mcp-unknowncheatz 0.3.0

package/src/parsers/reputation.ts

@@ -0,0 +1,130 @@
+import type { Cheerio, CheerioAPI } from "cheerio";
+import type { Element } from "domhandler";
+
+export type RepSign = "positive" | "negative" | "neutral" | "unknown";
+export type RepTier =
+  | "legend"
+  | "excellent"
+  | "great"
+  | "good"
+  | "average"
+  | "positive"
+  | "novice"
+  | "neutral"
+  | "negative"
+  | "unknown";
+
+export interface AuthorReputation {
+  score: number | null;
+  power: number | null;
+  dots: number;
+  positiveDots: number;
+  highPositiveDots: number;
+  negativeDots: number;
+  neutralDots: number;
+  sign: RepSign;
+  tier: RepTier;
+  description?: string;
+  trustScore: number;
+}
+
+function classifyBySrc(src: string): "highpos" | "pos" | "neg" | "neutral" | "unknown" {
+  const s = src.toLowerCase();
+  if (s.includes("reputation_highpos") || s.includes("rep_highpos")) return "highpos";
+  if (s.includes("reputation_pos") || s.includes("rep_pos")) return "pos";
+  if (s.includes("reputation_neg") || s.includes("rep_neg")) return "neg";
+  if (s.includes("reputation_bar") || s.includes("rep_bar")) return "neutral";
+  return "unknown";
+}
+
+function computeTier(score: number | null, sign: RepSign, negativeDots: number): RepTier {
+  if (sign === "negative" || negativeDots > 0) return "negative";
+  if (score === null) return sign === "neutral" ? "neutral" : "unknown";
+
+  if (score >= 5_000) return "legend";
+  if (score >= 2_000) return "excellent";
+  if (score >= 1_000) return "great";
+  if (score >= 500) return "good";
+  if (score >= 100) return "average";
+  if (score >= 10) return "positive";
+  if (score > 0) return "novice";
+  if (score === 0) return "neutral";
+  return "negative";
+}
+
+function computeTrustScore(score: number | null, negativeDots: number, dots: number): number {
+  if (negativeDots > 0) return 5;
+
+  const base = score ?? 0;
+  if (base <= 0) return dots > 0 ? 25 : 15;
+
+  const log = Math.log10(base + 1) * 20;
+  const clamped = Math.min(100, Math.round(log));
+  return Math.max(15, clamped);
+}
+
+export function parseReputationInPost($: CheerioAPI, postCell: Cheerio<Element>): AuthorReputation {
+  const infoText = postCell.find(".info").text();
+
+  const scoreMatch = infoText.match(/Reputation:\s*([-\d,]+)/i);
+  const score = scoreMatch ? parseInt(scoreMatch[1].replace(/,/g, ""), 10) : null;
+
+  const powerMatch = infoText.match(/Rep\s*Power:\s*([-\d,]+)/i);
+  const power = powerMatch ? parseInt(powerMatch[1].replace(/,/g, ""), 10) : null;
+
+  let highPositiveDots = 0;
+  let positiveDots = 0;
+  let negativeDots = 0;
+  let neutralDots = 0;
+  let description: string | undefined;
+
+  const dotsContainer = postCell.find("[id^='repinfoDots_']").first();
+  const dotsScope = dotsContainer.length > 0 ? dotsContainer : postCell;
+
+  dotsScope.find("img").each((_, el) => {
+    const src = $(el).attr("src") ?? "";
+    const kind = classifyBySrc(src);
+    if (kind === "highpos") highPositiveDots++;
+    else if (kind === "pos") positiveDots++;
+    else if (kind === "neg") negativeDots++;
+    else if (kind === "neutral") neutralDots++;
+
+    if (!description && kind !== "unknown") {
+      const alt = ($(el).attr("alt") ?? "").trim();
+      if (alt && !/^add to|^take from/i.test(alt)) {
+        description = alt;
+      }
+    }
+  });
+
+  const dots = highPositiveDots + positiveDots + negativeDots + neutralDots;
+
+  let sign: RepSign;
+  if (negativeDots > 0) sign = "negative";
+  else if (highPositiveDots > 0 || positiveDots > 0) sign = "positive";
+  else if (neutralDots > 0) sign = "neutral";
+  else if (score !== null) {
+    if (score > 0) sign = "positive";
+    else if (score < 0) sign = "negative";
+    else sign = "neutral";
+  } else {
+    sign = "unknown";
+  }
+
+  const tier = computeTier(score, sign, negativeDots);
+  const trustScore = computeTrustScore(score, negativeDots, dots);
+
+  return {
+    score,
+    power,
+    dots,
+    positiveDots,
+    highPositiveDots,
+    negativeDots,
+    neutralDots,
+    sign,
+    tier,
+    description,
+    trustScore,
+  };
+}

