Spaces:
Sleeping
Sleeping
Zeggai Abdellah
commited on
Commit
·
3552aba
1
Parent(s):
659455b
update the app.py and index.html to support mulit api key
Browse files- app.py +306 -58
- index.html +493 -218
app.py
CHANGED
|
@@ -10,12 +10,12 @@ from huggingface_hub import HfApi # For file persistence in Spaces
|
|
| 10 |
import os
|
| 11 |
import threading
|
| 12 |
import glob
|
|
|
|
|
|
|
| 13 |
|
| 14 |
# Load environment variables from .env file
|
| 15 |
load_dotenv()
|
| 16 |
|
| 17 |
-
from langchain_google_genai import GoogleGenerativeAI
|
| 18 |
-
|
| 19 |
app = FastAPI()
|
| 20 |
|
| 21 |
# Global variables to track generation status
|
|
@@ -27,11 +27,104 @@ generation_status = {
|
|
| 27 |
"questions_generated": 0,
|
| 28 |
"completed": False,
|
| 29 |
"result_file": None,
|
| 30 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
}
|
| 32 |
|
| 33 |
generation_lock = threading.Lock()
|
| 34 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
def estimate_difficulty(question: str, q_type: str) -> str:
|
| 36 |
"""
|
| 37 |
Estimate question difficulty based on type and content.
|
|
@@ -49,9 +142,10 @@ def estimate_difficulty(question: str, q_type: str) -> str:
|
|
| 49 |
return "medium"
|
| 50 |
return "hard" # applied
|
| 51 |
|
| 52 |
-
def generate_questions_for_chunk(chunk: str, chunk_id: int, model="gemini-2.0-flash") -> List[Dict]:
|
| 53 |
"""
|
| 54 |
Generate French questions for a given document chunk using the Gemini API.
|
|
|
|
| 55 |
"""
|
| 56 |
prompt = f"""
|
| 57 |
À partir du texte suivant d'un guide sur les vaccins en français, g��nérez 3 questions variées (factual, conceptual, applied) qui couvrent le contenu de manière exhaustive.
|
|
@@ -78,59 +172,91 @@ def generate_questions_for_chunk(chunk: str, chunk_id: int, model="gemini-2.0-fl
|
|
| 78 |
```
|
| 79 |
"""
|
| 80 |
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 86 |
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
questions = json.loads(questions_text)
|
| 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 |
def generate_questions_in_background(chunks: List[str]):
|
| 132 |
"""
|
| 133 |
Generate questions in a background thread and update status.
|
|
|
|
| 134 |
"""
|
| 135 |
global generation_status
|
| 136 |
|
|
@@ -141,17 +267,27 @@ def generate_questions_in_background(chunks: List[str]):
|
|
| 141 |
generation_status["total_chunks"] = len(chunks)
|
| 142 |
generation_status["processed_chunks"] = 0
|
| 143 |
generation_status["questions_generated"] = 0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 144 |
|
| 145 |
for i, chunk in enumerate(chunks):
|
| 146 |
print(f"Processing chunk {i+1}/{len(chunks)}...")
|
| 147 |
questions = generate_questions_for_chunk(chunk, i)
|
| 148 |
-
|
|
|
|
|
|
|
| 149 |
|
| 150 |
with generation_lock:
|
| 151 |
generation_status["processed_chunks"] = i + 1
|
| 152 |
|
| 153 |
-
|
|
|
|
|
|
|
| 154 |
|
|
|
|
| 155 |
dataset = {
|
| 156 |
"dataset_info": {
|
| 157 |
"title": "Vaccine Guide Question-Answer Dataset",
|
|
@@ -161,13 +297,17 @@ def generate_questions_in_background(chunks: List[str]):
|
|
| 161 |
"source": "Guide-pratique-de-mise-en-oeuvre-du-calendrier-national-de-vaccination-2023.pdf",
|
| 162 |
"generated_by": "Gemini API",
|
| 163 |
"total_questions": len(all_questions),
|
| 164 |
-
"intended_use": "Fine-tuning medical language models for knowledge recall and reasoning"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 165 |
},
|
| 166 |
"questions": all_questions
|
| 167 |
}
|
| 168 |
|
| 169 |
-
# Save the dataset
|
| 170 |
-
filename = f"
|
| 171 |
with open(f"./{filename}", 'w', encoding='utf-8') as f:
|
| 172 |
json.dump(dataset, f, indent=4, ensure_ascii=False)
|
| 173 |
|
|
@@ -176,12 +316,21 @@ def generate_questions_in_background(chunks: List[str]):
|
|
| 176 |
generation_status["completed"] = True
|
| 177 |
generation_status["is_running"] = False
|
| 178 |
generation_status["result_file"] = filename
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 179 |
|
| 180 |
except Exception as e:
|
| 181 |
print(f"Error in background generation: {e}")
|
| 182 |
with generation_lock:
|
| 183 |
generation_status["error"] = str(e)
|
| 184 |
generation_status["is_running"] = False
|
|
|
|
|
|
|
|
|
|
| 185 |
|
| 186 |
def save_dataset_to_space(dataset: Dict, filename: str):
|
| 187 |
"""
|
|
@@ -191,12 +340,12 @@ def save_dataset_to_space(dataset: Dict, filename: str):
|
|
| 191 |
with open(persistent_path, 'w', encoding='utf-8') as f:
|
| 192 |
json.dump(dataset, f, indent=4, ensure_ascii=False)
|
| 193 |
print(f"Dataset saved to {persistent_path}")
|
| 194 |
-
|
| 195 |
|
| 196 |
@app.get("/generate-questions")
|
| 197 |
async def generate_questions():
|
| 198 |
"""
|
| 199 |
Endpoint to generate questions from all JSON files in the data folder
|
|
|
|
| 200 |
"""
|
| 201 |
global generation_status
|
| 202 |
|
|
@@ -210,6 +359,10 @@ async def generate_questions():
|
|
| 210 |
}
|
| 211 |
|
| 212 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 213 |
# Reset status
|
| 214 |
with generation_lock:
|
| 215 |
generation_status["is_running"] = True
|
|
@@ -218,13 +371,17 @@ async def generate_questions():
|
|
| 218 |
generation_status["questions_generated"] = 0
|
| 219 |
generation_status["completed"] = False
|
| 220 |
generation_status["result_file"] = None
|
|
|
|
| 221 |
generation_status["error"] = None
|
|
|
|
|
|
|
|
|
|
| 222 |
|
| 223 |
# Load all JSON files from data folder
|
| 224 |
json_files = glob.glob("./chunk/*.json")
|
| 225 |
|
| 226 |
if not json_files:
|
| 227 |
-
raise HTTPException(status_code=404, detail="No JSON files found in
|
| 228 |
|
| 229 |
all_chunks = []
|
| 230 |
for json_file in json_files:
|
|
@@ -255,6 +412,7 @@ async def generate_questions():
|
|
| 255 |
return {
|
| 256 |
"status": "started",
|
| 257 |
"message": f"Question generation started for {len(json_files)} JSON files with {len(all_chunks)} chunks",
|
|
|
|
| 258 |
"current_status": generation_status
|
| 259 |
}
|
| 260 |
except Exception as e:
|
|
@@ -267,21 +425,111 @@ async def generate_questions():
|
|
| 267 |
async def get_generation_status():
|
| 268 |
"""
|
| 269 |
Endpoint to check the current status of generation
|
|
|
|
| 270 |
"""
|
| 271 |
with generation_lock:
|
| 272 |
status_copy = generation_status.copy()
|
| 273 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 274 |
return status_copy
|
| 275 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 276 |
@app.get("/download/{filename}")
|
| 277 |
async def download_file(filename: str):
|
| 278 |
"""
|
| 279 |
Endpoint to download generated files
|
|
|
|
| 280 |
"""
|
| 281 |
file_path = f"./{filename}"
|
| 282 |
if os.path.exists(file_path):
|
| 283 |
return FileResponse(file_path, media_type="application/json", filename=filename)
|
| 284 |
-
raise HTTPException(status_code=404, detail="File not found")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 285 |
|
| 286 |
@app.get("/")
|
| 287 |
async def root():
|
|
|
|
| 10 |
import os
|
| 11 |
import threading
|
| 12 |
import glob
|
| 13 |
+
import random
|
| 14 |
+
from langchain_google_genai import GoogleGenerativeAI
|
| 15 |
|
| 16 |
# Load environment variables from .env file
|
| 17 |
load_dotenv()
|
| 18 |
|
|
|
|
|
|
|
| 19 |
app = FastAPI()
|
| 20 |
|
| 21 |
# Global variables to track generation status
|
|
|
|
| 27 |
"questions_generated": 0,
|
| 28 |
"completed": False,
|
| 29 |
"result_file": None,
|
| 30 |
+
"progress_file": None, # New: track progress file
|
| 31 |
+
"error": None,
|
| 32 |
+
"current_api_key_index": 0, # New: track current API key
|
| 33 |
+
"failed_chunks": [], # New: track failed chunks for retry
|
| 34 |
+
"partial_results": [] # New: store partial results
|
| 35 |
}
|
| 36 |
|
| 37 |
generation_lock = threading.Lock()
|
| 38 |
|
| 39 |
+
def get_api_keys() -> List[str]:
|
| 40 |
+
"""
|
| 41 |
+
Get all available Google API keys from environment variables.
|
| 42 |
+
Supports GOOGLE_API_KEY, GOOGLE_API_KEY_1, GOOGLE_API_KEY_2, etc.
|
| 43 |
+
"""
|
| 44 |
+
api_keys = []
|
| 45 |
+
|
| 46 |
+
# Check for primary key
|
| 47 |
+
primary_key = os.getenv("GOOGLE_API_KEY")
|
| 48 |
+
if primary_key:
|
| 49 |
+
api_keys.append(primary_key)
|
| 50 |
+
|
| 51 |
+
# Check for numbered keys
|
| 52 |
+
i = 1
|
| 53 |
+
while True:
|
| 54 |
+
key = os.getenv(f"GOOGLE_API_KEY_{i}")
|
| 55 |
+
if key:
|
| 56 |
+
api_keys.append(key)
|
| 57 |
+
i += 1
|
| 58 |
+
else:
|
| 59 |
+
break
|
| 60 |
+
|
| 61 |
+
if not api_keys:
|
| 62 |
+
raise ValueError("No Google API keys found in environment variables")
|
| 63 |
+
|
| 64 |
+
return api_keys
|
| 65 |
+
|
| 66 |
+
def get_next_api_key() -> tuple[str, int]:
|
| 67 |
+
"""
|
| 68 |
+
Get the next API key in rotation and update the current index.
