Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
File size: 6,025 Bytes
81cdd5f |
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 |
# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import os
import sys
from flask import Flask, send_from_directory
import case_util
import config
from llm_client import VertexAILLMClient
from llm_client import HuggingFaceLLMClient
from background_task_manager import BackgroundTaskManager
from cache_manager import CacheManager
from rag.knowledge_base import KnowledgeBase
from rag.model_manager import ModelManager
from rag.rag_context_engine import RAGContextEngine, format_context_messages_to_string
from routes import main_bp
def _get_llm_client():
"""Initializes the LLM client and handles exit on failure."""
logger = logging.getLogger(__name__)
if config.MEDGEMMA_LOCATION == 'HUGGING_FACE':
logger.info("HUGGING_FACE MedGemma end point initialized.")
return HuggingFaceLLMClient(config.HF_TOKEN, config.MEDGEMMA_ENDPOINT_URL)
elif config.MEDGEMMA_LOCATION == 'VERTEX_AI':
logger.info("Vertex AI MedGemma end point initialized.")
return VertexAILLMClient(config.GCLOUD_SA_KEY, config.MEDGEMMA_ENDPOINT_URL)
logger.critical("LLM client failed to initialize. API calls will fail.")
sys.exit("Exiting: LLM client initialization failed.")
def _initialize_rag_system(flask_app: Flask):
"""Checks for persistent cache and initializes the RAG system."""
logger = logging.getLogger(__name__)
rag_context_cache = {}
# RAG Run is not needed if cache is present.
if config.USE_CACHE:
cache_manager = flask_app.config['DEMO_CACHE']
if len(cache_manager.cache) > 0:
logger.warning(f"The cache is not empty, so not initialising the RAG system.")
return
else:
logger.info(f"The cache is empty, so resuming the RAG initialisation")
try:
logger.info("--- Initializing RAG System and pre-fetching context... ---")
rag_model_manager = ModelManager()
rag_models = rag_model_manager.load_models()
if not rag_models.get("embedder"): raise RuntimeError("RAG embedder failed to load.")
knowledge_base = KnowledgeBase(models=rag_models)
knowledge_base.build(pdf_filepath=config.GUIDELINE_PDF_PATH)
if not knowledge_base.retriever: raise RuntimeError("Failed to build the RAG retriever.")
rag_engine = RAGContextEngine(knowledge_base=knowledge_base)
all_cases = flask_app.config.get("AVAILABLE_REPORTS", {})
for case_id, case_data in all_cases.items():
ground_truth_labels = case_data.ground_truth_labels
if not ground_truth_labels: continue
rag_queries = [label.lower() for label in ground_truth_labels.keys()]
if "normal" in rag_queries: continue
retrieved_docs = rag_engine.retrieve_context_docs_for_simple_queries(rag_queries)
citations = sorted(list(
set(doc.metadata.get("page_number") for doc in retrieved_docs if doc.metadata.get("page_number"))))
context_messages, _ = rag_engine.build_context_messages(retrieved_docs)
context_string = format_context_messages_to_string(context_messages)
rag_context_cache[case_id] = {"context_string": context_string, "citations": citations}
logger.info("✅ RAG System ready.")
except Exception as e:
logger.critical(f"FATAL: RAG System failed to initialize: {e}", exc_info=True)
sys.exit("Exiting: RAG system initialization failed.")
flask_app.config['RAG_CONTEXT_CACHE'] = rag_context_cache
def _initialize_demo_cache(flask_app: Flask):
"""Initializes the disk cache for MCQs and summary templates."""
logger = logging.getLogger(__name__)
if config.USE_CACHE:
cache_dir = os.getenv('CACHE_DIR', config.BASE_DIR / "persistent_cache")
cache_manager = CacheManager(cache_dir)
flask_app.config['DEMO_CACHE'] = cache_manager
logger.info("✅ Cache Setup Complete.")
else:
logger.warning("⚠️ Caching is DISABLED.")
flask_app.config['DEMO_CACHE'] = None
def _register_routes(flask_app: Flask):
"""Registers blueprints and defines static file serving."""
flask_app.register_blueprint(main_bp)
@flask_app.route('/', defaults={'path': ''})
@flask_app.route('/<path:path>')
def serve(path):
if path != "" and os.path.exists(os.path.join(flask_app.static_folder, path)):
return send_from_directory(flask_app.static_folder, path)
else:
return send_from_directory(flask_app.static_folder, 'index.html')
def create_app():
"""Creates and configures the Flask application by calling modular helper functions."""
application = Flask(__name__, static_folder=config.STATIC_DIR)
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - [%(name)s] - %(message)s')
# Sequentially call setup functions
application.config["LLM_CLIENT"] = _get_llm_client()
application.config["AVAILABLE_REPORTS"] = case_util.get_available_reports(config.MANIFEST_CSV_PATH)
_initialize_demo_cache(application)
task_manager = BackgroundTaskManager()
application.config['TASK_MANAGER'] = task_manager
# RAG and Cache initialization in the background
task_manager.start_task(key="rag_system", target_func=_initialize_rag_system, flask_app=application)
_register_routes(application)
return application
app = create_app()
if __name__ == '__main__':
app.run(host='0.0.0.0', port=7860, debug=True)
|