File size: 8,494 Bytes
fee7318
 
9ff25e7
fee7318
 
fdb9eae
 
fee7318
 
03da455
 
fee7318
fdb9eae
fee7318
 
 
 
 
 
 
f9c128f
 
fee7318
 
 
 
 
 
 
 
fdb9eae
 
fee7318
fdb9eae
 
fee7318
fdb9eae
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f9c128f
fdb9eae
fee7318
fdb9eae
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fee7318
fdb9eae
 
 
 
 
 
 
fee7318
fdb9eae
 
 
 
 
 
 
fee7318
fdb9eae
 
 
 
 
 
 
fee7318
fdb9eae
fee7318
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fdb9eae
fee7318
 
 
fdb9eae
f9c128f
fdb9eae
fee7318
f9c128f
fee7318
 
fdb9eae
 
fee7318
 
 
 
 
 
 
 
 
fdb9eae
 
 
 
 
 
 
 
 
 
 
 
fee7318
 
fdb9eae
 
03da455
fee7318
fdb9eae
 
 
03da455
fdb9eae
 
03da455
 
fdb9eae
 
 
 
 
 
 
 
 
 
03da455
 
 
 
 
fdb9eae
 
03da455
fdb9eae
 
 
 
 
 
 
 
fee7318
 
03da455
fdb9eae
 
 
 
 
 
fee7318
fdb9eae
 
 
03da455
fdb9eae
 
03da455
fdb9eae
 
fee7318
fdb9eae
fee7318
 
 
 
fdb9eae
 
fee7318
 
 
fdb9eae
fee7318
 
 
 
fdb9eae
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
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
import os
import re
import time
import requests
import tempfile
from pathlib import Path
from typing import Optional

from dotenv import load_dotenv
# --- MODIFIED: Switched ToolCallingAgent to CodeAgent ---
from smolagents import CodeAgent, tool, LiteLLMModel

# Import your custom logic
from tools import (
    EnhancedSearchTool,
    EnhancedWikipediaTool,
    excel_to_markdown,
    image_file_info,
    audio_file_info,
    code_file_read,
    extract_youtube_info
)

# Load environment variables
load_dotenv()

# --- Constants ---
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
FILE_PATH = f"{DEFAULT_API_URL}/files/"

# ----------- 1. Rate-Limited Model Wrapper -----------
class RateLimitedModel(LiteLLMModel):
    """
    Wraps the standard smolagents model to enforce a strict 4-second delay 
    between LLM calls to prevent 429 Too Many Requests errors on free APIs.
    """
    def __call__(self, messages, stop_sequences=None, grammar=None, **kwargs):
        print("\n⏳ [Rate Limit Protection] Pausing for 4 seconds before next LLM call...")
        time.sleep(4.0)
        return super().__call__(
            messages,
            stop_sequences=stop_sequences,
            grammar=grammar,
            **kwargs
        )

# ----------- 2. smolagents Tool Definitions -----------
@tool
def enhanced_web_search(query: str) -> str:
    """Enhanced web search with intelligent query processing. Use for recent/broad web info.
    Args:
        query: The specific search query string to look up on the web.
    """
    return EnhancedSearchTool().run(query)

@tool
def enhanced_wikipedia(query: str) -> str:
    """Enhanced Wikipedia search. Use this strictly for factual or encyclopedic knowledge.
    Args:
        query: The entity or subject to search for on Wikipedia.
    """
    return EnhancedWikipediaTool().run(query)

@tool
def process_excel(excel_path: str, sheet_name: Optional[str] = None) -> str:
    """Enhanced Excel analysis. Use for spreadsheet-related files (.xlsx, .csv).
    Args:
        excel_path: The absolute local file path to the excel or csv file.
        sheet_name: Optional specific sheet name to analyze.
    """
    return excel_to_markdown(excel_path=excel_path, sheet_name=sheet_name)

@tool
def analyze_image(image_path: str, question: str) -> str:
    """Enhanced image file analysis. Use for images (.png, .jpg, etc.).
    Args:
        image_path: The absolute local file path to the image.
        question: What you want to know about the image.
    """
    return image_file_info(image_path=image_path, question=question)

@tool
def process_audio(audio_path: str) -> str:
    """Enhanced audio processing. Use for sound files (.mp3, .wav, etc.) or transcription.
    Args:
        audio_path: The absolute local file path to the audio file.
    """
    return audio_file_info(audio_path=audio_path)

@tool
def analyze_code(file_path: str) -> str:
    """Enhanced code file analysis. Use when files like .py, .js, .html are mentioned.
    Args:
        file_path: The absolute local file path to the code file.
    """
    return code_file_read(file_path=file_path)

@tool
def extract_youtube(url: str) -> str:
    """Extracts transcription from a YouTube video link.
    Args:
        url: The full YouTube URL to extract text from.
    """
    return extract_youtube_info(url)

