60 lines
1.8 KiB
TypeScript
60 lines
1.8 KiB
TypeScript
import { socialGraphInstance } from "@snort/system";
|
|
import { useMemo } from "react";
|
|
|
|
import fuzzySearch from "@/Db/FuzzySearch";
|
|
|
|
export default function useProfileSearch(search: string) {
|
|
const results = useMemo(() => {
|
|
return userSearch(search);
|
|
}, [search]);
|
|
|
|
return results;
|
|
}
|
|
|
|
export function userSearch(search: string) {
|
|
const searchString = search.trim();
|
|
const fuseResults = fuzzySearch.search(searchString);
|
|
|
|
const followDistanceNormalizationFactor = 3;
|
|
const seenIds = new Set();
|
|
|
|
const combinedResults = fuseResults
|
|
.map(result => {
|
|
const fuseScore = result.score === undefined ? 1 : result.score;
|
|
|
|
const followDistance = wotScore(result.item.pubkey) / followDistanceNormalizationFactor;
|
|
|
|
const startsWithSearchString = [result.item.name, result.item.display_name, result.item.nip05].some(
|
|
field => field && field.toLowerCase?.().startsWith(searchString.toLowerCase()),
|
|
);
|
|
|
|
const boostFactor = startsWithSearchString ? 0.25 : 1;
|
|
|
|
const weightForFuseScore = 0.8;
|
|
const weightForFollowDistance = 0.2;
|
|
|
|
const combinedScore = (fuseScore * weightForFuseScore + followDistance * weightForFollowDistance) * boostFactor;
|
|
|
|
return { ...result, combinedScore };
|
|
})
|
|
// Sort by combined score, lower is better
|
|
.sort((a, b) => a.combinedScore - b.combinedScore)
|
|
.filter(r => {
|
|
// for some reason we get duplicates even though fuzzySearch should be removing existing profile on update
|
|
if (seenIds.has(r.item.pubkey)) {
|
|
return false;
|
|
}
|
|
seenIds.add(r.item.pubkey);
|
|
return true;
|
|
});
|
|
|
|
return combinedResults.map(r => r.item);
|
|
}
|
|
|
|
export function wotScore(pubkey: string) {
|
|
return socialGraphInstance.getFollowDistance(pubkey);
|
|
}
|
|
|
|
export function sortByWoT(pubkeys: Array<string>) {
|
|
return pubkeys.sort((a, b) => wotScore(a) > wotScore(b) ? 1 : -1);
|
|
} |