xiaoxia-saas/tests/unit/test_douyin_url_extract.py

"""抖音分享文本 URL 提取单测。"""

from __future__ import annotations

import pytest

# 直接 import 模块，用 _extract_url_from_text / _extract_and_validate_douyin_url 测试
from app.api.routes.scripts_ai import (
    _extract_and_validate_douyin_url,
    _extract_url_from_text,
)
from fastapi import HTTPException, status


class TestExtractUrlFromText:
    def test_pure_url(self):
        assert _extract_url_from_text("https://v.douyin.com/iZ7vU2qH/") == "https://v.douyin.com/iZ7vU2qH/"

    def test_share_text_with_prefix_suffix(self):
        """典型"复制链接"场景：包含中文+表情+链接+话题标签。"""
        s = "这个视频太搞笑了  https://v.douyin.com/iZ7vU2qH/ 快来看看！#搞笑 #日常"
        assert _extract_url_from_text(s) == "https://v.douyin.com/iZ7vU2qH/"

    def test_share_text_no_http_prefix(self):
        s = "复制此链接，打开Dou音搜索，直接观看视频！v.douyin.com/iZ7vU2qH/"
        # Should pick up v.douyin.com/... and add https:// prefix
        url = _extract_url_from_text(s)
        assert url and url.endswith("v.douyin.com/iZ7vU2qH/")

    def test_long_url_www(self):
        s = "https://www.douyin.com/video/7234567890123456789?previous_page=web_code_link"
        assert _extract_url_from_text(s) == s

    def test_empty_input(self):
        assert _extract_url_from_text("") is None
        assert _extract_url_from_text(None) is None  # type: ignore[arg-type]

    def test_no_url(self):
        assert _extract_url_from_text("这个视频很好看，但是没有链接") is None

    def test_trailing_punct_stripped(self):
        s = "https://v.douyin.com/iZ7vU2qH/。"
        assert _extract_url_from_text(s) == "https://v.douyin.com/iZ7vU2qH/"


class TestValidateUrl:
    def test_pure_short_url_ok(self):
        assert _extract_and_validate_douyin_url("https://v.douyin.com/iZ7vU2qH/").startswith("https://")

    def test_share_text_ok(self):
        s = "这个视频太搞笑了 https://v.douyin.com/abcdefG/ 快来看看！"
        url = _extract_and_validate_douyin_url(s)
        assert "douyin.com" in url

    def test_empty_raises_400(self):
        with pytest.raises(HTTPException) as ei:
            _extract_and_validate_douyin_url("")
        assert ei.value.status_code == 400

    def test_no_url_raises_400(self):
        with pytest.raises(HTTPException) as ei:
            _extract_and_validate_douyin_url("这个视频没有链接")
        assert ei.value.status_code == 400

    def test_non_douyin_raises_400(self):
        with pytest.raises(HTTPException) as ei:
            _extract_and_validate_douyin_url("https://www.bilibili.com/video/BV1xx411c7mD")
        assert ei.value.status_code == 400

    def test_scheme_added_when_missing(self):
        """只输入 v.douyin.com/xxx 时，补 https://。"""
        url = _extract_and_validate_douyin_url("v.douyin.com/iZ7vU2qH/")
        assert url.startswith("https://")