Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Converting a webpage into clean Markdown for an LLM is a three-stage job: fetch the source HTML, extract the page’s useful content, then serialize that reduced HTML as Markdown. A local serializer can finish in a few milliseconds when the HTML is already in memory, but downloading a URL and running a browser usually take far longer. Treat “5ms” as a conversion-stage target, not a promise for the complete URL-to-Markdown trip.
The most dependable Node.js design uses Mozilla Readability for article extraction and Turndown for HTML-to-Markdown conversion. Keep static fetching as the default, use browser rendering only for JavaScript-built pages, preserve the original URL for resolving links and images, and test the output on the page types your application actually processes.
What does a webpage-to-Markdown pipeline actually do?
A URL-to-Markdown converter solves several different problems that are easy to confuse. The network layer obtains HTML, the extraction layer decides which part is the main content, and the serialization layer expresses that content in Markdown. Keeping those responsibilities separate makes speed and quality failures much easier to diagnose.
1. Fetch the page and retain its URL
Start with the original URL and the response HTML. The URL is needed later to turn relative links such as /guide/install and relative images into absolute addresses. Follow redirects deliberately and record the final URL if your downstream system needs canonical metadata.
#1 Best Overall
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
2. Extract useful content from the DOM
Most pages contain navigation, cookie banners, footers, related-content cards, advertisements, and hidden interface text. Mozilla Readability analyzes a DOM and returns article-oriented fields such as title, byline, excerpt, site name, language, direction, and cleaned content. It is an extractor, not a Markdown serializer.
Readability can modify the DOM supplied to it. The Mozilla Readability documentation states that “The parse() method works by modifying the DOM.” Pass a clone when another part of your application must retain the untouched document.
3. Serialize the extracted HTML as Markdown
Turndown accepts an HTML string or DOM node and converts it to Markdown. Its rules and plugins determine how headings, links, tables, code blocks, unknown elements, blank elements, kept elements, and removed elements are handled. A converter’s defaults are not a guarantee that every site’s special structure will survive, so add rules for structures that matter to your corpus.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
How do I build the conversion in Node.js?
The following implementation keeps static fetching fast, extracts the main content, resolves relative URLs, and emits Markdown plus useful metadata. It uses Node.js with a fetch implementation available in current Node releases, jsdom, Mozilla Readability, and Turndown.
Install the conversion dependencies
npm install jsdom @mozilla/readability turndownUse a maintained Node.js release with a standards-compatible fetch. If your runtime does not provide one, add a fetch library and keep that dependency choice explicit in your deployment documentation.
Fetch, clone, extract, and convert
import { JSDOM } from 'jsdom';
import { Readability } from '@mozilla/readability';
import TurndownService from 'turndown';
const sourceUrl = 'https://example.com/article';
async function pageToMarkdown(url) {
const response = await fetch(url, {
headers: { 'User-Agent': 'Mozilla/5.0 content-extractor/1.0' }
});
if (!response.ok) {
throw new Error(`HTTP ${response.status} while fetching ${url}`);
}
const html = await response.text();
const dom = new JSDOM(html, { url });
const readerDocument = dom.window.document.cloneNode(true);
const article = new Readability(readerDocument).parse();
if (!article || !article.content) {
throw new Error('Main content could not be extracted');
}
const contentDom = new JSDOM(article.content, { url });
for (const element of contentDom.window.document.querySelectorAll('[href], [src]')) {
if (element.hasAttribute('href')) {
element.setAttribute('href', new URL(element.getAttribute('href'), url).href);
}
if (element.hasAttribute('src')) {
element.setAttribute('src', new URL(element.getAttribute('src'), url).href);
}
}
const turndown = new TurndownService({
headingStyle: 'atx',
codeBlockStyle: 'fenced',
bulletListMarker: '-'
});
const markdown = turndown.turndown(contentDom.window.document.body.innerHTML);
return {
sourceUrl: url,
title: article.title,
byline: article.byline,
excerpt: article.excerpt,
markdown
};
}
pageToMarkdown(sourceUrl).then(result => {
console.log(result.markdown);
});The example intentionally preserves metadata separately from the Markdown body. A retrieval or indexing system can store the source URL, title, byline, excerpt, language, and fetch timestamp as fields instead of forcing every attribute into prose.
