Pull tables and links from any page

Add outputs=tables to a Fetch request and Zenrows returns every table on the page as JSON rows keyed by column heading. A few lines of Python or Node.js turn those rows into CSV files you can open in a spreadsheet. The same parameter returns links, headings, images and metadata, so one request gives you the page’s structure with no selectors to write. This uses Fetch’s outputs parameter, not Extract’s extract=auto. Uses: Fetch with outputs and Adaptive Stealth Mode (mode=auto)

Ask your AI agent

With the Zenrows MCP server connected to Claude, Cursor or any MCP client, paste this prompt. Your agent picks the right Zenrows tools and settings on its own.

Use Zenrows to read every table on https://en.wikipedia.org/wiki/List_of_tallest_buildings
and save each table that has a header row as its own CSV file.

The same prompt works across hundreds of pages: list the URLs and ask for “every table from each page, as CSV”.

Get tables, links and more in one request

outputs takes a comma-separated list. This request asks for tables, links, headings, images and metadata at once. The response is JSON with one key per filter.

import json
import requests

params = {
    "apikey": "YOUR_ZENROWS_API_KEY",
    "url": "https://www.scrapingcourse.com/table-parsing",
    "mode": "auto",
    "outputs": "tables,links,headings,images,metadata",
}

response = requests.get("https://api.zenrows.com/v1/", params=params, timeout=180)
response.raise_for_status()
print(json.dumps(response.json(), indent=2))

const params = new URLSearchParams({
  apikey: "YOUR_ZENROWS_API_KEY",
  url: "https://www.scrapingcourse.com/table-parsing",
  mode: "auto",
  outputs: "tables,links,headings,images,metadata",
});

const response = await fetch(`https://api.zenrows.com/v1/?${params}`, {
  signal: AbortSignal.timeout(180_000),
});
if (!response.ok) throw new Error(`${response.status}: ${await response.text()}`);

console.log(JSON.stringify(await response.json(), null, 2));

curl "https://api.zenrows.com/v1/?apikey=YOUR_ZENROWS_API_KEY&url=https%3A%2F%2Fwww.scrapingcourse.com%2Ftable-parsing&mode=auto&outputs=tables,links,headings,images,metadata"

zenrows extract https://www.scrapingcourse.com/table-parsing --outputs tables,headings --out tables.json

In the CLI, zenrows extract --outputs sends the outputs parameter, see the CLI guide. The response is JSON with one key per filter, and a filter with nothing to return on a page, such as metadata here, is left out. Each table carries Dimensions (row and column counts, and whether it has a header row), Headings (the column names in page order) and Content (one object per row keyed by heading). See the returned table in Get common data with outputs.

Save every table as CSV

Loop over tables, skip layout tables that have no header row, and write each data table to its own file. Using Headings as the column order keeps the columns as they appear on the page.

import csv
import requests

params = {
    "apikey": "YOUR_ZENROWS_API_KEY",
    "url": "https://www.scrapingcourse.com/table-parsing",
    "mode": "auto",
    "outputs": "tables",
}

response = requests.get("https://api.zenrows.com/v1/", params=params, timeout=180)
response.raise_for_status()
tables = response.json()["tables"]

saved = 0
for table in tables:
    rows = table["Content"]
    if not table["Dimensions"]["Headings"] or len(rows) < 2:
        continue  # skip layout tables with no header row
    columns = list(dict.fromkeys(table["Headings"] + [key for row in rows for key in row]))
    saved += 1
    with open(f"table-{saved}.csv", "w", newline="", encoding="utf-8") as f:
        writer = csv.DictWriter(f, fieldnames=columns)
        writer.writeheader()
        writer.writerows(rows)
    print(f"table-{saved}.csv: {len(rows)} rows, columns {columns}")

import { writeFile } from "node:fs/promises";

const params = new URLSearchParams({
  apikey: "YOUR_ZENROWS_API_KEY",
  url: "https://www.scrapingcourse.com/table-parsing",
  mode: "auto",
  outputs: "tables",
});

const response = await fetch(`https://api.zenrows.com/v1/?${params}`, {
  signal: AbortSignal.timeout(180_000),
});
if (!response.ok) throw new Error(`${response.status}: ${await response.text()}`);
const { tables } = await response.json();

