diff options
Diffstat (limited to 'services/blog/nuldoc-src/djot')
| -rw-r--r-- | services/blog/nuldoc-src/djot/djot2ndoc.ts | 604 | ||||
| -rw-r--r-- | services/blog/nuldoc-src/djot/document.ts | 75 | ||||
| -rw-r--r-- | services/blog/nuldoc-src/djot/parse.ts | 33 | ||||
| -rw-r--r-- | services/blog/nuldoc-src/djot/to_html.ts | 499 |
4 files changed, 0 insertions, 1211 deletions
diff --git a/services/blog/nuldoc-src/djot/djot2ndoc.ts b/services/blog/nuldoc-src/djot/djot2ndoc.ts deleted file mode 100644 index 627e8d6..0000000 --- a/services/blog/nuldoc-src/djot/djot2ndoc.ts +++ /dev/null @@ -1,604 +0,0 @@ -import { - Block as DjotBlock, - BlockQuote as DjotBlockQuote, - BulletList as DjotBulletList, - CodeBlock as DjotCodeBlock, - Definition as DjotDefinition, - DefinitionList as DjotDefinitionList, - DefinitionListItem as DjotDefinitionListItem, - Delete as DjotDelete, - DisplayMath as DjotDisplayMath, - Div as DjotDiv, - Doc as DjotDoc, - DoubleQuoted as DjotDoubleQuoted, - Email as DjotEmail, - Emph as DjotEmph, - FootnoteReference as DjotFootnoteReference, - HardBreak as DjotHardBreak, - Heading as DjotHeading, - Image as DjotImage, - Inline as DjotInline, - InlineMath as DjotInlineMath, - Insert as DjotInsert, - Link as DjotLink, - ListItem as DjotListItem, - Mark as DjotMark, - NonBreakingSpace as DjotNonBreakingSpace, - OrderedList as DjotOrderedList, - Para as DjotPara, - RawBlock as DjotRawBlock, - RawInline as DjotRawInline, - Section as DjotSection, - SingleQuoted as DjotSingleQuoted, - SmartPunctuation as DjotSmartPunctuation, - SoftBreak as DjotSoftBreak, - Span as DjotSpan, - Str as DjotStr, - Strong as DjotStrong, - Subscript as DjotSubscript, - Superscript as DjotSuperscript, - Symb as DjotSymb, - Table as DjotTable, - TaskList as DjotTaskList, - TaskListItem as DjotTaskListItem, - Term as DjotTerm, - ThematicBreak as DjotThematicBreak, - Url as DjotUrl, - Verbatim as DjotVerbatim, -} from "@djot/djot"; -import { addClass, elem, Element, Node, rawHTML, text } from "../dom.ts"; - -function processBlock(node: DjotBlock): Element { - switch (node.tag) { - case "section": - return processSection(node); - case "para": - return processPara(node); - case "heading": - return processHeading(node); - case "thematic_break": - return processThematicBreak(node); - case "block_quote": - return processBlockQuote(node); - case "code_block": - return processCodeBlock(node); - case "bullet_list": - return processBulletList(node); - case "ordered_list": - return processOrderedList(node); - case "task_list": - return processTaskList(node); - case "definition_list": - return processDefinitionList(node); - case "table": - return processTable(node); - case "div": - return processDiv(node); - case "raw_block": - return processRawBlock(node); - } -} - -function processSection(node: DjotSection): Element { - return elem( - "section", - node.attributes, - ...node.children.map(processBlock), - ); -} - -function processPara(node: DjotPara): Element { - return elem( - "p", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processHeading(node: DjotHeading): Element { - return elem("h", node.attributes, ...node.children.map(processInline)); -} - -function processThematicBreak(node: DjotThematicBreak): Element { - return elem("hr", node.attributes); -} - -function processBlockQuote(node: DjotBlockQuote): Element { - return elem( - "blockquote", - node.attributes, - ...node.children.map(processBlock), - ); -} - -function processCodeBlock(node: DjotCodeBlock): Element { - const attributes = node.attributes || {}; - if (node.lang) { - attributes.language = node.lang; - } - if (node.attributes?.filename) { - attributes.filename = node.attributes.filename; - } - if (node.attributes?.numbered) { - attributes.numbered = "true"; - } - return elem("codeblock", attributes, text(node.text)); -} - -function processBulletList(node: DjotBulletList): Element { - const attributes = node.attributes || {}; - attributes.