Skip to main content

FreeAI::Alice - Alice AI Scraper

Alice

Scraper overview​

The Alice scraper retrieves responses from the Alice AI via the alice.yandex.ru website, along with a list of sources it relies on. The result includes the response text and source cards: address, title, and description.

Queries are written in natural language, just like a message in the Alice chat. No registration is required: the scraper opens the main page and obtains the session identifier automatically.

If a complete response is not received within 2 minutes, the query is retried. The number of attempts is determined by the global Proxy retries setting.

Results can be saved in the desired format using the Template Toolkit engine: text, CSV, or JSON.

Data collected​

  • Response text
  • Links, titles, and descriptions of sources

Capabilities​

  • Natural language queries
  • Advanced response mode
  • Collecting sources separately from the response text

Use cases​

  • Collecting answers for thematic queries for knowledge bases, content plans, and FAQs
  • Extracting source links with titles and descriptions
  • Monitoring topics with reference to pages cited by Alice
  • Quickly checking which sites Alice considers sources for a query

Queries​

Queries are specified as text, as if written in the Alice chat, for example:

Dollar to ruble exchange rate today
Can you give links to new movies 2026 any movies
What is a scraper?

Results​

By default, the query and answer are output:

$query
$answer

Example:

Can you give links to new movies 2026 any movies
Sure! Here are some interesting movies expected in 2026:

1. **Josephine** - drama, USA
2. **Wild Horse** - drama, UK
3. **Spider-Man: Brand New Day** - action, USA
...

Sources are not included in this text. They are located in the $p1.sources array.

Output results examples​

Answer and sources​

Result format:

ANSWER:
$p1.answer

SOURCES:
$p1.sources.format('Type:$type \n Link: $link \n $anchor \n Snippet:$snippet\n')

Example result:

ANSWER:
Sure! Here are some interesting movies expected in 2026:
...

SOURCES:
Type:source
 Link: https://www.film.ru/a-z/movies/2026
  Best movies of 2026
  Snippet:We have collected 2026 movies for you based on ratings from Film.ru authors
Type:source
 Link: https://www.afisha.ru/movie/y2026/
  Movies and series of 2026 - Afisha-Kino
  Snippet:Movies and series of 2026: choose cinema by genre or rating...

Exporting a list of links​

Result format:

$p1.sources.format('$link\n')

Example result:

https://www.film.ru/a-z/movies/2026
https://www.afisha.ru/movie/y2026/
https://www.kinoafisha.info/releases/2026/
https://ru.wikipedia.org/wiki/Категория:Фильмы_2026_года

CSV output of links, titles, and descriptions​

Result format:

[% FOREACH item IN p1.sources;
tools.CSVline(loop.count, item.link, item.anchor, item.snippet);
END %]

Example result:

1,https://www.film.ru/a-z/movies/2026,"Best movies of 2026","We have collected 2026 movies for you based on ratings from Film.ru authors"
2,https://www.afisha.ru/movie/y2026/,"Movies and series of 2026 - Afisha-Kino","Movies and series of 2026: choose cinema by genre or rating"

Specify the csv extension in the results filename.

A single query waits for a response for no longer than 2 minutes. Set the task timeout with a margin greater than 2 minutes multiplied by the number of retries.