xiaoxia-saas/tests/unit/test_text_splitter.py

"""文本分段工具单元测试."""

from __future__ import annotations

import pytest

from packages.application.tts_job.text_splitter import split_text


class TestSplitText:
    """split_text 函数测试"""

    def test_empty_string_returns_empty_list(self):
        """空字符串返回空列表"""
        assert split_text("") == []

    def test_whitespace_only_returns_empty_list(self):
        """纯空白字符返回空列表"""
        assert split_text("   \n  \t  ") == []

    def test_short_text_returns_single_segment(self):
        """短文本直接返回单段"""
        text = "这是一段短文本。"
        result = split_text(text, max_chars=500)
        assert result == [text]

    def test_text_length_equals_max_chars(self):
        """文本长度恰好等于 max_chars 时返回单段"""
        text = "a" * 100
        result = split_text(text, max_chars=100)
        assert len(result) == 1
        assert len(result[0]) == 100

    def test_splits_on_sentence_boundary(self):
        """在句子边界处分段"""
        # 构造长文本，确保超过 max_chars
        sentences = ["今天天气真好。我们一起去公园散步吧。", "公园里有很多花。还有很多小朋友在玩耍。"] * 10
        text = "".join(sentences)

        result = split_text(text, max_chars=200)

        assert len(result) >= 2
        # 每段都不超过 max_chars
        for seg in result:
            assert len(seg) <= 200

    def test_all_segments_within_max_chars(self):
        """所有分段都不超过 max_chars"""
        text = "这是第一句话。这是第二句话。这是第三句话。这是第四句话。这是第五句话。" * 10

        result = split_text(text, max_chars=100)

        for seg in result:
            assert len(seg) <= 100

    def test_long_single_sentence_hard_cut(self):
        """超长单句会被硬切"""
        text = "a" * 1000  # 没有标点

        result = split_text(text, max_chars=200)

        assert len(result) > 1
        for seg in result:
            assert len(seg) <= 200

    def test_newline_is_sentence_end(self):
        """换行符作为句子结束符"""
        text = "第一行内容\n第二行内容\n第三行内容" * 10

        result = split_text(text, max_chars=50)

        assert len(result) > 1
        for seg in result:
            assert len(seg) <= 50

    def test_chinese_punctuation(self):
        """中文标点（。！？；）作为句子结束符"""
        text = "你好！今天吃什么？我吃米饭；你呢？我也吃米饭。" * 10

        result = split_text(text, max_chars=80)

        for seg in result:
            assert len(seg) <= 80

    def test_english_punctuation(self):
        """英文标点（.!?;）作为句子结束符"""
        text = "Hello! How are you? I'm fine; thank you. Good bye." * 10

        result = split_text(text, max_chars=80)

        for seg in result:
            assert len(seg) <= 80

    def test_merged_short_segments(self):
        """过短的段落会被合并"""
        # 构造很多短句
        text = "你好。再见。谢谢。抱歉。好的。不行。可以。去吧。" * 5  # 每句3-4字

        result = split_text(text, max_chars=100)

        # 合并后段数应该比单纯按句切的少
        assert len(result) < len(text) // 3  # 粗略估计
        for seg in result:
            assert len(seg) <= 100

    def test_preserves_content(self):
        """分段后内容总和与原文基本一致（忽略strip的空白）"""
        text = "这是测试文本。包含多个句子。用来验证分段正确性。" * 5

        result = split_text(text, max_chars=50)

        # 合并所有分段，去掉空白后应该与原文去掉空白后基本一致
        combined = "".join(result).replace(" ", "")
        original = text.strip().replace(" ", "")
        assert combined == original

    def test_custom_max_chars(self):
        """支持自定义 max_chars"""
        text = "测试" * 100  # 200字

        result_50 = split_text(text, max_chars=50)
        result_100 = split_text(text, max_chars=100)

        # max_chars 越小，段数应该越多
        assert len(result_50) >= len(result_100)

    def test_single_char_text(self):
        """单字符文本"""
        assert split_text("好", max_chars=10) == ["好"]

    def test_text_with_only_punctuation(self):
        """纯标点文本"""
        text = "。。。。。。。。。。"  # 10个句号
        result = split_text(text, max_chars=5)

        assert len(result) >= 1
        for seg in result:
            assert len(seg) <= 5

    def test_mixed_content(self):
        """中英文混合内容"""
        text = "今天的天气是 sunny and warm。我们去了 park 玩。真的很开心！" * 5

        result = split_text(text, max_chars=80)

        for seg in result:
            assert len(seg) <= 80