|
| 69 |
+
Returns tuple of (api_key, key_index)
|
| 70 |
+
"""
|
| 71 |
+
global generation_status
|
| 72 |
+
|
| 73 |
+
api_keys = get_api_keys()
|
| 74 |
+
|
| 75 |
+
with generation_lock:
|
| 76 |
+
current_index = generation_status["current_api_key_index"]
|
| 77 |
+
next_index = (current_index + 1) % len(api_keys)
|
| 78 |
+
generation_status["current_api_key_index"] = next_index
|
| 79 |
+
|
| 80 |
+
return api_keys[next_index], next_index
|
| 81 |
+
|
| 82 |
+
def save_progress_file():
|
| 83 |
+
"""
|
| 84 |
+
Save current progress to a file that can be downloaded at any time.
|
| 85 |
+
"""
|
| 86 |
+
global generation_status
|
| 87 |
+
|
| 88 |
+
with generation_lock:
|
| 89 |
+
progress_data = {
|
| 90 |
+
"generation_info": {
|
| 91 |
+
"status": "in_progress" if generation_status["is_running"] else "completed",
|
| 92 |
+
"start_time": generation_status["start_time"],
|
| 93 |
+
"processed_chunks": generation_status["processed_chunks"],
|
| 94 |
+
"total_chunks": generation_status["total_chunks"],
|
| 95 |
+
"questions_generated": generation_status["questions_generated"],
|
| 96 |
+
"completed": generation_status["completed"],
|
| 97 |
+
"current_time": datetime.utcnow().isoformat(),
|
| 98 |
+
"failed_chunks": generation_status["failed_chunks"].copy(),
|
| 99 |
+
"error": generation_status["error"]
|
| 100 |
+
},
|
| 101 |
+
"partial_dataset": {
|
| 102 |
+
"dataset_info": {
|
| 103 |
+
"title": "Vaccine Guide Question-Answer Dataset (Partial)",
|
| 104 |
+
"description": "Partial dataset of question-answer pairs generated from a vaccine guide.",
|
| 105 |
+
"version": "1.1.0",
|
| 106 |
+
"created_date": generation_status["start_time"],
|
| 107 |
+
"source": "Guide-pratique-de-mise-en-oeuvre-du-calendrier-national-de-vaccination-2023.pdf",
|
| 108 |
+
"generated_by": "Gemini API",
|
| 109 |
+
"total_questions": len(generation_status["partial_results"]),
|
| 110 |
+
"intended_use": "Fine-tuning medical language models for knowledge recall and reasoning",
|
| 111 |
+
"note": "This is a partial dataset. Generation may still be in progress."
|
| 112 |
+
},
|
| 113 |
+
"questions": generation_status["partial_results"].copy()
|
| 114 |
+
}
|
| 115 |
+
}
|
| 116 |
+
|
| 117 |
+
# Save progress file
|
| 118 |
+
progress_filename = f"vaccine_questions_progress_{int(time.time())}.json"
|
| 119 |
+
generation_status["progress_file"] = progress_filename
|
| 120 |
+
|
| 121 |
+
try:
|
| 122 |
+
with open(f"./{progress_filename}", 'w', encoding='utf-8') as f:
|
| 123 |
+
json.dump(progress_data, f, indent=4, ensure_ascii=False)
|
| 124 |
+
print(f"Progress saved to {progress_filename}")
|
| 125 |
+
except Exception as e:
|
| 126 |
+
print(f"Error saving progress file: {e}")
|
| 127 |
+
|
| 128 |
def estimate_difficulty(question: str, q_type: str) -> str:
|
| 129 |
"""
|
| 130 |
Estimate question difficulty based on type and content.
|
|
|
|
| 142 |
return "medium"
|
| 143 |
return "hard" # applied
|
| 144 |
|
| 145 |
+
def generate_questions_for_chunk(chunk: str, chunk_id: int, model="gemini-2.0-flash", max_retries=3) -> List[Dict]:
|
| 146 |
"""
|
| 147 |
Generate French questions for a given document chunk using the Gemini API.
|
| 148 |
+
Now includes retry logic with different API keys.
|
| 149 |
"""
|
| 150 |
prompt = f"""
|
| 151 |
À partir du texte suivant d'un guide sur les vaccins en français, g��nérez 3 questions variées (factual, conceptual, applied) qui couvrent le contenu de manière exhaustive.
|
|
|
|
| 172 |
```
|
| 173 |
"""
|
| 174 |
|
| 175 |
+
last_error = None
|
| 176 |
+
|
| 177 |
+
for attempt in range(max_retries):
|
| 178 |
+
try:
|
| 179 |
+
# Get next API key for this attempt
|
| 180 |
+
api_key, key_index = get_next_api_key()
|
| 181 |
+
print(f"Chunk {chunk_id}, attempt {attempt + 1}: Using API key index {key_index}")
|
| 182 |
+
|
| 183 |
+
llm = GoogleGenerativeAI(
|
| 184 |
+
model=model,
|
| 185 |
+
google_api_key=api_key
|
| 186 |
+
)
|
| 187 |
|
| 188 |
+
response = llm.invoke(prompt)
|
| 189 |
+
|
| 190 |
+
questions_text = str(response) # Convert response to string
|
| 191 |
+
|
| 192 |
+
# Strip Markdown code fences
|
| 193 |
+
if questions_text.startswith("```json\n") and questions_text.endswith("\n```"):
|
| 194 |
+
questions_text = questions_text[7:-4].strip()
|
| 195 |
+
elif questions_text.startswith("```") and questions_text.endswith("```"):
|
| 196 |
+
questions_text = questions_text[3:-3].strip()
|
| 197 |
+
|
| 198 |
+
if not questions_text:
|
| 199 |
+
raise ValueError(f"Empty response for chunk {chunk_id}")
|
| 200 |
+
|
| 201 |
+
questions = json.loads(questions_text)
|
|
|
|
| 202 |
|
| 203 |
+
formatted_questions = []
|
| 204 |
+
for q in questions:
|
| 205 |
+
question_id = str(uuid.uuid4())
|
| 206 |
+
difficulty = estimate_difficulty(q["question"], q["type"])
|
| 207 |
+
formatted_questions.append({
|
| 208 |
+
"question_id": question_id,
|
| 209 |
+
"chunk_id": chunk_id,
|
| 210 |
+
"chunk_text": chunk,
|
| 211 |
+
"question": q["question"],
|
| 212 |
+
"type": q["type"],
|
| 213 |
+
"difficulty": difficulty,
|
| 214 |
+
"training_purpose": "Knowledge Recall" if q["type"] == "factual" else "Reasoning",
|
| 215 |
+
"validated": False,
|
| 216 |
+
"api_key_used": key_index, # Track which key was used
|
| 217 |
+
"generation_attempt": attempt + 1
|
| 218 |
+
})
|
| 219 |
+
|
| 220 |
+
# Update the global status and add to partial results
|
| 221 |
+
with generation_lock:
|
| 222 |
+
generation_status["questions_generated"] += len(formatted_questions)
|
| 223 |
+
generation_status["partial_results"].extend(formatted_questions)
|
| 224 |
+
|
| 225 |
+
# Save progress after each successful chunk
|
| 226 |
+
save_progress_file()
|
| 227 |
+
|
| 228 |
+
print(f"Successfully generated {len(formatted_questions)} questions for chunk {chunk_id}")
|
| 229 |
+
return formatted_questions
|
| 230 |
+
|
| 231 |
+
except Exception as e:
|
| 232 |
+
last_error = e
|
| 233 |
+
print(f"Attempt {attempt + 1} failed for chunk {chunk_id}: {e}")
|
| 234 |
+
|
| 235 |
+
# If this is not the last attempt, wait before retrying
|
| 236 |
+
if attempt < max_retries - 1:
|
| 237 |
+
wait_time = (attempt + 1) * 5 # Increasing wait time
|
| 238 |
+
print(f"Waiting {wait_time} seconds before retry...")
|
| 239 |
+
time.sleep(wait_time)
|
| 240 |
+
|
| 241 |
+
continue
|
| 242 |
|
| 243 |
+
# All attempts failed
|
| 244 |
+
print(f"All {max_retries} attempts failed for chunk {chunk_id}. Last error: {last_error}")
|
| 245 |
+
|
| 246 |
+
# Add to failed chunks list
|
| 247 |
+
with generation_lock:
|
| 248 |
+
generation_status["failed_chunks"].append({
|
| 249 |
+
"chunk_id": chunk_id,
|
| 250 |
+
"error": str(last_error),
|
| 251 |
+
"attempts": max_retries
|
| 252 |
+
})
|
| 253 |
+
|
| 254 |
+
return []
|
| 255 |
|
| 256 |
def generate_questions_in_background(chunks: List[str]):
|
| 257 |
"""
|
| 258 |
Generate questions in a background thread and update status.
|
| 259 |
+
Enhanced with better error handling and progress tracking.
|
| 260 |
"""
|
| 261 |
global generation_status
|
| 262 |
|
|
|
|
| 267 |
generation_status["total_chunks"] = len(chunks)
|
| 268 |
generation_status["processed_chunks"] = 0
|
| 269 |
generation_status["questions_generated"] = 0
|
| 270 |
+
generation_status["partial_results"] = []
|
| 271 |
+
generation_status["failed_chunks"] = []
|
| 272 |
+
|
| 273 |
+
# Save initial progress file
|
| 274 |
+
save_progress_file()
|
| 275 |
|
| 276 |
for i, chunk in enumerate(chunks):
|
| 277 |
print(f"Processing chunk {i+1}/{len(chunks)}...")
