feat: add support for uploading PPT(x) files

src/lib/misc/isLimitError.test.ts

@@ -1,21 +1,22 @@
-import { PDF_EXCEEDS_MAX_PAGE_LIMIT } from '../parser/pdf/convertPDFToImages';
+import { PDF_EXCEEDS_MAX_PAGE_LIMIT } from '../pdf/convertPDFToImages';
 import { isLimitError } from './isLimitError';
 
-const MOCK_MSG = "<div class=\"content\"><h3 class=\"title is-3\">Your request has hit the limit</h3><ul><li>Split your request into multiple smaller ones (i.e.) make your upload size smaller.</li><li><div class=\"is-flex is-align-items-center\"><a class=\"button is-success is-medium mr-2\" href=\"https://buy.stripe.com/eVadTGcCI6Ny73qfZ0\">Subscribe</a> for only $2 per month to remove all the limits.</div></li><li>Or <a href=\"https://alemayhu.com/patreon\">Become a patron</a> to support me.</li></ul><p>If you already have an account, please <a href=\"/login?redirect=/upload\">login</a> and try again. If you are still experiencing issues, please contact <a href=\"mailto:[email protected]\">[email protected]</a>.</p></div>";
+const MOCK_MSG =
+  '<div class="content"><h3 class="title is-3">Your request has hit the limit</h3><ul><li>Split your request into multiple smaller ones (i.e.) make your upload size smaller.</li><li><div class="is-flex is-align-items-center"><a class="button is-success is-medium mr-2" href="https://buy.stripe.com/eVadTGcCI6Ny73qfZ0">Subscribe</a> for only $2 per month to remove all the limits.</div></li><li>Or <a href="https://alemayhu.com/patreon">Become a patron</a> to support me.</li></ul><p>If you already have an account, please <a href="/login?redirect=/upload">login</a> and try again. If you are still experiencing issues, please contact <a href="mailto:[email protected]">[email protected]</a>.</p></div>';
 
 describe('isLimitError', () => {
   it('returns true ', () => {
-    expect(isLimitError(new Error("File too large"))).toBe(true);
-    expect(isLimitError(new Error("You can only add 100 cards"))).toBe(true);
+    expect(isLimitError(new Error('File too large'))).toBe(true);
+    expect(isLimitError(new Error('You can only add 100 cards'))).toBe(true);
     expect(isLimitError(new Error(PDF_EXCEEDS_MAX_PAGE_LIMIT))).toBe(true);
   });
 
   it('returns true for html', () => {
     expect(isLimitError(new Error(MOCK_MSG))).toBe(true);
-  })
+  });
 
-  it("returns false", () => {
-    expect(isLimitError(new Error("File too small"))).toBe(false);
+  it('returns false', () => {
+    expect(isLimitError(new Error('File too small'))).toBe(false);
     expect(isLimitError()).toBe(false);
-  })
+  });
 });

src/lib/misc/isLimitError.ts

@@ -1,4 +1,4 @@
-import { PDF_EXCEEDS_MAX_PAGE_LIMIT } from '../parser/pdf/convertPDFToImages';
+import { PDF_EXCEEDS_MAX_PAGE_LIMIT } from '../pdf/convertPDFToImages';
 
 const LIMIT_MESSAGES = [
   'File too large',

src/lib/parser/PrepareDeck.ts

@@ -1,9 +1,10 @@
 import getDeckFilename from '../anki/getDeckFilename';
 import { DeckParser, DeckParserInput } from './DeckParser';
 import Deck from './Deck';
-import { isPDFFile } from '../storage/checks';
+import { isPDFFile, isPPTFile } from '../storage/checks';
 import { convertPDFToHTML } from './experimental/VertexAPI/convertPDFToHTML';
-import { convertPDFToImages } from './pdf/convertPDFToImages';
+import { convertPDFToImages } from '../pdf/convertPDFToImages';
+import { convertPPTToPDF } from '../pdf/ConvertPPTToPDF';
 
 interface PrepareDeckResult {
   name: string;
@@ -15,11 +16,24 @@ export async function PrepareDeck(
   input: DeckParserInput
 ): Promise<PrepareDeckResult> {
   for (const file of input.files) {
-    if (!isPDFFile(file.name) || !file.contents) continue;
+    if ((!isPDFFile(file.name) && !isPPTFile(file.name)) || !file.contents)
+      continue;
 
-    if (input.noLimits && input.settings.vertexAIPDFQuestions) {
+    if (
+      isPDFFile(file.name) &&
+      input.noLimits &&
+      input.settings.vertexAIPDFQuestions
+    ) {
       file.contents = await convertPDFToHTML(file.contents.toString('base64'));
     } else {
+      if (isPPTFile(file.name)) {
+        file.contents = await convertPPTToPDF(
+          file.name,
+          file.contents,
+          input.workspace
+        );
+      }
+
       file.contents = await convertPDFToImages({
         name: file.name,
         workspace: input.workspace,

