add github workflow

Author: user786875

.github/workflows/linting-and-testing.yml

@@ -0,0 +1,44 @@
+on:
+  push:
+    branches: [main]
+  pull_request:
+    types: [opened, synchronize]
+
+name: linting and testing
+
+jobs:
+  test:
+    name: Automatic linting and testing
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v2
+
+      - name: Setup Python 3.10.5
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.10.5
+
+      - name: Setup Poetry 1.5.1
+        run: pip install poetry==1.5.1
+
+      - name: Check pyproject.toml
+        run: poetry check
+
+      - name: Install python dependencies with Poetry
+        run: poetry install
+
+      - name: Check files with isort
+        run: poetry run isort . --check
+
+      - name: Check files with Black
+        run: poetry run black . --check
+
+      - name: Lint files with flake8
+        run: poetry run flake8
+
+      - name: Check types with mypy
+        run: poetry run mypy
+
+      - name: Run tests
+        run: poetry run pytest

Dockerfile

@@ -0,0 +1,53 @@
+FROM python:3.10.5-slim as base
+
+# Python settings
+ENV PYTHONFAULTHANDLER=1 \
+    PYTHONUNBUFFERED=1 \
+    PYTHONHASHSEED=random \
+    PYTHONDONTWRITEBYTECODE=1
+
+
+WORKDIR /app
+
+
+FROM base as builder
+
+ENV \
+    # Pip settings
+    PIP_DEFAULT_TIMEOUT=100 \
+    PIP_DISABLE_PIP_VERSION_CHECK=1 \
+    PIP_NO_CACHE_DIR=1 \
+    # Poetry settings
+    POETRY_NO_INTERACTION=1 \
+    POETRY_VERSION=1.5.1
+
+# Setup poetry
+RUN pip install "poetry==$POETRY_VERSION"
+RUN python -m venv /venv
+
+# Copy relevant files
+COPY pyproject.toml poetry.lock .
+COPY vector_embedding_server ./vector_embedding_server
+RUN  touch /app/README.md # required by poetry somehow
+#CMD ["tail", "-f", "/dev/null"]
+
+
+
+
+RUN . /venv/bin/activate && \
+    # Install dependencies
+    poetry install -n --only main --no-root && \
+    # Install root package
+    poetry build -f wheel -n && \
+    pip install --no-deps dist/*.whl && \
+    rm -rf dist *.egg-info
+
+
+FROM base as final
+
+ENV PATH="/venv/bin:$PATH"
+
+COPY --from=builder /venv /venv
+COPY ./vector_embedding_server/server.py ./server.py
+COPY ./vector_embedding_server/templates ./templates
+CMD uvicorn server:app --host 0.0.0.0 --port 8080

README.md

@@ -1 +1,18 @@
-# vector-embedding-server
+# vector-embedding-server
+
+
+https://huggingface.co/intfloat/e5-large-v2/blob/main/README.md
+
+
+
+
+
+
+pip install fastapi-security python-jose[cryptography] passlib[bcrypt]
+
+
+-H "Authorization: Bearer $OPENAI_API_KEY" \
+
+
+
+TODO include fastapi-jwt

docker-compose.yml

@@ -0,0 +1,12 @@
+version: "3"
+
+services:
+  vector-embedding-server:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: vector-embedding-server:latest
+    environment:
+      TZ: Europe/Berlin
+    ports:
+      - "8080:8080"

vector_embedding_server/__init__py

@@ -0,0 +1,44 @@
+from typing import Optional
+from datetime import datetime, timedelta
+from jose import jwt
+from passlib.context import CryptContext
+from fastapi import Depends, HTTPException, status
+from fastapi.security import OAuth2PasswordBearer, OAuth2PasswordRequestForm
+
+
+oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token")
+
+pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
+
+FAKE_USERS_DB = {
+    "BCH": {
+        "username": "BCH",
+        "hashed_password": "$2b$12$YP6UgESiJ6.3c0EwnxNEnu9Ts075Jz82AcqawG7fxvFiMSUgs6cWK",
+        "disabled": False,
+    }
+}
+
+def authenticate_user(fake_db, username: str, password: str):
+    user = get_user(username)
+    print(user)
+    print(password)
+    if not user:
+        return False
+    if not pwd_context.verify(password, user["hashed_password"]):
+        return False
+    return user
+
+def get_user(username: str):
+    if username in FAKE_USERS_DB:
+        user_dict = FAKE_USERS_DB[username]
+        return user_dict
+
+def create_access_token(data: dict, expires_delta: Optional[timedelta] = None):
+    to_encode = data.copy()
+    if expires_delta:
+        expire = datetime.utcnow() + expires_delta
+    else:
+        expire = datetime.utcnow() + timedelta(minutes=15)
+    to_encode.update({"exp": expire})
+    encoded_jwt = jwt.encode(to_encode, "SECRET_KEY", algorithm="HS256")  # Ersetzen Sie SECRET_KEY durch Ihren geheimen Schlüssel
+    return encoded_jwt