|
| 278 |
questions = generate_questions_for_chunk(chunk, i)
|
| 279 |
+
|
| 280 |
+
if questions: # Only add if generation was successful
|
| 281 |
+
all_questions.extend(questions)
|
| 282 |
|
| 283 |
with generation_lock:
|
| 284 |
generation_status["processed_chunks"] = i + 1
|
| 285 |
|
| 286 |
+
# Rate limiting - slightly randomized to avoid hitting limits
|
| 287 |
+
sleep_time = random.uniform(8, 11) # Random between 8-11 seconds
|
| 288 |
+
time.sleep(sleep_time)
|
| 289 |
|
| 290 |
+
# Create final dataset
|
| 291 |
dataset = {
|
| 292 |
"dataset_info": {
|
| 293 |
"title": "Vaccine Guide Question-Answer Dataset",
|
|
|
|
| 297 |
"source": "Guide-pratique-de-mise-en-oeuvre-du-calendrier-national-de-vaccination-2023.pdf",
|
| 298 |
"generated_by": "Gemini API",
|
| 299 |
"total_questions": len(all_questions),
|
| 300 |
+
"intended_use": "Fine-tuning medical language models for knowledge recall and reasoning",
|
| 301 |
+
"total_chunks_processed": len(chunks),
|
| 302 |
+
"successful_chunks": len(chunks) - len(generation_status["failed_chunks"]),
|
| 303 |
+
"failed_chunks": len(generation_status["failed_chunks"]),
|
| 304 |
+
"failed_chunk_details": generation_status["failed_chunks"].copy()
|
| 305 |
},
|
| 306 |
"questions": all_questions
|
| 307 |
}
|
| 308 |
|
| 309 |
+
# Save the final dataset
|
| 310 |
+
filename = f"vaccine_questions_final_{int(time.time())}.json"
|
| 311 |
with open(f"./{filename}", 'w', encoding='utf-8') as f:
|
| 312 |
json.dump(dataset, f, indent=4, ensure_ascii=False)
|
| 313 |
|
|
|
|
| 316 |
generation_status["completed"] = True
|
| 317 |
generation_status["is_running"] = False
|
| 318 |
generation_status["result_file"] = filename
|
| 319 |
+
|
| 320 |
+
# Save final progress file
|
| 321 |
+
save_progress_file()
|
| 322 |
+
|
| 323 |
+
success_rate = (len(chunks) - len(generation_status["failed_chunks"])) / len(chunks) * 100
|
| 324 |
+
print(f"Generation completed! Success rate: {success_rate:.1f}% ({len(all_questions)} questions generated)")
|
| 325 |
|
| 326 |
except Exception as e:
|
| 327 |
print(f"Error in background generation: {e}")
|
| 328 |
with generation_lock:
|
| 329 |
generation_status["error"] = str(e)
|
| 330 |
generation_status["is_running"] = False
|
| 331 |
+
|
| 332 |
+
# Save progress even if there was an error
|
| 333 |
+
save_progress_file()
|
| 334 |
|
| 335 |
def save_dataset_to_space(dataset: Dict, filename: str):
|
| 336 |
"""
|
|
|
|
| 340 |
with open(persistent_path, 'w', encoding='utf-8') as f:
|
| 341 |
json.dump(dataset, f, indent=4, ensure_ascii=False)
|
| 342 |
print(f"Dataset saved to {persistent_path}")
|
|
|
|
| 343 |
|
| 344 |
@app.get("/generate-questions")
|
| 345 |
async def generate_questions():
|
| 346 |
"""
|
| 347 |
Endpoint to generate questions from all JSON files in the data folder
|
| 348 |
+
Enhanced with multi-key support validation
|
| 349 |
"""
|
| 350 |
global generation_status
|
| 351 |
|
|
|
|
| 359 |
}
|
| 360 |
|
| 361 |
try:
|
| 362 |
+
# Validate API keys before starting
|
| 363 |
+
api_keys = get_api_keys()
|
| 364 |
+
print(f"Found {len(api_keys)} API keys for rotation")
|
| 365 |
+
|
| 366 |
# Reset status
|
| 367 |
with generation_lock:
|
| 368 |
generation_status["is_running"] = True
|
|
|
|
| 371 |
generation_status["questions_generated"] = 0
|
| 372 |
generation_status["completed"] = False
|
| 373 |
generation_status["result_file"] = None
|
| 374 |
+
generation_status["progress_file"] = None
|
| 375 |
generation_status["error"] = None
|
| 376 |
+
generation_status["current_api_key_index"] = 0
|
| 377 |
+
generation_status["failed_chunks"] = []
|
| 378 |
+
generation_status["partial_results"] = []
|
| 379 |
|
| 380 |
# Load all JSON files from data folder
|
| 381 |
json_files = glob.glob("./chunk/*.json")
|
| 382 |
|
| 383 |
if not json_files:
|
| 384 |
+
raise HTTPException(status_code=404, detail="No JSON files found in chunk folder")
|
| 385 |
|
| 386 |
all_chunks = []
|
| 387 |
for json_file in json_files:
|
|
|
|
| 412 |
return {
|
| 413 |
"status": "started",
|
| 414 |
"message": f"Question generation started for {len(json_files)} JSON files with {len(all_chunks)} chunks",
|
| 415 |
+
"api_keys_available": len(api_keys),
|
| 416 |
"current_status": generation_status
|
| 417 |
}
|
| 418 |
except Exception as e:
|
|
|
|
| 425 |
async def get_generation_status():
|
| 426 |
"""
|
| 427 |
Endpoint to check the current status of generation
|
| 428 |
+
Enhanced with more detailed status information
|
| 429 |
"""
|
| 430 |
with generation_lock:
|
| 431 |
status_copy = generation_status.copy()
|
| 432 |
|
| 433 |
+
# Calculate additional metrics
|
| 434 |
+
if status_copy["total_chunks"] > 0:
|
| 435 |
+
progress_percentage = (status_copy["processed_chunks"] / status_copy["total_chunks"]) * 100
|
| 436 |
+
status_copy["progress_percentage"] = round(progress_percentage, 2)
|
| 437 |
+
else:
|
| 438 |
+
status_copy["progress_percentage"] = 0
|
| 439 |
+
|
| 440 |
+
# Add estimated time remaining if generation is running
|
| 441 |
+
if status_copy["is_running"] and status_copy["start_time"] and status_copy["processed_chunks"] > 0:
|
| 442 |
+
start_time = datetime.fromisoformat(status_copy["start_time"].replace('Z', '+00:00'))
|
| 443 |
+
elapsed_time = (datetime.utcnow() - start_time.replace(tzinfo=None)).total_seconds()
|
| 444 |
+
chunks_per_second = status_copy["processed_chunks"] / elapsed_time if elapsed_time > 0 else 0
|
| 445 |
+
|
| 446 |
+
if chunks_per_second > 0:
|
| 447 |
+
remaining_chunks = status_copy["total_chunks"] - status_copy["processed_chunks"]
|
| 448 |
+
estimated_remaining_seconds = remaining_chunks / chunks_per_second
|
| 449 |
+
status_copy["estimated_remaining_minutes"] = round(estimated_remaining_seconds / 60, 2)
|
| 450 |
+
else:
|
| 451 |
+
status_copy["estimated_remaining_minutes"] = None
|
| 452 |
+
|
| 453 |
return status_copy
|
| 454 |
|
| 455 |
+
@app.get("/download-progress")
|
| 456 |
+
async def download_progress():
|
| 457 |
+
"""
|
| 458 |
+
New endpoint to download current progress at any time
|
| 459 |
+
"""
|
| 460 |
+
global generation_status
|
| 461 |
+
|
| 462 |
+
# Force save current progress
|
| 463 |
+
save_progress_file()
|
| 464 |
+
|
| 465 |
+
with generation_lock:
|
| 466 |
+
progress_file = generation_status["progress_file"]
|
| 467 |
+
|
| 468 |
+
if progress_file and os.path.exists(f"./{progress_file}"):
|
| 469 |
+
return FileResponse(f"./{progress_file}", media_type="application/json", filename=progress_file)
|
| 470 |
+
else:
|
| 471 |
+
raise HTTPException(status_code=404, detail="No progress file available")
|
| 472 |
+
|
| 473 |
@app.get("/download/{filename}")
|
| 474 |
async def download_file(filename: str):
|
| 475 |
"""
|
| 476 |
Endpoint to download generated files
|
| 477 |
+
Enhanced with better error handling
|
| 478 |
"""
|
| 479 |
file_path = f"./{filename}"
|
| 480 |
if os.path.exists(file_path):
|
| 481 |
return FileResponse(file_path, media_type="application/json", filename=filename)
|
| 482 |
+
raise HTTPException(status_code=404, detail=f"File {filename} not found")
|
| 483 |
+
|
| 484 |
+
@app.get("/retry-failed")
|
| 485 |
+
async def retry_failed_chunks():
|
| 486 |
+
"""
|
| 487 |
+
New endpoint to retry only the failed chunks
|
| 488 |
+
"""
|
| 489 |
+
global generation_status
|
| 490 |
+
|
| 491 |
+
with generation_lock:
|
| 492 |
+
if generation_status["is_running"]:
|
| 493 |
+
return {
|
| 494 |
+
"status": "error",
|
| 495 |
+
"message": "Cannot retry while generation is running"
|
| 496 |
+
}
|
| 497 |
+
|
| 498 |
+
failed_chunks = generation_status["failed_chunks"].copy()
|
| 499 |
+
|
| 500 |
+
if not failed_chunks:
|
| 501 |
+
return {
|
| 502 |
+
"status": "success",
|
| 503 |
+
"message": "No failed chunks to retry"
|
| 504 |
+
}
|
| 505 |
+
|
| 506 |
+
# This would require implementing the retry logic
|
| 507 |
+
# For now, just return the failed chunks info
|
