ftnext

from rouge_score.rouge_scorer import RougeScorer
from rouge_score.tokenize import SPACES_RE
from rouge_score.tokenizers import Tokenizer


class NonAlphaNumericSupportTokenizer(Tokenizer):
    """
    >>> NonAlphaNumericSupportTokenizer().tokenize("いぬ ねこ")
    ['いぬ', 'ねこ']
    """

ftnext

"""
>>> HogeTranslator().foo(1)
MyTranslator foo
108
>>> HogeTranslator().foo(-1)
MyTranslator foo
Translator foo
42
"""

ftnext

# Code with bug. ref: https://stackoverflow.com/q/32831150
class A:
    def __init__(self, v, v2):
        self.v = v
        self.v2 = v2


class B(A):
    def __init__(self, v, v2):
        print(self.__class__)

ftnext

# ref: https://github.com/asciidwango/ExpertPython3_Source/blob/887b151bc5f20b6b6e742eff1f18d4ab6e86d872/chapter17/interfaces_abc.py
from abc import ABCMeta, abstractmethod


class RectangleInterface(metaclass=ABCMeta):
    @abstractmethod
    def area(self):
        """面積を返す"""

ftnext

# https://github.com/ftnext/unko-by-rye/blob/500e6a8cb5f9c510c99820a9ed2148e60a77e45f/docker/Dockerfile
FROM python:3.12-bookworm AS builder
WORKDIR /work
COPY requirements.lock requirements.lock
RUN <<install_python_packages
sed -i '/^-e/d' requirements.lock
python -m pip install --no-cache-dir -r requirements.lock
install_python_packages
COPY . .
RUN python -m pip install --no-cache-dir .

ftnext

# ref: https://docs.pydantic.dev/2.6/concepts/unions/#discriminated-unions
from typing import Literal

from pydantic import BaseModel, Field


class Cat(BaseModel):
    pet_type: Literal["cat"]
    meows: int

ftnext

# Based on https://gist.github.com/ftnext/b0f4db8dc71333f7b663c4f5da9ec16f
import numpy as np
from openai import OpenAI

client = OpenAI()


def get_embedding(text, model):
    response = client.embeddings.create(input=text, model=model)  # dimensions=256
    embedding = response.data[0].embedding

ftnext

# Based on https://gist.github.com/ftnext/74e19bc4e478fd02bfed72de3552512c
import numpy as np
from openai import OpenAI

client = OpenAI()


def get_embedding(text, model):
    response = client.embeddings.create(input=text, model=model)
    embedding = response.data[0].embedding

ftnext

# ref: https://huggingface.co/blog/how-to-generate#sampling
from transformers import AutoTokenizer, AutoModelForCausalLM, set_seed

tokenizer = AutoTokenizer.from_pretrained("gpt2")
model = AutoModelForCausalLM.from_pretrained("gpt2")


def generate(prompt):
    model_inputs = tokenizer(prompt, return_tensors="pt")
    output = model.generate(

ftnext

# See ref: https://github.com/ftnext/transcendent-book-py/blob/ce7acd2daa4e8ed0770e8739dabe2de9b0046388/chapter8/composite_example.py
class Node(metaclass=ABCMeta):
    def __init__(self, name: str) -> None:
        self.name = name
        self.parent: Branch | None = None

    @abstractmethod
    def __str__(self) -> str:
        ...