src/lib/pdf/ConvertPPTToPDF.ts

@@ -0,0 +1,50 @@
+import { S3 } from 'aws-sdk';
+import Workspace from '../parser/WorkSpace';
+import path from 'path';
+import fs from 'fs/promises';
+import { execFile } from 'child_process';
+
+export function convertPPTToPDF(
+  name: string,
+  contents: S3.Body,
+  workspace: Workspace
+): Promise<Buffer> {
+  return new Promise((resolve, reject) => {
+    const sofficeBin =
+      process.platform === 'darwin'
+        ? '/Applications/LibreOffice.app/Contents/MacOS/soffice'
+        : '/usr/bin/soffice';
+    const tempFile = path.join(workspace.location, name);
+
+    fs.writeFile(tempFile, Buffer.from(contents as Buffer))
+      .then(() => {
+        const pdfFile = path.join(
+          workspace.location,
+          path.basename(name, path.extname(name)) + '.pdf'
+        );
+
+        execFile(
+          sofficeBin,
+          ['--headless', '--convert-to', 'pdf', tempFile],
+          {
+            cwd: workspace.location,
+          },
+          async (error, stdout, stderr) => {
+            await fs.writeFile(
+              path.join(workspace.location, 'stdout.log'),
+              stdout
+            );
+            await fs.writeFile(
+              path.join(workspace.location, 'stderr.log'),
+              stderr
+            );
+            if (error) {
+              reject(new Error(error.message || 'Conversion failed'));
+            }
+            resolve(await fs.readFile(pdfFile));
+          }
+        );
+      })
+      .catch((err) => reject(new Error(err.message || 'File write failed')));
+  });
+}

src/lib/pdf/combineIntoHTML.ts

@@ -0,0 +1,26 @@
+import path from 'path';
+
+export function combineIntoHTML(imagePaths: string[], title: string): string {
+  const html = `<!DOCTYPE html>
+<html>
+<head><title>${title}</title></head>
+<body>
+  ${Array.from({ length: imagePaths.length / 2 }, (_, i) => {
+    const front = path.basename(imagePaths[i * 2]);
+    const back = path.basename(imagePaths[i * 2 + 1]);
+    return `<ul class="toggle">
+    <li>
+      <details>
+        <summary>
+        <img src="${front}" />
+        </summary>
+        <img src="${back}" />
+      </details>
+    </li>
+    </ul>`;
+  }).join('\n')}
+</body>
+</html>`;
+
+  return html;
+}

src/lib/pdf/convertPDFToImages.ts

@@ -0,0 +1,40 @@
+import { writeFile } from 'fs/promises';
+import path from 'path';
+import Workspace from '../parser/WorkSpace';
+import { S3 } from 'aws-sdk';
+import { getPageCount } from './getPageCount';
+import { convertPage } from './convertPage';
+import { combineIntoHTML } from './combineIntoHTML';
+
+interface ConvertPDFToImagesInput {
+  workspace: Workspace;
+  noLimits: boolean;
+  contents?: S3.Body;
+  name?: string;
+}
+
+export const PDF_EXCEEDS_MAX_PAGE_LIMIT =
+  'PDF exceeds maximum page limit of 100 for free and anonymous users.';
+
+export async function convertPDFToImages(
+  input: ConvertPDFToImagesInput
+): Promise<Buffer> {
+  const { contents, workspace, noLimits, name } = input;
+  const pdfPath = path.join(workspace.location, name ?? 'Default.pdf');
+  await writeFile(pdfPath, Buffer.from(contents as Buffer));
+
+  const pageCount = await getPageCount(pdfPath);
+  const title = path.basename(pdfPath);
+  if (!noLimits && pageCount > 100) {
+    throw new Error(PDF_EXCEEDS_MAX_PAGE_LIMIT);
+  }
+
+  const imagePaths = await Promise.all(
+    Array.from({ length: pageCount }, (_, i) =>
+      convertPage(pdfPath, i + 1, pageCount)
+    )
+  );
+
+  const html = combineIntoHTML(imagePaths, title);
+  return Buffer.from(html);
+}

src/lib/pdf/convertPage.ts

@@ -0,0 +1,50 @@
+import { execFile } from 'child_process';
+import os from 'os';
+
+export function convertPage(
+  pdfPath: string,
+  pageNumber: number,
+  totalPageCount: number
+): Promise<string> {
+  const outputFileNameBase = `${pdfPath}-page${pageNumber}`;
+
+  const determinePaddingLength = (pageCount: number): number => {
+    if (pageCount >= 1000) return 4;
+    if (pageCount >= 100) return 3;
+    if (pageCount >= 10) return 2;
+    return 1;
+  };
+
+  const paddedPageNumber = String(pageNumber).padStart(
+    determinePaddingLength(totalPageCount),
+    '0'
+  );
+
+  const pdftoppmPath =
+    os.platform() === 'darwin'
+      ? '/usr/local/bin/pdftoppm'
+      : '/usr/bin/pdftoppm';
+
+  return new Promise((resolve, reject) => {
+    execFile(
+      pdftoppmPath,
+      [
+        '-png',
+        '-f',
+        pageNumber.toString(),
+        '-l',
+        pageNumber.toString(),
+        pdfPath,
+        outputFileNameBase,
+      ],
+      (error) => {
+        if (error) {
+          return reject(
+            new Error(`Failed to convert page ${pageNumber} to PNG`)
+          );
+        }
+        resolve(`${outputFileNameBase}-${paddedPageNumber}.png`);
+      }
+    );
+  });
+}