__tight = node.tight ? "true" : "false"; - return elem("ul", attributes, ...node.children.map(processListItem)); -} - -function processOrderedList(node: DjotOrderedList): Element { - const attributes = node.attributes || {}; - attributes.__tight = node.tight ? "true" : "false"; - if (node.start !== undefined && node.start !== 1) { - attributes.start = node.start.toString(); - } - return elem("ol", attributes, ...node.children.map(processListItem)); -} - -function processTaskList(node: DjotTaskList): Element { - const attributes = node.attributes || {}; - attributes.type = "task"; - attributes.__tight = node.tight ? "true" : "false"; - return elem("ul", attributes, ...node.children.map(processTaskListItem)); -} - -function processListItem(node: DjotListItem): Element { - return elem( - "li", - node.attributes, - ...node.children.map(processBlock), - ); -} - -function processTaskListItem(node: DjotTaskListItem): Element { - const attributes = node.attributes || {}; - attributes.checked = node.checkbox === "checked" ? "true" : "false"; - return elem("li", attributes, ...node.children.map(processBlock)); -} - -function processDefinitionList(node: DjotDefinitionList): Element { - return elem( - "dl", - node.attributes, - ...node.children.flatMap(processDefinitionListItem), - ); -} - -function processDefinitionListItem(node: DjotDefinitionListItem): Element[] { - return [ - processTerm(node.children[0]), - processDefinition(node.children[1]), - ]; -} - -function processTerm(node: DjotTerm): Element { - return elem( - "dt", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processDefinition(node: DjotDefinition): Element { - return elem( - "dd", - node.attributes, - ...node.children.map(processBlock), - ); -} - -function processTable(node: DjotTable): Element { - // Tables in Djot have a caption as first child and then rows - // For now, we'll create a basic table structure and ignore caption - const tableElement = elem("table", node.attributes); - - // Process caption if it exists (first child) - if (node.children.length > 0 && node.children[0].tag === "caption") { - const caption = elem( - "caption", - undefined, - ...node.children[0].children.map(processInline), - ); - tableElement.children.push(caption); - } - - // Group rows into thead, tbody based on head property - const headerRows: Element[] = []; - const bodyRows: Element[] = []; - - // Start from index 1 to skip caption - for (let i = 1; i < node.children.length; i++) { - const row = node.children[i]; - if (row.tag === "row") { - const rowElement = elem( - "tr", - row.attributes, - ...row.children.map((cell) => { - const cellAttributes = cell.attributes || {}; - // Set alignment attribute if needed - if (cell.align !== "default") { - cellAttributes.align = cell.align; - } - return elem( - cell.head ? "th" : "td", - cellAttributes, - ...cell.children.map(processInline), - ); - }), - ); - - if (row.head) { - headerRows.push(rowElement); - } else { - bodyRows.push(rowElement); - } - } - } - - // Add thead and tbody if needed - if (headerRows.length > 0) { - tableElement.children.push(elem("thead", undefined, ...headerRows)); - } - - if (bodyRows.length > 0) { - tableElement.children.push(elem("tbody", undefined, ...bodyRows)); - } - - return tableElement; -} - -function processInline(node: DjotInline): Node { - switch (node.tag) { - case "str": - return processStr(node); - case "soft_break": - return processSoftBreak(node); - case "hard_break": - return processHardBreak(node); - case "verbatim": - return processVerbatim(node); - case "emph": - return processEmph(node); - case "strong": - return processStrong(node); - case "link": - return processLink(node); - case "image": - return processImage(node); - case "mark": - return processMark(node); - case "superscript": - return processSuperscript(node); - case "subscript": - return processSubscript(node); - case "insert": - return processInsert(node); - case "delete": - return processDelete(node); - case "email": - return processEmail(node); - case "footnote_reference": - return processFootnoteReference(node); - case "url": - return processUrl(node); - case "span": - return processSpan(node); - case "inline_math": - return processInlineMath(node); - case "display_math": - return processDisplayMath(node); - case "non_breaking_space": - return processNonBreakingSpace(node); - case "symb": - return processSymb(node); - case "raw_inline": - return processRawInline(node); - case "double_quoted": - return processDoubleQuoted(node); - case "single_quoted": - return processSingleQuoted(node); - case "smart_punctuation": - return processSmartPunctuation(node); - } -} - -function processStr(node: DjotStr): Node { - return text(node.text); -} - -function processSoftBreak(_node: DjotSoftBreak): Node { - return text("\n"); -} - -function processHardBreak(_node: DjotHardBreak): Node { - return elem("br"); -} - -function processVerbatim(node: DjotVerbatim): Element { - return elem("code", node.attributes, text(node.text)); -} - -function processEmph(node: DjotEmph): Element { - return elem( - "em", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processStrong(node: DjotStrong): Element { - return elem( - "strong", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processLink(node: DjotLink): Element { - const attributes = node.attributes || {}; - if (node.destination !== undefined) { - attributes.href = node.destination; - } - return elem("a", attributes, ...node.children.map(processInline)); -} - -function processImage(node: DjotImage): Element { - const attributes = node.attributes || {}; - if (node.destination !== undefined) { - attributes.src = node.destination; - } - - // Alt text is derived from children in Djot - const alt = node.children - .map((child) => { - if (child.tag === "str") { - return child.text; - } - return ""; - }) - .join(""); - - if (alt) { - attributes.alt = alt; - } - - return elem("img", attributes); -} - -function processMark(node: DjotMark): Element { - return elem( - "mark", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processSuperscript(node: DjotSuperscript): Element { - return elem( - "sup", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processSubscript(node: DjotSubscript): Element { - return elem( - "sub", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processInsert(node: DjotInsert): Element { - return elem( - "ins", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processDelete(node: DjotDelete): Element { - return elem( - "del", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processEmail(node: DjotEmail): Element { - return elem("email", node.attributes, text(node.text)); -} - -function processFootnoteReference(node: DjotFootnoteReference): Element { - return elem("footnoteref", { reference: node.text }); -} - -function processUrl(node: DjotUrl): Element { - const e = elem( - "a", - { - href: node.text, - ...node.attributes, - }, - text(node.text), - ); - addClass(e, "url"); - return e; -} - -function processSpan(node: DjotSpan): Element { - return elem( - "span", - node.attributes, - ...node.children.map(processInline), - ); -} - -function processInlineMath(node: DjotInlineMath): Element { - // For inline math, we'll wrap it in a span with a class - return elem( - "span", - { - class: "math inline", - ...node.attributes, - }, - text(node.text), - ); -} - -function