vector_embedding_server/auth.py

@@ -0,0 +1,24 @@
+# copied mostly from https://huggingface.co/intfloat/e5-large-v2/blob/main/README.md
+
+import torch.nn.functional as F
+from torch import Tensor
+from transformers import AutoModel, AutoTokenizer
+
+
+def average_pool(last_hidden_states: Tensor, attention_mask: Tensor) -> Tensor:
+    last_hidden = last_hidden_states.masked_fill(~attention_mask[..., None].bool(), 0.0)
+    return last_hidden.sum(dim=1) / attention_mask.sum(dim=1)[..., None]
+
+
+def predict(input_text: str) -> tuple[list[float], int]:
+    tokenizer = AutoTokenizer.from_pretrained("intfloat/e5-large-v2")
+    model = AutoModel.from_pretrained("intfloat/e5-large-v2")
+
+    # Tokenize the input texts
+    batch_dict = tokenizer(
+        [input_text], max_length=512, padding=True, truncation=True, return_tensors="pt"
+    )
+
+    outputs = model(**batch_dict)
+    embeddings = average_pool(outputs.last_hidden_state, batch_dict["attention_mask"])
+    return embeddings.tolist()[0], len(tokenizer.all_special_ids)

vector_embedding_server/e5_large_v2.py

@@ -0,0 +1,30 @@
+from typing import List
+from pydantic import BaseModel
+from enum import Enum
+
+
+class EmbeddingData(BaseModel):
+    object: str
+    embedding: List[float]
+    index: int
+
+
+class Usage(BaseModel):
+    prompt_tokens: int
+    total_tokens: int
+
+
+class EmbeddingResponse(BaseModel):
+    object: str
+    data: List[EmbeddingData]
+    model: str
+    usage: Usage
+
+
+class ModelName(str, Enum):
+    e5_large_v2 = "e5-large-v2"
+
+
+class EmbeddingInput(BaseModel):
+    model: ModelName
+    input: str

vector_embedding_server/openai_like_api_models.py

@@ -0,0 +1,106 @@
+import os
+from pathlib import Path
+from typing import Optional
+
+from fastapi import FastAPI, Query, Request, Depends, HTTPException, status
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import HTMLResponse
+from fastapi.templating import Jinja2Templates
+from datetime import datetime, timedelta
+from vector_embedding_server.auth import authenticate_user, create_access_token, FAKE_USERS_DB, OAuth2PasswordRequestForm, get_user
+
+from vector_embedding_server.openai_like_api_models import (
+    EmbeddingResponse,
+    EmbeddingInput,
+    ModelName,
+    EmbeddingData,
+    Usage,
+)
+from vector_embedding_server import e5_large_v2
+from pydantic import BaseModel
+from fastapi import Depends, HTTPException, status
+from fastapi.security import OAuth2PasswordBearer
+from jose import JWTError, jwt
+
+
+BASE_DIR = Path(__file__).resolve().parent
+
+templates = Jinja2Templates(directory=str(BASE_DIR / "templates"))
+
+app = FastAPI(title="Vector Embedding Server", docs_url=None)
+
+
+class Credentials(BaseModel):
+
+    username: str
+    password: str
+
+
+oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token")
+
+
+def get_current_user(token: str = Depends(oauth2_scheme)):
+    credentials_exception = HTTPException(
+        status_code=status.HTTP_401_UNAUTHORIZED,
+        detail="Could not validate credentials",
+        headers={"WWW-Authenticate": "Bearer"},
+    )
+    try:
+        payload = jwt.decode(token, "SECRET_KEY", algorithms=["HS256"])
+        print(payload)
+        username: str = payload.get("sub")
+        if username is None:
+            raise credentials_exception
+    except JWTError:
+        raise credentials_exception
+    print("username:udirtae", username)
+    user = get_user(username)  # Hier sollten Sie die Funktion zur Abrufung des Benutzers aus Ihrer Benutzerdatenbank aufrufen
+    if user is None:
+        raise credentials_exception
+    return user
+
+@app.post("/token")
+def login(credentials: Credentials):
+    user = authenticate_user(FAKE_USERS_DB, credentials.username, credentials.password)
+    if not user:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="Incorrect username or password",
+            headers={"WWW-Authenticate": "Bearer"},
+        )
+    access_token_expires = timedelta(minutes=15)
+    access_token = create_access_token(
+        data={"sub": user["username"]}, expires_delta=access_token_expires
+    )
+    return {"access_token": access_token}
+
+
+@app.post("/v1/embeddings", response_model=EmbeddingResponse)
+async def create_embedding(embedding_input: EmbeddingInput, current_user: str = Depends(get_current_user)):
+    if embedding_input.model == ModelName.e5_large_v2:
+        embedding, prompt_tokens = e5_large_v2.predict(embedding_input.input)
+    else:
+        raise NotImplemented
+
+    embedding_data = EmbeddingData(
+        object="embedding",
+        embedding=embedding,
+        index=0,
+    )
+
+    usage = Usage(prompt_tokens=prompt_tokens, total_tokens=prompt_tokens)
+
+    embedding_response = EmbeddingResponse(
+        model=embedding_input.model.name,
+        object="list",
+        data=[embedding_data],
+        usage=usage,
+    )
+    return embedding_response
+
+
+@app.get("/docs", response_class=HTMLResponse)
+async def docs(request: Request):  # type: ignore
+    return templates.TemplateResponse(
+        "stoplight-element-api-doc.html", {"request": request}
+    )