File size: 12,881 Bytes
7022d5d
 
 
 
 
 
 
da00dda
 
 
 
7022d5d
 
 
 
 
 
 
 
 
 
 
00259b9
 
 
 
 
7022d5d
 
 
 
da00dda
 
7022d5d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
00259b9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
da00dda
 
 
 
 
 
 
 
 
 
 
f6ff4be
 
 
 
 
 
 
 
 
 
 
 
 
 
00259b9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7022d5d
 
 
 
 
 
 
 
 
 
 
f6ff4be
 
 
 
 
 
 
 
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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
from langchain_core.messages import HumanMessage
from langchain_core.tools import tool
from langchain_community.tools import (
    DuckDuckGoSearchRun,
    WikipediaQueryRun,
    ArxivQueryRun
)
from langchain_google_community.search import (
    GoogleSearchAPIWrapper,
    GoogleSearchRun
)
from langchain_community.utilities import WikipediaAPIWrapper, ArxivAPIWrapper
from langchain_openai import ChatOpenAI

import base64
import pandas as pd
import os

import os
from huggingface_hub import InferenceClient
import json
import requests
from youtube_transcript_api import YouTubeTranscriptApi
from ultralytics import YOLO
import cv2

import re

from dotenv import load_dotenv
load_dotenv()
HF_TOKEN = os.getenv("HF_TOKEN")
GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")
GOOGLE_CSE_ID = os.getenv("GOOGLE_CSE_ID")
client = InferenceClient(
    provider="hf-inference",
    api_key=HF_TOKEN,
)

llm = ChatOpenAI(model="o4-mini")
vision_llm = ChatOpenAI(model="gpt-4o")

@tool
def analyze_image(img_path: str, question: str) -> str:
    """Analyze an image and answer a question about it."""
    try:
        with open(img_path, "rb") as image_file:
            image_bytes = image_file.read()
        
        image_base64 = base64.b64encode(image_bytes).decode("utf-8")
        
        message = [
            HumanMessage(
                content=[
                    {"type": "text", "text": question},
                    {
                        "type": "image_url",
                        "image_url": {"url": f"data:image/jpeg;base64,{image_base64}"}
                    }
                ]
            )
        ]
        
        response = vision_llm.invoke(message)
        return response.content
        
    except Exception as e:
        return f"Error analyzing image: {str(e)}"

@tool
def read_excel_file(file_path: str, question: str) -> str:
    """Read and analyze an Excel file to answer a question."""
    try:
        # Read Excel file
        df = pd.read_excel(file_path)
        
        df_dict = df.to_dict(orient='records')
        info = json.dumps(df_dict)        
        return info
        
    except Exception as e:
        return f"Error reading Excel file: {str(e)}"

@tool
def read_python_file(file_path: str, question: str) -> str:
    """Read and analyze a Python file to answer a question."""
    try:
        with open(file_path, 'r', encoding='utf-8') as f:
            code_content = f.read()
        
        prompt = f"""Here is Python code from a file:

        ```python
        {code_content}
        ```

        Question: {question}

        Please analyze the code and answer the question."""
        
        response = llm.invoke([HumanMessage(content=prompt)])
        return response.content
        
    except Exception as e:
        return f"Error reading Python file: {str(e)}"

@tool
def transcribe_audio(file_path: str, question: str) -> str:
    """Transcribe audio file."""
    try:
        headers = {
            "Authorization": f"Bearer {HF_TOKEN}",
            "Content-Type": "audio/mpeg"  # Add this line for MP3 files
        }
        API_URL =  "https://router.huggingface.co/hf-inference/models/openai/whisper-large-v3"

        def query(filename):
            with open(filename, "rb") as f:
                data = f.read()
            response = requests.request("POST", API_URL, headers=headers, data=data)
            return json.loads(response.content.decode("utf-8"))

        data = query(file_path)
        return data
        
    except Exception as e:
        return f"Error transcribing audio: {str(e)}"

