Update api.py
Browse files
api.py
CHANGED
|
@@ -1,24 +1,47 @@
|
|
| 1 |
import os
|
| 2 |
import traceback
|
|
|
|
|
|
|
| 3 |
from flask import Flask, request, jsonify
|
| 4 |
from flask_cors import CORS
|
| 5 |
from dotenv import load_dotenv
|
| 6 |
-
from openai import OpenAI
|
| 7 |
from langdetect import detect
|
| 8 |
from deep_translator import GoogleTranslator
|
| 9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
# Patch huggingface_hub automatically if Gradio overwrote it
|
| 12 |
try:
|
| 13 |
import huggingface_hub
|
| 14 |
if not hasattr(huggingface_hub, "cached_download"):
|
| 15 |
subprocess.run(
|
| 16 |
-
[
|
| 17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
)
|
| 19 |
print("✅ Downgraded huggingface-hub for sentence-transformers compatibility.")
|
| 20 |
except Exception as e:
|
| 21 |
print("⚠️ Could not auto-patch huggingface_hub:", e)
|
|
|
|
| 22 |
from sentence_transformers import SentenceTransformer
|
| 23 |
from pinecone import Pinecone
|
| 24 |
|
|
@@ -37,7 +60,11 @@ CORS(app, resources={r"/ask": {"origins": "*"}})
|
|
| 37 |
# ---------- OpenAI Client ----------
|
| 38 |
client = None
|
| 39 |
if OPENAI_API_KEY:
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
else:
|
| 42 |
print("⚠️ OPENAI_API_KEY is missing in .env")
|
| 43 |
|
|
@@ -63,11 +90,13 @@ try:
|
|
| 63 |
results = []
|
| 64 |
for match in matches:
|
| 65 |
meta = match.get("metadata", {})
|
| 66 |
-
results.append(
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
|
|
|
|
|
|
| 71 |
return results
|
| 72 |
|
| 73 |
retriever = PineconeRetriever(index, embedder)
|
|
@@ -94,7 +123,9 @@ def detect_language(question: str) -> str:
|
|
| 94 |
|
| 95 |
def normalize_language(lang: str, text: str) -> str:
|
| 96 |
"""Fix incorrect language detection like 'wer is' → German."""
|
| 97 |
-
if lang == "nl" and any(
|
|
|
|
|
|
|
| 98 |
return "de"
|
| 99 |
return lang
|
| 100 |
|
|
@@ -120,18 +151,24 @@ def format_answers(question: str, answer: str, results):
|
|
| 120 |
pages = [f"Seite {r.get('page', '')}" for r in results if r.get("page")]
|
| 121 |
source = ", ".join(pages) if pages else "No source"
|
| 122 |
top_score = max([r.get("score", 0.0) for r in results], default=0.0)
|
| 123 |
-
return {
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
|
| 125 |
# ---------- Routes ----------
|
| 126 |
@app.route("/", methods=["GET"])
|
| 127 |
def health():
|
| 128 |
-
return jsonify(
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
|
|
|
|
|
|
| 135 |
|
| 136 |
@app.route("/ask", methods=["POST", "OPTIONS"])
|
| 137 |
def ask():
|
|
@@ -181,7 +218,7 @@ def ask():
|
|
| 181 |
model="gpt-4o-mini",
|
| 182 |
messages=[
|
| 183 |
{"role": "system", "content": sys_prompt},
|
| 184 |
-
{"role": "user", "content": user_content}
|
| 185 |
],
|
| 186 |
max_tokens=700,
|
| 187 |
)
|
|
|
|
| 1 |
import os
|
| 2 |
import traceback
|
| 3 |
+
import subprocess
|
| 4 |
+
import importlib
|
| 5 |
from flask import Flask, request, jsonify
|
| 6 |
from flask_cors import CORS
|
| 7 |
from dotenv import load_dotenv
|
|
|
|
| 8 |
from langdetect import detect
|
| 9 |
from deep_translator import GoogleTranslator
|
| 10 |
+
|
| 11 |
+
# ---------- Fix OpenAI version issue ----------
|
| 12 |
+
try:
|
| 13 |
+
# Ensure modern OpenAI SDK is installed and loaded
|
| 14 |
+
subprocess.run(
|
| 15 |
+
["pip", "install", "--no-cache-dir", "--upgrade", "openai==1.12.0"],
|
| 16 |
+
check=True,
|
| 17 |
+
)
|
| 18 |
+
import openai
|
| 19 |
+
importlib.reload(openai)
|
| 20 |
+
from openai import OpenAI
|
| 21 |
+
print("✅ Using OpenAI SDK version 1.12.0")
|
| 22 |
+
except Exception as e:
|
| 23 |
+
print("⚠️ Could not ensure correct OpenAI version:", e)
|
| 24 |
+
from openai import OpenAI # fallback import
|
| 25 |
|
| 26 |
# Patch huggingface_hub automatically if Gradio overwrote it
|
| 27 |
try:
|
| 28 |
import huggingface_hub
|
| 29 |
if not hasattr(huggingface_hub, "cached_download"):
|
| 30 |
subprocess.run(
|
| 31 |
+
[
|
| 32 |
+
"pip",
|
| 33 |
+
"install",
|
| 34 |
+
"--no-cache-dir",
|
| 35 |
+
"huggingface-hub==0.24.5",
|
| 36 |
+
"transformers==4.30.2",
|
| 37 |
+
"sentence-transformers==2.2.2",
|
| 38 |
+
],
|
| 39 |
+
check=True,
|
| 40 |
)
|
| 41 |
print("✅ Downgraded huggingface-hub for sentence-transformers compatibility.")
