ordinaceprojekt/SběrDatRůzné/SudokuKiller/preskumaj_killer_data.py

"""
Průzkumný skript: zkouší najít strukturovaná data puzzle
na dailykillersudoku.com (cage definice, řešení).
"""

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/70000"


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()

        # Zachytávat network requesty
        api_responses = []
        async def on_response(response):
            url = response.url
            if any(k in url for k in ["api", "puzzle", "data", "json", "cage", "grid"]):
                try:
                    body = await response.text()
                    api_responses.append({"url": url, "status": response.status, "body": body[:2000]})
                except:
                    api_responses.append({"url": url, "status": response.status, "body": "(could not read)"})

        page.on("response", on_response)

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

        # 1) Network requesty
        print("\n=== Zachycené API/data requesty ===")
        for r in api_responses:
            print(f"\n  URL: {r['url']}")
            print(f"  Status: {r['status']}")
            if r['body'] and len(r['body']) < 2000:
                print(f"  Body: {r['body'][:500]}")

        # 2) Globální JS proměnné
        print("\n=== Globální proměnné ===")
        globals_check = await page.evaluate("""() => {
            const names = ['puzzle', 'puzzleData', 'gameData', 'game', 'board',
                           'grid', 'cages', 'cells', 'solution', 'killerData',
                           'sudoku', 'level', 'data', 'config', 'state',
                           'app', 'store', 'vuex', '__NUXT__', '__NEXT_DATA__',
                           'initialData', 'pageData', 'props', 'serverData'];
            const found = {};
            for (const name of names) {
                if (typeof window[name] !== 'undefined') {
                    const val = window[name];
                    found[name] = {
                        type: typeof val,
                        keys: typeof val === 'object' && val !== null ? Object.keys(val).slice(0, 20) : null
                    };
                }
            }
            return found;
        }""")
        print(json.dumps(globals_check, indent=2))

        # 3) SVG/Canvas analýza
        print("\n=== SVG/Canvas elementy ===")
        svg_info = await page.evaluate("""() => {
            const svgs = document.querySelectorAll('svg');
            const canvases = document.querySelectorAll('canvas');
            return {
                svg_count: svgs.length,
                canvas_count: canvases.length,
                svg_ids: Array.from(svgs).map(s => s.id || s.className || '(no id)').slice(0, 5),
                canvas_ids: Array.from(canvases).map(c => c.id || c.className || '(no id)').slice(0, 5)
            };
        }""")
        print(json.dumps(svg_info, indent=2))

        # 4) Data atributy
        print("\n=== Elementy s data- atributy ===")
        data_attrs = await page.evaluate("""() => {
            const all = document.querySelectorAll('[data-cage], [data-cell], [data-sum], [data-group], [data-value], [data-row], [data-col]');
            return {
                count: all.length,
                samples: Array.from(all).slice(0, 5).map(el => ({
                    tag: el.tagName,
                    attrs: Object.fromEntries(Array.from(el.attributes).filter(a => a.name.startsWith('data-')).map(a => [a.name, a.value]))
                }))
            };
        }""")
        print(json.dumps(data_attrs, indent=2))

        # 5) Tabulky a mřížky
        print("\n=== Tabulky / grid struktury ===")
        tables = await page.evaluate("""() => {
            const tables = document.querySelectorAll('table');
            const grids = document.querySelectorAll('[class*=grid], [class*=puzzle], [class*=board], [class*=cage], [class*=cell], [id*=grid], [id*=puzzle], [id*=board]');
            return {
                table_count: tables.length,
                grid_elements: Array.from(grids).slice(0, 10).map(el => ({
                    tag: el.tagName,
                    id: el.id,
                    class: el.className.toString().substring(0, 100),
                    children: el.children.length
                }))
            };
        }""")
        print(json.dumps(tables, indent=2))

        # 6) Script tagy s daty
        print("\n=== Script tagy s daty ===")
        scripts = await page.evaluate("""() => {
            const scripts = document.querySelectorAll('script');
            const results = [];
            for (const s of scripts) {
                const text = s.textContent || '';
                if (text.length > 10 && text.length < 50000) {
                    const keywords = ['puzzle', 'cage', 'cell', 'grid', 'solution', 'board', 'sum'];
                    const found = keywords.filter(k => text.toLowerCase().includes(k));
                    if (found.length > 0) {
                        results.push({
                            keywords: found,
                            length: text.length,
                            snippet: text.substring(0, 500)
                        });
                    }
                }
            }
            return results;
        }""")
        print(json.dumps(scripts, indent=2, ensure_ascii=False)[:5000])

        # 7) Vue/React/Angular state
        print("\n=== Framework state ===")
        framework = await page.evaluate("""() => {
            // Vue
            const vueEl = document.querySelector('[data-v-app]') || document.querySelector('#app') || document.querySelector('#__nuxt');
            let vueData = null;
            if (vueEl && vueEl.__vue_app__) {
                vueData = 'Vue 3 app found';
            } else if (vueEl && vueEl.__vue__) {
                vueData = 'Vue 2 app found';
                try {
                    const d = vueEl.__vue__.$data;
                    vueData = {type: 'Vue 2', keys: Object.keys(d)};
                } catch(e) {}
            }
            // __NUXT__
            if (typeof __NUXT__ !== 'undefined') {
                try { vueData = {type: 'Nuxt', keys: Object.keys(__NUXT__)}; } catch(e) {}
            }
            // React
            let reactData = null;
            const reactRoot = document.querySelector('#__next') || document.querySelector('#root');
            if (reactRoot) {
                const fiberKey = Object.keys(reactRoot).find(k => k.startsWith('__reactFiber') || k.startsWith('__reactInternalInstance'));
                if (fiberKey) reactData = 'React app found';
            }
            return {vue: vueData, react: reactData};
        }""")
        print(json.dumps(framework, indent=2))

        # 8) Všechny window properties (custom)
        print("\n=== Custom window properties ===")
        custom_props = await page.evaluate("""() => {
            const iframe = document.createElement('iframe');
            document.body.appendChild(iframe);
            const defaultKeys = new Set(Object.keys(iframe.contentWindow));
            document.body.removeChild(iframe);
            const custom = Object.keys(window).filter(k => !defaultKeys.has(k) && !k.startsWith('__'));
            return custom.slice(0, 50);
        }""")
        print(json.dumps(custom_props, indent=2))

        await browser.close()


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