keeps with its table); code blocks
+  stay whole up to 30 lines. Found printing a real note.
+- **Fixed (PDF):** emoji and pictographs printed as empty boxes (the embedded
+  fonts have no glyphs for them). They are left out of the PDF and announced
+  once with the `glyph-substituted` notice; Word, EPUB and HTML keep them.
+- Consumer action: none — refresh to latest.
+
+## 0.7.0
+
+- **Added: `@ai-matrx/print/document` — print-grade documents from markdown.**
+  One parsed tree (CommonMark + GFM via micromark/mdast) renders to **PDF**
+  (real text PDF via pdfmake), **DOCX** (docx), **EPUB 3** and a standalone
+  **HTML** file, so every format carries the same layout, numbering and
+  references. `exportDocument(markdown, "pdf" | "docx" | "epub" | "html" |
+  "markdown")` is the one call.
+- **Document settings in YAML frontmatter** (pandoc names, any spelling):
+  page size (letter, legal, tabloid, executive, A3–A5, B5), margins (`1in`,
+  `2.5cm`, pandoc `geometry`), orientation, cover page (title, subtitle,
+  author, date, logo), running header/footer templates with `{page}` /
+  `{pages}` / `{title}` / `{date}` ("Page X of Y"), a different first page,
+  1–3 columns, section numbering, table of contents (depth, title), citation
+  style, references, font size, hyphenation.
+- **Section breaks** — `<!-- section landscape columns=2 -->` starts a new
+  page and switches orientation/columns from there on (Word's "section break,
+  next page"). Added to `/directives` (`parseSectionBreakLine`,
+  `isBreakLine`, `SECTION_BREAK_MARKDOWN`); `splitAtPageBreaks` and both
+  markdown skins treat a section break as a page break.
+- **Table of contents** — automatic, with real page numbers in PDF, a
+  pre-filled Word TOC field (page numbers from the PDF layout; Word refreshes
+  on open), the EPUB navigation document, and links in HTML. Placement:
+  `toc: true` or a `[[toc]]` line.
+- **Captions, numbering, cross-references** (pandoc-crossref grammar):
+  `![Caption](url){#fig:id}`, `Table: Caption {#tbl:id}`, `## Title
+  {#sec:id}`, and `@fig:id` / `@tbl:id` / `@sec:id` → "Figure 2", "Table 1",
+  "Section 3" (linked; Word gets SEQ fields + bookmarks).
+- **Citations and bibliography** — pandoc citation syntax (`[@key, p. 4]`,
+  `[@a; @b]`, `[-@key]`, bare `@key`) over CSL-JSON references in the
+  frontmatter, formatted APA 7, MLA 9 or Chicago author-date, with the
+  reference list appended (or placed at `<!-- bibliography -->`).
+  Footnotes print as endnotes.
+- **Print quality** — keep-with-next headings (a heading is never the last
+  line on a page), short paragraphs never split, table rows never split and
+  header rows repeat, code/figures/tables kept whole when they fit, widow and
+  orphan control, hyphenation (HTML/EPUB CSS, Word auto-hyphenation), always
+  light.
+- **Added: `DocumentPrintPreview`** in `/react` — the real PDF pages, a web
+  view with a clickable TOC, Print, and one-click PDF / Word / EPUB / HTML /
+  Markdown downloads; unresolved references and missing images are listed
+  with their fix.
+- Nothing fails silently: unresolved cross-references/citations, unusable
+  settings, dropped raw HTML and unavailable images are reported as notices
+  with a remedy.
+- Consumer action: none required. New dependencies (`pdfmake`, `docx`,
+  `jszip`, `yaml`, mdast/micromark) load only when a document is exported.
+
 ## 0.6.2
 
 - **Fixed:** a page-break directive indented four or more columns is code (as

package/README.md

@@ -195,3 +195,41 @@ drawer are bundled dependencies.
 ## License
 
 MIT
+
+### `@ai-matrx/print/document` — print-grade documents (PDF, Word, EPUB, HTML)
+
+```ts
+import { exportDocument, downloadDocumentExport } from "@ai-matrx/print/document";
+
+const exp = await exportDocument(markdown, "docx"); // "pdf" | "docx" | "epub" | "html" | "markdown"
+downloadDocumentExport(exp);                         // browser: saves "<title>.docx"
+```
+
+Settings ride in the markdown's frontmatter:
+
+```yaml
+---
+title: Quarterly Review
+subtitle: Operations
+author: Jane Park
+date: today
+cover: true
+toc: true                      # or { depth: 2, title: Contents }
+page: { size: a4, orientation: portrait, margins: 1in }
+header: "{title} | | {date}"   # left | center | right
+footer: "Page {page} of {pages}"
+columns: 1
+numberSections: true
+csl: apa                       # apa | mla | chicago
+references:
+  - { id: smith2020, title: Paper Cuts, author: "Smith, Jane", issued: 2020, publisher: Acme }
+---
+```
+
+In the body: `<!-- pagebreak -->`, `<!-- section landscape columns=2 -->`,
+`[[toc]]`, `![Caption](url){#fig:chart}`, `Table: Caption {#tbl:sales}`,
+`## Methods {#sec:methods}`, `see @fig:chart`, `[@smith2020, p. 4]`.
+All four formats come from one parsed tree (`parseDocument`), so they agree.
+`DocumentPrintPreview` (in `/react`) shows the real PDF pages with Print and
+every download.
+

