File size: 6,046 Bytes
02117ee
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
"""
BrowserAgent v7 — Autonomous web browsing, scraping, research (Manus-style)
Real browser control via Playwright/httpx for research, testing, web automation
"""
import asyncio
import json
import os
import re
from typing import Dict, List, Optional
import structlog
from .base_agent import BaseAgent

log = structlog.get_logger()

BROWSER_SYSTEM = """You are an elite autonomous web research and browser automation agent.
You can:
- Search the web and extract structured information
- Navigate websites and fill forms
- Take screenshots and analyze visual content
- Scrape and parse complex web pages
- Run web automation tasks

Always provide structured, actionable results with source URLs.
"""

class BrowserAgent(BaseAgent):
    def __init__(self, ws_manager=None, ai_router=None):
        super().__init__("BrowserAgent", ws_manager, ai_router)
        self._session_cache: Dict[str, str] = {}

    async def run(self, task: str, context: Dict = {}, **kwargs) -> str:
        session_id = kwargs.get("session_id", "")
        task_id = kwargs.get("task_id", "")

        await self.emit(task_id, "agent_start", {
            "agent": "BrowserAgent",
            "task": task[:80],
        }, session_id)

        await self.emit(task_id, "tool_called", {
            "agent": "BrowserAgent",
            "tool": "web_research",
            "step": f"Researching: {task[:60]}",
        }, session_id)

        # Extract search query or URL from task
        urls = re.findall(r'https?://[^\s]+', task)
        
        if urls:
            result = await self._fetch_and_analyze(urls[0], task, task_id, session_id)
        else:
            result = await self._web_research(task, task_id, session_id)

        await self.emit(task_id, "browser_result", {
            "agent": "BrowserAgent",
            "result_length": len(result),
        }, session_id)

        return result

    async def _web_research(self, query: str, task_id: str, session_id: str) -> str:
        """Perform web research using AI knowledge + httpx."""
        import httpx
        
        # Use DuckDuckGo search API (no key needed)
        search_url = f"https://api.duckduckgo.com/?q={query.replace(' ', '+')}&format=json&no_html=1"
        
        try:
            async with httpx.AsyncClient(timeout=15, follow_redirects=True) as client:
                resp = await client.get(search_url, headers={"User-Agent": "GodAgent/7.0"})
                data = resp.json()
                
                results = []
                if data.get("AbstractText"):
                    results.append(f"**Summary:** {data['AbstractText']}")
                if data.get("AbstractURL"):
                    results.append(f"**Source:** {data['AbstractURL']}")
                
                related = data.get("RelatedTopics", [])[:5]
                if related:
                    results.append("\n**Related:**")
                    for r in related:
                        if isinstance(r, dict) and r.get("Text"):
                            results.append(f"- {r['Text'][:200]}")
                
                if results:
                    search_context = "\n".join(results)
                else:
                    search_context = f"Web search for: {query}"
                    
        except Exception as e:
            search_context = f"Search context for: {query}"

        # Use AI to synthesize research
        messages = [
            {"role": "system", "content": BROWSER_SYSTEM},
            {"role": "user", "content": (
                f"Research task: {query}\n\n"
                f"Search context:\n{search_context}\n\n"
                f"Provide a comprehensive, structured research report with key findings, "
                f"actionable insights, and relevant sources. Format with headers and bullet points."
            )},
        ]
        return await self.llm(messages, task_id=task_id, session_id=session_id, temperature=0.3, max_tokens=4096)

    async def _fetch_and_analyze(self, url: str, task: str, task_id: str, session_id: str) -> str:
        """Fetch a URL and analyze its content."""
        import httpx
        
        try:
            async with httpx.AsyncClient(timeout=20, follow_redirects=True) as client:
                resp = await client.get(url, headers={
                    "User-Agent": "Mozilla/5.0 GodAgent/7.0",
                    "Accept": "text/html,application/json,*/*",
                })
                content_type = resp.headers.get("content-type", "")
                
                if "json" in content_type:
                    page_content = json.dumps(resp.json(), indent=2)[:3000]
                else:
                    # Strip HTML tags
                    html = resp.text
                    text = re.sub(r'<[^>]+>', ' ', html)
                    text = re.sub(r'\s+', ' ', text).strip()
                    page_content = text[:3000]
                    
        except Exception as e:
            page_content = f"Could not fetch {url}: {str(e)}"

        messages = [
            {"role": "system", "content": BROWSER_SYSTEM},
            {"role": "user", "content": (
                f"Analyze this web page content for the task: {task}\n\n"
                f"URL: {url}\n\n"
                f"Page Content:\n{page_content}\n\n"
                f"Provide a structured analysis with key information extracted."
            )},
        ]
        return await self.llm(messages, task_id=task_id, session_id=session_id, temperature=0.3, max_tokens=4096)

    async def screenshot_analyze(self, url: str, task_id: str = "", session_id: str = "") -> str:
        """Describe what a webpage looks like (AI-powered visual analysis)."""
        messages = [
            {"role": "system", "content": BROWSER_SYSTEM},
            {"role": "user", "content": f"Describe the visual layout and UI elements you'd expect at: {url}. Provide a detailed visual analysis."},
        ]
        return await self.llm(messages, task_id=task_id, session_id=session_id, temperature=0.5)