Keep the original DOM when another task needs it
Constructing readerDocument from dom.window.document.cloneNode(true) protects the fetched document from Readability’s in-place changes. If you do not need the original tree, parsing directly is simpler, but make that a conscious trade-off rather than an accidental side effect.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Add converter rules only for observed structures
Turndown’s custom-rule API is useful for callouts, definition lists, embedded widgets, or site-specific code containers. Add a rule when a representative page demonstrates a real loss, then add that page to regression tests. Avoid a large collection of speculative rules: each one can change output on pages that previously converted correctly.
When should I use a browser instead of static HTML?
Use static fetching first. Escalate to a browser renderer when the meaningful text is absent from the initial response because JavaScript builds the page after load. Rendering every URL by default increases latency, memory use, operational complexity, and failure modes.
Rank #2
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
Static path
- Fetch the response HTML.
- Parse it with a DOM implementation.
- Run Readability and Turndown.
- Record stage timings and output diagnostics.
This path is usually the cheapest and fastest option for server-rendered articles, documentation, reference pages, and listings whose content is present in the response.
Browser fallback
- Run the static attempt and inspect whether the extracted text is empty, implausibly short, or missing expected headings.
- Open the page in a controlled browser only for those failures.
- Wait for the application condition your site requires, such as a known content selector or a network-idle policy, rather than applying an unlimited sleep.
- Capture the rendered DOM, then pass that HTML through the same extraction and Markdown stages.
- Apply timeouts, concurrency limits, and cancellation so a stalled page cannot consume a worker indefinitely.
A browser should be a fallback, not a quality shortcut. It can reveal client-rendered text, but it may also expose transient UI, personalization, consent dialogs, or duplicated navigation. Keep extraction and validation in place after rendering.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Does “5ms” mean a URL can become Markdown in five milliseconds?
No. The phrase normally describes local conversion work after HTML has already been obtained. Network transfer, redirects, TLS, JavaScript execution, and browser startup commonly dominate end-to-end latency.
| Measurement | Scope and conditions | What it demonstrates |
|---|---|---|
| 1.793 ms average | h2m-parser project benchmark across 95 files, without Readability; publication date was not stated in the reported result. | Raw conversion can be very fast on a fixed local corpus. |
| 13.927 ms average | h2m-parser project benchmark across the same 95 files, with Readability; publication date was not stated in the reported result. | Extraction adds measurable CPU time even when no network is involved. |
| 6.270 ms mean | h2m-parser benchmark on one 88 KB fixture with Readability; publication date was not stated in the reported result. | A particular input can land near a 5 ms conversion target, but it is not a universal rate. |
| 0.4–0.5 seconds per page | RenderScreenshot’s 20-page static pipeline, including download, reported October 18, 2026. | Retrieval makes the complete static path much slower than a local serializer. |
| About 1.5 seconds to network idle plus about 0.1 seconds to convert | RenderScreenshot’s Chrome-rendered path on the same 20-page test, reported October 18, 2026. | Browser execution adds substantial latency before conversion begins. |
Report these stages separately: fetch, browser render, extraction, serialization, and total elapsed time. Include input size, library versions, runtime, hardware or container limits, and whether the measurement includes page retrieval. A single “milliseconds per page” number hides the bottleneck your team needs to fix.
Which extraction approach produces the cleanest Markdown?
There is no universal winner. Coverage, boilerplate leakage, structural fidelity, JavaScript support, latency, and operational cost can point to different choices on different page sets.
| Approach | Published result | Metric and qualification | How to interpret it |
|---|---|---|---|
| Deterministic DOM walk | 87.2% visible-text coverage; 11.0% unseen-text leakage | URLpipe benchmark, September 2026, across 33 pages; implementations were re-created against the same fetched pages rather than measured as hosted vendor APIs. | Strong control and speed may come with missed article structure; inspect page-type variance. |
| Readability | 88.2% visible-text coverage; 21.3% unseen-text leakage | URLpipe benchmark, September 2026, same 33-page methodology. | Higher measured coverage in that sample did not eliminate hidden or non-visible text leakage. |
| Naive static conversion | 0.74 mean structural score | RenderScreenshot’s 20-page test in 2026; score tracks preservation of words, headings, code blocks, and tables, not general semantic correctness. | Useful baseline, not a quality guarantee. |
| DOM-cleaned static conversion | 0.79 mean structural score | Same RenderScreenshot test and scoring method. | Cleaning the DOM improved measured structure in that sample. |
| DOM-cleaned conversion with browser rendering | 0.84 mean structural score | Same RenderScreenshot test and scoring method; individual pages varied. | Rendering helped the tested pages, while adding browser latency and complexity. |
Use these figures as examples of methodology, not as a universal ranking. Re-test on your own pages and define acceptance thresholds for main-text coverage, navigation leakage, heading order, code fences, table structure, links, metadata, and end-to-end latency.
