Extract runs and text

demo/85-template-document.ts

@@ -4,7 +4,12 @@ import * as fs from "fs";
 import { Paragraph, patchDocument, TextRun } from "../build";
 
 patchDocument(fs.readFileSync("demo/assets/simple-template.docx"), {
-    children: [new Paragraph("ff"), new TextRun("fgf")],
+    patches: [
+        {
+            children: [new Paragraph("ff"), new TextRun("fgf")],
+            text: "{{ name }}",
+        },
+    ],
 }).then((doc) => {
     fs.writeFileSync("My Document.docx", doc);
 });

src/templater/from-docx.ts

@@ -1,25 +1,31 @@
 import * as JSZip from "jszip";
-import { xml2js, ElementCompact, js2xml } from "xml-js";
+import { xml2js, Element, js2xml } from "xml-js";
 import { replacer } from "./replacer";
+import { findLocationOfText } from "./traverser";
 
 // eslint-disable-next-line functional/prefer-readonly-type
 type InputDataType = Buffer | string | number[] | Uint8Array | ArrayBuffer | Blob | NodeJS.ReadableStream;
 
-export interface PatchDocumentOptions {
+export interface IPatch {
     readonly children: any[];
+    readonly text: string;
+}
+export interface PatchDocumentOptions {
+    readonly patches: readonly IPatch[];
 }
 
 export const patchDocument = async (data: InputDataType, options: PatchDocumentOptions): Promise<Buffer> => {
     const zipContent = await JSZip.loadAsync(data);
 
-    const map = new Map<string, ElementCompact>();
+    const map = new Map<string, Element>();
-    console.log(options);
 
     for (const [key, value] of Object.entries(zipContent.files)) {
         const json = toJson(await value.async("text"));
         if (key === "word/document.xml") {
-            console.log(json);
+            for (const patch of options.patches) {
-            replacer(json, options);
+                findLocationOfText(json, patch.text);
+                replacer(json, patch);
+            }
         }
 
         map.set(key, json);
@@ -42,12 +48,12 @@ export const patchDocument = async (data: InputDataType, options: PatchDocumentO
     return zipData;
 };
 
-const toJson = (xmlData: string): ElementCompact => {
+const toJson = (xmlData: string): Element => {
-    const xmlObj = xml2js(xmlData, { compact: false }) as ElementCompact;
+    const xmlObj = xml2js(xmlData, { compact: false }) as Element;
     return xmlObj;
 };
 
-const toXml = (jsonObj: ElementCompact): string => {
+const toXml = (jsonObj: Element): string => {
     const output = js2xml(jsonObj);
     return output;
 };

src/templater/replacer.ts

@@ -1,8 +1,8 @@
 import { Paragraph, TextRun } from "@file/paragraph";
 import { ElementCompact } from "xml-js";
-import { PatchDocumentOptions } from "./from-docx";
+import { IPatch } from "./from-docx";
 
-export const replacer = (json: ElementCompact, options: PatchDocumentOptions): ElementCompact => {
+export const replacer = (json: ElementCompact, options: IPatch): ElementCompact => {
     for (const child of options.children) {
         if (child instanceof Paragraph) {
             console.log("is para");

src/templater/run-renderer.ts

@@ -0,0 +1,78 @@
+import { Element } from "xml-js";
+
+export interface IRenderedParagraphNode {
+    readonly text: string;
+    readonly runs: readonly IRenderedRunNode[];
+}
+
+interface IParts {
+    readonly text: string;
+    readonly index: number;
+}
+
+export interface IRenderedRunNode {
+    readonly text: string;
+    readonly parts: readonly IParts[];
+    readonly index: number;
+}
+
+export const renderParagraphNode = (node: Element): IRenderedParagraphNode => {
+    if (node.name !== "w:p") {
+        throw new Error(`Invalid node type: ${node.name}`);
+    }
+
+    if (!node.elements) {
+        return {
+            text: "",
+            runs: [],
+        };
+    }
+
+    const runs = node.elements
+        .map((element, i) => ({ element, i }))
+        .filter(({ element }) => element.name === "w:r")
+        .map(({ element, i }) => renderRunNode(element, i))
+        .filter((e) => !!e)
+        .map((e) => e as IRenderedRunNode);
+
+    const text = runs.reduce((acc, curr) => acc + curr.text, "");
+
+    return {
+        text,
+        runs,
+    };
+};
+
+const renderRunNode = (node: Element, index: number): IRenderedRunNode => {
+    if (node.name !== "w:r") {
+        throw new Error(`Invalid node type: ${node.name}`);
+    }
+
+    if (!node.elements) {
+        return {
+            text: "",
+            parts: [],
+            index: -1,
+        };
+    }
+
+    const parts = node.elements
+        .map((element, i: number) =>
+            element.name === "w:t" && element.elements
+                ? {
+                      text: element.elements[0].text?.toString() ?? "",
+                      index: i,
+                  }
+                : undefined,
+        )
+        .filter((e) => !!e)
+        .map((e) => e as IParts);
+
+    const text = parts.reduce((acc, curr) => acc + curr.text, "");
+
+    return {
+        text,
+        parts,
+        index,
+    };
+};

src/templater/traverser.ts

@@ -0,0 +1,43 @@
+import { Element } from "xml-js";
+import { IRenderedParagraphNode, renderParagraphNode } from "./run-renderer";
+
+export interface ILocationOfText {
+    readonly parent: Element;
+    readonly startIndex: number;
+    readonly endIndex: number;
+    readonly currentText: string;
+    // This is optional because the text could start in the middle of a tag
+    readonly startElement?: Element;
+    // This is optional because the text could end in the middle of a tag
+    readonly endElement?: Element;
+}
+
+export const findLocationOfText = (node: Element, text: string): void => {
+    let renderedParagraphs: readonly IRenderedParagraphNode[] = [];
+
+    // eslint-disable-next-line functional/prefer-readonly-type
+    const queue: Element[] = [...(node.elements ?? [])];
+
+    // eslint-disable-next-line functional/immutable-data
+    let currentNode: Element | undefined;
+    while (queue.length > 0) {
+        // eslint-disable-next-line functional/immutable-data
+        currentNode = queue.shift();
+
+        if (!currentNode) {
+            break;
+        }
+
+        if (currentNode.name === "w:p") {
+            renderedParagraphs = [...renderedParagraphs, renderParagraphNode(currentNode)];
+        } else {
+            // eslint-disable-next-line functional/immutable-data
+            queue.push(...(currentNode.elements ?? []));
+        }
+    }
+
+    const filteredParagraphs = renderedParagraphs.filter((p) => p.text.includes(text));
+
+    console.log("paragrapghs", JSON.stringify(filteredParagraphs, null, 2));
+    return undefined;
+};