feat: use office-text-extractor to extract text

package.json

@@ -1,5 +1,5 @@
 {
-  "name": "docen",
+  "name": "docen-edge",
   "version": "0.0.0",
   "description": "Programmatically and command-line implementation of document formatting, powered by Demo Macro.",
   "private": true,

packages/docen/bundle.mjs

@@ -6,16 +6,16 @@ execSync("pnpm prepack", { stdio: "inherit" });
 execSync("pnpm ncc build dist/cli.cjs -o dist/ncc", { stdio: "inherit" });
 
 const version = readFileSync("package.json", "utf-8").match(
-  /"version": "(.*?)"/
+  /"version": "(.*?)"/,
 )[1];
 
-execSync(`cp dist/ncc/index.cjs dist/ncc/docen-${version}.cjs`, {
+execSync(`mv dist/ncc/index.cjs dist/ncc/docen-${version}.cjs`, {
   stdio: "inherit",
 });
 
 execSync(
-  `pnpm dlx pkg dist/ncc/docen-${version}.cjs --out-path dist/bundle -C GZip`,
+  `pnpm dlx pkg dist/ncc/docen-${version}.cjs --out-path dist/bundle -c pkg.config.json -C GZip`,
   {
     stdio: "inherit",
-  }
+  },
 );

packages/docen/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docen",
-  "version": "0.0.0",
+  "version": "0.0.1",
   "description": "Programmatically and command-line implementation of document formatting, powered by Demo Macro.",
   "main": "dist/index.cjs",
   "types": "dist/index.d.ts",
@@ -41,6 +41,6 @@
   "dependencies": {
     "@funish/cli": "0.0.4",
     "file-type": "18.5.0",
-    "mammoth": "1.6.0"
+    "office-text-extractor": "^3.0.1"
   }
 }

packages/docen/pkg.config.json

@@ -0,0 +1,5 @@
+{
+  "pkg": {
+    "scripts": "dist/ncc/*"
+  }
+}

packages/docen/src/cli.ts

@@ -1,5 +1,5 @@
-import { CLI } from "@funish/cli";
 import { docen } from ".";
+import { CLI } from "@funish/cli";
 
 const cli = new CLI("docen");
 

packages/docen/src/index.ts

@@ -1,16 +1,22 @@
+import { fileTypeFromBuffer } from "file-type";
 import { readFileSync, writeFileSync } from "fs";
-import { extractRawText } from "mammoth";
+import { getTextExtractor } from "office-text-extractor";
 import { extname } from "path";
-import { fileTypeFromBuffer } from "file-type";
 
 export async function docen(source: string, target: string): Promise<void> {
   const sourceContent = readFileSync(source);
   const sourceType = await fileTypeFromBuffer(sourceContent);
 
-  if (sourceType?.ext === "docx" && extname(target) === ".txt") {
-    extractRawText({ buffer: sourceContent }).then((result) => {
-      writeFileSync(target, result.value);
+  if (
+    (sourceType?.ext === "docx" || "pptx" || "xlsx" || "pdf") &&
+    extname(target) === ".txt"
+  ) {
+    const extractor = getTextExtractor();
+    const text = await extractor.extractText({
+      input: sourceContent,
+      type: "buffer",
     });
+    writeFileSync(target, text);
   } else {
     console.error(`Unsupported source file type: ${sourceType?.ext}`);
   }