|
| 42 |
except Exception as e:
|
| 43 |
print("⚠️ Could not auto-patch huggingface_hub:", e)
|
| 44 |
+
|
| 45 |
from sentence_transformers import SentenceTransformer
|
| 46 |
from pinecone import Pinecone
|
| 47 |
|
|
|
|
| 60 |
# ---------- OpenAI Client ----------
|
| 61 |
client = None
|
| 62 |
if OPENAI_API_KEY:
|
| 63 |
+
try:
|
| 64 |
+
client = OpenAI(api_key=OPENAI_API_KEY)
|
| 65 |
+
print("✅ OpenAI client initialized successfully.")
|
| 66 |
+
except Exception as e:
|
| 67 |
+
print("❌ Failed to initialize OpenAI client:", e)
|
| 68 |
else:
|
| 69 |
print("⚠️ OPENAI_API_KEY is missing in .env")
|
| 70 |
|
|
|
|
| 90 |
results = []
|
| 91 |
for match in matches:
|
| 92 |
meta = match.get("metadata", {})
|
| 93 |
+
results.append(
|
| 94 |
+
{
|
| 95 |
+
"context": meta.get("context", ""),
|
| 96 |
+
"page": meta.get("page"),
|
| 97 |
+
"score": match.get("score", 0),
|
| 98 |
+
}
|
| 99 |
+
)
|
| 100 |
return results
|
| 101 |
|
| 102 |
retriever = PineconeRetriever(index, embedder)
|
|
|
|
| 123 |
|
| 124 |
def normalize_language(lang: str, text: str) -> str:
|
| 125 |
"""Fix incorrect language detection like 'wer is' → German."""
|
| 126 |
+
if lang == "nl" and any(
|
| 127 |
+
word in text.lower() for word in ["wer", "was", "wie", "javid", "coaching"]
|
| 128 |
+
):
|
| 129 |
return "de"
|
| 130 |
return lang
|
| 131 |
|
|
|
|
| 151 |
pages = [f"Seite {r.get('page', '')}" for r in results if r.get("page")]
|
| 152 |
source = ", ".join(pages) if pages else "No source"
|
| 153 |
top_score = max([r.get("score", 0.0) for r in results], default=0.0)
|
| 154 |
+
return {
|
| 155 |
+
"answers": [
|
| 156 |
+
{"question": question, "answer": answer, "source": source, "bm25_score": top_score}
|
| 157 |
+
]
|
| 158 |
+
}
|
| 159 |
|
| 160 |
# ---------- Routes ----------
|
| 161 |
@app.route("/", methods=["GET"])
|
| 162 |
def health():
|
| 163 |
+
return jsonify(
|
| 164 |
+
{
|
| 165 |
+
"status": "running",
|
| 166 |
+
"retriever_ready": bool(retriever),
|
| 167 |
+
"openai_key_loaded": bool(OPENAI_API_KEY),
|
| 168 |
+
"pinecone_key_loaded": bool(PINECONE_API_KEY),
|
| 169 |
+
"index_name": PINECONE_INDEX_NAME,
|
| 170 |
+
}
|
| 171 |
+
)
|
| 172 |
|
| 173 |
@app.route("/ask", methods=["POST", "OPTIONS"])
|
| 174 |
def ask():
|
|
|
|
| 218 |
model="gpt-4o-mini",
|
| 219 |
messages=[
|
| 220 |
{"role": "system", "content": sys_prompt},
|
| 221 |
+
{"role": "user", "content": user_content},
|
| 222 |
],
|
| 223 |
max_tokens=700,
|
| 224 |
)
|