processDisplayMath(node: DjotDisplayMath): Element { - // For display math, we'll wrap it in a div with a class - return elem( - "div", - { - class: "math display", - ...node.attributes, - }, - text(node.text), - ); -} - -function processNonBreakingSpace(_node: DjotNonBreakingSpace): Node { - return text("\u00A0"); // Unicode non-breaking space -} - -function processSymb(node: DjotSymb): Node { - // Map symbol aliases to their Unicode characters - const symbolMap: Record<string, string> = { - "->": "→", - "<-": "←", - "<->": "↔", - "=>": "⇒", - "<=": "⇐", - "<=>": "⇔", - "--": "–", // en dash - "---": "—", // em dash - "...": "…", // ellipsis - // Add more symbol mappings as needed - }; - - const symbolText = symbolMap[node.alias] || node.alias; - - return text(symbolText); -} - -function processRawInline(node: DjotRawInline): Node { - // If the format is HTML, return as raw HTML - if (node.format === "html" || node.format === "HTML") { - return rawHTML(node.text); - } - - // For other formats, just return as text - return text(node.text); -} - -function processDoubleQuoted(node: DjotDoubleQuoted): Node { - const children = node.children.map(processInline); - const attributes = node.attributes || {}; - - if ( - children.length === 1 && children[0].kind === "text" && - Object.keys(attributes).length === 0 - ) { - const content = children[0].content; - return text(`\u201C${content}\u201D`); - } else { - return elem("span", node.attributes, ...children); - } -} - -function processSingleQuoted(node: DjotSingleQuoted): Node { - const children = node.children.map(processInline); - const attributes = node.attributes || {}; - - if ( - children.length === 1 && children[0].kind === "text" && - Object.keys(attributes).length === 0 - ) { - const content = children[0].content; - return text(`\u2018${content}\u2019`); - } else { - return elem("span", node.attributes, ...children); - } -} - -function processSmartPunctuation(node: DjotSmartPunctuation): Node { - // Map smart punctuation types to Unicode characters - const punctuationMap: Record<string, string> = { - left_single_quote: "\u2018", // ' - right_single_quote: "\u2019", // ' - left_double_quote: "\u201C", // " - right_double_quote: "\u201D", // " - ellipses: "\u2026", // … - em_dash: "\u2014", // — - en_dash: "\u2013", // – - }; - - return text(punctuationMap[node.type] || node.text); -} - -function processDiv(node: DjotDiv): Element { - if (node.attributes?.class === "note") { - delete node.attributes.class; - return elem( - "note", - node.attributes, - ...node.children.map(processBlock), - ); - } - - if (node.attributes?.class === "edit") { - delete node.attributes.class; - return elem( - "note", - node.attributes, - ...node.children.map(processBlock), - ); - } - - return elem( - "div", - node.attributes, - ...node.children.map(processBlock), - ); -} - -function processRawBlock(node: DjotRawBlock): Element { - // If the format is HTML, wrap the HTML content in a div - if (node.format === "html" || node.format === "HTML") { - return elem("div", { class: "raw-html" }, rawHTML(node.text)); - } - - // For other formats, wrap in a pre tag - return elem("pre", { "data-format": node.format }, text(node.text)); -} - -export function djot2ndoc(doc: DjotDoc): Element { - const children: Node[] = []; - for (const child of doc.children) { - children.push(processBlock(child)); - } - - // Process footnotes if any exist - if (doc.footnotes && Object.keys(doc.footnotes).length > 0) { - const footnoteSection = elem("section", { class: "footnotes" }); - - for (const [id, footnote] of Object.entries(doc.footnotes)) { - const footnoteElement = elem( - "footnote", - { id }, - ...footnote.children.map(processBlock), - ); - footnoteSection.children.push(footnoteElement); - } - - children.push(footnoteSection); - } - - return