3 files changed, 448 insertions, 0 deletions
diff --git a/vhosts/blog/nuldoc-src/docbook/document.ts b/vhosts/blog/nuldoc-src/docbook/document.ts
new file mode 100644
index 00000000..677c8275
--- /dev/null
+++ b/vhosts/blog/nuldoc-src/docbook/document.ts
@@ -0,0 +1,108 @@
+import { join } from "std/path/mod.ts";
+import { Config } from "../config.ts";
+import { DocBookError } from "../errors.ts";
+import { Revision, stringToDate } from "../revision.ts";
+import {
+  Element,
+  findChildElements,
+  findFirstChildElement,
+  innerText,
+} from "../dom.ts";
+
+export type Document = {
+  root: Element;
+  sourceFilePath: string;
+  link: string;
+  title: string;
+  summary: string; // TODO: should it be markup text?
+  tags: string[];
+  revisions: Revision[];
+};
+
+export function createNewDocumentFromRootElement(
+  root: Element,
+  sourceFilePath: string,
+  config: Config,
+): Document {
+  const article = findFirstChildElement(root, "article");
+  if (!article) {
+    throw new DocBookError(
+      `[docbook.new] <article> element not found`,
+    );
+  }
+  const info = findFirstChildElement(article, "info");
+  if (!info) {
+    throw new DocBookError(
+      `[docbook.new] <info> element not found`,
+    );
+  }
+
+  const titleElement = findFirstChildElement(info, "title");
+  if (!titleElement) {
+    throw new DocBookError(
+      `[docbook.new] <title> element not found`,
+    );
+  }
+  const title = innerText(titleElement).trim();
+  const abstractElement = findFirstChildElement(info, "abstract");
+  if (!abstractElement) {
+    throw new DocBookError(
+      `[docbook.new] <abstract> element not found`,
+    );
+  }
+  const summary = innerText(abstractElement).trim();
+  const keywordsetElement = findFirstChildElement(info, "keywordset");
+  let tags: string[];
+  if (!keywordsetElement) {
+    tags = [];
+  } else {
+    tags = findChildElements(keywordsetElement, "keyword").map((x) =>
+      innerText(x).trim()
+    );
+  }
+  const revhistoryElement = findFirstChildElement(info, "revhistory");
+  if (!revhistoryElement) {
+    throw new DocBookError(
+      `[docbook.new] <revhistory> element not found`,
+    );
+  }
+  const revisions = findChildElements(revhistoryElement, "revision").map(
+    (x, i) => {
+      const dateElement = findFirstChildElement(x, "date");
+      if (!dateElement) {
+        throw new DocBookError(
+          `[docbook.new] <date> element not found`,
+        );
+      }
+      const revremarkElement = findFirstChildElement(x, "revremark");
+      if (!revremarkElement) {
+        throw new DocBookError(
+          `[docbook.new] <revremark> element not found`,
+        );
+      }
+      return {
+        number: i + 1,
+        date: stringToDate(innerText(dateElement).trim()),
+        remark: innerText(revremarkElement).trim(),
+      };
+    },
+  );
+  if (revisions.length === 0) {
+    throw new DocBookError(
+      `[docbook.new] <revision> element not found`,
+    );
+  }
+
+  const cwd = Deno.cwd();
+  const contentDir = join(cwd, config.locations.contentDir);
+  const link = sourceFilePath.replace(contentDir, "").replace(".xml", "/");
+  return {
+    root: root,
+    sourceFilePath: sourceFilePath,
+    link: link,
+    title: title,
+    summary: summary,
+    tags: tags,
+    revisions: revisions,
+  };
+}
diff --git a/vhosts/blog/nuldoc-src/docbook/parse.ts b/vhosts/blog/nuldoc-src/docbook/parse.ts
new file mode 100644
index 00000000..bce317e6
--- /dev/null
+++ b/vhosts/blog/nuldoc-src/docbook/parse.ts
@@ -0,0 +1,21 @@
+import { Config } from "../config.ts";
+import { parseXmlFile } from "../xml.ts";
+import { DocBookError, XmlParseError } from "../errors.ts";
+import { createNewDocumentFromRootElement, Document } from "./document.ts";
+import toHtml from "./to_html.ts";
+
+export async function parseDocBookFile(
+  filePath: string,
