ordinaceprojekt/SběrDatRůzné/SudokuKiller/preskumaj_killer_data2.py

"""
Průzkumný skript v2: zkouší najít strukturovaná data puzzle
na dailykillersudoku.com — prozkoumá DKS objekt a platný puzzle.
"""

import asyncio
import json
import sys

sys.stdout.reconfigure(encoding="utf-8")

from playwright.async_api import async_playwright

URL = "https://www.dailykillersudoku.com/puzzle/376"


async def main():
    async with async_playwright() as p:
        browser = await p.chromium.launch(headless=True)
        context = await browser.new_context(viewport={"width": 1280, "height": 900})

        page = await context.new_page()

        api_responses = []
        async def on_response(response):
            url = response.url
            if "dailykillersudoku" in url and url != URL:
                try:
                    ct = response.headers.get("content-type", "")
                    if "json" in ct or "javascript" in ct or "text" in ct:
                        body = await response.text()
                        if len(body) < 5000:
                            api_responses.append({"url": url, "status": response.status, "body": body[:2000]})
                        else:
                            api_responses.append({"url": url, "status": response.status, "body": f"({len(body)} chars)"})
                except:
                    pass

        page.on("response", on_response)

        print(f"Načítám {URL} ...")
        await page.goto(URL, wait_until="networkidle", timeout=60_000)

        # 1) DKS objekt — klíče
        print("\n=== DKS objekt — klíče ===")
        dks = await page.evaluate("""() => {
            if (typeof DKS === 'undefined') return null;
            const result = {};
            for (const key of Object.keys(DKS)) {
                const val = DKS[key];
                const t = typeof val;
                if (t === 'function') {
                    result[key] = 'function';
                } else if (t === 'object' && val !== null) {
                    result[key] = {type: 'object', keys: Object.keys(val).slice(0, 15)};
                } else {
                    result[key] = val;
                }
            }
            return result;
        }""")
        if dks:
            print(json.dumps(dks, indent=2, ensure_ascii=False)[:5000])

        # 2) DKS.board nebo podobné puzzle objekty
        print("\n=== DKS puzzle-related data ===")
        puzzle_data = await page.evaluate("""() => {
            if (typeof DKS === 'undefined') return null;
            const result = {};
            const interesting = ['board', 'puzzle', 'game', 'grid', 'cages', 'cells',
                                 'solution', 'currentPuzzle', 'puzzleData', 'data',
                                 'sudoku', 'killer', 'state'];
            for (const key of Object.keys(DKS)) {
                if (interesting.some(i => key.toLowerCase().includes(i))) {
                    try {
                        result[key] = JSON.parse(JSON.stringify(DKS[key]));
                    } catch(e) {
                        result[key] = String(DKS[key]).substring(0, 200);
                    }
                }
            }
            return result;
        }""")
        if puzzle_data:
            print(json.dumps(puzzle_data, indent=2, ensure_ascii=False)[:8000])
        else:
            print("  žádné puzzle data")

        # 3) Script tagy s puzzle daty
        print("\n=== Script tagy s puzzle daty ===")
        scripts = await page.evaluate("""() => {
            const scripts = document.querySelectorAll('script');
            const results = [];
            for (const s of scripts) {
                const text = s.textContent || '';
                if (text.includes('cage') || text.includes('cell') || text.includes('solution')
                    || text.includes('group') || text.includes('sum') || text.includes('Board')
                    || text.includes('Puzzle')) {
                    results.push({
                        length: text.length,
                        snippet: text.substring(0, 1000)
                    });
                }
            }
            return results;
        }""")
        print(json.dumps(scripts, indent=2, ensure_ascii=False)[:8000])

        # 4) Zachycené requesty
        print("\n=== Zachycené requesty (dailykillersudoku) ===")
        for r in api_responses:
            print(f"\n  URL: {r['url']}")
            print(f"  Status: {r['status']}")
            print(f"  Body: {r['body'][:500]}")

        # 5) SVG obsah — puzzle mřížka
        print("\n=== SVG puzzle mřížka ===")
        svg_data = await page.evaluate("""() => {
            const svgs = document.querySelectorAll('svg');
            const results = [];
            for (const svg of svgs) {
                const html = svg.outerHTML;
                if (html.length > 1000) {
                    // Pravděpodobně puzzle mřížka
                    const texts = svg.querySelectorAll('text');
                    const textContent = Array.from(texts).map(t => ({
                        text: t.textContent,
                        x: t.getAttribute('x'),
                        y: t.getAttribute('y'),
                        class: t.getAttribute('class')
                    }));
                    const paths = svg.querySelectorAll('path');
                    results.push({
                        size: html.length,
                        width: svg.getAttribute('width'),
                        height: svg.getAttribute('height'),
                        viewBox: svg.getAttribute('viewBox'),
                        text_count: texts.length,
                        path_count: paths.length,
                        texts: textContent.slice(0, 30)
                    });
                }
            }
            return results;
        }""")
        print(json.dumps(svg_data, indent=2, ensure_ascii=False)[:5000])

        # 6) Hledej inline JS s daty puzzle
        print("\n=== Inline JS s puzzle daty ===")
        inline_data = await page.evaluate("""() => {
            const html = document.documentElement.innerHTML;
            // Hledej vzory jako JSON pole, cage definice apod.
            const patterns = [
                /DKS\.\w+\s*=\s*(\{[^}]{20,}\})/g,
                /DKS\.\w+\s*=\s*(\[[^\]]{20,}\])/g,
                /var\s+\w+\s*=\s*(\{[^}]{50,}\})/g,
                /puzzl\w*\s*[:=]\s*["'{[]/gi
            ];
            const found = [];
            for (const p of patterns) {
                let m;
                while ((m = p.exec(html)) !== null) {
                    found.push(m[0].substring(0, 300));
                }
            }
            return found;
        }""")
        print(json.dumps(inline_data, indent=2, ensure_ascii=False)[:3000])

        await browser.close()


if __name__ == "__main__":
    asyncio.run(main())