package/dist/directives.cjs

@@ -20,10 +20,17 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 // src/directives.ts
 var directives_exports = {};
 __export(directives_exports, {
+  BIBLIOGRAPHY_MARKDOWN: () => BIBLIOGRAPHY_MARKDOWN,
   PAGE_BREAK_CLASS: () => PAGE_BREAK_CLASS,
   PAGE_BREAK_LABEL: () => PAGE_BREAK_LABEL,
   PAGE_BREAK_MARKDOWN: () => PAGE_BREAK_MARKDOWN,
+  SECTION_BREAK_MARKDOWN: () => SECTION_BREAK_MARKDOWN,
+  TOC_MARKDOWN: () => TOC_MARKDOWN,
+  isBibliographyLine: () => isBibliographyLine,
+  isBreakLine: () => isBreakLine,
   isPageBreakLine: () => isPageBreakLine,
+  isTocLine: () => isTocLine,
+  parseSectionBreakLine: () => parseSectionBreakLine,
   splitAtPageBreaks: () => splitAtPageBreaks
 });
 module.exports = __toCommonJS(directives_exports);
@@ -52,6 +59,36 @@ function isPageBreakLine(line) {
   if (!div) return false;
   return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? "");
 }
+var SECTION_BREAK_MARKDOWN = "<!-- section -->";
+var SECTION_FORM = /^<!--\s*section(?:[\s_-]?break)?\b\s*:?\s*([^>]*?)\s*-->$/i;
+function parseSectionBreakLine(line) {
+  if (indentColumns(line) >= 4) return null;
+  const m = SECTION_FORM.exec(line.trim());
+  if (!m) return null;
+  const words = (m[1] ?? "").toLowerCase();
+  const out = {};
+  if (/\blandscape\b/.test(words)) out.orientation = "landscape";
+  else if (/\bportrait\b/.test(words)) out.orientation = "portrait";
+  const cols = /\b(?:columns?|cols?)\s*[=:]\s*(\d)\b/.exec(words) ?? /\b(\d)\s*[-\s]?columns?\b/.exec(words);
+  if (cols) {
+    const n = Math.min(3, Math.max(1, Number(cols[1])));
+    out.columns = n;
+  }
+  return out;
+}
+var TOC_MARKDOWN = "[[toc]]";
+var BIBLIOGRAPHY_MARKDOWN = "<!-- bibliography -->";
+var TOC_FORM = /^(?:\[\[\s*toc\s*\]\]|\[toc\]|\$\{toc\}|<!--\s*toc\s*-->|\\tableofcontents)$/i;
+var BIBLIOGRAPHY_FORM = /^(?:\[\[\s*(?:bibliography|references)\s*\]\]|<!--\s*(?:bibliography|references)\s*-->|\\printbibliography|\\bibliography\{[^}]*\})$/i;
+function isTocLine(line) {
+  return indentColumns(line) < 4 && TOC_FORM.test(line.trim());
+}
+function isBibliographyLine(line) {
+  return indentColumns(line) < 4 && BIBLIOGRAPHY_FORM.test(line.trim());
+}
+function isBreakLine(line) {
+  return isPageBreakLine(line) || parseSectionBreakLine(line) !== null;
+}
 function splitAtPageBreaks(markdown) {
   const lines = markdown.split(/\r?\n/);
   const pieces = [];
@@ -63,7 +100,7 @@ function splitAtPageBreaks(markdown) {
       const marker = fenceMatch[1];
       if (fence === null) fence = marker;
       else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;
-    } else if (fence === null && isPageBreakLine(line)) {
+    } else if (fence === null && isBreakLine(line)) {
       if (current.join("\n").trim()) pieces.push(current.join("\n"));
       current = [];
       continue;