How do I preserve headings, tables, code, and links?
Preserve structure deliberately because LLM retrieval quality depends on more than visible words. A clean paragraph stream that loses headings, code boundaries, or table relationships can be less useful than a slightly longer document with faithful structure.
Headings
Check that the extracted title is not repeated as an unrelated heading and that heading levels do not jump because a site used styling rather than semantic elements. Keep the original order; do not invent a hierarchy from font sizes.
Rank #3
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
Code blocks
Verify that fenced code retains language hints where the source provides them, and that indentation is not collapsed into ordinary paragraphs. Test documentation pages with inline code, multiline examples, shell commands, and tabs or spaces that carry meaning.
Tables
Markdown tables work best for simple rectangular data. Complex cells, row spans, nested lists, and layout tables may need a custom Turndown rule or an alternate representation. Compare column counts and header text before and after conversion rather than assuming a visually correct HTML table will serialize cleanly.
Links and images
Resolve relative href and src values against the retained source URL. Decide whether your downstream model needs image URLs, alt text, captions, or all three. Remove tracking parameters only if your product policy explicitly permits URL rewriting; otherwise preserve the canonical target exactly.
Metadata
Store title, byline, excerpt, language, direction, source URL, final URL after redirects, and retrieval time beside the Markdown. Metadata is easier to filter and cite when it is not buried in a generated preamble.
How should I validate conversion quality?
Build a small, permanent corpus that represents the pages your system receives. Re-run it whenever you change extractor settings, converter versions, browser versions, or custom rules.
Include these page types
- Long-form articles with author and publication metadata.
- Reference tables with headers, multiline cells, and links.
- Developer documentation containing nested headings and code fences.
- Listings with pagination, filters, cards, and repeated navigation.
- Client-rendered pages whose initial HTML contains little meaningful text.
Check each output automatically and manually
- The title and first heading are sensible and not duplicated boilerplate.
- Main text is present, while menus, footers, cookie prompts, and recommendation rails are absent.
- Heading order is intact and links resolve to absolute URLs.
- Code fences, inline code, lists, tables, images, captions, and block quotes survive.
- No hidden text, script content, style content, or repeated navigation enters the Markdown.
- Metadata fields are populated or explicitly null rather than guessed.
- Static failure triggers browser fallback only when a defined condition is met.
Keep golden outputs for a few pages so reviewers can see exactly what changed. Structural scores can reveal regressions, but they do not establish semantic correctness; a human review of representative failures remains necessary.
How do I make the pipeline safe for untrusted pages?
Readability is not a sanitizer. Its output can contain markup that is unsafe to place directly into an executable or rendered context. Mozilla’s documentation recommends sanitizing Readability output before using it where markup could execute.
- Isolate fetching and parsing from systems that execute HTML, scripts, or embedded URLs.
- Sanitize extracted HTML before passing it to any renderer or HTML-producing step.
- Keep Markdown as data; do not assume that Markdown rendering removes dangerous links or embedded HTML.
- Apply outbound request controls, size limits, timeouts, and redirect policies to prevent a page from turning the fetcher into an internal-network proxy.
- Log the source URL and failure reason without storing sensitive response bodies in unrestricted logs.
The exact sanitizer configuration depends on where the result is rendered. A Markdown document used only for tokenization has a different threat model from one displayed in a browser, but both should be treated as untrusted input until policy checks are complete.
Rank #4
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Broad Compatible: The usb c to usb adapter widely supports all USB C/ USB A devices like laptops, tablets, cellphones, car chargers, and phone chargers. Such as compatible with MacBook Pro/Air 2023/2022, Thunderbolt 4/3 Devices,Apple MagSafe Watch 9/8/7/SE/Ultra, iPad Pro 2022/2021, Samsung Galaxy S23/S20/S10, and iPhone 17/16/15 Pro. Plug and play
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
Should I use a hosted webpage-to-Markdown endpoint?
A hosted endpoint can remove browser orchestration, queueing, and maintenance from your application. Cloudflare documents a Markdown endpoint that accepts either a URL or HTML, and webforai describes reader-mode conversion with JavaScript rendering. Those published capabilities do not prove that either service is best for every site, nor do they establish affiliate availability.
