@segnavia/html 0.3.0 → 0.5.0

package/dist/book/book-document.d.ts

@@ -0,0 +1,5 @@
+import { BookDocument, BookDocumentOptions } from "./types.js";
+import { Book } from "@segnavia/format";
+//#region src/book/book-document.d.ts
+export declare function bookDocument(book: Book, options?: BookDocumentOptions): BookDocument;
+//#endregion

package/dist/book/book-document.js

@@ -0,0 +1,217 @@
+import { elements, parse, serialize } from "../sanitize.js";
+import { element, linkChildren, text, textContent } from "./dom.js";
+import { noteState, transformNotes } from "./notes.js";
+import { bookPrintPart } from "./types.js";
+import { isTag } from "domhandler";
+import { isBookContents, noteMarkerIssues, resolveDirection } from "@segnavia/format";
+//#region src/book/book-document.ts
+const PRINT_CSS = ":where(.segnavia-contents-list){list-style:none;padding:0;margin:0}:where(.segnavia-contents-entry){display:flex}:where(.segnavia-contents-page){margin-inline-start:auto}:where(.segnavia-contents-entry[data-level=\"2\"]){padding-inline-start:1em}:where(.segnavia-contents-entry[data-level=\"3\"]){padding-inline-start:2em}:where(.segnavia-contents-entry[data-level=\"4\"]){padding-inline-start:3em}:where(.segnavia-contents-entry[data-level=\"5\"]){padding-inline-start:4em}:where(.segnavia-contents-entry[data-level=\"6\"]){padding-inline-start:5em}@media print{:where(.segnavia-front,.segnavia-contents)~:where(.segnavia-front,.segnavia-contents),:where(.segnavia-chapter,.segnavia-endnotes,.segnavia-blank){break-before:page}:where(.segnavia-blank){break-after:page}:where(.segnavia-targets){position:absolute;inline-size:1px;block-size:1px;overflow:hidden;clip-path:inset(50%)}}";
+function direction(book, value) {
+	return resolveDirection(book.dir, book.lang, value);
+}
+function sectionNode(kind, id, children = []) {
+	return element("section", {
+		class: `segnavia-section segnavia-${kind}`,
+		id
+	}, children);
+}
+function authoredSection(section, id) {
+	const node = sectionNode("front", id, parse(section.html).children);
+	return {
+		info: {
+			id,
+			kind: "front",
+			title: section.title,
+			runningHead: section.title
+		},
+		node
+	};
+}
+function contentsSection(title) {
+	const id = "segnavia-contents";
+	const children = title !== void 0 ? [element("h1", {
+		class: "segnavia-contents-title",
+		dir: "ltr"
+	}, [text(title)])] : [];
+	return {
+		info: {
+			id,
+			kind: "contents",
+			title,
+			runningHead: title
+		},
+		node: sectionNode("contents", id, children)
+	};
+}
+function chapterSection(chapter, index) {
+	const id = `segnavia-chapter-${index}`;
+	const document = parse(chapter.html);
+	return {
+		info: {
+			id,
+			kind: "chapter",
+			title: chapter.title,
+			runningHead: chapter.runningHead ?? chapter.title
+		},
+		node: sectionNode("chapter", id, document.children)
+	};
+}
+function chapterHeadings(section, chapter, depth) {
+	let generated = 0;
+	return elements(section.node.children).flatMap((heading) => {
+		const match = /^h([2-6])$/.exec(heading.name);
+		if (!match || heading.attribs["data-contents"] === "exclude") return [];
+		const level = Number(match[1]);
+		if (level > depth) return [];
+		generated += 1;
+		const id = heading.attribs.id ?? `segnavia-h-${chapter}-${generated}`;
+		heading.attribs.id = id;
+		return [{
+			id,
+			level,
+			text: textContent(heading, true).replace(/\s+/g, " ").trim(),
+			sectionId: section.info.id
+		}];
+	});
+}
+function contentsList(book, headings, options) {
+	const entries = headings.map((heading) => {
+		const page = book.contents.pageNumbers ? options.pageNumbers?.[heading.id] ?? "" : "";
+		return element("li", {
+			class: "segnavia-contents-entry",
+			"data-level": String(heading.level)
+		}, [element("a", {
+			href: options.crossLinks ? `segnavia:target/${heading.id}` : `#${heading.id}`,
+			dir: direction(book, heading.text)
+		}, [text(heading.text)]), element("span", {
+			class: "segnavia-contents-page",
+			"data-target": heading.id,
+			dir: direction(book, page)
+		}, [text(page)])]);
+	});
+	return element("ol", { class: "segnavia-contents-list" }, entries);
+}
+function addContents(book, section, headings, options) {
+	const title = section.info.title;
+	const titleNode = section.node.children.find((child) => isTag(child) && child.name === "h1");
+	if (title !== void 0 && titleNode && isTag(titleNode)) titleNode.attribs.dir = direction(book, title);
+	section.node.children.push(contentsList(book, headings, options));