package/dist/directives.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/directives.ts"],"sourcesContent":["/**\n * @ai-matrx/print/directives — the ONE definition of the print directives\n * markdown can carry, shared by every surface that reads markdown.\n *\n * Why a separate, zero-dependency entry: the directive syntax has three\n * readers — this package's two converters (print window / PDF / HTML page),\n * the chat renderer's remark plugin (the on-screen preview), and any editor\n * that inserts one. If each re-derived the regex they would drift the first\n * time someone added an alias. `/markdown` carries ~20KB of stylesheet, so the\n * chat bundle imports the syntax from here instead and pays nothing.\n *\n * ── Page break ────────────────────────────────────────────────────────────\n * A line that is exactly one of these (case-insensitive, surrounding\n * whitespace ignored) forces the next content onto a new printed page:\n *\n *   <!-- pagebreak -->          ← canonical; invisible in any other renderer\n *   <!-- page-break -->   <!-- page break -->   <!-- newpage -->\n *   \\pagebreak   \\newpage       ← LaTeX / pandoc habit\n *   <div style=\"page-break-after: always\"></div>   ← Typora / VS Code habit\n *   (also `page-break-before`, `break-after: page`, `break-before: page`,\n *    and `<div class=\"page-break\"></div>` / `class=\"pagebreak\"`)\n *\n * On screen every surface draws the same idea: a dashed rule labelled\n * \"Page break\" with generous padding. In print it is a zero-height element\n * with `break-after: page`.\n */\n\n/** The form editors insert and docs recommend. */\nexport const PAGE_BREAK_MARKDOWN = \"<!-- pagebreak -->\";\n\n/** The class every surface puts on a rendered page break. */\nexport const PAGE_BREAK_CLASS = \"matrx-page-break\";\n\n/** Screen label for a rendered page break. */\nexport const PAGE_BREAK_LABEL = \"Page break\";\n\nconst COMMENT_FORM = /^<!--\\s*(?:page[\\s_-]?break|new[\\s_-]?page)\\s*-->$/i;\nconst LATEX_FORM = /^\\\\(?:pagebreak|newpage|clearpage)\\s*(?:\\{\\s*\\})?$/i;\nconst DIV_FORM = /^<div\\b([^>]*)>\\s*<\\/div>$|^<div\\b([^>]*)\\/>$/i;\nconst DIV_BREAK_ATTR =\n  /(?:page-break-(?:after|before)\\s*:\\s*always|break-(?:after|before)\\s*:\\s*page|class\\s*=\\s*[\"'](?:[^\"']*\\s)?page-?break(?:\\s[^\"']*)?[\"'])/i;\n\n/** Columns of leading indentation (tab = 4, NBSP counts as a space). */\nfunction indentColumns(line: string): number {\n  let cols = 0;\n  for (const ch of line) {\n    if (ch === \" \" || ch === \"\\u00A0\") cols += 1;\n    else if (ch === \"\\t\") cols += 4;\n    else break;\n  }\n  return cols;\n}\n\n/**\n * Is this single line a page-break directive? Indented four or more columns\n * it is not — in markdown that is a code block, and every surface must agree.\n */\nexport function isPageBreakLine(line: string): boolean {\n  if (indentColumns(line) >= 4) return false;\n  const t = line.trim();\n  if (t.length < 8 || t.length > 200) return false;\n  if (COMMENT_FORM.test(t) || LATEX_FORM.test(t)) return true;\n  const div = DIV_FORM.exec(t);\n  if (!div) return false;\n  return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? \"\");\n}\n\n/**\n * Split markdown into the pieces that belong on separate pages. Fence-aware:\n * a directive inside a ``` / ~~~ code block is code, not a break. Always\n * returns at least one piece; empty pieces (two breaks in a row, a break at\n * the very start or end) are kept out so no blank page is produced.\n */\nexport function splitAtPageBreaks(markdown: string): string[] {\n  const lines = markdown.split(/\\r?\\n/);\n  const pieces: string[] = [];\n  let current: string[] = [];\n  let fence: string | null = null;\n\n  for (const line of lines) {\n    // Any indent — the converters accept list-nested fences, so must this.\n    const fenceMatch = /^\\s*(`{3,}|~{3,})/.exec(line);\n    if (fenceMatch) {\n      const marker = fenceMatch[1] as string;\n      if (fence === null) fence = marker;\n      else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;\n    } else if (fence === null && isPageBreakLine(line)) {\n      if (current.join(\"\\n\").trim()) pieces.push(current.join(\"\\n\"));\n      current = [];\n      continue;\n    }\n    current.push(line);\n  }\n  if (current.join(\"\\n\").trim() || pieces.length === 0) pieces.push(current.join(\"\\n\"));\n  return pieces;\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA4BO,IAAM,sBAAsB;AAG5B,IAAM,mBAAmB;AAGzB,IAAM,mBAAmB;AAEhC,IAAM,eAAe;AACrB,IAAM,aAAa;AACnB,IAAM,WAAW;AACjB,IAAM,iBACJ;AAGF,SAAS,cAAc,MAAsB;AAC3C,MAAI,OAAO;AACX,aAAW,MAAM,MAAM;AACrB,QAAI,OAAO,OAAO,OAAO,OAAU,SAAQ;AAAA,aAClC,OAAO,IAAM,SAAQ;AAAA,QACzB;AAAA,EACP;AACA,SAAO;AACT;AAMO,SAAS,gBAAgB,MAAuB;AACrD,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,KAAK,KAAK;AACpB,MAAI,EAAE,SAAS,KAAK,EAAE,SAAS,IAAK,QAAO;AAC3C,MAAI,aAAa,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,EAAG,QAAO;AACvD,QAAM,MAAM,SAAS,KAAK,CAAC;AAC3B,MAAI,CAAC,IAAK,QAAO;AACjB,SAAO,eAAe,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,EAAE;AACnD;AAQO,SAAS,kBAAkB,UAA4B;AAC5D,QAAM,QAAQ,SAAS,MAAM,OAAO;AACpC,QAAM,SAAmB,CAAC;AAC1B,MAAI,UAAoB,CAAC;AACzB,MAAI,QAAuB;AAE3B,aAAW,QAAQ,OAAO;AAExB,UAAM,aAAa,oBAAoB,KAAK,IAAI;AAChD,QAAI,YAAY;AACd,YAAM,SAAS,WAAW,CAAC;AAC3B,UAAI,UAAU,KAAM,SAAQ;AAAA,eACnB,OAAO,CAAC,MAAM,MAAM,CAAC,KAAK,OAAO,UAAU,MAAM,OAAQ,SAAQ;AAAA,IAC5E,WAAW,UAAU,QAAQ,gBAAgB,IAAI,GAAG;AAClD,UAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC7D,gBAAU,CAAC;AACX;AAAA,IACF;AACA,YAAQ,KAAK,IAAI;AAAA,EACnB;AACA,MAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,KAAK,OAAO,WAAW,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AACpF,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../src/directives.ts"],"sourcesContent":["/**\n * @ai-matrx/print/directives — the ONE definition of the print directives\n * markdown can carry, shared by every surface that reads markdown.\n *\n * Why a separate, zero-dependency entry: the directive syntax has three\n * readers — this package's two converters (print window / PDF / HTML page),\n * the chat renderer's remark plugin (the on-screen preview), and any editor\n * that inserts one. If each re-derived the regex they would drift the first\n * time someone added an alias. `/markdown` carries ~20KB of stylesheet, so the\n * chat bundle imports the syntax from here instead and pays nothing.\n *\n * ── Page break ────────────────────────────────────────────────────────────\n * A line that is exactly one of these (case-insensitive, surrounding\n * whitespace ignored) forces the next content onto a new printed page:\n *\n *   <!-- pagebreak -->          ← canonical; invisible in any other renderer\n *   <!-- page-break -->   <!-- page break -->   <!-- newpage -->\n *   \\pagebreak   \\newpage       ← LaTeX / pandoc habit\n *   <div style=\"page-break-after: always\"></div>   ← Typora / VS Code habit\n *   (also `page-break-before`, `break-after: page`, `break-before: page`,\n *    and `<div class=\"page-break\"></div>` / `class=\"pagebreak\"`)\n *\n * On screen every surface draws the same idea: a dashed rule labelled\n * \"Page break\" with generous padding. In print it is a zero-height element\n * with `break-after: page`.\n */\n\n/** The form editors insert and docs recommend. */\nexport const PAGE_BREAK_MARKDOWN = \"<!-- pagebreak -->\";\n\n/** The class every surface puts on a rendered page break. */\nexport const PAGE_BREAK_CLASS = \"matrx-page-break\";\n\n/** Screen label for a rendered page break. */\nexport const PAGE_BREAK_LABEL = \"Page break\";\n\nconst COMMENT_FORM = /^<!--\\s*(?:page[\\s_-]?break|new[\\s_-]?page)\\s*-->$/i;\nconst LATEX_FORM = /^\\\\(?:pagebreak|newpage|clearpage)\\s*(?:\\{\\s*\\})?