feat: poppler working for some formats

package.json

@@ -12,7 +12,8 @@
     "@elysiajs/html": "^1.0.2",
     "@elysiajs/jwt": "^1.0.2",
     "@elysiajs/static": "^1.0.3",
-    "elysia": "^1.0.22"
+    "elysia": "^1.0.22",
+    "node-poppler": "^7.2.0"
   },
   "module": "src/index.tsx",
   "bun-create": {

src/converters/main.ts

@@ -20,6 +20,8 @@ import {
   properties as propertiesPdflatex,
 } from "./pdflatex";
 
+import { convert as convertPoppler, properties as propertiesPoppler } from "./poppler";
+
 import { normalizeFiletype } from "../helpers/normalizeFiletype";
 
 // This should probably be reconstructed so that the functions are not imported instead the functions hook into this to make the converters more modular
@@ -58,6 +60,10 @@ const properties: {
     properties: propertiesPdflatex,
     converter: convertPdflatex,
   },
+  poppler: {
+    properties: propertiesPoppler,
+    converter: convertPoppler,
+  },
   pandoc: {
     properties: propertiesPandoc,
     converter: convertPandoc,

src/converters/poppler.ts

@@ -0,0 +1,115 @@
+const { Poppler } = require("node-poppler");
+const poppler = new Poppler();
+
+export const properties = {
+  from: {
+    text: ["pdf"],
+  },
+  to: {
+    text: [
+      "jpeg",
+      "png",
+      "tiff",
+      "eps",
+      "icc",
+      "pdf",
+      "svg",
+      "ps",
+      "html",
+      "text",
+    ],
+  },
+};
+
+export function convert(
+  filePath: string,
+  fileType: string,
+  convertTo: string,
+  targetPath: string,
+  // biome-ignore lint/suspicious/noExplicitAny: <explanation>
+  options?: any,
+): Promise<string> {
+  return new Promise((resolve, reject) => {
+    const cairoFiles = [
+      "jpeg",
+      "png",
+      "tiff",
+      "eps",
+      "icc",
+      "pdf",
+      "svg",
+      "ps",
+    ];
+
+    if (cairoFiles.includes(convertTo)) {
+      const popplerOptions: {
+        jpegFile?: boolean;
+        pngFile?: boolean;
+        tiffFile?: boolean;
+        epsFile?: boolean;
+        iccFile?: boolean;
+        pdfFile?: boolean;
+        svgFile?: boolean;
+        psFile?: boolean;
+      } = {};
+
+      switch (convertTo) {
+        case "jpeg":
+          popplerOptions.jpegFile = true;
+          break;
+        case "png":
+          popplerOptions.pngFile = true;
+          break;
+        case "tiff":
+          popplerOptions.tiffFile = true;
+          break;
+        case "eps":
+          popplerOptions.epsFile = true;
+          break;
+        case "icc":
+          popplerOptions.iccFile = true;
+          break;
+        case "pdf":
+          popplerOptions.pdfFile = true;
+          break;
+        case "svg":
+          popplerOptions.svgFile = true;
+          break;
+        case "ps":
+          popplerOptions.psFile = true;
+          break;
+        default:
+          reject(`Invalid convertTo option: ${convertTo}`);
+      }
+
+      poppler
+        .pdfToCairo(filePath, targetPath, popplerOptions)
+        .then(() => {
+          resolve("success");
+        })
+        .catch((err: Error) => {
+          reject(err);
+        });
+    } else if (convertTo === "html") {
+      poppler
+        .pdfToHtml(filePath, targetPath)
+        .then(() => {
+          resolve("success");
+        })
+        .catch((err: Error) => {
+          reject(err);
+        });
+    } else if (convertTo === "text") {
+      poppler
+        .pdfToText(filePath, targetPath)
+        .then(() => {
+          resolve("success");
+        })
+        .catch((err: Error) => {
+          reject(err);
+        });
+    } else {
+      reject(`Invalid convertTo option: ${convertTo}`);
+    }
+  });
+}

src/helpers/normalizeFiletype.ts

@@ -25,6 +25,8 @@ export const normalizeOutputFiletype = (filetype: string): string => {
       return "tex";
     case "markdown":
       return "md";
+    case "text":
+      return "txt";
     default:
       return lowercaseFiletype;
   }