Claude Code. Any Model. The most powerful AI coding agent now speaks every language.
/**
* Calculate fuzzy match score for a string against a query
* Returns a score from 0 to 1 (1 being perfect match)
* Returns 0 if no match found
*/
export function fuzzyScore(text: string, query: string): number {
if (!text || !query) return 0;
const t = text.toLowerCase();
const q = query.toLowerCase();
// Exact match
if (t === q) return 1.0;
// Start match
if (t.startsWith(q)) return 0.9;
// Word boundary match (e.g. "claude-3" matches "3")
if (t.includes(` ${q}`) || t.includes(`-${q}`) || t.includes(`/${q}`)) return 0.8;
// Contains match
if (t.includes(q)) return 0.6; // base score for inclusion
// Separator-normalized match: treat spaces, hyphens, dots, underscores as equivalent
// This lets "glm 5" match "glm-5", "gpt4o" match "gpt-4o", etc.
const normSep = (s: string) => s.replace(/[\s\-_.]/g, "");
const tn = normSep(t);
const qn = normSep(q);
if (tn === qn) return 0.95;
if (tn.startsWith(qn)) return 0.85;
if (tn.includes(qn)) return 0.65;
// Subsequence match (fuzzy)
let score = 0;
let tIdx = 0;
let qIdx = 0;
let consecutive = 0;
while (tIdx < t.length && qIdx < q.length) {
if (t[tIdx] === q[qIdx]) {
score += 1 + consecutive * 0.5; // Bonus for consecutive matches
consecutive++;
qIdx++;
} else {
consecutive = 0;
}
tIdx++;
}
// Only count as match if we matched all query chars
if (qIdx === q.length) {
// Normalize score between 0.1 and 0.5 depending on compactness
// Higher score if match spans shorter distance
const compactness = q.length / (tIdx + 1); // +1 to avoid division by zero, though tIdx always >= 1 here
return 0.1 + 0.4 * compactness * (score / (q.length * 2)); // Heuristic
}
return 0;
}
/**
* Format a number as currency
*/
export function formatCurrency(amount: number): string {
if (amount === 0) return "FREE";
return `$${amount.toFixed(2)}`;
}