+	linkChildren(section.node);
+}
+function targetLinks(sections) {
+	const ids = sections.flatMap((section) => [section.info.id, ...elements(section.node.children).flatMap((node) => node.attribs.id ? [node.attribs.id] : [])]);
+	return element("div", {
+		class: "segnavia-targets",
+		"aria-hidden": "true"
+	}, [...new Set(ids)].map((id) => element("a", { href: `#${id}` }, [text("​")])));
+}
+function duplicateIdIssues(sections, headings) {
+	const listed = new Set(headings.map((heading) => heading.id));
+	const occurrences = /* @__PURE__ */ new Map();
+	for (const section of sections) {
+		const ids = [section.info.id, ...elements(section.node.children).flatMap((node) => node.attribs.id ? [node.attribs.id] : [])];
+		for (const id of ids) occurrences.set(id, [...occurrences.get(id) ?? [], section.info]);
+	}
+	return [...occurrences].flatMap(([id, owners]) => {
+		if (!listed.has(id) || owners.length < 2) return [];
+		const named = owners.slice(0, 2).map((owner) => `${owner.title ?? owner.id} (${owner.id})`);
+		return [`destination id ${id} is used in both ${named[0]} and ${named[1]}; listed heading and generated destination ids must be unique`];
+	});
+}
+function reservedIdIssues(section) {
+	return elements(section.node.children).flatMap((node) => {
+		const id = node.attribs.id;
+		if (!id?.startsWith("segnavia-")) return [];
+		return [`authored id ${id} in ${section.info.title ?? section.info.id} (${section.info.id}) uses the reserved segnavia- prefix`];
+	});
+}
+function endnotesSection(book, paragraphs) {
+	const id = "segnavia-endnotes";
+	const titleValue = book.endnotes.title;
+	const titleNode = element("h1", {
+		class: "segnavia-endnotes-title",
+		dir: direction(book, titleValue)
+	}, [text(titleValue)]);
+	return {
+		info: {
+			id,
+			kind: "endnotes",
+			title: titleValue,
+			runningHead: titleValue
+		},
+		node: sectionNode("endnotes", id, [titleNode, ...paragraphs])
+	};
+}
+function withBlankPages(sections, blankBefore) {
+	const counts = /* @__PURE__ */ new Map();
+	for (const id of blankBefore ?? []) counts.set(id, (counts.get(id) ?? 0) + 1);
+	return sections.flatMap((section) => [...Array.from({ length: counts.get(section.info.id) ?? 0 }, () => element("div", {
+		class: "segnavia-blank",
+		"aria-hidden": "true"
+	})), section.node]);
+}
+function bookDocument(book, options = {}) {
+	const notes = noteState();
+	const sections = [];
+	const headings = [];
+	const issues = [];
+	let frontNumber = 0;
+	let contents;
+	for (const entry of book.frontMatter) {
+		if (isBookContents(entry)) {
+			contents = contentsSection(entry.title);
+			sections.push(contents);
+			continue;
+		}
+		frontNumber += 1;
+		const section = authoredSection(entry, `segnavia-front-${frontNumber}`);
+		sections.push(section);
+		issues.push(...reservedIdIssues(section));
+		if (entry.contents && entry.title !== void 0) headings.push({
+			id: section.info.id,
+			level: 1,
+			text: entry.title,
+			sectionId: section.info.id
+		});
+		transformNotes(section.node, book, 0, notes);
+	}
+	for (const [offset, chapter] of book.chapters.entries()) {
+		const number = offset + 1;
+		const section = chapterSection(chapter, number);
+		sections.push(section);
+		issues.push(...reservedIdIssues(section));
+		headings.push({
+			id: section.info.id,
+			level: 1,
+			text: chapter.title,
+			sectionId: section.info.id
+		}, ...chapterHeadings(section, number, book.contents.depth));
+		transformNotes(section.node, book, number, notes);
+	}
+	if (notes.endnotes) sections.push(endnotesSection(book, notes.endnoteParagraphs));
+	if (book.endnotes.placement === "chapter") issues.push("endnotes at the end of each chapter belong to paginated mode, which is not built yet");
+	if (contents) addContents(book, contents, headings, options);
+	const visible = sections.filter((section) => options.part === void 0 || bookPrintPart(section.info.kind) === options.part);
+	const document = parse("");
+	document.children = [
+		element("style", {}, [text(PRINT_CSS)]),
+		...options.targets ? [targetLinks(visible)] : [],
+		...withBlankPages(visible, options.blankBefore)
+	];
+	linkChildren(document);
+	return {
+		html: serialize(document),
+		sections: sections.map((section) => section.info),
+		headings,
+		issues: [
+			...noteMarkerIssues(book, {
+				footnotes: notes.footnotes,
+				endnotes: notes.endnotes
+			}),
+			...issues,
+			...duplicateIdIssues(sections, headings)
+		]
+	};
+}
+//#endregion
+export { bookDocument };