Before adopting a hosted service, verify its page-size limits, authentication model, retention policy, geographic processing, JavaScript support, timeout behavior, output schema, retry semantics, and treatment of robots or access controls. Send a fixed evaluation corpus through the service and compare the same quality axes you use for an in-house pipeline. Keep a local fallback if a provider outage would block ingestion.
Free tools Windows power users keep installed
One-click scans. No signup required.
What commonly goes wrong, and how do I fix it?
The Markdown contains navigation and footer text
Extraction probably ran on the full document or the page’s semantic cues were weak. Confirm that Readability received a document rather than an HTML fragment, inspect the extracted HTML before serialization, and add a site-specific cleanup rule only after confirming the pattern across several pages.
The result is almost empty
The content may be client-rendered, blocked by an access response, or loaded inside an embedded frame. Check the fetched status, response size, and extracted text length, then route the page to a browser fallback when the initial HTML lacks the expected content.
Relative links point to the wrong place
The converter was given no base URL or the URL was lost during a redirect. Construct the DOM with the source or final URL, resolve every relative href and src before Turndown, and retain the URL in output metadata.
Tables or code blocks collapse
Inspect the extracted HTML, not only the final Markdown. If the extractor removed the structure, adjust extraction or add a targeted cleanup rule; if the HTML is correct, configure Turndown rules and test the target Markdown parser’s table and fence behavior.
Recommended Free Tools
Browser rendering is too slow
Do not render every page. Gate the fallback on measurable static failures, reuse browser workers safely, cap concurrency, apply navigation and total-job timeouts, and measure browser time separately from extraction and serialization.
Output changes after a dependency upgrade
Pin versions, retain representative golden outputs, and review diffs for headings, links, tables, code, and boilerplate. A small extractor change can alter both coverage and leakage, so treat upgrades as quality changes rather than routine formatting updates.
Best Value
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
What should I measure in production?
Capture a timing and quality record for every conversion class without retaining more page content than your policy allows.
- Fetch status, response bytes, redirect count, and fetch duration.
- Whether static or browser rendering was used, including render duration and timeout reason.
- Extraction duration, extracted character or word count, and whether Readability returned content.
- Serialization duration, Markdown size, and counts of headings, links, code fences, tables, and images.
- Warnings for unresolved URLs, duplicate navigation, missing title, suspicious hidden text, or sanitizer removals.
- Library and browser versions so benchmark results remain reproducible.
These measurements let you answer practical questions: Is the five-millisecond target being missed in serialization, or was the page slow to download? Did a browser fallback improve coverage, or merely add interface text? Which templates deserve custom rules?
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWhat is the practical architecture for a reliable converter?
Use a staged pipeline with explicit handoffs: fetcher, optional renderer, extractor, URL normalizer, sanitizer, Markdown serializer, validator, and metadata store. Each stage should return both its result and diagnostics so failures are explainable instead of appearing as one generic “conversion failed” error.
Start with static HTML, Readability, and Turndown. Add browser rendering only for proven client-rendered cases. Preserve headings, tables, code, links, images, and metadata; validate against a representative corpus; and report local conversion time separately from retrieval and rendering. That design delivers the speed implied by “5ms” where it is technically realistic without pretending that an arbitrary public URL can be fetched, rendered, extracted, and serialized in five milliseconds.
Frequently Asked Questions
Can I convert HTML that is already stored in a database without fetching a URL?
Yes. Pass the stored HTML to the DOM parser and keep a separate base URL if the document contains relative links or images. Without a base URL, those references cannot be resolved reliably, even though text conversion itself can proceed.
How can I keep source citations when the model receives Markdown?
Store the canonical or final source URL and page metadata beside the Markdown record, then attach stable document IDs during chunking. This avoids adding noisy citation prose to every chunk while preserving a traceable origin for retrieval and answer generation.
What is a good way to decide whether a page needs browser rendering?
Use a measurable gate such as missing expected selectors, very low extracted text, or a known client-rendered template. Keep the gate template-aware and log every escalation so you can remove unnecessary browser work as site behavior changes.
Should Markdown conversion happen before or after chunking for an LLM?
Convert and validate the complete extracted document first, then chunk it using heading and code-block boundaries where possible. Chunking raw HTML can split tags and table structure, while chunking after serialization gives the indexer a stable representation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