# Simple math tools
@tool
def add(a: float, b: float) -> float:
    """Add two numbers."""
    return a + b

@tool
def sum_list(numbers: list) -> float:
    """Sum a list of numbers."""
    return sum(numbers)

# Simple data tools
@tool
def extract_values(data: str, column: str) -> list:
    """Extract all values from a column in JSON data."""
    parsed = json.loads(data)
    values = []
    for row in parsed:
        for key, value in row.items():
            if column.lower() in key.lower():
                try:
                    values.append(float(value))
                except:
                    pass
    return values

@tool
def filter_rows(data: str, exclude_words: list) -> str:
    """Remove rows containing any of the exclude words."""
    parsed = json.loads(data)
    filtered = []
    for row in parsed:
        row_text = " ".join(str(v).lower() for v in row.values())
        if not any(word.lower() in row_text for word in exclude_words):
            filtered.append(row)
    return json.dumps(filtered)

@tool
def read_excel(file_path: str) -> str:
    """Read any Excel file and return as JSON."""
    df = pd.read_excel(file_path)
    return json.dumps(df.to_dict(orient='records'))

@tool
def object_detection(video_url: str) -> str:
    """Analyze objects and visual content in a YouTube video."""
    try:        
        model = YOLO("yolo11n.pt")  # Load an official Detect model
        results = model.track(video_url)
        
        # Track objects across frames
        frame_objects = []
        for i, result in enumerate(results):
            if result.boxes is not None:
                objects_in_frame = []
                for j in range(len(result.boxes)):
                    class_name = result.names[int(result.boxes.cls[j].item())]
                    confidence = float(result.boxes.conf[j].item())
                    if confidence > 0.5:  # Only high confidence detections
                        objects_in_frame.append(class_name)
                
                frame_objects.append({
                    "frame": i,
                    "objects": objects_in_frame,
                    "unique_objects": list(set(objects_in_frame))
                })
        
        return json.dumps(frame_objects, indent=2)
        
    except Exception as e:
        return f"Error analyzing video: {str(e)}"

@tool
def get_youtube_transcript(video_url: str) -> str:
    """Get transcript from a YouTube video."""
    try:
        # Extract video ID
        video_id_match = re.search(r'(?:v=|\/)([0-9A-Za-z_-]{11}).*', video_url)
        if not video_id_match:
            return "Error: Could not extract video ID"
        
        video_id = video_id_match.group(1)
        transcript = YouTubeTranscriptApi.get_transcript(video_id)
        
        # Format with timestamps
        formatted_transcript = []
        for entry in transcript:
            formatted_transcript.append({
                "start": entry['start'],
                "duration": entry['duration'], 
                "text": entry['text']
            })
        
        return json.dumps(formatted_transcript, indent=2)
        
    except Exception as e:
        return f"Error getting transcript: {str(e)}"

    # @tool
def analyze_video_content(video_url: str, question: str = "", max_vision_frames: int = 1) -> str:
    """Analyze video content using YOLO for object detection and vision LLM for detailed analysis."""
    try:
        model = YOLO("yolo11n.pt")
        results = model.track(video_url)
        
        # Step 1: YOLO analysis for all frames
        frame_objects = []
        frames_with_content = []
        
        for i, result in enumerate(results):
            frame_data = {
                "frame": i,
                "objects": [],
                "unique_objects": [],
                "object_counts": {}
            }
            
            if result.boxes is not None:
                objects_in_frame = []
                for j in range(len(result.boxes)):
                    class_name = result.names[int(result.boxes.cls[j].item())]
                    confidence = float(result.boxes.conf[j].item())
                    if confidence > 0.5:
                        objects_in_frame.append(class_name)
                
                # Count objects
                for obj in objects_in_frame:
                    frame_data["object_counts"][obj] = frame_data["object_counts"].get(obj, 0) + 1
                
                frame_data["objects"] = objects_in_frame
                frame_data["unique_objects"] = list(set(objects_in_frame))
                