# ----------- 3. Enhanced File Processing -----------
def detect_file_type(file_path: str) -> Optional[str]:
    ext = Path(file_path).suffix.lower()
    file_type_mapping = {
        '.xlsx': 'excel', '.xls': 'excel', '.csv': 'excel',
        '.png': 'image', '.jpg': 'image', '.jpeg': 'image', 
        '.bmp': 'image', '.gif': 'image', '.tiff': 'image', '.webp': 'image',
        '.mp3': 'audio', '.wav': 'audio', '.ogg': 'audio', 
        '.flac': 'audio', '.m4a': 'audio', '.aac': 'audio',
        '.py': 'code', '.ipynb': 'code', '.js': 'code', '.html': 'code',
        '.css': 'code', '.java': 'code', '.cpp': 'code', '.c': 'code',
        '.sql': 'code', '.r': 'code', '.json': 'code', '.xml': 'code',
        '.txt': 'text', '.md': 'text', '.pdf': 'document',
        '.doc': 'document', '.docx': 'document'
    }
    return file_type_mapping.get(ext)

def process_file(task_id: str, question_text: str) -> str:
    file_url = f"{FILE_PATH}{task_id}"
    try:
        print(f"[{task_id}] Attempting download: {file_url}")
        response = requests.get(file_url, timeout=30)
        response.raise_for_status()
    except requests.exceptions.RequestException as exc:
        print(f"[{task_id}] No file downloaded: {str(exc)}")
        return question_text  

    content_disposition = response.headers.get("content-disposition", "")
    filename = task_id  
    filename_match = re.search(r'filename[*]?=(?:"([^"]+)"|([^;]+))', content_disposition)
    if filename_match:
        filename = (filename_match.group(1) or filename_match.group(2)).strip()

    temp_storage_dir = Path(tempfile.gettempdir()) / "gaia_enhanced_files" / task_id
    temp_storage_dir.mkdir(parents=True, exist_ok=True)
    
    file_path = temp_storage_dir / filename
    file_path.write_bytes(response.content)
    
    file_size = len(response.content)
    file_type = detect_file_type(filename)
    
    enhanced_question = (
        f"{question_text}\n\n"
        f"==================================================\n"
        f"FILE INFORMATION:\n"
        f"A file was downloaded for this task and saved locally at:\n"
        f"{str(file_path)}\n"
        f"File details:\n"
        f"- Name: {filename}\n"
        f"- Size: {file_size:,} bytes\n"
        f"- Type: {file_type or 'unknown'}\n"
        f"==================================================\n"
    )
    return enhanced_question

# ----------- 4. Agent Class -----------
class GaiaAgent:
    """GAIA Agent powered by smolagents CodeAgent"""
    
    def __init__(self):
        self.model = RateLimitedModel(
            model_id=os.getenv("GEMINI_MODEL", "gemini/gemini-2.5-flash"),
            api_key=os.getenv("GEMINI_API_KEY")
        )
        
        # --- UPGRADED: Instantiated as CodeAgent with dependency authorization ---
        self.agent = CodeAgent(
            tools=[
                enhanced_web_search, 
                enhanced_wikipedia, 
                process_excel,
                analyze_image, 
                process_audio, 
                analyze_code, 
                extract_youtube
            ],
            model=self.model,
            # Increased max_steps to 12. Code-writing agents require a few extra internal 
            # iterations to verify, execute, and format complex multi-modal answers.
            max_steps=12, 
            # Crucial for GAIA spreadsheet parsing and data manipulation questions
            additional_authorized_imports=["pandas", "numpy", "re", "math", "json", "collections"]
        )
        
        print("✓ smolagents CodeAgent Architecture initialized")
        print("✓ 4-Second Rate Limit Protection Active")

    def __call__(self, task_id: str, question: str) -> str:
        print(f"\n{'='*60}")
        print(f"[{task_id}] PROCESSING: {question}")
        
        # 1. Download file and attach context to the prompt
        processed_question = process_file(task_id, question)
        
        try:
            # 2. Executing code loop sequence
            result = self.agent.run(processed_question)
            
            print(f"[{task_id}] FINAL ANSWER: {result}")
            print(f"{'='*60}")
            return str(result)
            
        except Exception as e:
            error_msg = f"Critical error in execution: {str(e)}"
            print(f"[{task_id}] {error_msg}")
            
            # Fallback direct query if agent framework loop experiences structural issues
            try:
                print("Attempting fallback direct response...")
                return self.model(messages=[{"role": "user", "content": question}]).content
            except:
                return error_msg

# ----------- Testing -----------
if __name__ == "__main__":
    agent = GaiaAgent()
    
    sample_questions = [
        "What is the current population of Tokyo?",
        "Tell me about the history of machine learning.",
    ]
    
    print("\n" + "="*80)
    print("SMOLAGENTS GAIA DEMONSTRATION")
    print("="*80)
    
    for i, question in enumerate(sample_questions):
        print(f"\nExample {i+1}: {question}")
        result = agent(f"demo_{i}", question)