| 508 |
+
return {
|
| 509 |
+
"status": "info",
|
| 510 |
+
"message": f"Found {len(failed_chunks)} failed chunks",
|
| 511 |
+
"failed_chunks": failed_chunks,
|
| 512 |
+
"note": "Retry functionality can be implemented based on requirements"
|
| 513 |
+
}
|
| 514 |
+
|
| 515 |
+
@app.get("/api-keys-status")
|
| 516 |
+
async def get_api_keys_status():
|
| 517 |
+
"""
|
| 518 |
+
New endpoint to check API keys status
|
| 519 |
+
"""
|
| 520 |
+
try:
|
| 521 |
+
api_keys = get_api_keys()
|
| 522 |
+
return {
|
| 523 |
+
"status": "success",
|
| 524 |
+
"total_keys": len(api_keys),
|
| 525 |
+
"current_key_index": generation_status["current_api_key_index"],
|
| 526 |
+
"message": f"{len(api_keys)} API keys configured for rotation"
|
| 527 |
+
}
|
| 528 |
+
except Exception as e:
|
| 529 |
+
return {
|
| 530 |
+
"status": "error",
|
| 531 |
+
"message": str(e)
|
| 532 |
+
}
|
| 533 |
|
| 534 |
@app.get("/")
|
| 535 |
async def root():
|
index.html
CHANGED
|
@@ -3,64 +3,102 @@
|
|
| 3 |
<head>
|
| 4 |
<meta charset="UTF-8">
|
| 5 |
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 6 |
-
<title>Générateur de Questions sur les Vaccins
|
| 7 |
<style>
|
| 8 |
body {
|
| 9 |
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
| 10 |
line-height: 1.6;
|
| 11 |
color: #333;
|
| 12 |
-
max-width:
|
| 13 |
margin: 0 auto;
|
| 14 |
padding: 20px;
|
| 15 |
-
background
|
|
|
|
| 16 |
}
|
| 17 |
h1 {
|
| 18 |
-
color:
|
| 19 |
text-align: center;
|
| 20 |
-
|
| 21 |
-
|
|
|
|
| 22 |
}
|
| 23 |
.container {
|
| 24 |
-
background
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
|
|
|
| 28 |
margin-bottom: 20px;
|
|
|
|
| 29 |
}
|
| 30 |
button {
|
| 31 |
-
background
|
| 32 |
color: white;
|
| 33 |
border: none;
|
| 34 |
-
padding:
|
| 35 |
-
border-radius:
|
| 36 |
cursor: pointer;
|
| 37 |
font-size: 16px;
|
| 38 |
-
|
|
|
|
|
|
|
| 39 |
}
|
| 40 |
button:hover {
|
| 41 |
-
|
|
|
|
| 42 |
}
|
| 43 |
button:disabled {
|
| 44 |
-
background
|
| 45 |
cursor: not-allowed;
|
|
|
|
|
|
|
| 46 |
}
|
| 47 |
.download-btn {
|
| 48 |
-
background
|
|
|
|
| 49 |
}
|
| 50 |
.download-btn:hover {
|
| 51 |
-
|
| 52 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
#statusContainer {
|
| 54 |
margin-top: 20px;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
padding: 15px;
|
| 56 |
-
border-
|
| 57 |
-
|
|
|
|
| 58 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
.loader {
|
| 60 |
display: inline-block;
|
| 61 |
width: 20px;
|
| 62 |
height: 20px;
|
| 63 |
-
border: 3px solid rgba(
|
| 64 |
border-radius: 50%;
|
| 65 |
border-top-color: #3498db;
|
| 66 |
animation: spin 1s ease infinite;
|
|
@@ -70,6 +108,7 @@
|
|
| 70 |
@keyframes spin {
|
| 71 |
to { transform: rotate(360deg); }
|
| 72 |
}
|
|
|
|
| 73 |
.stats-container {
|
| 74 |
display: grid;
|
| 75 |
grid-template-columns: repeat(auto-fill, minmax(200px, 1fr));
|
|
@@ -77,14 +116,18 @@
|
|
| 77 |
margin-top: 20px;
|
| 78 |
}
|
| 79 |
.stat-card {
|
| 80 |
-
background
|
| 81 |
-
padding:
|
| 82 |
-
border-radius:
|
| 83 |
-
box-shadow: 0
|
| 84 |
text-align: center;
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
}
|
| 86 |
.stat-value {
|
| 87 |
-
font-size:
|
| 88 |
font-weight: bold;
|
| 89 |
color: #2c3e50;
|
| 90 |
margin: 10px 0;
|
|
@@ -92,80 +135,181 @@
|
|
| 92 |
.stat-label {
|
| 93 |
color: #7f8c8d;
|
| 94 |
font-size: 14px;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
}
|
|
|
|
| 96 |
#questionsPreview {
|
| 97 |
margin-top: 20px;
|
| 98 |
-
max-height:
|
| 99 |
overflow-y: auto;
|
| 100 |
-
background
|
| 101 |
-
border-radius:
|
| 102 |
-
box-shadow: 0
|
| 103 |
}
|
| 104 |
table {
|
| 105 |
width: 100%;
|
| 106 |
border-collapse: collapse;
|
| 107 |
}
|
| 108 |
th, td {
|
| 109 |
-
padding:
|
| 110 |
text-align: left;
|
| 111 |
-
border-bottom: 1px solid #
|
| 112 |
}
|
| 113 |
th {
|
| 114 |
-
background
|
| 115 |
font-weight: 600;
|
|
|
|
|
|
|
|
|
|
| 116 |
}
|
| 117 |
tr:hover {
|
| 118 |
-
background-color:
|
| 119 |
}
|
|
|
|
| 120 |
.badge {
|
| 121 |
display: inline-block;
|
| 122 |
-
padding:
|
| 123 |
-
border-radius:
|
| 124 |
-
font-size:
|
| 125 |
font-weight: 600;
|
| 126 |
text-transform: uppercase;
|
| 127 |
color: white;
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
.badge-
|
| 131 |
-
.badge-
|
| 132 |
-
.badge-
|
| 133 |
-
.badge-
|
| 134 |
-
.badge-
|
|
|
|
| 135 |
|
| 136 |
-
.
|
| 137 |
-
margin-top: 15px;
|
| 138 |
-
background-color: #ecf0f1;
|
| 139 |
-
border-radius: 4px;
|
| 140 |
-
height: 20px;
|
| 141 |
-
overflow: hidden;
|
| 142 |
-
}
|
| 143 |
-
.progress-bar {
|
| 144 |
-
height: 100%;
|
| 145 |
-
background-color: #3498db;
|
| 146 |
-
transition: width 0.3s ease;
|
| 147 |
display: flex;
|
|
|
|
|
|
|
|
|
|
| 148 |
align-items: center;
|
| 149 |
-
justify-content: center;
|
| 150 |
-
color: white;
|
| 151 |
-
font-size: 12px;
|
| 152 |
-
font-weight: bold;
|
| 153 |
}
|
| 154 |
-
|
| 155 |
-
|
|
|
|
|
|
|
|
|
|
| 156 |
}
|
| 157 |
-
.
|
| 158 |
-
background-color: #
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
}
|
| 160 |
</style>
|
| 161 |
</head>
|
| 162 |
<body>
|
| 163 |
-
<h1
|
| 164 |
|
|
|
|
| 165 |
<div class="container">
|
| 166 |
-
<
|
| 167 |
-
<
|
| 168 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
|
| 170 |
<div id="statusContainer" style="display: none;">
|
| 171 |
<div id="statusHeader">
|
|
@@ -177,25 +321,39 @@
|
|
| 177 |
<div id="progressBar" class="progress-bar" style="width: 0%">0%</div>
|
| 178 |
</div>
|
| 179 |
|
| 180 |
-
<div id="statusDetails" style="margin-top:
|
| 181 |
-
<div
|
| 182 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 183 |
</div>
|
| 184 |
</div>
|
| 185 |
</div>
|
| 186 |
|
|
|
|
| 187 |
<div class="container" id="resultsContainer" style="display: none;">
|
| 188 |
-
<h2
|
| 189 |
|
| 190 |
<div class="stats-container" id="statsContainer">
|
| 191 |
<!-- Stats will be populated here -->
|
| 192 |
</div>
|
| 193 |
|
| 194 |
-
<div style="margin-top:
|
| 195 |
-
<button id="downloadBtn" class="download-btn"
|
| 196 |
</div>
|
| 197 |
|
| 198 |
-
<h3 style="margin-top: 30px;"
|
| 199 |
<div id="questionsPreview">
|
| 200 |
<table>
|
| 201 |
<thead>
|
|
@@ -204,6 +362,8 @@
|
|
| 204 |
<th>Type</th>
|
| 205 |
<th>Difficulté</th>
|
| 206 |
<th>But d'Entraînement</th>
|
|
|
|
|
|
|
| 207 |
</tr>
|
| 208 |
</thead>
|
| 209 |
<tbody id="questionsTableBody">
|
|
@@ -214,105 +374,144 @@
|
|
| 214 |
</div>
|
| 215 |
|
| 216 |
<script>
|
| 217 |
-
const apiBaseUrl = window.location.origin;
|
| 218 |
let generatedDataset = null;
|
| 219 |
let downloadUrl = '';
|
| 220 |
let statusCheckInterval = null;
|
|
|
|
| 221 |
|
|
|
|
| 222 |
document.getElementById('generateBtn').addEventListener('click', startGeneration);
|
| 223 |
document.getElementById('downloadBtn').addEventListener('click', downloadDataset);
|
|
|
|
|
|
|
|
|
|
| 224 |
|
| 225 |
-
// Check for ongoing generation when the page loads
|
| 226 |
-
window.addEventListener('load',
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
|
| 230 |
-
|
| 231 |
-
|
| 232 |
-
|
| 233 |
-
|
| 234 |
-
|
| 235 |
-
|
| 236 |
-
|
| 237 |
-
|
| 238 |
-
|
| 239 |
-
|
| 240 |
-
|
| 241 |
-
|
| 242 |
-
|
| 243 |
-
|
| 244 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 245 |
}
|
| 246 |
|
| 247 |
-
function
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 248 |
const generateBtn = document.getElementById('generateBtn');
|
| 249 |
const statusContainer = document.getElementById('statusContainer');
|
| 250 |
const resultsContainer = document.getElementById('resultsContainer');
|
| 251 |
|
| 252 |
-
// Disable button and show status
|
| 253 |
generateBtn.disabled = true;
|
| 254 |
statusContainer.style.display = 'block';
|
| 255 |
resultsContainer.style.display = 'none';
|
| 256 |
|
| 257 |
-
|
| 258 |
-
|
| 259 |
-
|
| 260 |
-
|
| 261 |
-
|
| 262 |
-
|
| 263 |
-
|
| 264 |
-
|
| 265 |
-
.