                # Store frame for potential vision analysis
                if objects_in_frame:  # Only store frames with detected objects
                    frames_with_content.append({
                        "frame_index": i,
                        "objects": objects_in_frame,
                        "object_counts": frame_data["object_counts"],
                        "total_objects": len(objects_in_frame),
                        "image": result.orig_img
                    })
            
            frame_objects.append(frame_data)
        
        # Step 2: If there's a specific question, use vision LLM on selected frames
        detailed_analyses = []
        if question.strip():
            # Sort frames by total objects and select top frames
            frames_with_content.sort(key=lambda x: x["total_objects"], reverse=True)
            selected_frames = frames_with_content[:max_vision_frames]
            
            for frame_data in selected_frames:
                try:
                    # Encode frame directly to base64
                    _, buffer = cv2.imencode('.jpg', frame_data["image"])
                    image_bytes = buffer.tobytes()
                    image_base64 = base64.b64encode(image_bytes).decode("utf-8")
                    
                    message = [
                        HumanMessage(
                            content=[
                                {"type": "text", "text": question},
                                {
                                    "type": "image_url",
                                    "image_url": {"url": f"data:image/jpeg;base64,{image_base64}"}
                                }
                            ]
                        )
                    ]
                    
                    vision_response = vision_llm.invoke(message)
                    
                    detailed_analyses.append({
                        "frame_index": frame_data["frame_index"],
                        "yolo_objects": frame_data["objects"],
                        "yolo_counts": frame_data["object_counts"],
                        "vision_analysis": vision_response.content
                    })
                
                except Exception as vision_error:
                    detailed_analyses.append({
                        "frame_index": frame_data["frame_index"],
                        "yolo_objects": frame_data["objects"],
                        "yolo_counts": frame_data["object_counts"],
                        "vision_analysis": f"Vision analysis failed: {str(vision_error)}"
                    })
        
        # Combine results
        result_data = {
            "video_url": video_url,
            "question": question,
            "total_frames": len(frame_objects),
            "yolo_analysis": frame_objects,
            "frames_with_objects": len(frames_with_content)
        }
        
        if detailed_analyses:
            result_data["detailed_vision_analysis"] = detailed_analyses
            result_data["vision_frames_analyzed"] = len(detailed_analyses)
        
        return json.dumps(result_data, indent=2)
        
    except Exception as e:
        return f"Error analyzing video content: {str(e)}"
@tool
def google_search():
    """Google search tool"""
    api_wrapper = GoogleSearchAPIWrapper(
    google_api_key=GOOGLE_API_KEY,
    google_cse_id=GOOGLE_CSE_ID,
    k=10,  # Number of results
    siterestrict=False  # Site restrictions
)
    google_search = GoogleSearchRun(api_wrapper=api_wrapper)
    return google_search

@tool
def wiki_search():
    """Google search tool"""
    api_wrapper = WikipediaAPIWrapper()
    search = WikipediaQueryRun(api_wrapper=api_wrapper)
    return search

@tool
def arxiv_search():
    """Google search tool"""
    api_wrapper = ArxivAPIWrapper()
    search = ArxivQueryRun(api_wrapper=api_wrapper)
    return search
def general_tools():    
    tools = [
        analyze_image,
        read_python_file,
        transcribe_audio,
    ]
    return tools

def analyze_video_tools():
    tools = [object_detection, analyze_video_content]
    return tools

def youtube_transcript_tools():
    tools = [get_youtube_transcript]
    return tools

def file_agent_tools():
    tools = [read_excel]
    return tools

def math_agent_tools():
    tools = [add, sum_list]
    return tools

def data_agent_tools():
    tools = [extract_values, filter_rows]
    return tools

def search_agen_tools():
    tools = [
        google_search,
        ArxivQueryRun(api_wrapper=ArxivAPIWrapper()),
        WikipediaQueryRun(api_wrapper=WikipediaAPIWrapper())
    ]
    return tools