处理字号（主入口） Args: text_blocks: 文字块列表 unify: 是否执行聚类统一 vertical_threshold_ratio: 垂直距离阈值比例 font_diff_threshold: 字号差异阈值 Returns: 处理后的文字块列表
(
self,
text_blocks: List[Dict[str, Any]],
unify: bool = True,
vertical_threshold_ratio: float = 0.5,
font_diff_threshold: float = 5.0
)
| 15 | self.text_offset = text_offset |
| 16 | |
| 17 | def process( |
| 18 | self, |
| 19 | text_blocks: List[Dict[str, Any]], |
| 20 | unify: bool = True, |
| 21 | vertical_threshold_ratio: float = 0.5, |
| 22 | font_diff_threshold: float = 5.0 |
| 23 | ) -> List[Dict[str, Any]]: |
| 24 | """ |
| 25 | 处理字号（主入口） |
| 26 | |
| 27 | Args: |
| 28 | text_blocks: 文字块列表 |
| 29 | unify: 是否执行聚类统一 |
| 30 | vertical_threshold_ratio: 垂直距离阈值比例 |
| 31 | font_diff_threshold: 字号差异阈值 |
| 32 | |
| 33 | Returns: |
| 34 | 处理后的文字块列表 |
| 35 | """ |
| 36 | # 步骤 1: 计算初始字号 |
| 37 | blocks = self.calculate_font_sizes(text_blocks) |
| 38 | |
| 39 | # 步骤 2: 聚类统一 |
| 40 | if unify and len(blocks) > 1: |
| 41 | blocks = self.unify_by_clustering( |
| 42 | blocks, |
| 43 | vertical_threshold_ratio, |
| 44 | font_diff_threshold |
| 45 | ) |
| 46 | |
| 47 | return blocks |
| 48 | |
| 49 | def calculate_font_sizes(self, text_blocks: List[Dict[str, Any]]) -> List[Dict[str, Any]]: |
| 50 | """Set font_size from block height (text: height - offset; formula: height * ratio).""" |
no test coverage detected