elem("__root__", undefined, elem("article", undefined, ...children)); -} diff --git a/services/blog/nuldoc-src/djot/document.ts b/services/blog/nuldoc-src/djot/document.ts deleted file mode 100644 index 3e8cd92..0000000 --- a/services/blog/nuldoc-src/djot/document.ts +++ /dev/null @@ -1,75 +0,0 @@ -import { Doc as DjotDoc } from "@djot/djot"; -import { join } from "@std/path"; -import { z } from "zod/mod.ts"; -import { Config } from "../config.ts"; -import { Element } from "../dom.ts"; -import { Revision, stringToDate } from "../revision.ts"; -import { djot2ndoc } from "./djot2ndoc.ts"; - -export const PostMetadataSchema = z.object({ - article: z.object({ - uuid: z.string(), - title: z.string(), - description: z.string(), - tags: z.array(z.string()), - toc: z.boolean().optional(), - revisions: z.array(z.object({ - date: z.string(), - remark: z.string(), - isInternal: z.boolean().optional(), - })), - }), -}); - -export type PostMetadata = z.infer<typeof PostMetadataSchema>; - -export type TocEntry = { - id: string; - text: string; - level: number; - children: TocEntry[]; -}; - -export type TocRoot = { - entries: TocEntry[]; -}; - -export type Document = { - root: Element; - sourceFilePath: string; - uuid: string; - link: string; - title: string; - description: string; // TODO: should it be markup text? - tags: string[]; - revisions: Revision[]; - toc?: TocRoot; - isTocEnabled: boolean; -}; - -export function createNewDocumentFromDjotDocument( - root: DjotDoc, - meta: PostMetadata, - sourceFilePath: string, - config: Config, -): Document { - const cwd = Deno.cwd(); - const contentDir = join(cwd, config.locations.contentDir); - const link = sourceFilePath.replace(contentDir, "").replace(".xml", "/"); - return { - root: djot2ndoc(root), - sourceFilePath, - uuid: meta.article.uuid, - link: link, - title: meta.article.title, - description: meta.article.description, - tags: meta.article.tags, - revisions: meta.article.revisions.map((r, i) => ({ - number: i, - date: stringToDate(r.date), - remark: r.remark, - isInternal: !!r.isInternal, - })), - isTocEnabled: meta.article.toc !== false, - }; -} diff --git a/services/blog/nuldoc-src/djot/parse.ts b/services/blog/nuldoc-src/djot/parse.ts deleted file mode 100644 index c79a670..0000000 --- a/services/blog/nuldoc-src/djot/parse.ts +++ /dev/null @@ -1,33 +0,0 @@ -import { parse as parseDjot } from "@djot/djot"; -import { parse as parseToml } from "@std/toml"; -import { Config } from "../config.ts"; -import { - createNewDocumentFromDjotDocument, - Document, - PostMetadata, - PostMetadataSchema, -} from "./document.ts"; -import toHtml from "./to_html.ts"; - -export async function parseDjotFile( - filePath: string, - config: Config, -): Promise<Document> { - try { - const fileContent = await Deno.readTextFile(filePath); - const [, frontmatter, ...rest] = fileContent.split(/^---$/m); - const meta = parseMetadata(frontmatter); - const root = parseDjot(rest.join("\n")); - const doc = createNewDocumentFromDjotDocument(root, meta, filePath, config); - return await toHtml(doc); - } catch (e) { - if (e instanceof Error) { - e.message = `${e.message} in ${filePath}`; - } - throw e; - } -} - -function parseMetadata(s: string): PostMetadata { - return PostMetadataSchema.parse(parseToml(s)); -} diff --git a/services/blog/nuldoc-src/djot/to_html.ts b/services/blog/nuldoc-src/djot/to_html.ts deleted file mode 100644 index 8219b74..0000000 --- a/services/blog/nuldoc-src/djot/to_html.ts +++ /dev/null @@ -1,499 +0,0 @@ -import { BundledLanguage, bundledLanguages, codeToHtml } from "shiki"; -import { Document, TocEntry } from "./document.ts"; -import { NuldocError } from "../errors.ts"; -import { - addClass, - elem, - Element, - forEachChild, - forEachChildRecursively, - forEachChildRecursivelyAsync, - forEachElementOfType, - innerText, - Node, - processTextNodesInElement, - RawHTML, - rawHTML, - Text, - text, -} from "../dom.ts"; - -export default async function toHtml(doc: Document): Promise<Document> { - mergeConsecutiveTextNodes(doc); - removeUnnecessaryTextNode(doc); - transformLinkLikeToAnchorElement(doc); - transformSectionIdAttribute(doc); - setSectionTitleAnchor(doc); - transformSectionTitleElement(doc); - transformNoteElement(doc); - addAttributesToExternalLinkElement(doc); - traverseFootnotes(doc); - removeUnnecessaryParagraphNode(doc); - await transformAndHighlightCodeBlockElement(doc); - mergeConsecutiveTextNodes(doc); - generateTableOfContents(doc); - removeTocAttributes(doc); - return doc; -} - -function mergeConsecutiveTextNodes(doc: Document) { - forEachChildRecursively(doc.root, (n) => { - if (n.kind !== "element") { - return; - } - - const newChildren: Node[] = []; - let currentTextContent = ""; - - for (const child of n.children) { - if (child.kind === "text") { - currentTextContent += child.content; - } else { - if (currentTextContent !== "") { - newChildren.push(text(currentTextContent)); - currentTextContent = ""; - } - newChildren.push(child); - } - } - - if (currentTextContent !== "") { - newChildren.push(text(currentTextContent)); - } - - n.children = newChildren; - }); -} - -function removeUnnecessaryTextNode(doc: Document) { - forEachChildRecursively(doc.root, (n) => { - if (n.kind !== "element") { - return; - } - - let changed = true; - while (changed) { - changed = false; - if (n.children.length === 0) { - break; - } - const firstChild = n.children[0]; - if (firstChild.kind === "text" && firstChild.content.trim() === "") { - n.children.shift(); - changed = true; - } - if (n.children.length === 0) { - break; - } - const lastChild = n.children[n.children.length - 1]; - if (lastChild.kind === "text" && lastChild.content.trim() === "") { - n.children.pop(); - changed = true; - } - } - }); -} - -function transformLinkLikeToAnchorElement(doc: Document) { - forEachChildRecursively(doc.root, (n) => { - if ( - n.kind !== "element" || n.name === "a" || n.name === "code" || - n.name === "codeblock" - ) { - return; - } - - processTextNodesInElement(n, (content) => { - const nodes: Node[] = []; - let restContent = content; - while (restContent !== "") { - const match = /^(.*?)(https?:\/\/[^ \n]+)(.*)$/s.exec(restContent); - if (!match) { - nodes.push(text(restContent)); - restContent = ""; - break; - } - const [_, prefix, url, suffix] = match; - nodes.push(text(prefix)); - nodes.push(elem("a", { href: url, class: "url" }, text(url))); - restContent = suffix; - } - return nodes; - }); - }); -} - -function transformSectionIdAttribute(doc: Document) { - const sectionStack: string[] = []; - const usedIds = new Set<string>(); - - const processNode = (n: Node) => { - if (n.kind !== "element") { - return; - } - - if (n.name === "section") { - const idAttr = n.attributes.id; - if (!idAttr) { - return; - } - - let newId: string; - if (sectionStack.length === 0) { - newId = `section--${idAttr}`; - } else { - newId = `section--${sectionStack.join("--")}--${idAttr}`; - } - - if (usedIds.has(newId)) { - throw new NuldocError( - `[nuldoc.tohtml] Duplicate section ID: ${newId}`, - ); - } - - usedIds.add(newId); - n.attributes.id = newId; - sectionStack.push(idAttr); - - forEachChild(n, processNode); - - sectionStack.pop(); - } else { - forEachChild(n, processNode); - } - }; - - forEachChild(doc.root, processNode); -} - -function setSectionTitleAnchor(doc: Document) { - const sectionStack: Element[] = []; - const g = (c: Node) => { - if (c.kind !