$/i;\nconst DIV_FORM = /^<div\\b([^>]*)>\\s*<\\/div>$|^<div\\b([^>]*)\\/>$/i;\nconst DIV_BREAK_ATTR =\n  /(?:page-break-(?:after|before)\\s*:\\s*always|break-(?:after|before)\\s*:\\s*page|class\\s*=\\s*[\"'](?:[^\"']*\\s)?page-?break(?:\\s[^\"']*)?[\"'])/i;\n\n/** Columns of leading indentation (tab = 4, NBSP counts as a space). */\nfunction indentColumns(line: string): number {\n  let cols = 0;\n  for (const ch of line) {\n    if (ch === \" \" || ch === \"\\u00A0\") cols += 1;\n    else if (ch === \"\\t\") cols += 4;\n    else break;\n  }\n  return cols;\n}\n\n/**\n * Is this single line a page-break directive? Indented four or more columns\n * it is not — in markdown that is a code block, and every surface must agree.\n */\nexport function isPageBreakLine(line: string): boolean {\n  if (indentColumns(line) >= 4) return false;\n  const t = line.trim();\n  if (t.length < 8 || t.length > 200) return false;\n  if (COMMENT_FORM.test(t) || LATEX_FORM.test(t)) return true;\n  const div = DIV_FORM.exec(t);\n  if (!div) return false;\n  return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? \"\");\n}\n\n// ── Section break ─────────────────────────────────────────────────────────\n//\n// A section break is a page break that can also change the page layout for\n// everything after it (Word's \"Section break (next page)\"):\n//\n//   <!-- section -->                         ← new page, same layout\n//   <!-- section: landscape -->              ← new page, landscape from here on\n//   <!-- section landscape columns=2 -->     ← landscape AND two columns\n//   <!-- section portrait columns=1 -->      ← back to the default layout\n//\n// Recognised words: `landscape` / `portrait`, `columns=N` (or `cols=N`,\n// `N columns`, 1–3). Unknown words are ignored, never an error — the break\n// still happens. Every page-break reader treats a section break as a break.\n\n/** The form editors insert for a plain section break. */\nexport const SECTION_BREAK_MARKDOWN = \"<!-- section -->\";\n\nexport interface SectionBreakDirective {\n  /** Absent = inherit the document's orientation. */\n  orientation?: \"portrait\" | \"landscape\";\n  /** Absent = inherit the document's column count. 1–3. */\n  columns?: 1 | 2 | 3;\n}\n\nconst SECTION_FORM = /^<!--\\s*section(?:[\\s_-]?break)?\\b\\s*:?\\s*([^>]*?)\\s*-->$/i;\n\n/** Parse a section-break line; `null` when the line is not one. */\nexport function parseSectionBreakLine(line: string): SectionBreakDirective | null {\n  if (indentColumns(line) >= 4) return null;\n  const m = SECTION_FORM.exec(line.trim());\n  if (!m) return null;\n  const words = (m[1] ?? \"\").toLowerCase();\n  const out: SectionBreakDirective = {};\n  if (/\\blandscape\\b/.test(words)) out.orientation = \"landscape\";\n  else if (/\\bportrait\\b/.test(words)) out.orientation = \"portrait\";\n  const cols = /\\b(?:columns?|cols?)\\s*[=:]\\s*(\\d)\\b/.exec(words) ?? /\\b(\\d)\\s*[-\\s]?columns?\\b/.exec(words);\n  if (cols) {\n    const n = Math.min(3, Math.max(1, Number(cols[1])));\n    out.columns = n as 1 | 2 | 3;\n  }\n  return out;\n}\n\n// ── Table of contents / bibliography placement ───────────────────────────\n//\n// `[[toc]]`, `[toc]`, `[TOC]`, `${toc}`, `<!-- toc -->`, `\\tableofcontents`\n// put the table of contents at that spot (otherwise frontmatter `toc: true`\n// puts it at the start). `<!-- bibliography -->`, `[[bibliography]]`,\n// `\\printbibliography` put the reference list there (otherwise it closes the\n// document whenever something is cited).\n\nexport const TOC_MARKDOWN = \"[[toc]]\";\nexport const BIBLIOGRAPHY_MARKDOWN = \"<!-- bibliography -->\";\n\nconst TOC_FORM = /^(?:\\[\\[\\s*toc\\s*\\]\\]|\\[toc\\]|\\$\\{toc\\}|<!--\\s*toc\\s*-->|\\\\tableofcontents)$/i;\nconst BIBLIOGRAPHY_FORM =\n  /^(?:\\[\\[\\s*(?:bibliography|references)\\s*\\]\\]|<!--\\s*(?:bibliography|references)\\s*-->|\\\\printbibliography|\\\\bibliography\\{[^}]*\\})$/i;\n\nexport function isTocLine(line: string): boolean {\n  return indentColumns(line) < 4 && TOC_FORM.test(line.trim());\n}\n\nexport function isBibliographyLine(line: string): boolean {\n  return indentColumns(line) < 4 && BIBLIOGRAPHY_FORM.test(line.trim());\n}\n\n/** Any directive that ends the current printed page (page or section break). */\nexport function isBreakLine(line: string): boolean {\n  return isPageBreakLine(line) || parseSectionBreakLine(line) !== null;\n}\n\n/**\n * Split markdown into the pieces that belong on separate pages. Fence-aware:\n * a directive inside a ``` / ~~~ code block is code, not a break. Always\n * returns at least one piece; empty pieces (two breaks in a row, a break at\n * the very start or end) are kept out so no blank page is produced.\n */\nexport function splitAtPageBreaks(markdown: string): string[] {\n  const lines = markdown.split(/\\r?\\n/);\n  const pieces: string[] = [];\n  let current: string[] = [];\n  let fence: string | null = null;\n\n  for (const line of lines) {\n    // Any indent — the converters accept list-nested fences, so must this.\n    const fenceMatch = /^\\s*(`{3,}|~{3,})/.exec(line);\n    if (fenceMatch) {\n      const marker = fenceMatch[1] as string;\n      if (fence === null) fence = marker;\n      else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;\n    } else if (fence === null && isBreakLine(line)) {\n      if (current.join(\"\\n\").trim()) pieces.push(current.join(\"\\n\"));\n      current = [];\n      continue;\n    }\n    current.push(line);\n  }\n  if (current.join(\"\\n\").trim() || pieces.length === 0) pieces.push(current.join(\"\\n\"));\n  return pieces;\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA4BO,IAAM,sBAAsB;AAG5B,IAAM,mBAAmB;AAGzB,IAAM,mBAAmB;AAEhC,IAAM,eAAe;AACrB,IAAM,aAAa;AACnB,IAAM,WAAW;AACjB,IAAM,iBACJ;AAGF,SAAS,cAAc,MAAsB;AAC3C,MAAI,OAAO;AACX,aAAW,MAAM,MAAM;AACrB,QAAI,OAAO,OAAO,OAAO,OAAU,SAAQ;AAAA,aAClC,OAAO,IAAM,SAAQ;AAAA,QACzB;AAAA,EACP;AACA,SAAO;AACT;AAMO,SAAS,gBAAgB,MAAuB;AACrD,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,KAAK,KAAK;AACpB,MAAI,EAAE,SAAS,KAAK,EAAE,SAAS,IAAK,QAAO;AAC3C,MAAI,aAAa,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,EAAG,QAAO;AACvD,QAAM,MAAM,SAAS,KAAK,CAAC;AAC3B,MAAI,CAAC,IAAK,QAAO;AACjB,SAAO,eAAe,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,EAAE;AACnD;AAiBO,IAAM,yBAAyB;AAStC,IAAM,eAAe;AAGd,SAAS,sBAAsB,MAA4C;AAChF,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,aAAa,KAAK,KAAK,KAAK,CAAC;AACvC,MAAI,CAAC,EAAG,QAAO;AACf,QAAM,SAAS,EAAE,CAAC,KAAK,IAAI,YAAY;AACvC,QAAM,MAA6B,CAAC;AACpC,MAAI,gBAAgB,KAAK,KAAK,EAAG,KAAI,cAAc;AAAA,WAC1C,eAAe,KAAK,KAAK,EAAG,KAAI,cAAc;AACvD,QAAM,OAAO,uCAAuC,KAAK,KAAK,KAAK,4BAA4B,KAAK,KAAK;AACzG,MAAI,MAAM;AACR,UAAM,IAAI,KAAK,IAAI,GAAG,KAAK,IAAI,GAAG,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC;AAClD,QAAI,UAAU;AAAA,EAChB;AACA,SAAO;AACT;AAUO,IAAM,eAAe;AACrB,IAAM,wBAAwB;AAErC,IAAM,WAAW;AACjB,IAAM,oBACJ;AAEK,SAAS,UAAU,MAAuB;AAC/C,SAAO,cAAc,IAAI,IAAI,KAAK,SAAS,KAAK,KAAK,KAAK,CAAC;AAC7D;AAEO,SAAS,mBAAmB,MAAuB;AACxD,SAAO,cAAc,IAAI,IAAI,KAAK,kBAAkB,KAAK,KAAK,KAAK,CAAC;AACtE;AAGO,SAAS,YAAY,MAAuB;AACjD,SAAO,gBAAgB,IAAI,KAAK,sBAAsB,IAAI,MAAM;AAClE;AAQO,SAAS,kBAAkB,UAA4B;AAC5D,QAAM,QAAQ,SAAS,MAAM,OAAO;AACpC,QAAM,SAAmB,CAAC;AAC1B,MAAI,UAAoB,CAAC;AACzB,MAAI,QAAuB;AAE3B,aAAW,QAAQ,OAAO;AAExB,UAAM,aAAa,oBAAoB,KAAK,IAAI;AAChD,QAAI,YAAY;AACd,YAAM,SAAS,WAAW,CAAC;AAC3B,UAAI,UAAU,KAAM,SAAQ;AAAA,eACnB,OAAO,CAAC,MAAM,MAAM,CAAC,KAAK,OAAO,UAAU,MAAM,OAAQ,SAAQ;AAAA,IAC5E,WAAW,UAAU,QAAQ,YAAY,IAAI,GAAG;AAC9C,UAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC7D,gBAAU,CAAC;AACX;AAAA,IACF;AACA,YAAQ,KAAK,IAAI;AAAA,EACnB;AACA,MAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,KAAK,OAAO,WAAW,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AACpF,SAAO;AACT;","names":[]}

