Ship a batch web-page-to-PDF tool that converts a URL list to a ZIP of A4 PDFs, reusing the HTML-to-PDF fetch and capture pipeline.

Date: 2026-09-06 22:30 Summary: Ship a batch web-page-to-PDF tool that converts a URL list to a ZIP of A4 PDFs, reusing the HTML-to-PDF fetch and capture pipeline. Visibility: people

[question] 立项：batch convert web page to pdf，并实现（参考convert-html-to-pdf 的实现）

Tool links (English)

https://onlinefreetools.org/tools/batch-convert-web-pages-to-pdf Chinese: https://onlinefreetools.org/zh/tools/batch-convert-web-pages-to-pdf

[try to solve]

Context

The site already had Convert HTML to PDF (convert-html-to-pdf): paste HTML or load one URL, preview, download one PDF. The new request is a batch job: many public page URLs at once, one PDF per URL, take them away as a ZIP. That is a different primary control and output, so it is not a doorway retitle of the neighbor page.

Process

Wrote work-tasks/batch-convert-web-pages-to-pdf/ (00–03): direction A (browser JS), slug batch-convert-web-pages-to-pdf, H1 Batch convert web pages to PDF. Coverage table mapped secondary phrases (multiple webpages, URL list, ZIP) into description / FAQ / use cases. Intent review: satisfy the list+ZIP task; leave single HTML paste on the neighbor page; do not crawl a whole sitemap.
PDF Keyword Planner N4 (convert html pdf) already belongs to convert-html-to-pdf. Ads/Planner for this slug: not applicable; FAQ points single-page HTML/URL searchers to the neighbor tool.
coverage:gate --phase=0b then catalog shard, icon, page, English i18n, phase=2. Ten locales rewritten from local search phrases (zh H1「批量把网页转成 PDF」, es「Convertir varias páginas web a PDF」, ja「複数のウェブページをPDFにまとめて変換」, and so on), then phase=4 on en,zh,es,ja.
Implementation reuses GET /api/tools/convert-html-to-pdf and /asset (no new Worker route). Client capture follows the neighbor page: DOMPurify, hidden iframe, image proxy, overflow unlock, html2pdf A4. Successful rows go into a ZIP via fflate. Cap is 10 URLs, serial, failed rows skip.
Load sample fills two example.com URLs and converts local HTML stand-ins so the table and ZIP appear on first paint without fetching the live site (the single-page tool already showed that auto-fetching example.com can error on open). Convert all fetches the pasted addresses.
Isolation does not allow src/pages/site/*.ts for a single slug, so the capture helper is inlined in batchConvertWebPagesToPdfPage.ts. verify:tool ran coverage, full build:site, HTML smoke, SEO, vendor. Isolation still reports the pre-existing dirty convert-html-to-pdf files in the working tree (load-wait work from the same day); those were not edited for this slug.

Root cause / analysis

Single-URL HTML conversion and multi-URL archival are the same capture engine but different user tasks. Splitting them keeps H1 honest: people searching “convert HTML to PDF” stay on the editor page; people with a list of help-center or policy URLs get a queue, per-row status, and a ZIP. Auto-running a live fetch on entry would couple first paint to whether example.com (or any sample host) allows the Worker fetch.

Solution

New tool: /tools/batch-convert-web-pages-to-pdf
Related: convert-html-to-pdf, combine-files-into-one-pdf
localProcessing: false because URL fetch goes through the edge once and is not stored; PDF bytes are built in the tab
Home + README list entries added; English homepage card is present after build:site

Notes / boundaries

Not a site crawler: you paste URLs you already have (max 10).
Login walls, JS-rendered apps, and print CSS often will not match the live page (same limits as the neighbor tool).
Combining PDFs into one file is a different tool (combine-files-into-one-pdf).
Isolation green for this slug alone requires a clean working tree or CROSS_TOOL_UPDATE=1 if other tool shards are already dirty.
Live click-through of Convert all against arbitrary public sites was not run in a headed browser in this session; prerendered HTML contains loadSample() on first paint and lint:tool-page --require-html passed.

[actions]