package/src/parsers/subforums.ts

@@ -0,0 +1,45 @@
+import { load } from "cheerio";
+
+export interface Subforum {
+  slug: string;
+  label: string;
+  url: string;
+  description?: string;
+  threadCount?: number;
+  postCount?: number;
+}
+
+export function parseSubforums(html: string): Subforum[] {
+  const $ = load(html);
+  const map = new Map<string, Subforum>();
+
+  $("a[href*='/forum/']").each((_, el) => {
+    const link = $(el);
+    const href = (link.attr("href") ?? "").trim().replace(/^\/\//, "https://");
+    const match = href.match(/\/forum\/([a-z0-9][a-z0-9-]{1,})\/?(?:$|[?#])/i);
+    if (!match) return;
+
+    const slug = match[1].toLowerCase();
+    if (slug.endsWith(".php") || ["forum", "index", "portal", "downloads", "search", "misc", "usercp"].includes(slug)) return;
+
+    const label = link.text().trim();
+    if (!label || label.length > 80 || map.has(slug)) return;
+
+    const row = link.closest("tr, li.forumbit_post, .forumbit_nopost");
+    const description = row.find(".forumdescription, .smallfont.forumdescription").first().text().trim() || undefined;
+    const statText = row.find(".forumstats, td.alt2").text();
+    const threadMatch = statText.match(/Threads[:\s]+([\d,]+)/i);
+    const postMatch = statText.match(/Posts[:\s]+([\d,]+)/i);
+
+    map.set(slug, {
+      slug,
+      label,
+      url: `https://www.unknowncheats.me/forum/${slug}/`,
+      description,
+      threadCount: threadMatch ? parseInt(threadMatch[1].replace(/,/g, ""), 10) : undefined,
+      postCount: postMatch ? parseInt(postMatch[1].replace(/,/g, ""), 10) : undefined,
+    });
+  });
+
+  return [...map.values()];
+}