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
|
| 272 |
-
|
| 273 |
-
|
| 274 |
-
|
| 275 |
-
|
|
|
|
| 276 |
}
|
| 277 |
|
| 278 |
function setupStatusMonitoring() {
|
| 279 |
-
// Clear any existing interval
|
| 280 |
if (statusCheckInterval) {
|
| 281 |
clearInterval(statusCheckInterval);
|
| 282 |
}
|
| 283 |
|
| 284 |
-
// Show the status container
|
| 285 |
document.getElementById('statusContainer').style.display = 'block';
|
|
|
|
| 286 |
|
| 287 |
-
|
| 288 |
-
statusCheckInterval = setInterval(checkGenerationStatus, 5000);
|
| 289 |
-
|
| 290 |
-
// Do an immediate check
|
| 291 |
checkGenerationStatus();
|
| 292 |
}
|
| 293 |
|
| 294 |
-
function checkGenerationStatus() {
|
| 295 |
-
|
| 296 |
-
|
| 297 |
-
|
| 298 |
-
|
| 299 |
-
|
| 300 |
-
|
| 301 |
-
|
| 302 |
-
|
| 303 |
-
|
| 304 |
-
|
| 305 |
-
|
| 306 |
-
|
| 307 |
-
|
| 308 |
-
|
| 309 |
-
|
| 310 |
-
|
| 311 |
-
|
| 312 |
-
|
| 313 |
-
|
| 314 |
-
|
| 315 |
-
|
|
|
|
| 316 |
}
|
| 317 |
|
| 318 |
function updateStatusDisplay(status) {
|
|
@@ -321,54 +520,83 @@
|
|
| 321 |
const processedChunks = document.getElementById('processedChunks');
|
| 322 |
const totalChunks = document.getElementById('totalChunks');
|
| 323 |
const questionsGenerated = document.getElementById('questionsGenerated');
|
|
|
|
|
|
|
|
|
|
|
|
|
| 324 |
|
| 325 |
-
// Update text
|
| 326 |
if (status.error) {
|
| 327 |
-
statusText.textContent =
|
| 328 |
progressBar.classList.add('error');
|
| 329 |
} else if (status.completed) {
|
| 330 |
-
statusText.textContent = 'Génération terminée avec succès !';
|
| 331 |
progressBar.classList.add('complete');
|
| 332 |
} else if (status.is_running) {
|
| 333 |
-
statusText.textContent = 'Génération des questions en cours...';
|
| 334 |
}
|
| 335 |
|
| 336 |
-
// Update progress
|
| 337 |
-
processedChunks.textContent = status.processed_chunks;
|
| 338 |
-
totalChunks.textContent = status.total_chunks;
|
| 339 |
-
questionsGenerated.textContent = status.questions_generated;
|
|
|
|
|
|
|
| 340 |
|
| 341 |
-
//
|
| 342 |
if (status.total_chunks > 0) {
|
| 343 |
const percentage = Math.round((status.processed_chunks / status.total_chunks) * 100);
|
| 344 |
progressBar.style.width = `${percentage}%`;
|
| 345 |
progressBar.textContent = `${percentage}%`;
|
| 346 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 347 |
}
|
| 348 |
|
| 349 |
-
function
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 350 |
if (!downloadUrl) return;
|
| 351 |
|
| 352 |
-
|
| 353 |
-
|
| 354 |
-
|
| 355 |
-
|
| 356 |
-
|
| 357 |
-
|
| 358 |
-
|
| 359 |
-
|
| 360 |
-
|
| 361 |
-
|
| 362 |
-
|
| 363 |
-
|
| 364 |
-
|
| 365 |
-
|
| 366 |
-
|
| 367 |
-
document.getElementById('generateBtn').disabled = false;
|
| 368 |
-
})
|
| 369 |
-
.catch(error => {
|
| 370 |
-
console.error('Error loading results:', error);
|
| 371 |
-
});
|
| 372 |
}
|
| 373 |
|
| 374 |
function displayResults(dataset) {
|
|
@@ -376,56 +604,57 @@
|
|
| 376 |
const statsContainer = document.getElementById('statsContainer');
|
| 377 |
const questionsTableBody = document.getElementById('questionsTableBody');
|
| 378 |
|
| 379 |
-
//
|
| 380 |
statsContainer.innerHTML = '';
|
| 381 |
|
| 382 |
-
//
|
| 383 |
addStatCard(statsContainer, dataset.dataset_info.total_questions, 'Questions Totales');
|
|
|
|
|
|
|
| 384 |
|
| 385 |
-
//
|
| 386 |
const typeCount = countByProperty(dataset.questions, 'type');
|
| 387 |
for (const [type, count] of Object.entries(typeCount)) {
|
| 388 |
-
addStatCard(statsContainer, count,
|
| 389 |
}
|
| 390 |
|
| 391 |
-
//
|
| 392 |
const difficultyCount = countByProperty(dataset.questions, 'difficulty');
|
| 393 |
for (const [difficulty, count] of Object.entries(difficultyCount)) {
|
| 394 |
-
addStatCard(statsContainer, count,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 395 |
}
|
| 396 |
|
| 397 |
-
//
|
| 398 |
questionsTableBody.innerHTML = '';
|
| 399 |
-
dataset.questions.forEach(question => {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 400 |
const row = document.createElement('tr');
|
| 401 |
-
|
| 402 |
-
const questionCell = document.createElement('td');
|
| 403 |
-
questionCell.textContent = question.question;
|
| 404 |
-
|
| 405 |
-
const typeCell = document.createElement('td');
|
| 406 |
-
const typeBadge = document.createElement('span');
|
| 407 |
-
typeBadge.className = `badge badge-${question.type}`;
|
| 408 |
-
typeBadge.textContent = question.type;
|
| 409 |
-
typeCell.appendChild(typeBadge);
|
| 410 |
-
|
| 411 |
-
const difficultyCell = document.createElement('td');
|
| 412 |
-
const difficultyBadge = document.createElement('span');
|
| 413 |
-
difficultyBadge.className = `badge badge-${question.difficulty}`;
|
| 414 |
-
difficultyBadge.textContent = question.difficulty;
|
| 415 |
-
difficultyCell.appendChild(difficultyBadge);
|
| 416 |
-
|
| 417 |
-
const purposeCell = document.createElement('td');
|
| 418 |
-
purposeCell.textContent = question.training_purpose;
|
| 419 |
-
|
| 420 |
-
row.appendChild(questionCell);
|
| 421 |
-
row.appendChild(typeCell);
|
| 422 |
-
row.appendChild(difficultyCell);
|
| 423 |
-
row.appendChild(purposeCell);
|
| 424 |
-
|
| 425 |
questionsTableBody.appendChild(row);
|
| 426 |
-
}
|
| 427 |
|
| 428 |
-
// Show the results container
|
| 429 |
resultsContainer.style.display = 'block';
|
| 430 |
}
|
| 431 |
|
|
@@ -433,16 +662,11 @@
|
|
| 433 |
const card = document.createElement('div');
|
| 434 |
card.className = 'stat-card';
|
| 435 |
|
| 436 |
-
|
| 437 |
-
|
| 438 |
-
|
|
|
|
| 439 |
|
| 440 |
-
const labelElement = document.createElement('div');
|
| 441 |
-
labelElement.className = 'stat-label';
|
| 442 |
-
labelElement.textContent = label;
|
| 443 |
-
|
| 444 |
-
card.appendChild(valueElement);
|
| 445 |
-
card.appendChild(labelElement);
|
| 446 |
container.appendChild(card);
|
| 447 |
}
|
| 448 |
|
|
@@ -461,9 +685,60 @@
|
|
| 461 |
|
| 462 |
function downloadDataset() {
|
| 463 |
if (downloadUrl) {
|
| 464 |
-
window.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 465 |
}
|
| 466 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 467 |
</script>
|
| 468 |
</body>
|
| 469 |
</html>
|
|
|
|
| 3 |
<head>
|
| 4 |
<meta charset="UTF-8">
|
| 5 |
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 6 |
+
<title>Générateur de Questions sur les Vaccins - Amélioré</title>
|
| 7 |
<style>
|
| 8 |
body {
|
| 9 |
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
| 10 |
line-height: 1.6;
|
| 11 |
color: #333;
|
| 12 |
+
max-width: 1200px;
|
| 13 |
margin: 0 auto;
|
| 14 |
padding: 20px;
|
| 15 |
+
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
|
| 16 |
+
min-height: 100vh;
|
| 17 |
}
|
| 18 |
h1 {
|
| 19 |
+
color: white;
|
| 20 |
text-align: center;
|
| 21 |
+
font-size: 2.5em;
|
| 22 |
+
margin-bottom: 30px;
|
| 23 |
+
text-shadow: 2px 2px 4px rgba(0,0,0,0.3);
|
| 24 |
}
|
| 25 |
.container {
|
| 26 |
+
background: rgba(255, 255, 255, 0.95);
|
| 27 |
+
backdrop-filter: blur(10px);
|
| 28 |
+
border-radius: 15px;
|
| 29 |
+
box-shadow: 0 8px 32px rgba(0, 0, 0, 0.1);
|
| 30 |
+
padding: 25px;
|
| 31 |
margin-bottom: 20px;
|
| 32 |
+
border: 1px solid rgba(255, 255, 255, 0.2);
|
| 33 |
}
|
| 34 |
button {
|
| 35 |
+
background: linear-gradient(45deg, #3498db, #2980b9);
|
| 36 |
color: white;
|
| 37 |
border: none;
|
| 38 |
+
padding: 12px 24px;
|
| 39 |
+
border-radius: 8px;
|
| 40 |
cursor: pointer;
|
| 41 |
font-size: 16px;
|
| 42 |
+
font-weight: 600;
|
| 43 |
+
transition: all 0.3s ease;
|
| 44 |
+
box-shadow: 0 4px 15px rgba(52, 152, 219, 0.3);
|
| 45 |
}
|
| 46 |
button:hover {
|
| 47 |
+
transform: translateY(-2px);
|
| 48 |
+
box-shadow: 0 6px 20px rgba(52, 152, 219, 0.4);
|
| 49 |
}
|
| 50 |
button:disabled {
|
| 51 |
+
background: #95a5a6;
|
| 52 |
cursor: not-allowed;
|
| 53 |
+
transform: none;
|
| 54 |
+
box-shadow: none;
|
| 55 |
}
|
| 56 |
.download-btn {
|
| 57 |
+
background: linear-gradient(45deg, #27ae60, #219955);
|
| 58 |
+
box-shadow: 0 4px 15px rgba(39, 174, 96, 0.3);
|
| 59 |
}
|
| 60 |
.download-btn:hover {
|
| 61 |
+
box-shadow: 0 6px 20px rgba(39, 174, 96, 0.4);
|
| 62 |
}
|
| 63 |
+
.warning-btn {
|
| 64 |
+
background: linear-gradient(45deg, #f39c12, #e67e22);
|
| 65 |
+
box-shadow: 0 4px 15px rgba(243, 156, 18, 0.3);
|
| 66 |
+
}
|
| 67 |
+
.danger-btn {
|
| 68 |
+
background: linear-gradient(45deg, #e74c3c, #c0392b);
