eCourt-note/app/api/template/lib-tmpl.ts

import PizzaZip from "pizzip";
import {
  DOMParser,
  XMLSerializer,
  type Document as XmlDocument,
  type Element as XmlElement,
  type Node as XmlNode,
  type Attr as XmlAttr,
} from "@xmldom/xmldom";
import xpath from "xpath";

export interface ExtractedData {
  [tagName: string]: string;
}

const W_NS =
  "http://schemas.openxmlformats.org/wordprocessingml/2006/main";

const asXpathNode = (node: XmlNode): globalThis.Node =>
  node as unknown as globalThis.Node;

function createSelector() {
  return xpath.useNamespaces({ w: W_NS });
}

/**
 * อ่าน Content Control (w:sdt) จาก DOCX Buffer
 * ปัจจุบันอ่านจาก word/document.xml ตาม logic ของโค้ดต้นฉบับ
 */
export function extractContentControls(
  docxBuffer: Buffer,
): ExtractedData {
  const zip = new PizzaZip(docxBuffer);
  const docXmlFile = zip.file("word/document.xml");

  if (!docXmlFile) {
    throw new Error("Invalid docx file: word/document.xml not found.");
  }

  const parser = new DOMParser();
  const doc = parser.parseFromString(docXmlFile.asText(), "text/xml");
  const select = createSelector();
  const result: ExtractedData = {};

  const sdtNodes = select(
    "//w:sdt",
    asXpathNode(doc),
  ) as unknown as XmlElement[];

  sdtNodes.forEach((sdt) => {
    const tagAttr = select(
      "./w:sdtPr/w:tag/@w:val",
      asXpathNode(sdt),
      true,
    ) as unknown as XmlAttr | undefined;

    const tagName = tagAttr?.value;

    if (!tagName) return;

    const sdtContent = select(
      "./w:sdtContent",
      asXpathNode(sdt),
      true,
    ) as unknown as XmlElement | undefined;

    if (!sdtContent) {
      result[tagName] = "";
      return;
    }

    const nodes = select(
      ".//w:p | .//w:t | .//w:tab | .//w:br | .//w:noBreakHyphen | .//w:softHyphen | .//w:sym",
      asXpathNode(sdtContent),
    ) as unknown as XmlElement[];

    let extractedText = "";

    nodes.forEach((node) => {
      switch (node.nodeName) {
        case "w:p":
          if (
            extractedText.length > 0 &&
            !extractedText.endsWith("\n")
          ) {
            extractedText += "\n";
          }
          break;

        case "w:t":
          extractedText += node.textContent || "";
          break;

        case "w:tab":
          extractedText += "\t";
          break;

        case "w:br":
          extractedText += "\n";
          break;

        case "w:noBreakHyphen":
          extractedText += "\u2011";
          break;

        case "w:softHyphen":
          extractedText += "\u00ad";
          break;

        case "w:sym": {
          const charAttr = node.getAttribute("w:char");
          if (charAttr) {
            const charCode = Number.parseInt(charAttr, 16);
            extractedText += String.fromCharCode(charCode);
          }
          break;
        }
      }
    });

    result[tagName] = extractedText;
  });

  return result;
}

type SdtControlType = "plain" | "rich" | "other";
type SdtStructure = "block" | "inline";

function detectSdtControlType(
  sdt: XmlElement,
  select: xpath.XPathSelect,
): { type: SdtControlType; multiLine: boolean } {
  const sdtPr = select(
    "./w:sdtPr",
    asXpathNode(sdt),
    true,
  ) as unknown as XmlElement | undefined;

  if (!sdtPr) {
    return { type: "other", multiLine: false };
  }

  const textEl = select(
    "./w:text",
    asXpathNode(sdtPr),
    true,
  ) as unknown as XmlElement | undefined;

  if (textEl) {
    const multiLineAttr = select(
      "./@w:multiLine",
      asXpathNode(textEl),
      true,
    ) as unknown as XmlAttr | undefined;

    return {
      type: "plain",
      multiLine:
        multiLineAttr?.value === "1" ||
        multiLineAttr?.value === "true",
    };
  }

  const richTextEl = select(
    "./w:richText",
    asXpathNode(sdtPr),
    true,
  ) as unknown as XmlElement | undefined;