package/src/parsers/thread-list.ts

@@ -0,0 +1,155 @@
+import { load } from "cheerio";
+
+export interface ThreadListEntry {
+  title: string;
+  url: string;
+  threadId: string;
+  author?: string;
+  date?: string;
+  replies: number;
+  views: number;
+  subforum?: string;
+  snippet?: string;
+  isSticky: boolean;
+  prefix?: string;
+}
+
+const UC_BASE = "https://www.unknowncheats.me";
+
+function absoluteUrl(href: string): string {
+  if (href.startsWith("http")) return href;
+  if (href.startsWith("//")) return `https:${href}`;
+  return `${UC_BASE}${href.startsWith("/") ? "" : "/"}${href}`;
+}
+
+export function parseThreadList(html: string): ThreadListEntry[] {
+  const $ = load(html);
+  const results: ThreadListEntry[] = [];
+
+  $("a[id^='thread_title_']").each((_, el) => {
+    const link = $(el);
+    const title = link.text().trim();
+    const href = link.attr("href") ?? "";
+    const id = (link.attr("id") ?? "").replace("thread_title_", "");
+    if (!title || !href) return;
+
+    const row = link.closest("tr, div[id^='threadbit'], li[id^='thread_']");
+
+    const author =
+      row.find(".threadstarterinfo a, a.username, .username").first().text().trim() ||
+      row.find("div.smallfont").eq(0).text().trim() ||
+      undefined;
+
+    const date =
+      row.find(".threadlastpost .date, .time, .date").first().text().trim() || undefined;
+
+    const subforum =
+      row.find("a[href*='forumdisplay'], .forumtitle").first().text().trim() || undefined;
+
+    const cells = row.find("td");
+    let replies = 0;
+    let views = 0;
+    cells.each((_, td) => {
+      const text = $(td).text().trim();
+      const replyMatch = text.match(/(\d[\d,]*)\s*(?:Repl|repl)/);
+      const viewMatch = text.match(/(\d[\d,]*)\s*(?:View|view)/);
+      if (replyMatch) replies = parseInt(replyMatch[1].replace(/,/g, ""), 10);
+      if (viewMatch) views = parseInt(viewMatch[1].replace(/,/g, ""), 10);
+    });
+    if (replies === 0 && views === 0) {
+      const nums: number[] = [];
+      cells.each((_, td) => {
+        const text = $(td).text().trim().replace(/,/g, "");
+        if (/^\d+$/.test(text)) nums.push(parseInt(text, 10));
+      });
+      if (nums.length >= 2) {
+        replies = nums[nums.length - 2];
+        views = nums[nums.length - 1];
+      }
+    }
+
+    const snippetRaw = row
+      .find(".threadpreview, .searchresult_text, .smallfont:not(:has(a))")
+      .first()
+      .text()
+      .trim();
+    const snippet = snippetRaw ? snippetRaw.slice(0, 200) : undefined;
+
+    const prefixRaw = row.find(".prefix, .threadprefix").first().text().trim();
+    const prefix = prefixRaw || undefined;
+
+    const parent =
+      row.parent().attr("id") ??
+      row.attr("id") ??
+      row.closest("[id]").attr("id") ??
+      "";
+    const rowClass = row.attr("class") ?? "";
+    const rowText = row.text();
+    const stickyIconPresent = row.find("img[src*='sticky'], img[alt*='Sticky' i]").length > 0;
+    const isSticky =
+      /sticky/i.test(parent) ||
+      /sticky/i.test(rowClass) ||
+      stickyIconPresent ||
+      /\bSticky:/i.test(rowText) ||
+      snippet?.toLowerCase() === "sticky";
+
+    results.push({
+      title,
+      url: absoluteUrl(href),
+      threadId: id,
+      author,
+      date,
+      replies,
+      views,
+      subforum,
+      snippet,
+      isSticky,
+      prefix,
+    });
+  });
+
+  return results;
+}
+
+export interface PaginationInfo {
+  currentPage: number;
+  totalPages: number;
+}
+
+export function parsePaginationInfo(html: string): PaginationInfo {
+  const $ = load(html);
+  const navText = $(".pagenav").first().text().trim();
+  const match = navText.match(/Page\s+(\d+)\s+of\s+(\d+)/i);
+  if (match) {
+    return { currentPage: parseInt(match[1], 10), totalPages: parseInt(match[2], 10) };
+  }
+  return { currentPage: 1, totalPages: 1 };
+}
+
+export interface ThreadFilter {
+  query?: string;
+  minReplies?: number;
+  minViews?: number;
+  prefix?: string;
+  author?: string;
+  includeSticky?: boolean;
+}
+
+export function filterThreads(threads: ThreadListEntry[], filter: ThreadFilter): ThreadListEntry[] {
+  const terms = filter.query?.toLowerCase().split(/\s+/).filter(Boolean) ?? [];
+  const authorLc = filter.author?.toLowerCase();
+  const prefixLc = filter.prefix?.toLowerCase();
+
+  return threads.filter((thread) => {
+    if (!filter.includeSticky && thread.isSticky) return false;
+    if (filter.minReplies !== undefined && thread.replies < filter.minReplies) return false;
+    if (filter.minViews !== undefined && thread.views < filter.minViews) return false;
+    if (authorLc && !(thread.author ?? "").toLowerCase().includes(authorLc)) return false;
+    if (prefixLc && !(thread.prefix ?? "").toLowerCase().includes(prefixLc)) return false;
+    if (terms.length > 0) {
+      const haystack = `${thread.title} ${thread.snippet ?? ""}`.toLowerCase();
+      if (!terms.every((term) => haystack.includes(term))) return false;
+    }
+    return true;
+  });
+}

