octagon-mcp/dist/docs/search.js

MCP server for Octagon API. Provides specialized AI agents for investment research of public and private markets.

const STOP_WORDS = new Set([
    "and",
    "api",
    "can",
    "complete",
    "docs",
    "example",
    "examples",
    "for",
    "from",
    "guide",
    "how",
    "integration",
    "the",
    "this",
    "use",
    "using",
    "with",
]);
function tokenize(value) {
    return Array.from(new Set(value
        .toLowerCase()
        .split(/[^a-z0-9]+/)
        .filter(token => token.length >= 2)));
}
function distinctiveTokens(tokens) {
    const distinctive = tokens.filter(token => token.length >= 3 && !STOP_WORDS.has(token));
    return distinctive.length > 0 ? distinctive : tokens;
}
function countMatches(value, tokens) {
    const lower = value.toLowerCase();
    return tokens.reduce((count, token) => count + (lower.includes(token) ? 1 : 0), 0);
}
function createSnippet(entry, tokens) {
    const haystack = entry.content ?? entry.summary;
    if (!haystack) {
        return undefined;
    }
    const lower = haystack.toLowerCase();
    const index = tokens
        .map(token => lower.indexOf(token))
        .filter(position => position >= 0)
        .sort((a, b) => a - b)[0];
    if (index === undefined) {
        return haystack.slice(0, 240).trim();
    }
    const start = Math.max(0, index - 100);
    const end = Math.min(haystack.length, index + 180);
    const prefix = start > 0 ? "..." : "";
    const suffix = end < haystack.length ? "..." : "";
    return `${prefix}${haystack.slice(start, end).trim()}${suffix}`;
}
export function searchDocsEntries(entries, { query, section, limit, includeSnippets, }) {
    const tokens = tokenize(query);
    const normalizedSection = section?.trim().toLowerCase();
    if (tokens.length === 0) {
        return [];
    }
    const requiredTokens = distinctiveTokens(tokens);
    return entries
        .filter(entry => normalizedSection
        ? entry.section.toLowerCase().includes(normalizedSection)
        : true)
        .map(entry => {
        const titleScore = countMatches(entry.title, tokens) * 8;
        const sectionScore = countMatches(entry.section, tokens) * 4;
        const summaryScore = countMatches(entry.summary ?? "", tokens) * 3;
        const contentScore = countMatches(entry.content ?? "", tokens);
        const distinctiveScore = countMatches(entry.title, requiredTokens) * 10 +
            countMatches(entry.section, requiredTokens) * 4 +
            countMatches(entry.summary ?? "", requiredTokens) * 3 +
            countMatches(entry.content ?? "", requiredTokens);
        const exactTitleBonus = entry.title
            .toLowerCase()
            .includes(query.toLowerCase())
            ? 12
            : 0;
        const hasDistinctiveMatch = distinctiveScore > 0;
        const score = hasDistinctiveMatch
            ? titleScore +
                sectionScore +
                summaryScore +
                contentScore +
                distinctiveScore +
                exactTitleBonus
            : 0;
        return {
            entry,
            score,
            snippet: includeSnippets ? createSnippet(entry, tokens) : undefined,
        };
    })
        .filter(result => result.score > 0)
        .sort((a, b) => b.score - a.score || a.entry.title.localeCompare(b.entry.title))
        .slice(0, limit);
}