Fix svg scaling for certain books [#5]

Author: garzj

src/item/BiBoxBook.ts

@@ -1,6 +1,7 @@
-import { delay, promisePool } from '../util/promise';
+import { promisePool } from '../util/promise';
 import { Book } from './Book';
 import { defDownloadOptions, DownloadOptions } from './download-options';
+import { getPdfOptions } from './get-pdf-options';
 
 export class BiBoxBook extends Book {
   async download(outDir: string, _options?: DownloadOptions) {
@@ -63,15 +64,7 @@ export class BiBoxBook extends Book {
           const pdfFile = this.getPdfPath(dir, pageNo);
 
           await page.pdf({
-            height: await page.evaluate(
-              () =>
-                window.document.querySelector('img')?.height ||
-                window.innerHeight
-            ),
-            width: await page.evaluate(
-              () =>
-                window.document.querySelector('img')?.width || window.innerWidth
-            ),
+            ...(await getPdfOptions(page, options)),
             path: pdfFile,
           });
         } finally {

src/item/DigiBook.ts

@@ -3,23 +3,29 @@ import { waitForGoto } from '../util/puppeteer';
 import { URL } from 'url';
 import { Book } from './Book';
 import { defDownloadOptions, DownloadOptions } from './download-options';
+import { SizeAttributes, getPdfOptions } from './get-pdf-options';
 
 export class DigiBook extends Book {
   async download(outDir: string, _options?: DownloadOptions) {
     const dir = await this.mkSubDir(outDir);
     const options = defDownloadOptions(_options);
+    options.format ??= 'a4';
 
     // Get url of 1st svg page
     const checkPage = await this.shelf.browser.newPage();
     let page1Url: string;
+    let sizeHint: SizeAttributes;
     try {
       await checkPage.goto(new URL(`?page=1`, this.url).toString(), {
         waitUntil: 'networkidle2',
         timeout: this.shelf.options.timeout,
       });
-      page1Url = await checkPage.$eval(
+      [page1Url, sizeHint] = await checkPage.$eval(
         '#pg1 > object',
-        (obj) => (obj as HTMLObjectElement).data
+        (obj: HTMLObjectElement): [string, SizeAttributes] => [
+          obj.data,
+          { width: obj.width, height: obj.height },
+        ]
       );
     } finally {
       await checkPage.close();
@@ -54,7 +60,7 @@ export class DigiBook extends Book {
         const pdfFile = this.getPdfPath(dir, pageNo);
 
         await page.pdf({
-          format: options.format,
+          ...(await getPdfOptions(page, options, sizeHint)),
           path: pdfFile,
         });
 

src/item/ScookBook.ts

@@ -2,6 +2,7 @@ import { ScrapeError } from '../error/ScrapeError';
 import { delay, promisePool } from '../util/promise';
 import { Book } from './Book';
 import { defDownloadOptions, DownloadOptions } from './download-options';
+import { getPdfOptions } from './get-pdf-options';
 
 export class ScookBook extends Book {
   async download(outDir: string, _options?: DownloadOptions) {
@@ -84,15 +85,7 @@ export class ScookBook extends Book {
           const pdfFile = this.getPdfPath(dir, pageNo);
 
           await page.pdf({
-            height: await page.evaluate(
-              () =>
-                window.document.querySelector('img')?.height ||
-                window.innerHeight
-            ),
-            width: await page.evaluate(
-              () =>
-                window.document.querySelector('img')?.width || window.innerWidth
-            ),
+            ...(await getPdfOptions(page, options)),
             path: pdfFile,
           });
         } finally {

src/item/download-options.ts

@@ -7,10 +7,9 @@ export interface DownloadOptions {
   format?: PaperFormat;
 }
 
-const defaultOptions: Required<DownloadOptions> = {
+const defaultOptions: DownloadOptions = {
   concurrency: 10,
   mergePdfs: true,
-  format: 'a4',
 };
 
 export function defDownloadOptions(_options?: DownloadOptions) {

src/item/get-pdf-options.ts

@@ -0,0 +1,44 @@
+import { PDFOptions, Page } from 'puppeteer';
+import { DownloadOptions } from './download-options';
+
+export interface SizeAttributes {
+  width: string;
+  height: string;
+}
+
+export async function getPdfOptions(
+  page: Page,
+  options: DownloadOptions,
+  sizeHint?: SizeAttributes
+): Promise<PDFOptions> {
+  const size = await page.evaluate(async (sizeHint): Promise<PDFOptions> => {
+    const root = document.documentElement;
+
+    if (root instanceof SVGGraphicsElement) {
+      if (sizeHint) {
+        root.setAttribute('width', sizeHint.width);
+        root.setAttribute('height', sizeHint.height);
+        return sizeHint;
+      }
+      const bbox = root.getBBox();
+      return { width: bbox.width, height: bbox.height };
+    }
+
+    if (sizeHint) return sizeHint;
+
+    return {
+      width:
+        document.querySelector<HTMLImageElement>('body > img')?.width ??
+        window.innerWidth,
+      height:
+        document.querySelector<HTMLImageElement>('body > img')?.height ??
+        window.innerHeight,
+    };
+  }, sizeHint);
+
+  return {
+    format: options.format,
+    printBackground: true,
+    ...size,
+  };
+}