package/dist/directives.d.cts

@@ -35,6 +35,22 @@ declare const PAGE_BREAK_LABEL = "Page break";
  * it is not — in markdown that is a code block, and every surface must agree.
  */
 declare function isPageBreakLine(line: string): boolean;
+/** The form editors insert for a plain section break. */
+declare const SECTION_BREAK_MARKDOWN = "<!-- section -->";
+interface SectionBreakDirective {
+    /** Absent = inherit the document's orientation. */
+    orientation?: "portrait" | "landscape";
+    /** Absent = inherit the document's column count. 1–3. */
+    columns?: 1 | 2 | 3;
+}
+/** Parse a section-break line; `null` when the line is not one. */
+declare function parseSectionBreakLine(line: string): SectionBreakDirective | null;
+declare const TOC_MARKDOWN = "[[toc]]";
+declare const BIBLIOGRAPHY_MARKDOWN = "<!-- bibliography -->";
+declare function isTocLine(line: string): boolean;
+declare function isBibliographyLine(line: string): boolean;
+/** Any directive that ends the current printed page (page or section break). */
+declare function isBreakLine(line: string): boolean;
 /**
  * Split markdown into the pieces that belong on separate pages. Fence-aware:
  * a directive inside a ``` / ~~~ code block is code, not a break. Always
@@ -43,4 +59,4 @@ declare function isPageBreakLine(line: string): boolean;
  */
 declare function splitAtPageBreaks(markdown: string): string[];
 
-export { PAGE_BREAK_CLASS, PAGE_BREAK_LABEL, PAGE_BREAK_MARKDOWN, isPageBreakLine, splitAtPageBreaks };
+export { BIBLIOGRAPHY_MARKDOWN, PAGE_BREAK_CLASS, PAGE_BREAK_LABEL, PAGE_BREAK_MARKDOWN, SECTION_BREAK_MARKDOWN, type SectionBreakDirective, TOC_MARKDOWN, isBibliographyLine, isBreakLine, isPageBreakLine, isTocLine, parseSectionBreakLine, splitAtPageBreaks };

package/dist/directives.d.ts

@@ -35,6 +35,22 @@ declare const PAGE_BREAK_LABEL = "Page break";
  * it is not — in markdown that is a code block, and every surface must agree.
  */
 declare function isPageBreakLine(line: string): boolean;
+/** The form editors insert for a plain section break. */
+declare const SECTION_BREAK_MARKDOWN = "<!-- section -->";
+interface SectionBreakDirective {
+    /** Absent = inherit the document's orientation. */
+    orientation?: "portrait" | "landscape";
+    /** Absent = inherit the document's column count. 1–3. */
+    columns?: 1 | 2 | 3;
+}
+/** Parse a section-break line; `null` when the line is not one. */
+declare function parseSectionBreakLine(line: string): SectionBreakDirective | null;
+declare const TOC_MARKDOWN = "[[toc]]";
+declare const BIBLIOGRAPHY_MARKDOWN = "<!-- bibliography -->";
+declare function isTocLine(line: string): boolean;
+declare function isBibliographyLine(line: string): boolean;
+/** Any directive that ends the current printed page (page or section break). */
+declare function isBreakLine(line: string): boolean;
 /**
  * Split markdown into the pieces that belong on separate pages. Fence-aware:
  * a directive inside a ``` / ~~~ code block is code, not a break. Always
@@ -43,4 +59,4 @@ declare function isPageBreakLine(line: string): boolean;
  */
 declare function splitAtPageBreaks(markdown: string): string[];
 
-export { PAGE_BREAK_CLASS, PAGE_BREAK_LABEL, PAGE_BREAK_MARKDOWN, isPageBreakLine, splitAtPageBreaks };
+export { BIBLIOGRAPHY_MARKDOWN, PAGE_BREAK_CLASS, PAGE_BREAK_LABEL, PAGE_BREAK_MARKDOWN, SECTION_BREAK_MARKDOWN, type SectionBreakDirective, TOC_MARKDOWN, isBibliographyLine, isBreakLine, isPageBreakLine, isTocLine, parseSectionBreakLine, splitAtPageBreaks };

package/dist/directives.js

@@ -24,6 +24,36 @@ function isPageBreakLine(line) {
   if (!div) return false;
   return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? "");
 }
+var SECTION_BREAK_MARKDOWN = "<!-- section -->";
+var SECTION_FORM = /^<!--\s*section(?:[\s_-]?break)?\b\s*:?\s*([^>]*?)\s*-->$/i;
+function parseSectionBreakLine(line) {
+  if (indentColumns(line) >= 4) return null;
+  const m = SECTION_FORM.exec(line.trim());
+  if (!m) return null;
+  const words = (m[1] ?? "").toLowerCase();
+  const out = {};
+  if (/\blandscape\b/.test(words)) out.orientation = "landscape";
+  else if (/\bportrait\b/.test(words)) out.orientation = "portrait";
+  const cols = /\b(?:columns?|cols?)\s*[=:]\s*(\d)\b/.exec(words) ?? /\b(\d)\s*[-\s]?columns?\b/.exec(words);
+  if (cols) {
+    const n = Math.min(3, Math.max(1, Number(cols[1])));
+    out.columns = n;
+  }
+  return out;
+}
+var TOC_MARKDOWN = "[[toc]]";
+var BIBLIOGRAPHY_MARKDOWN = "<!-- bibliography -->";
+var TOC_FORM = /^(?:\[\[\s*toc\s*\]\]|\[toc\]|\$\{toc\}|<!--\s*toc\s*-->|\\tableofcontents)$/i;
+var BIBLIOGRAPHY_FORM = /^(?:\[\[\s*(?:bibliography|references)\s*\]\]|<!--\s*(?:bibliography|references)\s*-->|\\printbibliography|\\bibliography\{[^}]*\})$/i;
+function isTocLine(line) {
+  return indentColumns(line) < 4 && TOC_FORM.test(line.trim());
+}
+function isBibliographyLine(line) {
+  return indentColumns(line) < 4 && BIBLIOGRAPHY_FORM.test(line.trim());
+}
+function isBreakLine(line) {
+  return isPageBreakLine(line) || parseSectionBreakLine(line) !== null;
+}
 function splitAtPageBreaks(markdown) {
   const lines = markdown.split(/\r?\n/);
   const pieces = [];
@@ -35,7 +65,7 @@ function splitAtPageBreaks(markdown) {
       const marker = fenceMatch[1];
       if (fence === null) fence = marker;
       else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;
-    } else if (fence === null && isPageBreakLine(line)) {
+    } else if (fence === null && isBreakLine(line)) {
       if (current.join("\n").trim()) pieces.push(current.join("\n"));
       current = [];
       continue;
@@ -46,10 +76,17 @@ function splitAtPageBreaks(markdown) {
   return pieces;
 }
 export {
+  BIBLIOGRAPHY_MARKDOWN,
   PAGE_BREAK_CLASS,
   PAGE_BREAK_LABEL,
   PAGE_BREAK_MARKDOWN,
+  SECTION_BREAK_MARKDOWN,
+  TOC_MARKDOWN,
+  isBibliographyLine,
+  isBreakLine,
   isPageBreakLine,
+  isTocLine,
+  parseSectionBreakLine,
   splitAtPageBreaks
 };
 //# sourceMappingURL=directives.js.map