|
| 69 |
+
box-shadow: 0 4px 15px rgba(231, 76, 60, 0.3);
|
| 70 |
+
}
|
| 71 |
+
.small-btn {
|
| 72 |
+
padding: 8px 16px;
|
| 73 |
+
font-size: 14px;
|
| 74 |
+
margin: 5px;
|
| 75 |
+
}
|
| 76 |
+
|
| 77 |
#statusContainer {
|
| 78 |
margin-top: 20px;
|
| 79 |
+
padding: 20px;
|
| 80 |
+
border-left: 5px solid #3498db;
|
| 81 |
+
background: linear-gradient(135deg, #e8f4fc 0%, #f0f8ff 100%);
|
| 82 |
+
border-radius: 10px;
|
| 83 |
+
}
|
| 84 |
+
|
| 85 |
+
.api-status {
|
| 86 |
+
background: linear-gradient(135deg, #f8f9fa 0%, #e9ecef 100%);
|
| 87 |
padding: 15px;
|
| 88 |
+
border-radius: 10px;
|
| 89 |
+
margin-bottom: 15px;
|
| 90 |
+
border-left: 4px solid #28a745;
|
| 91 |
}
|
| 92 |
+
.api-status.error {
|
| 93 |
+
border-left-color: #dc3545;
|
| 94 |
+
background: linear-gradient(135deg, #f8d7da 0%, #f5c6cb 100%);
|
| 95 |
+
}
|
| 96 |
+
|
| 97 |
.loader {
|
| 98 |
display: inline-block;
|
| 99 |
width: 20px;
|
| 100 |
height: 20px;
|
| 101 |
+
border: 3px solid rgba(255, 255, 255, 0.3);
|
| 102 |
border-radius: 50%;
|
| 103 |
border-top-color: #3498db;
|
| 104 |
animation: spin 1s ease infinite;
|
|
|
|
| 108 |
@keyframes spin {
|
| 109 |
to { transform: rotate(360deg); }
|
| 110 |
}
|
| 111 |
+
|
| 112 |
.stats-container {
|
| 113 |
display: grid;
|
| 114 |
grid-template-columns: repeat(auto-fill, minmax(200px, 1fr));
|
|
|
|
| 116 |
margin-top: 20px;
|
| 117 |
}
|
| 118 |
.stat-card {
|
| 119 |
+
background: linear-gradient(135deg, #ffffff 0%, #f8f9fa 100%);
|
| 120 |
+
padding: 20px;
|
| 121 |
+
border-radius: 12px;
|
| 122 |
+
box-shadow: 0 4px 15px rgba(0, 0, 0, 0.1);
|
| 123 |
text-align: center;
|
| 124 |
+
transition: transform 0.3s ease;
|
| 125 |
+
}
|
| 126 |
+
.stat-card:hover {
|
| 127 |
+
transform: translateY(-5px);
|
| 128 |
}
|
| 129 |
.stat-value {
|
| 130 |
+
font-size: 28px;
|
| 131 |
font-weight: bold;
|
| 132 |
color: #2c3e50;
|
| 133 |
margin: 10px 0;
|
|
|
|
| 135 |
.stat-label {
|
| 136 |
color: #7f8c8d;
|
| 137 |
font-size: 14px;
|
| 138 |
+
font-weight: 500;
|
| 139 |
+
}
|
| 140 |
+
|
| 141 |
+
.progress-container {
|
| 142 |
+
margin-top: 15px;
|
| 143 |
+
background-color: #ecf0f1;
|
| 144 |
+
border-radius: 10px;
|
| 145 |
+
height: 25px;
|
| 146 |
+
overflow: hidden;
|
| 147 |
+
position: relative;
|
| 148 |
+
}
|
| 149 |
+
.progress-bar {
|
| 150 |
+
height: 100%;
|
| 151 |
+
background: linear-gradient(45deg, #3498db, #2980b9);
|
| 152 |
+
transition: width 0.3s ease;
|
| 153 |
+
display: flex;
|
| 154 |
+
align-items: center;
|
| 155 |
+
justify-content: center;
|
| 156 |
+
color: white;
|
| 157 |
+
font-size: 12px;
|
| 158 |
+
font-weight: bold;
|
| 159 |
+
position: relative;
|
| 160 |
+
}
|
| 161 |
+
.progress-bar.complete {
|
| 162 |
+
background: linear-gradient(45deg, #27ae60, #219955);
|
| 163 |
+
}
|
| 164 |
+
.progress-bar.error {
|
| 165 |
+
background: linear-gradient(45deg, #e74c3c, #c0392b);
|
| 166 |
+
}
|
| 167 |
+
|
| 168 |
+
.time-estimate {
|
| 169 |
+
background: rgba(52, 152, 219, 0.1);
|
| 170 |
+
padding: 10px;
|
| 171 |
+
border-radius: 8px;
|
| 172 |
+
margin-top: 10px;
|
| 173 |
+
font-size: 14px;
|
| 174 |
+
color: #2980b9;
|
| 175 |
+
}
|
| 176 |
+
|
| 177 |
+
.failed-chunks {
|
| 178 |
+
background: rgba(231, 76, 60, 0.1);
|
| 179 |
+
border: 1px solid rgba(231, 76, 60, 0.2);
|
| 180 |
+
border-radius: 8px;
|
| 181 |
+
padding: 15px;
|
| 182 |
+
margin-top: 15px;
|
| 183 |
+
}
|
| 184 |
+
.failed-chunks h4 {
|
| 185 |
+
color: #e74c3c;
|
| 186 |
+
margin-top: 0;
|
| 187 |
+
}
|
| 188 |
+
.failed-chunk-item {
|
| 189 |
+
background: white;
|
| 190 |
+
padding: 10px;
|
| 191 |
+
margin: 5px 0;
|
| 192 |
+
border-radius: 5px;
|
| 193 |
+
border-left: 3px solid #e74c3c;
|
| 194 |
}
|
| 195 |
+
|
| 196 |
#questionsPreview {
|
| 197 |
margin-top: 20px;
|
| 198 |
+
max-height: 500px;
|
| 199 |
overflow-y: auto;
|
| 200 |
+
background: rgba(255, 255, 255, 0.9);
|
| 201 |
+
border-radius: 12px;
|
| 202 |
+
box-shadow: 0 4px 15px rgba(0, 0, 0, 0.1);
|
| 203 |
}
|
| 204 |
table {
|
| 205 |
width: 100%;
|
| 206 |
border-collapse: collapse;
|
| 207 |
}
|
| 208 |
th, td {
|
| 209 |
+
padding: 15px;
|
| 210 |
text-align: left;
|
| 211 |
+
border-bottom: 1px solid #e0e6ed;
|
| 212 |
}
|
| 213 |
th {
|
| 214 |
+
background: linear-gradient(135deg, #f8f9fa 0%, #e9ecef 100%);
|
| 215 |
font-weight: 600;
|
| 216 |
+
color: #495057;
|
| 217 |
+
position: sticky;
|
| 218 |
+
top: 0;
|
| 219 |
}
|
| 220 |
tr:hover {
|
| 221 |
+
background-color: rgba(52, 152, 219, 0.05);
|
| 222 |
}
|
| 223 |
+
|
| 224 |
.badge {
|
| 225 |
display: inline-block;
|
| 226 |
+
padding: 6px 12px;
|
| 227 |
+
border-radius: 20px;
|
| 228 |
+
font-size: 11px;
|
| 229 |
font-weight: 600;
|
| 230 |
text-transform: uppercase;
|
| 231 |
color: white;
|
| 232 |
+
letter-spacing: 0.5px;
|
| 233 |
+
}
|
| 234 |
+
.badge-easy { background: linear-gradient(45deg, #27ae60, #2ecc71); }
|
| 235 |
+
.badge-medium { background: linear-gradient(45deg, #f39c12, #e67e22); }
|
| 236 |
+
.badge-hard { background: linear-gradient(45deg, #e74c3c, #c0392b); }
|
| 237 |
+
.badge-factual { background: linear-gradient(45deg, #3498db, #2980b9); }
|
| 238 |
+
.badge-conceptual { background: linear-gradient(45deg, #9b59b6, #8e44ad); }
|
| 239 |
+
.badge-applied { background: linear-gradient(45deg, #e67e22, #d35400); }
|
| 240 |
|
| 241 |
+
.control-panel {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 242 |
display: flex;
|
| 243 |
+
flex-wrap: wrap;
|
| 244 |
+
gap: 10px;
|
| 245 |
+
margin-bottom: 20px;
|
| 246 |
align-items: center;
|
|
|
|
|
|
|
|
|
|
|
|
|
| 247 |
}
|
| 248 |
+
|
| 249 |
+
.alert {
|
| 250 |
+
padding: 15px;
|
| 251 |
+
border-radius: 8px;
|
| 252 |
+
margin: 10px 0;
|
| 253 |
}
|
| 254 |
+
.alert-success {
|
| 255 |
+
background-color: #d4edda;
|
| 256 |
+
border: 1px solid #c3e6cb;
|
| 257 |
+
color: #155724;
|
| 258 |
+
}
|
| 259 |
+
.alert-warning {
|
| 260 |
+
background-color: #fff3cd;
|
| 261 |
+
border: 1px solid #ffeaa7;
|
| 262 |
+
color: #856404;
|
| 263 |
+
}
|
| 264 |
+
.alert-error {
|
| 265 |
+
background-color: #f8d7da;
|
| 266 |
+
border: 1px solid #f5c6cb;
|
| 267 |
+
color: #721c24;
|
| 268 |
+
}
|
| 269 |
+
|
| 270 |
+
@media (max-width: 768px) {
|
| 271 |
+
body {
|
| 272 |
+
padding: 10px;
|
| 273 |
+
}
|
| 274 |
+
.stats-container {
|
| 275 |
+
grid-template-columns: repeat(auto-fill, minmax(150px, 1fr));
|
| 276 |
+
gap: 10px;
|
| 277 |
+
}
|
| 278 |
+
.control-panel {
|
| 279 |
+
flex-direction: column;
|
| 280 |
+
align-items: stretch;
|
| 281 |
+
}
|
| 282 |
+
button {
|
| 283 |
+
width: 100%;
|
| 284 |
+
margin: 5px 0;
|
| 285 |
+
}
|
| 286 |
}
|
| 287 |
</style>
|
| 288 |
</head>
|
| 289 |
<body>
|
| 290 |
+
<h1>🧬 Générateur de Questions sur les Vaccins</h1>
|
| 291 |
|
| 292 |
+
<!-- API Status Container -->
|
| 293 |
<div class="container">
|
| 294 |
+
<h3>🔑 État des Clés API</h3>
|
| 295 |
+
<div id="apiStatusContainer">
|
| 296 |
+
<div class="api-status">
|
| 297 |
+
<span>Vérification des clés API...</span>
|
| 298 |
+
</div>
|
| 299 |
+
</div>
|
| 300 |
+
<button id="checkApiBtn" class="small-btn">Vérifier les Clés</button>
|
| 301 |
+
</div>
|
| 302 |
+
|
| 303 |
+
<!-- Generation Control Container -->
|
| 304 |
+
<div class="container">
|
| 305 |
+
<h2>🚀 Génération de Questions</h2>
|
| 306 |
+
<p>Générez des questions à partir du guide de vaccination avec gestion multi-clés et suivi en temps réel.</p>
|
| 307 |
+
|
| 308 |
+
<div class="control-panel">
|
| 309 |
+
<button id="generateBtn">Générer des Questions</button>
|
| 310 |
+
<button id="downloadProgressBtn" class="download-btn small-btn" style="display: none;">Télécharger Progrès</button>
|
| 311 |
+
<button id="retryFailedBtn" class="warning-btn small-btn" style="display: none;">Voir Échecs</button>
|
| 312 |
+
</div>
|
| 313 |
|
| 314 |
<div id="statusContainer" style="display: none;">
|
| 315 |
<div id="statusHeader">
|
|
|
|
| 321 |
<div id="progressBar" class="progress-bar" style="width: 0%">0%</div>
|
| 322 |
</div>
|
| 323 |
|
| 324 |
+
<div id="statusDetails" style="margin-top: 15px;">
|
| 325 |
+
<div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 15px;">
|
| 326 |
+
<div><strong>Chunks traités:</strong> <span id="processedChunks">0</span>/<span id="totalChunks">0</span></div>
|
| 327 |
+