  if (richTextEl) {
    return { type: "rich", multiLine: true };
  }

  const otherTypeTags = [
    "w:date",
    "w:dropDownList",
    "w:comboBox",
    "w:checkbox",
    "w:citation",
    "w:group",
    "w:docPartObj",
    "w:docPartList",
  ];

  for (const tag of otherTypeTags) {
    const found = select(
      `./${tag}`,
      asXpathNode(sdtPr),
      true,
    ) as unknown as XmlElement | undefined;

    if (found) {
      return { type: "other", multiLine: false };
    }
  }

  return { type: "rich", multiLine: true };
}

function detectSdtStructure(
  sdtContent: XmlElement,
  select: xpath.XPathSelect,
): SdtStructure {
  const directP = select(
    "./w:p",
    asXpathNode(sdtContent),
  ) as unknown as XmlElement[];

  return directP.length > 0 ? "block" : "inline";
}

function buildRunChildren(
  doc: XmlDocument,
  text: string,
): XmlElement[] {
  const nodes: XmlElement[] = [];
  const tokens = text.split(/(\t|\n)/);

  tokens.forEach((token) => {
    if (token === "") return;

    if (token === "\t") {
      nodes.push(doc.createElementNS(W_NS, "w:tab"));
    } else if (token === "\n") {
      nodes.push(doc.createElementNS(W_NS, "w:br"));
    } else {
      const t = doc.createElementNS(W_NS, "w:t");
      t.setAttribute("xml:space", "preserve");
      t.textContent = token;
      nodes.push(t);
    }
  });

  if (nodes.length === 0) {
    const t = doc.createElementNS(W_NS, "w:t");
    t.setAttribute("xml:space", "preserve");
    t.textContent = "";
    nodes.push(t);
  }

  return nodes;
}

function buildRun(
  doc: XmlDocument,
  text: string,
  rPr?: XmlElement,
): XmlElement {
  const run = doc.createElementNS(W_NS, "w:r");

  if (rPr) {
    run.appendChild(rPr.cloneNode(true));
  }

  buildRunChildren(doc, text).forEach((node) =>
    run.appendChild(node),
  );

  return run;
}

function insertInlineContent(
  doc: XmlDocument,
  sdtContent: XmlElement,
  select: xpath.XPathSelect,
  text: string,
  allowLineBreak: boolean,
): void {
  const safeText = allowLineBreak
    ? text
    : text.replace(/\n/g, " ");

  const existingRuns = select(
    "./w:r",
    asXpathNode(sdtContent),
  ) as unknown as XmlElement[];

  const firstRPr =
    existingRuns.length > 0
      ? (select(
          "./w:rPr",
          asXpathNode(existingRuns[0]),
          true,
        ) as unknown as XmlElement | undefined)
      : undefined;

  existingRuns.forEach((r) => {
    if (r.parentNode) r.parentNode.removeChild(r);
  });

  sdtContent.appendChild(
    buildRun(doc, safeText, firstRPr),
  );
}

function insertBlockContent(
  doc: XmlDocument,
  sdtContent: XmlElement,
  select: xpath.XPathSelect,
  text: string,
  allowMultiParagraph: boolean,
): void {
  const allParagraphs = select(
    "./w:p",
    asXpathNode(sdtContent),
  ) as unknown as XmlElement[];

  if (allParagraphs.length === 0) {
    // ป้องกัน template ที่มีโครงสร้างผิดปกติ
    const p = doc.createElementNS(W_NS, "w:p");
    sdtContent.appendChild(p);
    allParagraphs.push(p);
  }

  const firstP = allParagraphs[0];

  const firstPPr = select(
    "./w:pPr",
    asXpathNode(firstP),
    true,
  ) as unknown as XmlElement | undefined;

  const firstRun = select(
    "./w:r",
    asXpathNode(firstP),
    true,
  ) as unknown as XmlElement | undefined;

  const firstRPr = firstRun
    ? (select(
        "./w:rPr",
        asXpathNode(firstRun),
        true,
      ) as unknown as XmlElement | undefined)
    : undefined;

  allParagraphs.forEach((p) => {
    if (p !== firstP && p.parentNode) {
      p.parentNode.removeChild(p);
    }
  });