package/dist/book/dom.js

@@ -0,0 +1,46 @@
+import { Element, Text, isTag } from "domhandler";
+//#region src/book/dom.ts
+function element(name, attribs = {}, children = []) {
+	const node = new Element(name, attribs, children);
+	linkChildren(node);
+	return node;
+}
+function text(value) {
+	return new Text(value);
+}
+function linkChildren(parent) {
+	for (const [index, child] of parent.children.entries()) {
+		child.parent = parent;
+		child.prev = parent.children[index - 1] ?? null;
+		child.next = parent.children[index + 1] ?? null;
+		if (isTag(child)) linkChildren(child);
+	}
+}
+function replace(node, replacement) {
+	const parent = node.parent;
+	if (!parent) return;
+	const index = parent.children.indexOf(node);
+	if (index < 0) return;
+	parent.children[index] = replacement;
+	linkChildren(parent);
+}
+function insertAfter(node, addition) {
+	const parent = node.parent;
+	if (!parent) return;
+	const index = parent.children.indexOf(node);
+	if (index < 0) return;
+	parent.children.splice(index + 1, 0, addition);
+	linkChildren(parent);
+}
+function append(parent, addition) {
+	parent.children.push(addition);
+	linkChildren(parent);
+}
+function textContent(node, omitNotes = false) {
+	if ("data" in node) return node.data;
+	if (!("children" in node)) return "";
+	if (omitNotes && isTag(node) && node.name === "span" && node.attribs["data-note"] !== void 0) return "";
+	return node.children.map((child) => textContent(child, omitNotes)).join("");
+}
+//#endregion
+export { append, element, insertAfter, linkChildren, replace, text, textContent };