+  config: Config,
+): Promise<Document> {
+  try {
+    const root = await parseXmlFile(filePath);
+    const doc = createNewDocumentFromRootElement(root, filePath, config);
+    return toHtml(doc);
+  } catch (e) {
+    if (e instanceof DocBookError || e instanceof XmlParseError) {
+      e.message = `${e.message} in ${filePath}`;
+    }
+    throw e;
+  }
+}
diff --git a/vhosts/blog/nuldoc-src/docbook/to_html.ts b/vhosts/blog/nuldoc-src/docbook/to_html.ts
new file mode 100644
index 00000000..4add912c
--- /dev/null
+++ b/vhosts/blog/nuldoc-src/docbook/to_html.ts
@@ -0,0 +1,319 @@
+// @deno-types="../types/highlight-js.d.ts"
+import hljs from "npm:highlight.js";
+import { Document } from "./document.ts";
+import { DocBookError } from "../errors.ts";
+import {
+  addClass,
+  Element,
+  findFirstChildElement,
+  forEachChild,
+  forEachChildRecursively,
+  Node,
+  RawHTML,
+  removeChildElements,
+  Text,
+} from "../dom.ts";
+
+export default function toHtml(doc: Document): Document {
+  removeArticleInfo(doc);
+  removeArticleAttributes(doc);
+  removeUnnecessaryTextNode(doc);
+  transformElementNames(doc, "emphasis", "em");
+  transformElementNames(doc, "informaltable", "table");
+  transformElementNames(doc, "itemizedlist", "ul");
+  transformElementNames(doc, "link", "a");
+  transformElementNames(doc, "listitem", "li");
+  transformElementNames(doc, "literal", "code");
+  transformElementNames(doc, "orderedlist", "ol");
+  transformElementNames(doc, "para", "p");
+  transformElementNames(doc, "superscript", "sup");
+  transformAttributeNames(doc, "xml:id", "id");
+  transformAttributeNames(doc, "xl:href", "href");
+  transformSectionIdAttribute(doc);
+  setSectionTitleAnchor(doc);
+  transformSectionTitleElement(doc);
+  transformProgramListingElement(doc);
+  transformLiteralLayoutElement(doc);
+  transformNoteElement(doc);
+  setDefaultLangAttribute(doc);
+  traverseFootnotes(doc);
+  highlightPrograms(doc);
+  return doc;
+}
+
+function removeArticleInfo(doc: Document) {
+  const article = findFirstChildElement(doc.root, "article");
+  if (!article) {
+    throw new DocBookError(
+      `[docbook.tohtml] <article> element not found`,
+    );
+  }
+  removeChildElements(article, "info");
+}
+
+function removeArticleAttributes(doc: Document) {
+  const article = findFirstChildElement(doc.root, "article");
+  if (!article) {
+    throw new DocBookError(
+      `[docbook.tohtml] <article> element not found`,
+    );
+  }
+  article.attributes.delete("xmlns");
+  article.attributes.delete("xmlns:xl");
+  article.attributes.delete("version");
+}
+
+function removeUnnecessaryTextNode(doc: Document) {
+  const g = (n: Node) => {
+    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;
+      }
+    }
+
+    forEachChild(n, g);
+  };
+  forEachChild(doc.root, g);
+}
+
+function transformElementNames(
+  doc: Document,
+  from: string,
+  to: string,
+) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind === "element" && n.name === from) {
+      n.name = to;
+    }
+  });
+}
+
+function transformAttributeNames(
+  doc: Document,
+  from: string,
+  to: string,
+) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element") {
+      return;
+    }
+    const value = n.attributes.get(from) as string;
+    if (value !== undefined) {
+      n.attributes.delete(from);
+      n.attributes.set(to, value);
+    }
+  });
+}
+
+function transformSectionIdAttribute(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "section") {
+      return;
+    }
+
+    const idAttr = n.attributes.get("id");
+    n.attributes.set("id", `section--${idAttr}`);
+  });
+}
+
+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 === "title") {
+      const currentSection = sectionStack[sectionStack.length - 1];
+      if (!currentSection) {
+        throw new DocBookError(
+          "[docbook.tohtml] <title> element must be inside <section>",