package/src/parsers/thread.ts

@@ -0,0 +1,84 @@
+import { load } from "cheerio";
+import type { ThreadData, ThreadPost } from "../types.js";
+import { parseReputationInPost } from "./reputation.js";
+
+function parseTotalPages($: ReturnType<typeof load>): number {
+  // .pagenav contains "Page X of Y"
+  const navText = $(".pagenav").text();
+  const match = navText.match(/Page\s+\d+\s+of\s+(\d+)/i);
+  if (match) return parseInt(match[1], 10);
+  return 1;
+}
+
+function parseTitle($: ReturnType<typeof load>): string {
+  const raw = $("title").first().text().trim();
+  // Strip trailing " - Page N" and site suffix " - unknowncheats.me"
+  return raw
+    .replace(/\s*-\s*unknowncheats\.me\s*$/i, "")
+    .replace(/\s*-\s*Page\s+\d+\s*$/i, "")
+    .trim() || "Unknown Thread";
+}
+
+export function parseThread(html: string, url: string, pageNum = 1): ThreadData {
+  const $ = load(html);
+
+  const title = parseTitle($);
+  const totalPages = parseTotalPages($);
+  const posts: ThreadPost[] = [];
+
+  // Each post is wrapped in table[id^='post'] where id is purely numeric (e.g. post4638271)
+  $("table[id]").each((_, el) => {
+    const tableEl = $(el);
+    const idAttr = tableEl.attr("id") ?? "";
+
+    // Only match post tables: id starts with 'post' followed by digits only (not 'post_message_')
+    if (!/^post\d+$/.test(idAttr)) return;
+
+    const postId = parseInt(idAttr.replace("post", ""), 10);
+
+    // Author: a.bigusername inside this post table
+    const author = tableEl.find("a.bigusername").first().text().trim();
+
+    // Reputation lives inside the left postbit cell (td.alt2)
+    const postbitCell = tableEl.find("td.alt2").first();
+    const reputation = postbitCell.length > 0 ? parseReputationInPost($, postbitCell) : undefined;
+
+    // Date: first td.thead text (strip whitespace and img alt text)
+    const dateRaw = tableEl.find("td.thead").first().clone();
+    dateRaw.find("img, a").remove();
+    const date = dateRaw.text().trim().replace(/\s+/g, " ");
+
+    // Content: div#post_message_NNNNN
+    const contentEl = tableEl.find(`div[id='post_message_${postId}']`).clone();
+    // Remove quoted blocks
+    contentEl.find("div[style*='margin']").has("div.smallfont").remove();
+    const content = contentEl.text().trim().replace(/\s+/g, " ");
+
+    // Extract links
+    const links: { text: string; url: string }[] = [];
+    contentEl.find("a[href]").each((_, a) => {
+      const href = $(a).attr("href") ?? "";
+      const text = $(a).text().trim();
+      if (href && !href.startsWith("#")) {
+        const url = href.startsWith("http") ? href : `https://www.unknowncheats.me${href}`;
+        links.push({ text: text || url, url });
+      }
+    });
+
+    // Extract images
+    const images: string[] = [];
+    contentEl.find("img[src]").each((_, img) => {
+      const src = $(img).attr("src") ?? "";
+      if (src && !src.includes("clear.gif") && !src.includes("spacer") && !src.includes("wol_error") && !src.includes("statusicon")) {
+        const url = src.startsWith("http") ? src : `https://www.unknowncheats.me${src}`;
+        images.push(url);
+      }
+    });
+
+    if (author || content) {
+      posts.push({ author, date, content, postNumber: postId, links, images, reputation });
+    }
+  });
+
+  return { title, posts, currentPage: pageNum, totalPages, url };
+}