== "element") { - return; - } - - if (c.name === "section") { - sectionStack.push(c); - } - forEachChild(c, g); - if (c.name === "section") { - sectionStack.pop(); - } - if (c.name === "h") { - const currentSection = sectionStack[sectionStack.length - 1]; - if (!currentSection) { - throw new NuldocError( - "[nuldoc.tohtml] <h> element must be inside <section>", - ); - } - const sectionId = currentSection.attributes.id; - const aElement = elem("a", undefined, ...c.children); - aElement.attributes.href = `#${sectionId}`; - c.children = [aElement]; - } - }; - forEachChild(doc.root, g); -} - -function transformSectionTitleElement(doc: Document) { - let sectionLevel = 1; - const g = (c: Node) => { - if (c.kind !== "element") { - return; - } - - if (c.name === "section") { - sectionLevel += 1; - c.attributes.__sectionLevel = sectionLevel.toString(); - } - forEachChild(c, g); - if (c.name === "section") { - sectionLevel -= 1; - } - if (c.name === "h") { - c.name = `h${sectionLevel}`; - } - }; - forEachChild(doc.root, g); -} - -function transformNoteElement(doc: Document) { - forEachElementOfType(doc.root, "note", (n) => { - const editatAttr = n.attributes?.editat; - const operationAttr = n.attributes?.operation; - const isEditBlock = editatAttr && operationAttr; - - const labelElement = elem( - "div", - { class: "admonition-label" }, - text(isEditBlock ? `${editatAttr} ${operationAttr}` : "NOTE"), - ); - const contentElement = elem( - "div", - { class: "admonition-content" }, - ...n.children, - ); - n.name = "div"; - addClass(n, "admonition"); - n.children = [labelElement, contentElement]; - }); -} - -function addAttributesToExternalLinkElement(doc: Document) { - forEachElementOfType(doc.root, "a", (n) => { - const href = n.attributes.href ?? ""; - if (!href.startsWith("http")) { - return; - } - n.attributes.target = "_blank"; - n.attributes.rel = "noreferrer"; - }); -} - -function traverseFootnotes(doc: Document) { - let footnoteCounter = 0; - const footnoteMap = new Map<string, number>(); - - forEachElementOfType(doc.root, "footnoteref", (n) => { - const reference = n.attributes.reference; - if (!reference) { - return; - } - - let footnoteNumber: number; - if (footnoteMap.has(reference)) { - footnoteNumber = footnoteMap.get(reference)!; - } else { - footnoteNumber = ++footnoteCounter; - footnoteMap.set(reference, footnoteNumber); - } - - n.name = "sup"; - delete n.attributes.reference; - n.attributes.class = "footnote"; - n.children = [ - elem( - "a", - { - id: `footnoteref--${reference}`, - class: "footnote", - href: `#footnote--${reference}`, - }, - text(`[${footnoteNumber}]`), - ), - ]; - }); - - forEachElementOfType(doc.root, "footnote", (n) => { - const id = n.attributes.id; - if (!id || !footnoteMap.has(id)) { - n.name = "span"; - n.children = []; - return; - } - - const footnoteNumber = footnoteMap.get(id)!; - - n.name = "div"; - delete n.attributes.id; - n.attributes.class = "footnote"; - n.attributes.id = `footnote--${id}`; - - n.children = [ - elem( - "a", - { href: `#footnoteref--${id}` }, - text(`${footnoteNumber}. `), - ), - ...n.children, - ]; - }); -} - -function removeUnnecessaryParagraphNode(doc: Document) { - forEachChildRecursively(doc.root, (n) => { - if (n.kind !== "element" || (n.name !== "ul" && n.name !== "ol")) { - return; - } - - const isTight = n.attributes.__tight === "true"; - if (!isTight) { - return; - } - - for (const child of n.children) { - if (child.kind !== "element" || child.name !== "li") { - continue; - } - const newGrandChildren: Node[] = []; - for (const grandChild of child.children) { - if (grandChild.kind === "element" && grandChild.name === "p") { - newGrandChildren.push(...grandChild.children); - } else { - newGrandChildren.push(grandChild); - } - } - child.children = newGrandChildren; - } - }); -} - -async function transformAndHighlightCodeBlockElement(doc: Document) { - await forEachChildRecursivelyAsync(doc.root, async (n) => { - if (n.kind !