Article Reader v1.2.0 - archive.today fallback

When a live fetch only recovers a teaser (or nothing) on a paywalled site, the reader now offers a one-click link to the newest archive.today snapshot of that URL - a frozen public copy that is often complete.

• Shown as a toolbar button and a prominent notice link.
• Appears only when the result looks incomplete: below the minimum useful length, or a known paywall domain (NYT/WaPo) under a teaser-sized threshold. Full articles don't show it.
• The app links to archive.today; it does not scrape it.

For content you're entitled to, sign in via 'Use my logged-in Chrome'; otherwise the archived copy is one click away.

Article Reader v1.1.1 - self-installing launcher

run.bat now sets itself up on first run (creates the Python environment, installs dependencies, fetches the headless browser) - so downloading the source zip and double-clicking run.bat now just works. Previously it failed with 'The system cannot find the path specified' because the zip has no .venv. Requires Python on PATH; clear guidance is printed if it's missing.

Article Reader v1.1.0 - Washington Post support

Washington Post support

The Washington Post doesn't just paywall — it resets connections from automated browsers at the network level, so every fast extraction tier is dead on arrival. v1.1 handles this class of site properly:

• Auto-engaged Chrome engine. Domains with network-level bot walls (NYT, WaPo) now use the real-Chrome engine automatically — no checkbox needed — and skip the connection-blocked tiers entirely. WaPo articles went from ~100 seconds of futile timeouts to a result in under 10 seconds; NYT got faster too.
• Safer overlay stripping. The stripper no longer removes elements that wrap real article content (WaPo nests the story near paywall-named containers).
• Container fallback extraction. When readability under-extracts on heavily templated pages, the largest <article>/<main> region is used instead.

As always: logged out, you get whatever the site sends free visitors (WaPo sends only about a paragraph). Sign in once in the dedicated Chrome window — subscription, free account, or library access — for full articles.

Full changelog: v1.0.0...v1.1.0

Article Reader v1.0.0

First release. A small, self-hosted web app: paste an article URL, get a clean, popup-free, reader-mode version you can read and print — no accounts, no cloud.

Highlights

• Fast engine (default): tiered extraction — direct fetch → headless Chromium with JavaScript disabled → overlay stripping → Wayback Machine fallback — with Mozilla-style readability isolation of the article body.
• "Use my logged-in Chrome" mode: for sites that serve automated browsers a bot-challenge wall (e.g. the New York Times). Drives a real Chrome with a dedicated profile; log in once and your session persists.
• Reader view with print / save-as-PDF, dark-mode aware, images and links resolved to absolute URLs.
• An "Extraction trail" on every page showing which method recovered the text.

Install (Windows)

git clone https://github.com/jp244btc/article-reader.git
cd article-reader
py -m venv .venv
.venv\Scripts\python.exe -m pip install -r requirements.txt
.venv\Scripts\python.exe -m playwright install chromium

Then double-click run.bat and paste an article URL at http://127.0.0.1:5000.

Known limitation

If a site sends only a teaser to logged-out visitors, the withheld text never reaches your machine and cannot be recovered client-side — sign in once in the dedicated Chrome window (subscription, free account, or library access) for full articles. See the README for details.