@gotza02/sequential-thinking 10000.0.1 → 10000.0.2

package/README.md

@@ -17,6 +17,12 @@ Forces the AI to think before acting using a strict block-based flow:
 - **Auto-Backup:** Automatically creates `.bak` backups before modifying any file via `deep_code_edit`.
 - **Destructive Action Protection:** Prevents accidental data loss during code refactoring.
 
+### ⚽ Sports Intelligence (Upgraded)
+- **Dual-Source Deep Dive:** Simultaneously scrapes news (BBC, Sky) and stats (Understat, FBref) for comprehensive coverage.
+- **Smart xG Extraction:** "Smart Extractor" technology specifically for Understat to get accurate Expected Goals data.
+- **Handicap Odds:** Dedicated search protocols for Asian Handicap and betting market analysis.
+- **Fallback Protection:** Intelligent warning system when API keys are missing.
+
 ### 🌐 Web Search Integration
 - Built-in support for **Exa**, **Brave**, and **Google Search**.
 - Allows the AI to "pause and research" during the thinking process.

package/dist/tools/sports/core/constants.d.ts

@@ -31,7 +31,8 @@ export declare const CACHE_CONFIG: {
     readonly MAX_SIZE: 1000;
 };
 export declare const SCRAPER_CONFIG: {
-    readonly PRIORITY_DOMAINS: readonly ["whoscored.com", "sofascore.com", "flashscore.com", "understat.com", "fbref.com", "bbc.co.uk/sport", "skysports.com", "sportsmole.co.uk", "goal.com", "thesquareball.net"];
+    readonly PRIORITY_DOMAINS: readonly ["understat.com", "bbc.co.uk/sport", "sportsmole.co.uk", "skysports.com", "goal.com", "thesquareball.net", "fbref.com", "whoscored.com", "sofascore.com", "flashscore.com"];
+    readonly NEWS_DOMAINS: readonly ["bbc.co.uk/sport", "skysports.com", "sportsmole.co.uk", "standard.co.uk/sport", "goal.com", "espn.co.uk/football", "talksport.com", "mirror.co.uk/sport", "dailymail.co.uk/sport"];
     readonly TIMEOUT: 15000;
     readonly USER_AGENT: "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36";
     readonly MAX_CONTENT_LENGTH: 50000;

package/dist/tools/sports/core/constants.js

@@ -44,16 +44,28 @@ export const CACHE_CONFIG = {
 export const SCRAPER_CONFIG = {
     // Priority domains for sports data
     PRIORITY_DOMAINS: [
-        'whoscored.com',
-        'sofascore.com',
-        'flashscore.com',
-        'understat.com',
-        'fbref.com',
+        'understat.com', // High accuracy (Smart Extractor)
+        'bbc.co.uk/sport', // Text-heavy, reliable
+        'sportsmole.co.uk', // Text-heavy, reliable
+        'skysports.com', // Good text/stats mix
+        'goal.com', // Text-heavy
+        'thesquareball.net', // Text-heavy
+        'fbref.com', // Structured data
+        'whoscored.com', // Hard to scrape (JS heavy)
+        'sofascore.com', // Hard to scrape (JS heavy)
+        'flashscore.com', // Hard to scrape (JS heavy)
+    ],
+    // News-focused domains (for lineups, injuries)
+    NEWS_DOMAINS: [
         'bbc.co.uk/sport',
         'skysports.com',
         'sportsmole.co.uk',
+        'standard.co.uk/sport',
         'goal.com',
-        'thesquareball.net'
+        'espn.co.uk/football',
+        'talksport.com',
+        'mirror.co.uk/sport',
+        'dailymail.co.uk/sport'
     ],
     // Scraping timeout in milliseconds
     TIMEOUT: 15000,

package/dist/tools/sports/providers/scraper.d.ts

@@ -20,7 +20,8 @@ export declare class ScraperProvider extends ScraperProviderBase {
      */
     protected scrape<T>(url: string, extractor?: (html: string) => T): Promise<APIResponse<T>>;
     /**
-     * Specialized extractor for Understat (Example)
+     * Specialized extractor for Understat
+     * Extracts xG, shots, and other advanced metrics from the JSON data embedded in script tags.
      */
     private extractUnderstatData;
     /**
@@ -72,3 +73,7 @@ export declare function scrapeMatchContent(url: string): Promise<APIResponse<str
  * Find best URL for match data from search results
  */
 export declare function findBestMatchUrl(urls: string[]): string | null;
+/**
+ * Find best URL for news/lineups from search results
+ */
+export declare function findBestNewsUrl(urls: string[]): string | null;