package/dist/directives.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/directives.ts"],"sourcesContent":["/**\n * @ai-matrx/print/directives — the ONE definition of the print directives\n * markdown can carry, shared by every surface that reads markdown.\n *\n * Why a separate, zero-dependency entry: the directive syntax has three\n * readers — this package's two converters (print window / PDF / HTML page),\n * the chat renderer's remark plugin (the on-screen preview), and any editor\n * that inserts one. If each re-derived the regex they would drift the first\n * time someone added an alias. `/markdown` carries ~20KB of stylesheet, so the\n * chat bundle imports the syntax from here instead and pays nothing.\n *\n * ── Page break ────────────────────────────────────────────────────────────\n * A line that is exactly one of these (case-insensitive, surrounding\n * whitespace ignored) forces the next content onto a new printed page:\n *\n *   <!-- pagebreak -->          ← canonical; invisible in any other renderer\n *   <!-- page-break -->   <!-- page break -->   <!-- newpage -->\n *   \\pagebreak   \\newpage       ← LaTeX / pandoc habit\n *   <div style=\"page-break-after: always\"></div>   ← Typora / VS Code habit\n *   (also `page-break-before`, `break-after: page`, `break-before: page`,\n *    and `<div class=\"page-break\"></div>` / `class=\"pagebreak\"`)\n *\n * On screen every surface draws the same idea: a dashed rule labelled\n * \"Page break\" with generous padding. In print it is a zero-height element\n * with `break-after: page`.\n */\n\n/** The form editors insert and docs recommend. */\nexport const PAGE_BREAK_MARKDOWN = \"<!-- pagebreak -->\";\n\n/** The class every surface puts on a rendered page break. */\nexport const PAGE_BREAK_CLASS = \"matrx-page-break\";\n\n/** Screen label for a rendered page break. */\nexport const PAGE_BREAK_LABEL = \"Page break\";\n\nconst COMMENT_FORM = /^<!--\\s*(?:page[\\s_-]?break|new[\\s_-]?page)\\s*-->$/i;\nconst LATEX_FORM = /^\\\\(?:pagebreak|newpage|clearpage)\\s*(?:\\{\\s*\\})?$/i;\nconst DIV_FORM = /^<div\\b([^>]*)>\\s*<\\/div>$|^<div\\b([^>]*)\\/>$/i;\nconst DIV_BREAK_ATTR =\n  /(?:page-break-(?:after|before)\\s*:\\s*always|break-(?:after|before)\\s*:\\s*page|class\\s*=\\s*[\"'](?:[^\"']*\\s)?page-?break(?:\\s[^\"']*)?[\"'])/i;\n\n/** Columns of leading indentation (tab = 4, NBSP counts as a space). */\nfunction indentColumns(line: string): number {\n  let cols = 0;\n  for (const ch of line) {\n    if (ch === \" \" || ch === \"\\u00A0\") cols += 1;\n    else if (ch === \"\\t\") cols += 4;\n    else break;\n  }\n  return cols;\n}\n\n/**\n * Is this single line a page-break directive? Indented four or more columns\n * it is not — in markdown that is a code block, and every surface must agree.\n */\nexport function isPageBreakLine(line: string): boolean {\n  if (indentColumns(line) >= 4) return false;\n  const t = line.trim();\n  if (t.length < 8 || t.length > 200) return false;\n  if (COMMENT_FORM.test(t) || LATEX_FORM.test(t)) return true;\n  const div = DIV_FORM.exec(t);\n  if (!div) return false;\n  return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? \"\");\n}\n\n/**\n * Split markdown into the pieces that belong on separate pages. Fence-aware:\n * a directive inside a ``` / ~~~ code block is code, not a break. Always\n * returns at least one piece; empty pieces (two breaks in a row, a break at\n * the very start or end) are kept out so no blank page is produced.\n */\nexport function splitAtPageBreaks(markdown: string): string[] {\n  const lines = markdown.split(/\\r?\\n/);\n  const pieces: string[] = [];\n  let current: string[] = [];\n  let fence: string | null = null;\n\n  for (const line of lines) {\n    // Any indent — the converters accept list-nested fences, so must this.\n    const fenceMatch = /^\\s*(`{3,}|~{3,})/.exec(line);\n    if (fenceMatch) {\n      const marker = fenceMatch[1] as string;\n      if (fence === null) fence = marker;\n      else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;\n    } else if (fence === null && isPageBreakLine(line)) {\n      if (current.join(\"\\n\").trim()) pieces.push(current.join(\"\\n\"));\n      current = [];\n      continue;\n    }\n    current.push(line);\n  }\n  if (current.join(\"\\n\").trim() || pieces.length === 0) pieces.push(current.join(\"\\n\"));\n  return pieces;\n}\n"],"mappings":";AA4BO,IAAM,sBAAsB;AAG5B,IAAM,mBAAmB;AAGzB,IAAM,mBAAmB;AAEhC,IAAM,eAAe;AACrB,IAAM,aAAa;AACnB,IAAM,WAAW;AACjB,IAAM,iBACJ;AAGF,SAAS,cAAc,MAAsB;AAC3C,MAAI,OAAO;AACX,aAAW,MAAM,MAAM;AACrB,QAAI,OAAO,OAAO,OAAO,OAAU,SAAQ;AAAA,aAClC,OAAO,IAAM,SAAQ;AAAA,QACzB;AAAA,EACP;AACA,SAAO;AACT;AAMO,SAAS,gBAAgB,MAAuB;AACrD,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,KAAK,KAAK;AACpB,MAAI,EAAE,SAAS,KAAK,EAAE,SAAS,IAAK,QAAO;AAC3C,MAAI,aAAa,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,EAAG,QAAO;AACvD,QAAM,MAAM,SAAS,KAAK,CAAC;AAC3B,MAAI,CAAC,IAAK,QAAO;AACjB,SAAO,eAAe,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,EAAE;AACnD;AAQO,SAAS,kBAAkB,UAA4B;AAC5D,QAAM,QAAQ,SAAS,MAAM,OAAO;AACpC,QAAM,SAAmB,CAAC;AAC1B,MAAI,UAAoB,CAAC;AACzB,MAAI,QAAuB;AAE3B,aAAW,QAAQ,OAAO;AAExB,UAAM,aAAa,oBAAoB,KAAK,IAAI;AAChD,QAAI,YAAY;AACd,YAAM,SAAS,WAAW,CAAC;AAC3B,UAAI,UAAU,KAAM,SAAQ;AAAA,eACnB,OAAO,CAAC,MAAM,MAAM,CAAC,KAAK,OAAO,UAAU,MAAM,OAAQ,SAAQ;AAAA,IAC5E,WAAW,UAAU,QAAQ,gBAAgB,IAAI,GAAG;AAClD,UAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC7D,gBAAU,CAAC;AACX;AAAA,IACF;AACA,YAAQ,KAAK,IAAI;AAAA,EACnB;AACA,MAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,KAAK,OAAO,WAAW,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AACpF,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../src/directives.ts"],"sourcesContent":["/**\n * @ai-matrx/print/directives — the ONE definition of the print directives\n * markdown can carry, shared by every surface that reads markdown.