<div><strong>Questions générées:</strong> <span id="questionsGenerated">0</span></div>
|
| 328 |
+
<div><strong>Clé API actuelle:</strong> <span id="currentApiKey">-</span></div>
|
| 329 |
+
<div><strong>Échecs:</strong> <span id="failedChunks">0</span></div>
|
| 330 |
+
</div>
|
| 331 |
+
</div>
|
| 332 |
+
|
| 333 |
+
<div id="timeEstimate" class="time-estimate" style="display: none;">
|
| 334 |
+
<strong>⏱️ Temps estimé restant:</strong> <span id="estimatedTime">Calcul en cours...</span>
|
| 335 |
+
</div>
|
| 336 |
+
|
| 337 |
+
<div id="failedChunksContainer" class="failed-chunks" style="display: none;">
|
| 338 |
+
<h4>❌ Chunks Échoués</h4>
|
| 339 |
+
<div id="failedChunksList"></div>
|
| 340 |
</div>
|
| 341 |
</div>
|
| 342 |
</div>
|
| 343 |
|
| 344 |
+
<!-- Results Container -->
|
| 345 |
<div class="container" id="resultsContainer" style="display: none;">
|
| 346 |
+
<h2>📊 Résultats de Génération</h2>
|
| 347 |
|
| 348 |
<div class="stats-container" id="statsContainer">
|
| 349 |
<!-- Stats will be populated here -->
|
| 350 |
</div>
|
| 351 |
|
| 352 |
+
<div style="margin-top: 25px; text-align: center;">
|
| 353 |
+
<button id="downloadBtn" class="download-btn">📥 Télécharger le Dataset Complet</button>
|
| 354 |
</div>
|
| 355 |
|
| 356 |
+
<h3 style="margin-top: 30px;">👀 Aperçu des Questions</h3>
|
| 357 |
<div id="questionsPreview">
|
| 358 |
<table>
|
| 359 |
<thead>
|
|
|
|
| 362 |
<th>Type</th>
|
| 363 |
<th>Difficulté</th>
|
| 364 |
<th>But d'Entraînement</th>
|
| 365 |
+
<th>Chunk ID</th>
|
| 366 |
+
<th>Clé API</th>
|
| 367 |
</tr>
|
| 368 |
</thead>
|
| 369 |
<tbody id="questionsTableBody">
|
|
|
|
| 374 |
</div>
|
| 375 |
|
| 376 |
<script>
|
| 377 |
+
const apiBaseUrl = window.location.origin;
|
| 378 |
let generatedDataset = null;
|
| 379 |
let downloadUrl = '';
|
| 380 |
let statusCheckInterval = null;
|
| 381 |
+
let currentStatus = null;
|
| 382 |
|
| 383 |
+
// Event listeners
|
| 384 |
document.getElementById('generateBtn').addEventListener('click', startGeneration);
|
| 385 |
document.getElementById('downloadBtn').addEventListener('click', downloadDataset);
|
| 386 |
+
document.getElementById('downloadProgressBtn').addEventListener('click', downloadProgress);
|
| 387 |
+
document.getElementById('checkApiBtn').addEventListener('click', checkApiKeys);
|
| 388 |
+
document.getElementById('retryFailedBtn').addEventListener('click', showFailedChunks);
|
| 389 |
|
| 390 |
+
// Check for ongoing generation and API status when the page loads
|
| 391 |
+
window.addEventListener('load', () => {
|
| 392 |
+
checkApiKeys();
|
| 393 |
+
checkOngoingGeneration();
|
| 394 |
+
});
|
| 395 |
+
|
| 396 |
+
async function checkApiKeys() {
|
| 397 |
+
try {
|
| 398 |
+
const response = await fetch(`${apiBaseUrl}/api-keys-status`);
|
| 399 |
+
const status = await response.json();
|
| 400 |
+
|
| 401 |
+
const container = document.getElementById('apiStatusContainer');
|
| 402 |
+
|
| 403 |
+
if (status.status === 'success') {
|
| 404 |
+
container.innerHTML = `
|
| 405 |
+
<div class="api-status">
|
| 406 |
+
<strong>✅ ${status.total_keys} clés API configurées</strong><br>
|
| 407 |
+
<small>Clé actuelle: Index ${status.current_key_index} | Rotation automatique activée</small>
|
| 408 |
+
</div>
|
| 409 |
+
`;
|
| 410 |
+
} else {
|
| 411 |
+
container.innerHTML = `
|
| 412 |
+
<div class="api-status error">
|
| 413 |
+
<strong>❌ Erreur de configuration API</strong><br>
|
| 414 |
+
<small>${status.message}</small>
|
| 415 |
+
</div>
|
| 416 |
+
`;
|
| 417 |
+
}
|
| 418 |
+
} catch (error) {
|
| 419 |
+
console.error('Error checking API keys:', error);
|
| 420 |
+
document.getElementById('apiStatusContainer').innerHTML = `
|
| 421 |
+
<div class="api-status error">
|
| 422 |
+
<strong>❌ Impossible de vérifier les clés API</strong><br>
|
| 423 |
+
<small>${error.message}</small>
|
| 424 |
+
</div>
|
| 425 |
+
`;
|
| 426 |
+
}
|
| 427 |
}
|
| 428 |
|
| 429 |
+
async function checkOngoingGeneration() {
|
| 430 |
+
try {
|
| 431 |
+
const response = await fetch(`${apiBaseUrl}/generation-status`);
|
| 432 |
+
const status = await response.json();
|
| 433 |
+
currentStatus = status;
|
| 434 |
+
|
| 435 |
+
if (status.is_running || status.completed) {
|
| 436 |
+
setupStatusMonitoring();
|
| 437 |
+
updateStatusDisplay(status);
|
| 438 |
+
}
|
| 439 |
+
|
| 440 |
+
if (status.completed && status.result_file) {
|
| 441 |
+
downloadUrl = `/download/${status.result_file}`;
|
| 442 |
+
await loadResults();
|
| 443 |
+
}
|
| 444 |
+
} catch (error) {
|
| 445 |
+
console.error('Error checking generation status:', error);
|
| 446 |
+
}
|
| 447 |
+
}
|
| 448 |
+
|
| 449 |
+
async function startGeneration() {
|
| 450 |
const generateBtn = document.getElementById('generateBtn');
|
| 451 |
const statusContainer = document.getElementById('statusContainer');
|
| 452 |
const resultsContainer = document.getElementById('resultsContainer');
|
| 453 |
|
|
|
|
| 454 |
generateBtn.disabled = true;
|
| 455 |
statusContainer.style.display = 'block';
|
| 456 |
resultsContainer.style.display = 'none';
|
| 457 |
|
| 458 |
+
try {
|
| 459 |
+
const response = await fetch(`${apiBaseUrl}/generate-questions`);
|
| 460 |
+
|
| 461 |
+
if (!response.ok) {
|
| 462 |
+
throw new Error(`HTTP error! Status: ${response.status}`);
|
| 463 |
+
}
|
| 464 |
+
|
| 465 |
+
const data = await response.json();
|
| 466 |
+
console.log('Generation started:', data);
|
| 467 |
+
|
| 468 |
+
// Show success alert
|
| 469 |
+
showAlert('Génération démarrée avec succès!', 'success');
|
| 470 |
+
|
| 471 |
+
setupStatusMonitoring();
|
| 472 |
+
} catch (error) {
|
| 473 |
+
console.error('Error starting generation:', error);
|
| 474 |
+
showAlert(`Erreur lors du démarrage: ${error.message}`, 'error');
|
| 475 |
+
document.getElementById('statusText').textContent = `Erreur: ${error.message}`;
|
| 476 |
+
generateBtn.disabled = false;
|
| 477 |
+
}
|
| 478 |
}
|
| 479 |
|
| 480 |
function setupStatusMonitoring() {
|
|
|
|
| 481 |
if (statusCheckInterval) {
|
| 482 |
clearInterval(statusCheckInterval);
|
| 483 |
}
|
| 484 |
|
|
|
|
| 485 |
document.getElementById('statusContainer').style.display = 'block';
|
| 486 |
+
document.getElementById('downloadProgressBtn').style.display = 'inline-block';
|
| 487 |
|
| 488 |
+
statusCheckInterval = setInterval(checkGenerationStatus, 3000);
|
|
|
|
|
|
|
|
|
|
| 489 |
checkGenerationStatus();
|
| 490 |
}
|
| 491 |
|
| 492 |
+
async function checkGenerationStatus() {
|
| 493 |
+
try {
|
| 494 |
+
const response = await fetch(`${apiBaseUrl}/generation-status`);
|
| 495 |
+
const status = await response.json();
|
| 496 |
+
currentStatus = status;
|
| 497 |
+
|
| 498 |
+
updateStatusDisplay(status);
|
| 499 |
+
|
| 500 |
+
if (status.completed) {
|
| 501 |
+
clearInterval(statusCheckInterval);
|
| 502 |
+
downloadUrl = `/download/${status.result_file}`;
|
| 503 |
+
await loadResults();
|
| 504 |
+
showAlert('Génération terminée avec succès!', 'success');
|
| 505 |
+
}
|
| 506 |
+
|
| 507 |
+
if (status.error) {
|
| 508 |
+
clearInterval(statusCheckInterval);
|
| 509 |
+
document.getElementById('generateBtn').disabled = false;
|
| 510 |
+
showAlert(`Erreur durant la génération: ${status.error}`, 'error');
|
| 511 |
+
}
|
| 512 |
+
} catch (error) {
|
| 513 |
+
console.error('Error checking status:', error);
|
| 514 |
+
}
|
| 515 |
}
|
| 516 |
|
| 517 |
function updateStatusDisplay(status) {
|
|
|
|
| 520 |
const processedChunks = document.getElementById('processedChunks');
|
| 521 |
const totalChunks = document.getElementById('totalChunks');
|
| 522 |
const questionsGenerated = document.getElementById('questionsGenerated');
|
| 523 |
+
const currentApiKey = document.getElementById('currentApiKey');
|
| 524 |
+
const failedChunks = document.getElementById('failedChunks');
|
| 525 |
+
const timeEstimate = document.getElementById('timeEstimate');
|
| 526 |
+
const estimatedTime = document.getElementById('estimatedTime');
|
| 527 |
|
| 528 |
+
// Update status text
|
| 529 |
if (status.error) {
|
| 530 |
+
statusText.textContent = `❌ Erreur: ${status.error}`;
|
| 531 |
progressBar.classList.add('error');
|
| 532 |
} else if (status.completed) {
|
| 533 |
+
statusText.textContent = '✅ Génération terminée avec succès !';
|
| 534 |
progressBar.classList.add('complete');
|
| 535 |
} else if (status.is_running) {
|
| 536 |
+