package/dist/tools/sports/providers/scraper.js

@@ -85,19 +85,62 @@ export class ScraperProvider extends ScraperProviderBase {
         }
     }
     /**
-     * Specialized extractor for Understat (Example)
+     * Specialized extractor for Understat
+     * Extracts xG, shots, and other advanced metrics from the JSON data embedded in script tags.
      */
     extractUnderstatData(html) {
         const dom = new JSDOM(html);
         const title = dom.window.document.title;
-        // Understat stores data in JSON strings within <script> tags
-        // This is a placeholder for actual regex-based JSON extraction
-        const scripts = Array.from(dom.window.document.querySelectorAll('script'));
-        const dataScript = scripts.find(s => s.textContent?.includes('JSON.parse'));
-        if (dataScript) {
-            return `## Understat Analysis: ${title}\n\n[Advanced xG Data Extracted from Script Tags]\nMatches, xG, xGA, and player positions parsed successfully.`;
+        let output = `## Understat Analysis: ${title}\n\n`;
+        // Helper to extract JSON from script variables
+        const extractJson = (varName) => {
+            const scripts = Array.from(dom.window.document.querySelectorAll('script'));
+            for (const script of scripts) {
+                const content = script.textContent || '';
+                const regex = new RegExp(`var ${varName}\\s*=\\s*JSON\\.parse\\('([^']+)'\\)`);
+                const match = content.match(regex);
+                if (match && match[1]) {
+                    try {
+                        // Decode hex sequences if any (Understat sometimes uses hex encoding)
+                        const jsonString = match[1].replace(/\\x([0-9A-Fa-f]{2})/g, (_, hex) => String.fromCharCode(parseInt(hex, 16)));
+                        return JSON.parse(jsonString);
+                    }
+                    catch (e) {
+                        return null;
+                    }
+                }
+            }
+            return null;
+        };
+        const matchData = extractJson('match_info');
+        const rosterData = extractJson('rostersData');
+        if (matchData) {
+            output += `### Match Overview\n`;
+            output += `**Team:** ${matchData.h_team} vs ${matchData.a_team}\n`;
+            output += `**Score:** ${matchData.h_goals} - ${matchData.a_goals}\n`;
+            output += `**xG:** ${matchData.h_xg} - ${matchData.a_xg}\n`;
+            output += `**Probabilities:** Home Win: ${matchData.h_win}, Draw: ${matchData.draw}, Away Win: ${matchData.a_win}\n\n`;
         }
-        return `Title: ${title}\n\n(Understat specialized extraction fallback)`;
+        if (rosterData) {
+            output += `### Key Player Stats (xG > 0.1)\n`;
+            const processPlayers = (players, teamName) => {
+                output += `**${teamName}:**\n`;
+                Object.values(players).forEach((p) => {
+                    if (parseFloat(p.xG) > 0.1) {
+                        output += `- ${p.player}: xG ${p.xG}, xA ${p.xA}, Shots: ${p.shots}\n`;
+                    }
+                });
+                output += '\n';
+            };
+            if (rosterData.h)
+                processPlayers(rosterData.h, matchData?.h_team || 'Home');
+            if (rosterData.a)
+                processPlayers(rosterData.a, matchData?.a_team || 'Away');
+        }
+        if (!matchData && !rosterData) {
+            return `Title: ${title}\n\n(Understat extraction failed - JSON data not found)`;
+        }
+        return output;
     }
     /**
      * Specialized extractor for WhoScored (Example)
@@ -216,3 +259,15 @@ export function findBestMatchUrl(urls) {
     // Return first URL if no priority match
     return urls[0] || null;
 }
+/**
+ * Find best URL for news/lineups from search results
+ */
+export function findBestNewsUrl(urls) {
+    const newsDomains = SCRAPER_CONFIG.NEWS_DOMAINS;
+    for (const url of urls) {
+        if (newsDomains.some(domain => url.includes(domain))) {
+            return url;
+        }
+    }
+    return null;
+}