package/dist/book/notes.js

@@ -0,0 +1,112 @@
+import { elements } from "../sanitize.js";
+import { append, element, insertAfter, replace, text } from "./dom.js";
+import { isTag } from "domhandler";
+import { noteMarker } from "@segnavia/format";
+//#region src/book/notes.ts
+const BLOCKS = /* @__PURE__ */ new Set([
+	"p",
+	"li",
+	"blockquote",
+	"dd",
+	"dt",
+	"figcaption",
+	"td",
+	"th",
+	"pre",
+	"h1",
+	"h2",
+	"h3",
+	"h4",
+	"h5",
+	"h6",
+	"div",
+	"section",
+	"article"
+]);
+const INSIDE_BLOCKS = /* @__PURE__ */ new Set([
+	"li",
+	"dd",
+	"dt",
+	"td",
+	"th"
+]);
+function noteState() {
+	return {
+		footnotes: false,
+		endnotes: false,
+		globalFootnote: 0,
+		chapterFootnotes: /* @__PURE__ */ new Map(),
+		endnoteCount: 0,
+		endnoteParagraphs: []
+	};
+}
+function nearestBlock(note) {
+	for (let parent = note.parent; parent; parent = parent.parent) if (isTag(parent) && BLOCKS.has(parent.name)) return parent;
+}
+function markerLink(id, href, marker) {
+	return element("a", {
+		id,
+		href: `#${href}`
+	}, [element("bdi", {}, [text(marker)])]);
+}
+function noteParagraph(className, noteId, refId, marker, content) {
+	return element("p", {
+		class: className,
+		id: noteId
+	}, [
+		element("a", { href: `#${refId}` }, [element("bdi", {}, [text(marker)])]),
+		text(" "),
+		...content
+	]);
+}
+function footnoteNumber(book, chapter, state) {
+	if (book.footnotes.restart === "book") {
+		state.globalFootnote += 1;
+		return state.globalFootnote;
+	}
+	const number = (state.chapterFootnotes.get(chapter) ?? 0) + 1;
+	state.chapterFootnotes.set(chapter, number);
+	return number;
+}
+function addFootnoteBlock(block, paragraph, blocks) {
+	const existing = blocks.get(block);
+	if (existing) {
+		append(existing, paragraph);
+		return;
+	}
+	const notes = element("div", { class: "segnavia-footnotes" }, [paragraph]);
+	blocks.set(block, notes);
+	if (INSIDE_BLOCKS.has(block.name)) append(block, notes);
+	else insertAfter(block, notes);
+}
+function transformFootnote(note, book, chapter, state, blocks) {
+	const block = nearestBlock(note);
+	if (!block) return;
+	const number = footnoteNumber(book, chapter, state);
+	const noteId = `segnavia-fn-${chapter}-${number}`;
+	const refId = `segnavia-fnref-${chapter}-${number}`;
+	const marker = noteMarker(number, book.footnotes.marker);
+	const content = [...note.children];
+	replace(note, element("sup", { class: "segnavia-note-ref" }, [markerLink(refId, noteId, marker)]));
+	addFootnoteBlock(block, noteParagraph("segnavia-footnote", noteId, refId, marker, content), blocks);
+	state.footnotes = true;
+}
+function transformEndnote(note, book, state) {
+	state.endnoteCount += 1;
+	const number = state.endnoteCount;
+	const noteId = `segnavia-en-${number}`;
+	const refId = `segnavia-enref-${number}`;
+	const marker = noteMarker(number, book.endnotes.marker);
+	const content = [...note.children];
+	replace(note, element("sup", { class: "segnavia-note-ref" }, [markerLink(refId, noteId, marker)]));
+	state.endnoteParagraphs.push(noteParagraph("segnavia-endnote", noteId, refId, marker, content));
+	state.endnotes = true;
+}
+function transformNotes(root, book, chapter, state) {
+	const blocks = /* @__PURE__ */ new Map();
+	const notes = elements(root.children).filter((node) => node.name === "span" && (node.attribs["data-note"] === "footnote" || node.attribs["data-note"] === "endnote"));
+	for (const note of notes) if (note.attribs["data-note"] === "footnote") transformFootnote(note, book, chapter, state, blocks);
+	else transformEndnote(note, book, state);
+}
+//#endregion
+export { noteState, transformNotes };

