Method process_text

utils/advanced_ocr.py:92–132

将原始文本标准化并生成多视图便于匹配

(self, raw_text: str)

        return self._engine is not None or self._engine_paddle is not None or self._engine_tess is not None

    def process_text(self, raw_text: str) -> ProcessedText:
        """将原始文本标准化并生成多视图便于匹配"""
        if not raw_text or not raw_text.strip():
            return ProcessedText(original='', cleaned='', no_spaces='', words=[], chars=[])

        # 正规化：全半角、大小写、常见混淆字符
        def to_half_width(s: str) -> str:
            res = []
            for ch in s:
                code = ord(ch)
                if code == 0x3000:
                    code = 32
                elif 0xFF01 <= code <= 0xFF5E:
                    code -= 0xFEE0
                res.append(chr(code))
            return ''.join(res)

        def normalize_confusions(s: str) -> str:
            mapping = {
                'Ｉ': 'I', 'Ｌ': 'L', 'Ｏ': 'O', 'Ｓ': 'S', 'Ｂ': 'B',
                '０': '0', '１': '1', '２': '2', '５': '5', '６': '6', '８': '8', '９': '9',
                # 常见 OCR 易混：
                'O': '0', 'o': '0', 'l': '1', 'I': '1', '丨': '1', '｜': '1',
                'Z': '2', 'S': '5', 'B': '8',
            }
            return ''.join(mapping.get(c, c) for c in s)

        # 1. 清理文本：保留中文、字母、数字、空格
        raw_text_norm = normalize_confusions(to_half_width(raw_text)).casefold()
        cleaned = re.sub(r'[^\u4e00-\u9fff\w\s]', ' ', raw_text_norm)
        cleaned = re.sub(r'\s+', ' ', cleaned).strip()
        
        # 2. 无空格版本：用于连续匹配
        no_spaces = re.sub(r'\s+', '', cleaned)
        
        # 3. 提取词语：按空格分割
        words = [w.strip() for w in cleaned.split() if w.strip()]
        
        # 4. 提取字符
        chars = list(no_spaces)
        return ProcessedText(original=raw_text, cleaned=cleaned, no_spaces=no_spaces, words=words, chars=chars)

    def smart_text_contains(self, processed_text: ProcessedText, keyword: str) -> bool:
        """