package/dist/tools/sports/tools/match.js

@@ -5,7 +5,7 @@
 import { z } from 'zod';
 import { createAPIProvider } from '../providers/api.js';
 import { getSearchProvider } from '../providers/search.js';
-import { scrapeMatchContent, findBestMatchUrl } from '../providers/scraper.js';
+import { scrapeMatchContent, findBestMatchUrl, findBestNewsUrl } from '../providers/scraper.js';
 import { getGlobalCache, CacheService } from '../core/cache.js';
 import { formatMatchesTable, formatScore, formatMatchStatus } from '../utils/formatter.js';
 import { CACHE_CONFIG, LEAGUES } from '../core/constants.js';
@@ -76,7 +76,7 @@ async function performMatchAnalysis(homeTeam, awayTeam, league, context) {
         { type: 'form', query: `${homeTeam} ${awayTeam} recent form last 5 matches${dateQuery}`, title: 'Recent Form' },
         { type: 'news', query: `${baseQuery} team news injuries lineups${dateQuery}`, title: 'Team News & Lineups' },
         { type: 'stats', query: `${baseQuery} xG expected goals stats${dateQuery}`, title: 'Advanced Metrics' },
-        { type: 'odds', query: `${baseQuery} odds movement dropping odds${dateQuery}`, title: 'Market & Odds' },
+        { type: 'odds', query: `${baseQuery} Asian Handicap odds prediction today${dateQuery}`, title: 'Latest Handicap Odds' },
         { type: 'fatigue', query: `${homeTeam} ${awayTeam} days rest fixture congestion${dateQuery}`, title: 'Fatigue & Schedule' },
         { type: 'setpieces', query: `${baseQuery} set pieces corners aerial duels${dateQuery}`, title: 'Set Pieces' },
     ];
@@ -111,24 +111,45 @@ async function performMatchAnalysis(homeTeam, awayTeam, league, context) {
         const batchResults = await Promise.all(batchPromises);
         combinedResults += batchResults.join('\n');
     }
-    // Deep dive: Scrape the best URL
+    // Deep dive: Scrape the best URLs (Dual-Source Strategy)
     if (candidateUrls.length > 0) {
-        const bestUrl = findBestMatchUrl(candidateUrls);
-        if (bestUrl) {
-            combinedResults += `\n--- DEEP DIVE ANALYSIS (${bestUrl}) ---\n`;
+        const bestStatsUrl = findBestMatchUrl(candidateUrls);
+        const bestNewsUrl = findBestNewsUrl(candidateUrls);
+        // 1. Scrape Stats Source (Priority)
+        if (bestStatsUrl) {
+            combinedResults += `\n--- DEEP DIVE ANALYSIS: STATS & DATA (${bestStatsUrl}) ---\n`;
             try {
-                const scrapedContent = await scrapeMatchContent(bestUrl);
+                const scrapedContent = await scrapeMatchContent(bestStatsUrl);
                 if (scrapedContent.success) {
                     combinedResults += scrapedContent.data || '';
                 }
                 else {
-                    combinedResults += `(Deep dive failed: ${scrapedContent.error})`;
+                    combinedResults += `(Stats scrape failed: ${scrapedContent.error})`;
                 }
             }
             catch (scrapeError) {
-                combinedResults += `(Deep dive failed: ${scrapeError instanceof Error ? scrapeError.message : String(scrapeError)})`;
+                combinedResults += `(Stats scrape failed: ${scrapeError instanceof Error ? scrapeError.message : String(scrapeError)})`;
             }
-            combinedResults += `\n--- END DEEP DIVE ---\n`;
+            combinedResults += `\n--- END STATS ---\n`;
+        }
+        // 2. Scrape News Source (if different from stats)
+        if (bestNewsUrl && bestNewsUrl !== bestStatsUrl) {
+            combinedResults += `\n--- DEEP DIVE ANALYSIS: NEWS & LINEUPS (${bestNewsUrl}) ---\n`;
+            try {
+                const scrapedContent = await scrapeMatchContent(bestNewsUrl);
+                if (scrapedContent.success) {
+                    // Truncate news to avoid blowing up context, keep first 3000 chars
+                    const newsContent = scrapedContent.data?.substring(0, 3000) || '';
+                    combinedResults += newsContent + (scrapedContent.data && scrapedContent.data.length > 3000 ? '\n...(truncated)' : '');
+                }
+                else {
+                    combinedResults += `(News scrape failed: ${scrapedContent.error})`;
+                }
+            }
+            catch (scrapeError) {
+                combinedResults += `(News scrape failed: ${scrapeError instanceof Error ? scrapeError.message : String(scrapeError)})`;
+            }
+            combinedResults += `\n--- END NEWS ---\n`;
         }
     }
     combinedResults += `\n--- END DATA ---\n\n`;
@@ -161,19 +182,16 @@ Market Intelligence, Referee stats, Weather, Fatigue analysis, Set Pieces.`, {
     }, async ({ homeTeam, awayTeam, league, context, useApi }) => {
         const errors = [];
         let analysisResult = null;
+        // Check API availability explicitly
+        const api = createAPIProvider();
+        const apiAvailable = api.getStatus().available;
+        const forcedFallback = useApi && !apiAvailable;
         // Try API providers first if requested
-        if (useApi) {
+        if (useApi && apiAvailable) {
             try {
-                const api = createAPIProvider();
-                const apiStatus = api.getStatus();
-                if (apiStatus.available) {
-                    // Could try to fetch match data from API here
-                    // For now, we'll use the search-based analysis as fallback
-                    analysisResult = await performMatchAnalysis(homeTeam, awayTeam, league, context);
-                }
-                else {
-                    errors.push('API providers not configured');
-                }
+                // Could try to fetch match data from API here
+                // For now, we'll use the search-based analysis as fallback
+                analysisResult = await performMatchAnalysis(homeTeam, awayTeam, league, context);
             }
             catch (error) {
                 errors.push(`API error: ${error instanceof Error ? error.message : String(error)}`);
@@ -183,6 +201,11 @@ Market Intelligence, Referee stats, Weather, Fatigue analysis, Set Pieces.`, {
         if (!analysisResult) {
             analysisResult = await performMatchAnalysis(homeTeam, awayTeam, league, context);
         }
+        // Add warning if running in forced fallback mode
+        if (forcedFallback && analysisResult) {
+            const warning = "⚠️ **WARNING: Fallback Mode Active**\nNo API keys configured. Using web search and scraping (accuracy may vary).\n\n";
+            analysisResult = warning + analysisResult;
+        }
         return {
             content: [{ type: 'text', text: analysisResult }],
         };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@gotza02/sequential-thinking",
-  "version": "10000.0.1",
+  "version": "10000.0.2",
   "publishConfig": {
     "access": "public"
   },