Q&A with Wikipedia 📚

---
title: "Q&A with Wikipedia 📚"
description: "Using code blocks to query Wikipedia"
input:
  - name: question
    type: string
    description: "The question to answer"
---

<!-- The code block is connected to the web so it can be used to hit external APIs too. This can be very useful in a number of ways. Here we show how it can be used to query for up-to-date information from Wikipedia. -->

Retrieval assisted (RAG) question answering using Wikipedia.
e.g. Try asking "When did OceanGate sink?", "How tall is the tallest penguin species?"

Question: {% $frontmatter.input.question %}

<!-- First we get the language model to generate a search term -->

Let's search Wikipedia to answer this question. What's the best search term to use? Output just the search term.

{% ai #search_term model="openai/gpt-4o-mini" /%}

<!-- This chunk of code will hit Wikipedia's API and extract the top 3 articles -->


`\`\`\js {% #wikipedia_results %}

// Read the term generated by the language model
let input;
try {
    input = aimVariables.search_term.result.trim().replace(/^"+|"+$/g, '');
} catch (error) {
    console.error("Error reading search term:", error);
    return { result: "", error: error.toString() };
}

console.log("Searching for the term", input);

try {
    const maxResults = 3;
    const searchTerm = encodeURIComponent(input.trim());
    const url = `https://en.wikipedia.org/w/api.php?format=json&action=query&list=search&srsearch=${searchTerm}&srlimit=${maxResults}&utf8=&origin=*`;
    const response = await fetch(url);
    const data = await response.json();

    const fetchExtract = async (title) => {
        console.log("Retrieving", title);
        const extractUrl = `https://en.wikipedia.org/w/api.php?format=json&action=query&prop=extracts&exintro&explaintext&titles=${encodeURIComponent(title)}&redirects=1&origin=*`;
        const extractResponse = await fetch(extractUrl);
        const extractData = await extractResponse.json();
        const pageId = Object.keys(extractData.query.pages)[0];
        const extract = extractData.query.pages[pageId].extract;
        return [title, extract];
    }

    console.log("data", data);

    try {
        if (data?.query?.search?.length > 0) {
            console.log("Got some results extracting top", data?.query?.search?.length);
            const extracts = await Promise.all(data?.query?.search?.map(result => fetchExtract(result.title)));
            return { result: extracts, error: null };
        } else {
            return { result: "No results found.", error: null };
        }
    } catch (error) {
        console.error("Error extracting Wikipedia results:", error);
        return { result: "", error: error.toString() };
    }
} catch (error) {
    console.error("Error processing Wikipedia results:", error);
    return { result: "", error: error.toString() };
}
    
`\`\`\
<!-- The code returns the top 3 Wikipedia articles that match the search term -->

## Answer:

Based on the information above give the most helpful answer to the question.


Results: {% debug($wikipedia_results) %}

Question: {% $frontmatter.input.question %}
{% ai #answer model="openai/gpt-4o-mini" /%}

{% $answer.result %}