\n *\n * Why a separate, zero-dependency entry: the directive syntax has three\n * readers — this package's two converters (print window / PDF / HTML page),\n * the chat renderer's remark plugin (the on-screen preview), and any editor\n * that inserts one. If each re-derived the regex they would drift the first\n * time someone added an alias. `/markdown` carries ~20KB of stylesheet, so the\n * chat bundle imports the syntax from here instead and pays nothing.\n *\n * ── Page break ────────────────────────────────────────────────────────────\n * A line that is exactly one of these (case-insensitive, surrounding\n * whitespace ignored) forces the next content onto a new printed page:\n *\n *   <!-- pagebreak -->          ← canonical; invisible in any other renderer\n *   <!-- page-break -->   <!-- page break -->   <!-- newpage -->\n *   \\pagebreak   \\newpage       ← LaTeX / pandoc habit\n *   <div style=\"page-break-after: always\"></div>   ← Typora / VS Code habit\n *   (also `page-break-before`, `break-after: page`, `break-before: page`,\n *    and `<div class=\"page-break\"></div>` / `class=\"pagebreak\"`)\n *\n * On screen every surface draws the same idea: a dashed rule labelled\n * \"Page break\" with generous padding. In print it is a zero-height element\n * with `break-after: page`.\n */\n\n/** The form editors insert and docs recommend. */\nexport const PAGE_BREAK_MARKDOWN = \"<!-- pagebreak -->\";\n\n/** The class every surface puts on a rendered page break. */\nexport const PAGE_BREAK_CLASS = \"matrx-page-break\";\n\n/** Screen label for a rendered page break. */\nexport const PAGE_BREAK_LABEL = \"Page break\";\n\nconst COMMENT_FORM = /^<!--\\s*(?:page[\\s_-]?break|new[\\s_-]?page)\\s*-->$/i;\nconst LATEX_FORM = /^\\\\(?:pagebreak|newpage|clearpage)\\s*(?:\\{\\s*\\})?$/i;\nconst DIV_FORM = /^<div\\b([^>]*)>\\s*<\\/div>$|^<div\\b([^>]*)\\/>$/i;\nconst DIV_BREAK_ATTR =\n  /(?:page-break-(?:after|before)\\s*:\\s*always|break-(?:after|before)\\s*:\\s*page|class\\s*=\\s*[\"'](?:[^\"']*\\s)?page-?break(?:\\s[^\"']*)?[\"'])/i;\n\n/** Columns of leading indentation (tab = 4, NBSP counts as a space). */\nfunction indentColumns(line: string): number {\n  let cols = 0;\n  for (const ch of line) {\n    if (ch === \" \" || ch === \"\\u00A0\") cols += 1;\n    else if (ch === \"\\t\") cols += 4;\n    else break;\n  }\n  return cols;\n}\n\n/**\n * Is this single line a page-break directive? Indented four or more columns\n * it is not — in markdown that is a code block, and every surface must agree.\n */\nexport function isPageBreakLine(line: string): boolean {\n  if (indentColumns(line) >= 4) return false;\n  const t = line.trim();\n  if (t.length < 8 || t.length > 200) return false;\n  if (COMMENT_FORM.test(t) || LATEX_FORM.test(t)) return true;\n  const div = DIV_FORM.exec(t);\n  if (!div) return false;\n  return DIV_BREAK_ATTR.test(div[1] ?? div[2] ?? \"\");\n}\n\n// ── Section break ─────────────────────────────────────────────────────────\n//\n// A section break is a page break that can also change the page layout for\n// everything after it (Word's \"Section break (next page)\"):\n//\n//   <!-- section -->                         ← new page, same layout\n//   <!-- section: landscape -->              ← new page, landscape from here on\n//   <!-- section landscape columns=2 -->     ← landscape AND two columns\n//   <!-- section portrait columns=1 -->      ← back to the default layout\n//\n// Recognised words: `landscape` / `portrait`, `columns=N` (or `cols=N`,\n// `N columns`, 1–3). Unknown words are ignored, never an error — the break\n// still happens. Every page-break reader treats a section break as a break.\n\n/** The form editors insert for a plain section break. */\nexport const SECTION_BREAK_MARKDOWN = \"<!-- section -->\";\n\nexport interface SectionBreakDirective {\n  /** Absent = inherit the document's orientation. */\n  orientation?: \"portrait\" | \"landscape\";\n  /** Absent = inherit the document's column count. 1–3. */\n  columns?: 1 | 2 | 3;\n}\n\nconst SECTION_FORM = /^<!--\\s*section(?:[\\s_-]?break)?\\b\\s*:?\\s*([^>]*?)