== "element" || n.name !== "codeblock") { - return; - } - - const language = n.attributes.language || "text"; - const filename = n.attributes.filename; - const numbered = n.attributes.numbered; - const sourceCodeNode = n.children[0] as Text | RawHTML; - const sourceCode = sourceCodeNode.kind === "text" - ? sourceCodeNode.content.trimEnd() - : sourceCodeNode.html.trimEnd(); - - const highlighted = await codeToHtml(sourceCode, { - lang: language in bundledLanguages ? language as BundledLanguage : "text", - theme: "github-light", - colorReplacements: { - "#fff": "#f5f5f5", - }, - }); - - n.name = "div"; - n.attributes.class = "codeblock"; - delete n.attributes.language; - - if (numbered === "true") { - delete n.attributes.numbered; - addClass(n, "numbered"); - } - if (filename) { - delete n.attributes.filename; - - n.children = [ - elem("div", { class: "filename" }, text(filename)), - rawHTML(highlighted), - ]; - } else { - if (sourceCodeNode.kind === "text") { - n.children[0] = rawHTML(highlighted); - } else { - sourceCodeNode.html = highlighted; - } - } - }); -} - -function generateTableOfContents(doc: Document) { - if (!doc.isTocEnabled) { - return; - } - const tocEntries: TocEntry[] = []; - const stack: TocEntry[] = []; - const excludedLevels: number[] = []; // Track levels to exclude - - const processNode = (node: Node) => { - if (node.kind !== "element") { - return; - } - - const match = node.name.match(/^h(\d+)$/); - if (match) { - const level = parseInt(match[1]); - - let parentSection: Element | null = null; - const findParentSection = (n: Node, target: Node): Element | null => { - if (n.kind !== "element") return null; - - for (const child of n.children) { - if (child === target && n.name === "section") { - return n; - } - const result = findParentSection(child, target); - if (result) return result; - } - return null; - }; - - parentSection = findParentSection(doc.root, node); - if (!parentSection) return; - - // Check if this section has toc=false attribute - const tocAttribute = parentSection.attributes.toc; - if (tocAttribute === "false") { - // Add this level to excluded levels and remove deeper levels - excludedLevels.length = 0; - excludedLevels.push(level); - return; - } - - // Check if this header should be excluded based on parent exclusion - const shouldExclude = excludedLevels.some((excludedLevel) => - level > excludedLevel - ); - if (shouldExclude) { - return; - } - - // Clean up excluded levels that are now at same or deeper level - while ( - excludedLevels.length > 0 && - excludedLevels[excludedLevels.length - 1] >= level - ) { - excludedLevels.pop(); - } - - const sectionId = parentSection.attributes.id; - if (!sectionId) return; - - let headingText = ""; - for (const child of node.children) { - if (child.kind === "element" && child.name === "a") { - headingText = innerText(child); - } - } - - const entry: TocEntry = { - id: sectionId, - text: headingText, - level: level, - children: [], - }; - - while (stack.length > 0 && stack[stack.length - 1].level >= level) { - stack.pop(); - } - - if (stack.length === 0) { - tocEntries.push(entry); - } else { - stack[stack.length - 1].children.push(entry); - } - - stack.push(entry); - } - - forEachChild(node, processNode); - }; - - forEachChild(doc.root, processNode); - - // Don't generate TOC if there's only one top-level section with no children - if (tocEntries.length === 1 && tocEntries[0].children.length === 0) { - return; - } - - doc.toc = { - entries: tocEntries, - }; -} - -function removeTocAttributes(doc: Document) { - forEachChildRecursively(doc.root, (node) => { - if (node.kind === "element" && node.name === "section") { - delete node.attributes.toc; - } - }); -} |