+        );
+      }
+      const sectionId = currentSection.attributes.get("id");
+      const aElement: Element = {
+        kind: "element",
+        name: "a",
+        attributes: new Map(),
+        children: c.children,
+      };
+      aElement.attributes.set("href", `#${sectionId}`);
+      c.children = [aElement];
+    }
+  };
+  forEachChild(doc.root, g);
+}
+
+function transformSectionTitleElement(doc: Document) {
+  let sectionLevel = 1;
+  const g = (c: Node) => {
+    if (c.kind !== "element") {
+      return;
+    }
+
+    if (c.name === "section") {
+      sectionLevel += 1;
+      c.attributes.set("--section-level", sectionLevel.toString());
+    }
+    forEachChild(c, g);
+    if (c.name === "section") {
+      sectionLevel -= 1;
+    }
+    if (c.name === "title") {
+      c.name = `h${sectionLevel}`;
+    }
+  };
+  forEachChild(doc.root, g);
+}
+
+function transformProgramListingElement(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "programlisting") {
+      return;
+    }
+
+    n.name = "pre";
+    addClass(n, "highlight");
+    const codeElement: Element = {
+      kind: "element",
+      name: "code",
+      attributes: new Map(),
+      children: n.children,
+    };
+    n.children = [codeElement];
+  });
+}
+
+function transformLiteralLayoutElement(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "literallayout") {
+      return;
+    }
+
+    n.name = "pre";
+    addClass(n, "highlight");
+    const codeElement: Element = {
+      kind: "element",
+      name: "code",
+      attributes: new Map(),
+      children: n.children,
+    };
+    n.children = [codeElement];
+  });
+}
+
+function transformNoteElement(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "note") {
+      return;
+    }
+
+    const labelElement: Element = {
+      kind: "element",
+      name: "div",
+      attributes: new Map([["class", "admonition-label"]]),
+      children: [{
+        kind: "text",
+        content: "NOTE",
+        raw: false,
+      }],
+    };
+    const contentElement: Element = {
+      kind: "element",
+      name: "div",
+      attributes: new Map([["class", "admonition-content"]]),
+      children: n.children,
+    };
+    n.name = "div";
+    addClass(n, "admonition");
+    n.children = [
+      labelElement,
+      contentElement,
+    ];
+  });
+}
+
+function setDefaultLangAttribute(_doc: Document) {
+  // TODO
+  // if (!e.attributes.has("lang")) {
+  //   e.attributes.set("lang", "ja-JP");
+  // }
+}
+
+function traverseFootnotes(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "footnote") {
+      return;
+    }
+
+    // TODO
+    // <footnote>x</footnote>
+    //
+    // <sup class="footnote">[<a id="_footnoteref_1" class="footnote" href="#_footnotedef_1">1</a>]</sup>
+    //
+    // <div class="footnote" id="_footnotedef_1">
+    //   <a href="#_footnoteref_1">1</a>. RAS syndrome
+    // </div>
+    n.name = "span";
+    n.children = [];
+  });
+}
+
+function highlightPrograms(doc: Document) {
+  forEachChildRecursively(doc.root, (n) => {
+    if (n.kind !== "element" || n.name !== "pre") {
+      return;
+    }
+    const preClass = n.attributes.get("class") || "";
+    if (!preClass.includes("highlight")) {
+      return;
+    }
+    const codeElement = findFirstChildElement(n, "code");
+    if (!codeElement) {
+      return;
+    }
+    const language = n.attributes.get("language");
+    if (!language) {
+      return;
+    }
+    const sourceCodeNode = codeElement.children[0] as Text | RawHTML;
+    const sourceCode = sourceCodeNode.content;
+
+    if (!hljs.getLanguage(language)) {
+      return;
+    }
+
+    const highlighted =
+      hljs.highlight(sourceCode, { language: language }).value;
+
+    sourceCodeNode.content = highlighted;
+    sourceCodeNode.raw = true;
+    codeElement.attributes.set("class", "highlight");
+  });
+}