Spaces:
Sleeping
Sleeping
File size: 12,216 Bytes
34abd7e b0ba8c2 |
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 |
import os
import re
import json
import time
import gradio as gr
import tempfile
from typing import Dict, Any, List, Optional
from transformers import AutoTokenizer
from sentence_transformers import SentenceTransformer
from pydantic import BaseModel, Field
from anthropic import Anthropic
from huggingface_hub import login
from src.prompts import SYSTEM_PROMPT, ANALYSIS_PROMPT_TEMPLATE_CLAUDE, ANALYSIS_PROMPT_TEMPLATE_GEMINI
CLAUDE_MODEL = "claude-3-5-sonnet-20241022"
OPENAI_MODEL = "gpt-4o"
GEMINI_MODEL = "gemini-2.0-flash"
DEFAULT_TEMPERATURE = 0.7
TOKENIZER_MODEL = "answerdotai/ModernBERT-base"
SENTENCE_TRANSFORMER_MODEL = "all-MiniLM-L6-v2"
hf_token = os.environ.get('HF_TOKEN', None)
login(token=hf_token)
tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_MODEL)
sentence_model = SentenceTransformer(SENTENCE_TRANSFORMER_MODEL)
class CourseInfo(BaseModel):
course_name: str = Field(description="Name of the course")
section_name: str = Field(description="Name of the course section")
lesson_name: str = Field(description="Name of the lesson")
class QuizOption(BaseModel):
text: str = Field(description="The text of the answer option")
correct: bool = Field(description="Whether this option is correct")
class QuizQuestion(BaseModel):
question: str = Field(description="The text of the quiz question")
options: List[QuizOption] = Field(description="List of answer options")
class Segment(BaseModel):
segment_number: int = Field(description="The segment number")
topic_name: str = Field(description="Unique and specific topic name that clearly differentiates it from other segments")
key_concepts: List[str] = Field(description="3-5 key concepts discussed in the segment")
summary: str = Field(description="Brief summary of the segment (3-5 sentences)")
quiz_questions: List[QuizQuestion] = Field(description="5 quiz questions based on the segment content")
class TextSegmentAnalysis(BaseModel):
course_info: CourseInfo = Field(description="Information about the course")
segments: List[Segment] = Field(description="List of text segments with analysis")
def clean_text(text):
text = re.sub(r'\[speaker_\d+\]', '', text)
text = re.sub(r'\s+', ' ', text).strip()
return text
def split_text_by_tokens(text, max_tokens=12000):
text = clean_text(text)
tokens = tokenizer.encode(text)
if len(tokens) <= max_tokens:
return [text]
split_point = len(tokens) // 2
sentences = re.split(r'(?<=[.!?])\s+', text)
first_half = []
second_half = []
current_tokens = 0
for sentence in sentences:
sentence_tokens = len(tokenizer.encode(sentence))
if current_tokens + sentence_tokens <= split_point:
first_half.append(sentence)
current_tokens += sentence_tokens
else:
second_half.append(sentence)
return [" ".join(first_half), " ".join(second_half)]
def generate_with_claude(text, api_key, course_name="", section_name="", lesson_name=""):
client = Anthropic(api_key=api_key)
segment_analysis_schema = TextSegmentAnalysis.model_json_schema()
tools = [
{
"name": "build_segment_analysis",
"description": "Build the text segment analysis with quiz questions",
"input_schema": segment_analysis_schema
}
]
prompt = ANALYSIS_PROMPT_TEMPLATE_CLAUDE.format(
course_name=course_name,
section_name=section_name,
lesson_name=lesson_name,
text=text
)
try:
response = client.messages.create(
model=CLAUDE_MODEL,
max_tokens=8192,
temperature=DEFAULT_TEMPERATURE,
system=SYSTEM_PROMPT,
messages=[
{
"role": "user",
"content": prompt
}
],
tools=tools,
tool_choice={"type": "tool", "name": "build_segment_analysis"}
)
# Extract the tool call content
if response.content and len(response.content) > 0 and hasattr(response.content[0], 'input'):
function_call = response.content[0].input
return function_call
else:
raise Exception("No valid tool call found in the response")
except Exception as e:
raise Exception(f"Error calling Anthropic API: {str(e)}")
def get_active_api_key(gemini_key, claude_key, language):
if language == "Uzbek" and claude_key:
return claude_key, "claude"
else:
return gemini_key, "gemini"
def segment_and_analyze_text(text: str, gemini_api_key: str, claude_api_key: str, language: str,
course_name="", section_name="", lesson_name="") -> Dict[str, Any]:
active_key, api_type = get_active_api_key(gemini_api_key, claude_api_key, language)
if api_type == "claude":
return generate_with_claude(text, active_key, course_name, section_name, lesson_name)
from langchain_google_genai import ChatGoogleGenerativeAI
os.environ["GOOGLE_API_KEY"] = active_key
llm = ChatGoogleGenerativeAI(
model=GEMINI_MODEL,
temperature=DEFAULT_TEMPERATURE,
max_retries=3
)
base_prompt = ANALYSIS_PROMPT_TEMPLATE_GEMINI.format(
course_name=course_name,
section_name=section_name,
lesson_name=lesson_name,
text=text
)
language_instruction = f"\nIMPORTANT: Generate ALL content (including topic names, key concepts, summaries, and quiz questions) in {language} language."