// Quote a CSV cell when it holds a comma, a quote or a line break
const cell = (value) => {
  const text = String(value ?? "");
  return /[",\n]/.test(text) ? `"${text.replaceAll('"', '""')}"` : text;
};

let saved = 0;
for (const table of tables) {
  const rows = table.Content;
  if (!table.Dimensions.Headings || rows.length < 2) continue; // skip layout tables
  const columns = [...new Set([...table.Headings, ...rows.flatMap((row) => Object.keys(row))])];
  const lines = [columns, ...rows.map((row) => columns.map((c) => row[c]))].map((r) => r.map(cell).join(","));
  saved += 1;
  await writeFile(`table-${saved}.csv`, lines.join("\n") + "\n");
  console.log(`table-${saved}.csv: ${rows.length} rows, columns ${columns.join(", ")}`);
}

Example output

The script prints one line per saved file and writes the table with its page column order:

Output (Python)

table-1.csv: 15 rows, columns ['Product ID', 'Name', 'Category', 'Price', 'In Stock']

table-1.csv (first rows)

Product ID,Name,Category,Price,In Stock
001,Laptop,Electronics,$999.99,Yes
002,Smartphone,Electronics,$599.99,Yes
003,Headphones,Audio,$149.99,No

Pages with many tables

The same script handles pages with dozens of tables. Pointed at https://en.wikipedia.org/wiki/List_of_tallest_buildings, it skips the legend and navigation tables and saves each data table:

Output (Python, trimmed)

table-1.csv: 43 rows, columns ['Rank', 'Building', 'City', 'Country', 'Height', 'Floors', 'Built']
table-2.csv: 20 rows, columns ['Building', 'City', 'Country', 'Height', 'Floors', 'Built']
table-3.csv: 64 rows, columns ['Rank', 'Building', 'City', 'Country', 'Height (m)', 'Height (ft)', 'Floors', 'Built', 'Reference']
table-4.csv: 46 rows, columns ['Building', 'Country', 'City', 'Planned architectural height', 'Floors', 'Planned completion', 'Ref.']

A table without a header row comes back with "Headings": false in Dimensions and generic keys such as column_1, column_2. Keep those too when you want them: drop the Headings check.

Turn links into a clean list

outputs=links returns every href on the page, as written in the HTML: relative paths, # anchors and repeats included. Resolve each against the page URL, drop the fragment and deduplicate. Here the catalog page’s 59 links become 29 unique URLs.

from urllib.parse import urljoin, urldefrag, urlparse
import requests

PAGE = "https://www.scrapingcourse.com/ecommerce/"

params = {
    "apikey": "YOUR_ZENROWS_API_KEY",
    "url": PAGE,
    "mode": "auto",
    "outputs": "links,images,metadata",
}

response = requests.get("https://api.zenrows.com/v1/", params=params, timeout=180)
response.raise_for_status()
data = response.json()

links = []
for href in data["links"]:
    url, _ = urldefrag(urljoin(PAGE, href))  # absolute, no #fragment
    if urlparse(url).scheme in {"http", "https"}:
        links.append(url)
unique = sorted(set(links))

site = urlparse(PAGE).netloc
internal = [u for u in unique if urlparse(u).netloc == site]
print(f"{len(data['links'])} links on the page, {len(unique)} unique, {len(internal)} on {site}")
print("\n".join(internal[:8]))
print(f"{len(set(data['images']))} images")
print(data["metadata"])

with open("links.txt", "w") as f:
    f.write("\n".join(unique) + "\n")

import { writeFile } from "node:fs/promises";

const PAGE = "https://www.scrapingcourse.com/ecommerce/";
const params = new URLSearchParams({
  apikey: "YOUR_ZENROWS_API_KEY",
  url: PAGE,
  mode: "auto",
  outputs: "links,images,metadata",
});