\\s*-->$/i;\n\n/** Parse a section-break line; `null` when the line is not one. */\nexport function parseSectionBreakLine(line: string): SectionBreakDirective | null {\n  if (indentColumns(line) >= 4) return null;\n  const m = SECTION_FORM.exec(line.trim());\n  if (!m) return null;\n  const words = (m[1] ?? \"\").toLowerCase();\n  const out: SectionBreakDirective = {};\n  if (/\\blandscape\\b/.test(words)) out.orientation = \"landscape\";\n  else if (/\\bportrait\\b/.test(words)) out.orientation = \"portrait\";\n  const cols = /\\b(?:columns?|cols?)\\s*[=:]\\s*(\\d)\\b/.exec(words) ?? /\\b(\\d)\\s*[-\\s]?columns?\\b/.exec(words);\n  if (cols) {\n    const n = Math.min(3, Math.max(1, Number(cols[1])));\n    out.columns = n as 1 | 2 | 3;\n  }\n  return out;\n}\n\n// ── Table of contents / bibliography placement ───────────────────────────\n//\n// `[[toc]]`, `[toc]`, `[TOC]`, `${toc}`, `<!-- toc -->`, `\\tableofcontents`\n// put the table of contents at that spot (otherwise frontmatter `toc: true`\n// puts it at the start). `<!-- bibliography -->`, `[[bibliography]]`,\n// `\\printbibliography` put the reference list there (otherwise it closes the\n// document whenever something is cited).\n\nexport const TOC_MARKDOWN = \"[[toc]]\";\nexport const BIBLIOGRAPHY_MARKDOWN = \"<!-- bibliography -->\";\n\nconst TOC_FORM = /^(?:\\[\\[\\s*toc\\s*\\]\\]|\\[toc\\]|\\$\\{toc\\}|<!--\\s*toc\\s*-->|\\\\tableofcontents)$/i;\nconst BIBLIOGRAPHY_FORM =\n  /^(?:\\[\\[\\s*(?:bibliography|references)\\s*\\]\\]|<!--\\s*(?:bibliography|references)\\s*-->|\\\\printbibliography|\\\\bibliography\\{[^}]*\\})$/i;\n\nexport function isTocLine(line: string): boolean {\n  return indentColumns(line) < 4 && TOC_FORM.test(line.trim());\n}\n\nexport function isBibliographyLine(line: string): boolean {\n  return indentColumns(line) < 4 && BIBLIOGRAPHY_FORM.test(line.trim());\n}\n\n/** Any directive that ends the current printed page (page or section break). */\nexport function isBreakLine(line: string): boolean {\n  return isPageBreakLine(line) || parseSectionBreakLine(line) !== null;\n}\n\n/**\n * Split markdown into the pieces that belong on separate pages. Fence-aware:\n * a directive inside a ``` / ~~~ code block is code, not a break. Always\n * returns at least one piece; empty pieces (two breaks in a row, a break at\n * the very start or end) are kept out so no blank page is produced.\n */\nexport function splitAtPageBreaks(markdown: string): string[] {\n  const lines = markdown.split(/\\r?\\n/);\n  const pieces: string[] = [];\n  let current: string[] = [];\n  let fence: string | null = null;\n\n  for (const line of lines) {\n    // Any indent — the converters accept list-nested fences, so must this.\n    const fenceMatch = /^\\s*(`{3,}|~{3,})/.exec(line);\n    if (fenceMatch) {\n      const marker = fenceMatch[1] as string;\n      if (fence === null) fence = marker;\n      else if (marker[0] === fence[0] && marker.length >= fence.length) fence = null;\n    } else if (fence === null && isBreakLine(line)) {\n      if (current.join(\"\\n\").trim()) pieces.push(current.join(\"\\n\"));\n      current = [];\n      continue;\n    }\n    current.push(line);\n  }\n  if (current.join(\"\\n\").trim() || pieces.length === 0) pieces.push(current.join(\"\\n\"));\n  return pieces;\n}\n"],"mappings":";AA4BO,IAAM,sBAAsB;AAG5B,IAAM,mBAAmB;AAGzB,IAAM,mBAAmB;AAEhC,IAAM,eAAe;AACrB,IAAM,aAAa;AACnB,IAAM,WAAW;AACjB,IAAM,iBACJ;AAGF,SAAS,cAAc,MAAsB;AAC3C,MAAI,OAAO;AACX,aAAW,MAAM,MAAM;AACrB,QAAI,OAAO,OAAO,OAAO,OAAU,SAAQ;AAAA,aAClC,OAAO,IAAM,SAAQ;AAAA,QACzB;AAAA,EACP;AACA,SAAO;AACT;AAMO,SAAS,gBAAgB,MAAuB;AACrD,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,KAAK,KAAK;AACpB,MAAI,EAAE,SAAS,KAAK,EAAE,SAAS,IAAK,QAAO;AAC3C,MAAI,aAAa,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,EAAG,QAAO;AACvD,QAAM,MAAM,SAAS,KAAK,CAAC;AAC3B,MAAI,CAAC,IAAK,QAAO;AACjB,SAAO,eAAe,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,EAAE;AACnD;AAiBO,IAAM,yBAAyB;AAStC,IAAM,eAAe;AAGd,SAAS,sBAAsB,MAA4C;AAChF,MAAI,cAAc,IAAI,KAAK,EAAG,QAAO;AACrC,QAAM,IAAI,aAAa,KAAK,KAAK,KAAK,CAAC;AACvC,MAAI,CAAC,EAAG,QAAO;AACf,QAAM,SAAS,EAAE,CAAC,KAAK,IAAI,YAAY;AACvC,QAAM,MAA6B,CAAC;AACpC,MAAI,gBAAgB,KAAK,KAAK,EAAG,KAAI,cAAc;AAAA,WAC1C,eAAe,KAAK,KAAK,EAAG,KAAI,cAAc;AACvD,QAAM,OAAO,uCAAuC,KAAK,KAAK,KAAK,4BAA4B,KAAK,KAAK;AACzG,MAAI,MAAM;AACR,UAAM,IAAI,KAAK,IAAI,GAAG,KAAK,IAAI,GAAG,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC;AAClD,QAAI,UAAU;AAAA,EAChB;AACA,SAAO;AACT;AAUO,IAAM,eAAe;AACrB,IAAM,wBAAwB;AAErC,IAAM,WAAW;AACjB,IAAM,oBACJ;AAEK,SAAS,UAAU,MAAuB;AAC/C,SAAO,cAAc,IAAI,IAAI,KAAK,SAAS,KAAK,KAAK,KAAK,CAAC;AAC7D;AAEO,SAAS,mBAAmB,MAAuB;AACxD,SAAO,cAAc,IAAI,IAAI,KAAK,kBAAkB,KAAK,KAAK,KAAK,CAAC;AACtE;AAGO,SAAS,YAAY,MAAuB;AACjD,SAAO,gBAAgB,IAAI,KAAK,sBAAsB,IAAI,MAAM;AAClE;AAQO,SAAS,kBAAkB,UAA4B;AAC5D,QAAM,QAAQ,SAAS,MAAM,OAAO;AACpC,QAAM,SAAmB,CAAC;AAC1B,MAAI,UAAoB,CAAC;AACzB,MAAI,QAAuB;AAE3B,aAAW,QAAQ,OAAO;AAExB,UAAM,aAAa,oBAAoB,KAAK,IAAI;AAChD,QAAI,YAAY;AACd,YAAM,SAAS,WAAW,CAAC;AAC3B,UAAI,UAAU,KAAM,SAAQ;AAAA,eACnB,OAAO,CAAC,MAAM,MAAM,CAAC,KAAK,OAAO,UAAU,MAAM,OAAQ,SAAQ;AAAA,IAC5E,WAAW,UAAU,QAAQ,YAAY,IAAI,GAAG;AAC9C,UAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AAC7D,gBAAU,CAAC;AACX;AAAA,IACF;AACA,YAAQ,KAAK,IAAI;AAAA,EACnB;AACA,MAAI,QAAQ,KAAK,IAAI,EAAE,KAAK,KAAK,OAAO,WAAW,EAAG,QAAO,KAAK,QAAQ,KAAK,IAAI,CAAC;AACpF,SAAO;AACT;","names":[]}