  const existingRuns = select(
    "./w:r",
    asXpathNode(firstP),
  ) as unknown as XmlElement[];

  existingRuns.forEach((r) => {
    if (r.parentNode) r.parentNode.removeChild(r);
  });

  if (!allowMultiParagraph) {
    firstP.appendChild(buildRun(doc, text, firstRPr));
    return;
  }

  text.split("\n").forEach((line, index) => {
    let targetP: XmlElement;

    if (index === 0) {
      targetP = firstP;
    } else {
      targetP = doc.createElementNS(W_NS, "w:p");

      if (firstPPr) {
        targetP.appendChild(firstPPr.cloneNode(true));
      }

      sdtContent.appendChild(targetP);
    }

    const run = doc.createElementNS(W_NS, "w:r");

    if (firstRPr) {
      run.appendChild(firstRPr.cloneNode(true));
    }

    buildRunChildren(doc, line).forEach((node) =>
      run.appendChild(node),
    );

    targetP.appendChild(run);
  });
}

function insertFallbackContent(
  sdtContent: XmlElement,
  select: xpath.XPathSelect,
  text: string,
): void {
  const textNodes = select(
    ".//w:t",
    asXpathNode(sdtContent),
  ) as unknown as XmlElement[];

  if (textNodes.length === 0) return;

  textNodes[0].textContent = text;

  for (let i = 1; i < textNodes.length; i++) {
    textNodes[i].textContent = "";
  }
}

export type TemplateData = Record<string, unknown>;

function normalizeValue(value: unknown): string {
  if (value === null || value === undefined) {
    return "";
  }

  if (typeof value === "string") {
    return value;
  }

  if (
    typeof value === "number" ||
    typeof value === "boolean" ||
    typeof value === "bigint"
  ) {
    return String(value);
  }

  // รองรับ JSON ที่ส่ง array/object มา แต่ Content Control ต้องการ text
  return JSON.stringify(value);
}

/**
 * นำ JSON data ไปแทนค่า Content Control ใน DOCX template
 * คืนค่าเป็น Buffer ของ DOCX ที่สร้างเสร็จแล้ว
 */
export function templateContentControls(
  templateBuffer: Buffer,
  data: TemplateData,
): Buffer {
  const zip = new PizzaZip(templateBuffer);
  const docXmlFile = zip.file("word/document.xml");

  if (!docXmlFile) {
    throw new Error("Invalid docx file: word/document.xml not found.");
  }

  const parser = new DOMParser();
  const doc = parser.parseFromString(docXmlFile.asText(), "text/xml");
  const select = createSelector();

  const sdtNodes = select(
    "//w:sdt",
    asXpathNode(doc),
  ) as unknown as XmlElement[];

  sdtNodes.forEach((sdt) => {
    const tagAttr = select(
      "./w:sdtPr/w:tag/@w:val",
      asXpathNode(sdt),
      true,
    ) as unknown as XmlAttr | undefined;

    const tagName = tagAttr?.value;

    if (!tagName || !Object.prototype.hasOwnProperty.call(data, tagName)) {
      return;
    }

    const newValue = normalizeValue(data[tagName]);

    const sdtContent = select(
      "./w:sdtContent",
      asXpathNode(sdt),
      true,
    ) as unknown as XmlElement | undefined;

    if (!sdtContent) return;

    const { type: controlType, multiLine } =
      detectSdtControlType(sdt, select);

    const structure = detectSdtStructure(sdtContent, select);

    if (controlType === "other") {
      insertFallbackContent(sdtContent, select, newValue);
      return;
    }

    if (structure === "inline") {
      const allowLineBreak =
        controlType === "plain" ? multiLine : true;

      insertInlineContent(
        doc,
        sdtContent,
        select,
        newValue,
        allowLineBreak,
      );
    } else {
      insertBlockContent(
        doc,
        sdtContent,
        select,
        newValue,
        controlType === "rich",
      );
    }
  });

  const serializer = new XMLSerializer();
  zip.file(
    "word/document.xml",
    serializer.serializeToString(doc),
  );

  return zip.generate({
    type: "nodebuffer",
    compression: "DEFLATE",
  });
}