const response = await fetch(`https://api.zenrows.com/v1/?${params}`, {
  signal: AbortSignal.timeout(180_000),
});
if (!response.ok) throw new Error(`${response.status}: ${await response.text()}`);
const data = await response.json();

const unique = new Set();
for (const href of data.links) {
  const url = new URL(href, PAGE); // absolute
  url.hash = ""; // no #fragment
  if (url.protocol === "http:" || url.protocol === "https:") unique.add(url.href);
}
const links = [...unique].sort();
const internal = links.filter((u) => new URL(u).host === new URL(PAGE).host);

console.log(`${data.links.length} links on the page, ${links.length} unique, ${internal.length} on ${new URL(PAGE).host}`);
await writeFile("links.txt", links.join("\n") + "\n");

Output (Python, trimmed)

59 links on the page, 29 unique, 28 on www.scrapingcourse.com
https://www.scrapingcourse.com/ecommerce/
https://www.scrapingcourse.com/ecommerce/?add-to-cart=2740
https://www.scrapingcourse.com/ecommerce/?add-to-cart=2765
https://www.scrapingcourse.com/ecommerce/cart/
https://www.scrapingcourse.com/ecommerce/checkout/
https://www.scrapingcourse.com/ecommerce/my-account/
https://www.scrapingcourse.com/ecommerce/page/10/
https://www.scrapingcourse.com/ecommerce/page/11/
16 images
['viewport: width=device-width, initial-scale=1', 'robots: max-image-preview:large', 'generator: WordPress 6.4.3', 'generator: WooCommerce 8.6.1']

metadata returns each <meta> tag as a name: content string. Split on the first : to get a dictionary.

Customize it

Change	How
Every filter at once	`outputs=*`, see Output filters
Typed numbers instead of `"$999.99"` strings	Validate the rows with a schema, see Validate extracted data with a schema
One specific table or column	Use `css_extractor` with a selector such as `table#prices td:nth-child(2)`, see Extract data with CSS selectors
Tables that load after the page	Use `js_render=true` with `wait_for=table` instead of `mode=auto`, see Wait for selector
Tables from thousands of pages	Put `"outputs": "tables"` in a Batch job’s `zenrows_params`, see Scrape a list of URLs
Every URL on a whole site	Use Crawl, see Map every URL on a website

Troubleshooting

tables is missing from the response: the page has no <table> elements in its final HTML. Data shown in a grid of <div>s is read with css_extractor instead.
A CSV has columns named column_1, column_2: that table has no header row. Rename the columns in your script, or skip it with the Headings check above.
A table repeats a column heading, such as two “Height” columns in different units: rows are keyed by heading. Read that table with css_extractor to get every column by position.
Links look like /cart/ or #content: links returns each href as written. Resolve them against the page URL as shown in Turn links into a clean list.
The request takes long to return: set your HTTP client timeout to 180 seconds or more so mode=auto has time to escalate.

Extract data with CSS selectors

Pick exact elements and attributes with css_extractor.

Validate extracted data with a schema

Turn price strings into numbers and Yes/No into booleans.

Map every URL on a website

Collect every page URL of a site with Crawl.

Turn search and listing pages into JSON

Get every item on search and category pages as JSON.

Start Here

Foundations Course

Fetch Recipes

Extract Recipes

Crawl Recipes

Batch Recipes

Browser Recipes

Pipelines

Use Cases

By Surface

Production

Pull tables and links from any page

Ask your AI agent

Get tables, links and more in one request

Save every table as CSV

Example output

Pages with many tables

Turn links into a clean list

Customize it

Troubleshooting

Extract data with CSS selectors

Validate extracted data with a schema

Map every URL on a website

Turn search and listing pages into JSON

​Ask your AI agent

​Get tables, links and more in one request

​Save every table as CSV

​Example output

​Pages with many tables

​Turn links into a clean list

​Customize it

​Troubleshooting

​Related recipes

Extract data with CSS selectors

Validate extracted data with a schema

Map every URL on a website

Turn search and listing pages into JSON

Ask your AI agent

Get tables, links and more in one request

Save every table as CSV

Example output

Pages with many tables

Turn links into a clean list

Customize it

Troubleshooting

Related recipes