prompt = base_prompt + language_instruction
try:
messages = [
{"role": "system", "content": SYSTEM_PROMPT},
{"role": "user", "content": prompt}
]
response = llm.invoke(messages)
try:
content = response.content
json_match = re.search(r'```json\s*([\s\S]*?)\s*```', content)
if json_match:
json_str = json_match.group(1)
else:
json_match = re.search(r'(\{[\s\S]*\})', content)
if json_match:
json_str = json_match.group(1)
else:
json_str = content
# Parse the JSON
function_call = json.loads(json_str)
return function_call
except json.JSONDecodeError:
raise Exception("Could not parse JSON from LLM response")
except Exception as e:
raise Exception(f"Error calling API: {str(e)}")
def format_quiz_for_display(results, language="English"):
output = []
if language == "Uzbek":
course_header = "KURS"
section_header = "BO'LIM"
lesson_header = "DARS"
segment_header = "QISM"
key_concepts_header = "ASOSIY TUSHUNCHALAR"
summary_header = "QISQACHA MAZMUN"
quiz_questions_header = "TEST SAVOLLARI"
elif language == "Russian":
course_header = "КУРС"
section_header = "РАЗДЕЛ"
lesson_header = "УРОК"
segment_header = "СЕГМЕНТ"
key_concepts_header = "КЛЮЧЕВЫЕ ПОНЯТИЯ"
summary_header = "КРАТКОЕ СОДЕРЖАНИЕ"
quiz_questions_header = "ТЕСТОВЫЕ ВОПРОСЫ"
else:
course_header = "COURSE"
section_header = "SECTION"
lesson_header = "LESSON"
segment_header = "SEGMENT"
key_concepts_header = "KEY CONCEPTS"
summary_header = "SUMMARY"
quiz_questions_header = "QUIZ QUESTIONS"
if "course_info" in results:
course_info = results["course_info"]
output.append(f"{'='*40}")
output.append(f"{course_header}: {course_info.get('course_name', 'N/A')}")
output.append(f"{section_header}: {course_info.get('section_name', 'N/A')}")
output.append(f"{lesson_header}: {course_info.get('lesson_name', 'N/A')}")
output.append(f"{'='*40}\n")
segments = results.get("segments", [])
for i, segment in enumerate(segments):
topic = segment["topic_name"]
segment_num = i + 1
output.append(f"\n\n{'='*40}")
output.append(f"{segment_header} {segment_num}: {topic}")
output.append(f"{'='*40}\n")
output.append(f"{key_concepts_header}:")
for concept in segment["key_concepts"]:
output.append(f"• {concept}")
output.append(f"\n{summary_header}:")
output.append(segment["summary"])
output.append(f"\n{quiz_questions_header}:")
for i, q in enumerate(segment["quiz_questions"]):
output.append(f"\n{i+1}. {q['question']}")
for j, option in enumerate(q['options']):
letter = chr(97 + j).upper()
correct_marker = " ✓" if option["correct"] else ""
output.append(f" {letter}. {option['text']}{correct_marker}")
return "\n".join(output)
def analyze_document(text, gemini_api_key, claude_api_key, course_name, section_name, lesson_name, language):
try:
start_time = time.time()
text_parts = split_text_by_tokens(text)
input_tokens = 0
output_tokens = 0
all_results = {
"course_info": {
"course_name": course_name,
"section_name": section_name,
"lesson_name": lesson_name
},
"segments": []
}
segment_counter = 1
# Process each part of the text
for part in text_parts:
if language == "Uzbek" and claude_api_key:
# from prompts import ANALYSIS_PROMPT_TEMPLATE_CLAUDE
prompt_template = ANALYSIS_PROMPT_TEMPLATE_CLAUDE
else:
# from prompts import ANALYSIS_PROMPT_TEMPLATE_GEMINI
prompt_template = ANALYSIS_PROMPT_TEMPLATE_GEMINI
# Format the prompt with actual values
actual_prompt = prompt_template.format(
course_name=course_name,
section_name=section_name,
lesson_name=lesson_name,
text=part
)
prompt_tokens = len(tokenizer.encode(actual_prompt))
input_tokens += prompt_tokens
analysis = segment_and_analyze_text(
text,
gemini_api_key,
claude_api_key,
language,
course_name=course_name,
section_name=section_name,
lesson_name=lesson_name
)
if "segments" in analysis:
for segment in analysis["segments"]:
segment["segment_number"] = segment_counter
all_results["segments"].append(segment)
segment_counter += 1
end_time = time.time()
total_time = end_time - start_time
print(f"Total quiz processing time: {total_time}s")
formatted_output = format_quiz_for_display(all_results, language)
output_tokens = len(tokenizer.encode(formatted_output))
token_info = f"Input tokens: {input_tokens}\nOutput tokens: {output_tokens}\nTotal tokens: {input_tokens + output_tokens}\n"
formatted_text = format_quiz_for_display(all_results, language)
formatted_text = f"Total quiz Processing time: {total_time:.2f}s\n{token_info}\n" + formatted_text
output_tokens = len(tokenizer.encode(formatted_output))
json_path = tempfile.mktemp(suffix='.json')
with open(json_path, 'w', encoding='utf-8') as json_file:
json.dump(all_results, json_file, indent=2)
txt_path = tempfile.mktemp(suffix='.txt')
with open(txt_path, 'w', encoding='utf-8') as txt_file:
txt_file.write(formatted_text)
return formatted_text, json_path, txt_path
except Exception as e:
error_message = f"Error processing document: {str(e)}"
return error_message, None, None |