package/src/search-fallback.ts

@@ -0,0 +1,81 @@
+import { load } from "cheerio";
+import { filterThreads, parseThreadList, type ThreadListEntry } from "./parsers/thread-list.js";
+
+function discoverSubforumSlugs(html: string): Array<{ slug: string; label: string }> {
+  const $ = load(html);
+  const slugs = new Map<string, string>();
+
+  $("a[href*='/forum/']").each((_, el) => {
+    const href = ($(el).attr("href") ?? "").replace(/^\/\//, "https://");
+    const match = href.match(/\/forum\/([a-z0-9][a-z0-9-]{1,})\/?(?:$|[?#])/i);
+    if (!match) return;
+
+    const slug = match[1].toLowerCase();
+    if (slug.endsWith(".php") || slug === "forum" || slug === "index") return;
+
+    const label = $(el).text().trim() || slug;
+    if (!slugs.has(slug)) slugs.set(slug, label);
+  });
+
+  return [...slugs.entries()].map(([slug, label]) => ({ slug, label }));
+}
+
+function rankSubforums(
+  subforums: Array<{ slug: string; label: string }>,
+  query: string
+): Array<{ slug: string; label: string; score: number }> {
+  const terms = query.toLowerCase().split(/\s+/).filter(Boolean);
+  const slugGuess = query.trim().toLowerCase().replace(/\s+/g, "-");
+
+  const ranked = subforums.map((entry) => {
+    const haystack = `${entry.slug} ${entry.label}`.toLowerCase();
+    let score = terms.filter((term) => haystack.includes(term)).length;
+    if (entry.slug === slugGuess) score += 10;
+    if (entry.slug.includes(slugGuess) || slugGuess.includes(entry.slug)) score += 3;
+    return { ...entry, score };
+  });
+
+  return ranked.filter((entry) => entry.score > 0).sort((a, b) => b.score - a.score);
+}
+
+export async function searchViaSubforums(
+  query: string,
+  fetchHtml: (url: string) => Promise<string>
+): Promise<{ results: ThreadListEntry[]; scannedSubforums: string[] }> {
+  const indexHtml = await fetchHtml("https://www.unknowncheats.me/forum/index.php");
+  const subforums = discoverSubforumSlugs(indexHtml);
+  const ranked = rankSubforums(subforums, query);
+
+  const candidates = ranked.length > 0
+    ? ranked.slice(0, 3)
+    : [{ slug: query.trim().toLowerCase().replace(/\s+/g, "-"), label: query, score: 1 }];
+
+  const seen = new Set<string>();
+  const results: ThreadListEntry[] = [];
+  const scannedSubforums: string[] = [];
+
+  for (const candidate of candidates) {
+    const url = `https://www.unknowncheats.me/forum/${candidate.slug}/`;
+    let html: string;
+
+    try {
+      html = await fetchHtml(url);
+    } catch {
+      continue;
+    }
+
+    scannedSubforums.push(candidate.slug);
+    const threads = filterThreads(parseThreadList(html), { query, includeSticky: true });
+
+    for (const thread of threads) {
+      if (seen.has(thread.url)) continue;
+      seen.add(thread.url);
+      results.push({
+        ...thread,
+        subforum: thread.subforum ?? candidate.label,
+      });
+    }
+  }
+
+  return { results, scannedSubforums };
+}