statusText.textContent = '⚡ Génération des questions en cours...';
|
| 537 |
}
|
| 538 |
|
| 539 |
+
// Update counts and progress
|
| 540 |
+
processedChunks.textContent = status.processed_chunks || 0;
|
| 541 |
+
totalChunks.textContent = status.total_chunks || 0;
|
| 542 |
+
questionsGenerated.textContent = status.questions_generated || 0;
|
| 543 |
+
currentApiKey.textContent = status.current_api_key_index || 0;
|
| 544 |
+
failedChunks.textContent = (status.failed_chunks || []).length;
|
| 545 |
|
| 546 |
+
// Update progress bar
|
| 547 |
if (status.total_chunks > 0) {
|
| 548 |
const percentage = Math.round((status.processed_chunks / status.total_chunks) * 100);
|
| 549 |
progressBar.style.width = `${percentage}%`;
|
| 550 |
progressBar.textContent = `${percentage}%`;
|
| 551 |
}
|
| 552 |
+
|
| 553 |
+
// Show time estimate
|
| 554 |
+
if (status.estimated_remaining_minutes) {
|
| 555 |
+
timeEstimate.style.display = 'block';
|
| 556 |
+
const minutes = Math.round(status.estimated_remaining_minutes);
|
| 557 |
+
estimatedTime.textContent = `${minutes} minutes`;
|
| 558 |
+
}
|
| 559 |
+
|
| 560 |
+
// Show failed chunks
|
| 561 |
+
if (status.failed_chunks && status.failed_chunks.length > 0) {
|
| 562 |
+
document.getElementById('retryFailedBtn').style.display = 'inline-block';
|
| 563 |
+
updateFailedChunksDisplay(status.failed_chunks);
|
| 564 |
+
}
|
| 565 |
}
|
| 566 |
|
| 567 |
+
function updateFailedChunksDisplay(failedChunks) {
|
| 568 |
+
const container = document.getElementById('failedChunksContainer');
|
| 569 |
+
const list = document.getElementById('failedChunksList');
|
| 570 |
+
|
| 571 |
+
if (failedChunks.length > 0) {
|
| 572 |
+
container.style.display = 'block';
|
| 573 |
+
list.innerHTML = failedChunks.map(chunk => `
|
| 574 |
+
<div class="failed-chunk-item">
|
| 575 |
+
<strong>Chunk ${chunk.chunk_id}:</strong> ${chunk.error}<br>
|
| 576 |
+
<small>Tentatives: ${chunk.attempts}</small>
|
| 577 |
+
</div>
|
| 578 |
+
`).join('');
|
| 579 |
+
}
|
| 580 |
+
}
|
| 581 |
+
|
| 582 |
+
async function loadResults() {
|
| 583 |
if (!downloadUrl) return;
|
| 584 |
|
| 585 |
+
try {
|
| 586 |
+
const response = await fetch(`${apiBaseUrl}${downloadUrl}`);
|
| 587 |
+
if (!response.ok) {
|
| 588 |
+
throw new Error(`HTTP error! Status: ${response.status}`);
|
| 589 |
+
}
|
| 590 |
+
|
| 591 |
+
const dataset = await response.json();
|
| 592 |
+
generatedDataset = dataset;
|
| 593 |
+
displayResults(dataset);
|
| 594 |
+
|
| 595 |
+
document.getElementById('generateBtn').disabled = false;
|
| 596 |
+
} catch (error) {
|
| 597 |
+
console.error('Error loading results:', error);
|
| 598 |
+
showAlert(`Erreur lors du chargement des résultats: ${error.message}`, 'error');
|
| 599 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 600 |
}
|
| 601 |
|
| 602 |
function displayResults(dataset) {
|
|
|
|
| 604 |
const statsContainer = document.getElementById('statsContainer');
|
| 605 |
const questionsTableBody = document.getElementById('questionsTableBody');
|
| 606 |
|
| 607 |
+
// Clear and populate stats
|
| 608 |
statsContainer.innerHTML = '';
|
| 609 |
|
| 610 |
+
// Main stats
|
| 611 |
addStatCard(statsContainer, dataset.dataset_info.total_questions, 'Questions Totales');
|
| 612 |
+
addStatCard(statsContainer, dataset.dataset_info.successful_chunks || 0, 'Chunks Réussis');
|
| 613 |
+
addStatCard(statsContainer, dataset.dataset_info.failed_chunks || 0, 'Chunks Échoués');
|
| 614 |
|
| 615 |
+
// Type distribution
|
| 616 |
const typeCount = countByProperty(dataset.questions, 'type');
|
| 617 |
for (const [type, count] of Object.entries(typeCount)) {
|
| 618 |
+
addStatCard(statsContainer, count, `${capitalizeFirstLetter(type)}`);
|
| 619 |
}
|
| 620 |
|
| 621 |
+
// Difficulty distribution
|
| 622 |
const difficultyCount = countByProperty(dataset.questions, 'difficulty');
|
| 623 |
for (const [difficulty, count] of Object.entries(difficultyCount)) {
|
| 624 |
+
addStatCard(statsContainer, count, `${capitalizeFirstLetter(difficulty)}`);
|
| 625 |
+
}
|
| 626 |
+
|
| 627 |
+
// Success rate
|
| 628 |
+
if (dataset.dataset_info.total_chunks_processed) {
|
| 629 |
+
const successRate = Math.round(((dataset.dataset_info.successful_chunks || 0) / dataset.dataset_info.total_chunks_processed) * 100);
|
| 630 |
+
addStatCard(statsContainer, `${successRate}%`, 'Taux de Réussite');
|
| 631 |
}
|
| 632 |
|
| 633 |
+
// Populate questions table
|
| 634 |
questionsTableBody.innerHTML = '';
|
| 635 |
+
dataset.questions.forEach((question, index) => {
|
| 636 |
+
if (index < 50) { // Limit display for performance
|
| 637 |
+
const row = document.createElement('tr');
|
| 638 |
+
|
| 639 |
+
row.innerHTML = `
|
| 640 |
+
<td>${question.question}</td>
|
| 641 |
+
<td><span class="badge badge-${question.type}">${question.type}</span></td>
|
| 642 |
+
<td><span class="badge badge-${question.difficulty}">${question.difficulty}</span></td>
|
| 643 |
+
<td>${question.training_purpose}</td>
|
| 644 |
+
<td>${question.chunk_id}</td>
|
| 645 |
+
<td>${question.api_key_used || 'N/A'}</td>
|
| 646 |
+
`;
|
| 647 |
+
|
| 648 |
+
questionsTableBody.appendChild(row);
|
| 649 |
+
}
|
| 650 |
+
});
|
| 651 |
+
|
| 652 |
+
if (dataset.questions.length > 50) {
|
| 653 |
const row = document.createElement('tr');
|
| 654 |
+
row.innerHTML = `<td colspan="6" style="text-align: center; font-style: italic; color: #666;">... et ${dataset.questions.length - 50} autres questions</td>`;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 655 |
questionsTableBody.appendChild(row);
|
| 656 |
+
}
|
| 657 |
|
|
|
|
| 658 |
resultsContainer.style.display = 'block';
|
| 659 |
}
|
| 660 |
|
|
|
|
| 662 |
const card = document.createElement('div');
|
| 663 |
card.className = 'stat-card';
|
| 664 |
|
| 665 |
+
card.innerHTML = `
|
| 666 |
+
<div class="stat-value">${value}</div>
|
| 667 |
+
<div class="stat-label">${label}</div>
|
| 668 |
+
`;
|
| 669 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 670 |
container.appendChild(card);
|
| 671 |
}
|
| 672 |
|
|
|
|
| 685 |
|
| 686 |
function downloadDataset() {
|
| 687 |
if (downloadUrl) {
|
| 688 |
+
window.open(`${apiBaseUrl}${downloadUrl}`, '_blank');
|
| 689 |
+
}
|
| 690 |
+
}
|
| 691 |
+
|
| 692 |
+
async function downloadProgress() {
|
| 693 |
+
try {
|
| 694 |
+
const response = await fetch(`${apiBaseUrl}/download-progress`);
|
| 695 |
+
if (response.ok) {
|
| 696 |
+
const blob = await response.blob();
|
| 697 |
+
const url = window.URL.createObjectURL(blob);
|
| 698 |
+
const a = document.createElement('a');
|
| 699 |
+
a.href = url;
|
| 700 |
+
a.download = `progress_${new Date().getTime()}.json`;
|
| 701 |
+
document.body.appendChild(a);
|
| 702 |
+
a.click();
|
| 703 |
+
window.URL.revokeObjectURL(url);
|
| 704 |
+
document.body.removeChild(a);
|
| 705 |
+
showAlert('Fichier de progrès téléchargé!', 'success');
|
| 706 |
+
} else {
|
| 707 |
+
throw new Error('Pas de fichier de progrès disponible');
|
| 708 |
+
}
|
| 709 |
+
} catch (error) {
|
| 710 |
+
showAlert(`Erreur de téléchargement: ${error.message}`, 'error');
|
| 711 |
+
}
|
| 712 |
+
}
|
| 713 |
+
|
| 714 |
+
async function showFailedChunks() {
|
| 715 |
+
try {
|
| 716 |
+
const response = await fetch(`${apiBaseUrl}/retry-failed`);
|
| 717 |
+
const data = await response.json();
|
| 718 |
+
|
| 719 |
+
if (data.failed_chunks && data.failed_chunks.length > 0) {
|
| 720 |
+
updateFailedChunksDisplay(data.failed_chunks);
|
| 721 |
+
document.getElementById('failedChunksContainer').style.display = 'block';
|
| 722 |
+
showAlert(`${data.failed_chunks.length} chunks ont échoué. Voir les détails ci-dessous.`, 'warning');
|
| 723 |
+
} else {
|
| 724 |
+
showAlert('Aucun chunk échoué trouvé!', 'success');
|
| 725 |
+
}
|
| 726 |
+
} catch (error) {
|
| 727 |
+
showAlert(`Erreur: ${error.message}`, 'error');
|
| 728 |
}
|
| 729 |
}
|
| 730 |
+
|
| 731 |
+
function showAlert(message, type) {
|
| 732 |
+
const alertDiv = document.createElement('div');
|
| 733 |
+
alertDiv.className = `alert alert-${type}`;
|
| 734 |
+
alertDiv.textContent = message;
|
| 735 |
+
|
| 736 |
+
document.body.insertBefore(alertDiv, document.body.firstChild);
|
| 737 |
+
|
| 738 |
+
setTimeout(() => {
|
| 739 |
+
alertDiv.remove();
|
| 740 |
+
}, 5000);
|
| 741 |
+
}
|
| 742 |
</script>
|
| 743 |
</body>
|
| 744 |
</html>
|