package/dist/book/types.d.ts

@@ -0,0 +1,29 @@
+//#region src/book/types.d.ts
+export interface BookDocumentOptions {
+  pageNumbers?: Record<string, string>;
+  blankBefore?: string[];
+  part?: 'front' | 'body';
+  crossLinks?: boolean;
+  targets?: boolean;
+}
+export interface BookSectionInfo {
+  id: string;
+  kind: 'front' | 'contents' | 'chapter' | 'endnotes';
+  title?: string;
+  runningHead?: string;
+}
+export type BookPrintPart = 'front' | 'body';
+export declare function bookPrintPart(kind: BookSectionInfo['kind']): BookPrintPart;
+export interface BookHeading {
+  id: string;
+  level: number;
+  text: string;
+  sectionId: string;
+}
+export interface BookDocument {
+  html: string;
+  sections: BookSectionInfo[];
+  headings: BookHeading[];
+  issues: string[];
+}
+//#endregion

package/dist/book/types.js

@@ -0,0 +1,6 @@
+//#region src/book/types.ts
+function bookPrintPart(kind) {
+	return kind === "front" || kind === "contents" ? "front" : "body";
+}
+//#endregion
+export { bookPrintPart };

package/dist/index.d.ts

@@ -1,8 +1,10 @@
 import { REFUSED_ELEMENTS, SanitizeOptions, findActiveContent, sanitize } from "./sanitize.js";
 import { BINDINGS, FillOptions, Filled, bindingNames, fill, hasBindings } from "./bindings.js";
+import { BookDocument, BookDocumentOptions, BookHeading, BookPrintPart, BookSectionInfo, bookPrintPart } from "./book/types.js";
+import { bookDocument } from "./book/book-document.js";
 //#region src/index.d.ts
 export type HtmlKind = 'plain' | 'live-data' | 'scripted' | 'refused';
 export declare function classify(html: string, prefix: string): HtmlKind;
 export declare function searchText(html: string): string;
 //#endregion
-export { BINDINGS, type FillOptions, type Filled, REFUSED_ELEMENTS, type SanitizeOptions, bindingNames, fill, findActiveContent, hasBindings, sanitize };
+export { BINDINGS, type BookDocument, type BookDocumentOptions, type BookHeading, type BookPrintPart, type BookSectionInfo, type FillOptions, type Filled, REFUSED_ELEMENTS, type SanitizeOptions, bindingNames, bookDocument, bookPrintPart, fill, findActiveContent, hasBindings, sanitize };

package/dist/index.js

@@ -1,5 +1,7 @@
 import { REFUSED_ELEMENTS, elements, findActiveContent, parse, sanitize } from "./sanitize.js";
 import { BINDINGS, bindingNames, fill, hasBindings } from "./bindings.js";
+import { bookPrintPart } from "./book/types.js";
+import { bookDocument } from "./book/book-document.js";
 //#region src/index.ts
 function classify(html, prefix) {
 	const scripted = elements(parse(html).children).some((element) => element.name.toLowerCase() === "script");
@@ -12,4 +14,4 @@ function searchText(html) {
 	return html.replace(/<(script|style)\b[^>]*>[\s\S]*?<\/\1\s*>/gi, " ").replace(/<[^>]+>/g, " ").replace(/&nbsp;/g, " ").replace(/\s+/g, " ").trim();
 }
 //#endregion
-export { BINDINGS, REFUSED_ELEMENTS, bindingNames, classify, fill, findActiveContent, hasBindings, sanitize, searchText };
+export { BINDINGS, REFUSED_ELEMENTS, bindingNames, bookDocument, bookPrintPart, classify, fill, findActiveContent, hasBindings, sanitize, searchText };

package/dist/sanitize.d.ts

@@ -1,4 +1,4 @@
-import "domhandler";
+import { Element } from "domhandler";
 //#region src/sanitize.d.ts
 export declare const REFUSED_ELEMENTS: Set<string>;
 export interface SanitizeOptions {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@segnavia/html",
-  "version": "0.3.0",
+  "version": "0.5.0",
   "type": "module",
   "license": "MIT",
   "repository": {
@@ -26,6 +26,7 @@
     }
   },
   "dependencies": {
+    "@segnavia/format": "0.5.0",
     "htmlparser2": "^10.0.0",
     "domhandler": "^5.0.3",
     "dom-serializer": "^2.0.0"