OCR MCP Service

"""Technical analysis generator for OCR results.""" from typing import Optional, Dict, Any, List from .models import OCRResult class AnalysisGenerator: """Generate technical analysis description for OCR results.""" def __init__(self): """Initialize the analysis generator.""" pass def generate_analysis( self, ocr_result: OCRResult, layout_info: Optional[Dict[str, Any]] = None ) -> str: """ Generate technical analysis description. Args: ocr_result: OCR recognition result layout_info: Optional layout analysis information Returns: Technical analysis description in Chinese """ lines = [] # Basic metrics basic_info = self._analyze_basic_metrics(ocr_result) lines.extend(basic_info) # Text statistics text_stats = self._analyze_text_statistics(ocr_result) if text_stats: lines.append("") lines.extend(text_stats) # Layout information (if available) if layout_info: layout_info_lines = self._analyze_layout_info(layout_info) if layout_info_lines: lines.append("") lines.extend(layout_info_lines) # Quality assessment quality_info = self._generate_quality_assessment(ocr_result) if quality_info: lines.append("") lines.extend(quality_info) return "\n".join(lines) def _analyze_basic_metrics(self, ocr_result: OCRResult) -> List[str]: """Analyze basic metrics.""" lines = [] lines.append("【技术解析】") lines.append(f"识别引擎: {ocr_result.engine}") lines.append(f"处理时间: {ocr_result.processing_time:.2f}秒") # Confidence rating confidence_rating = self._get_confidence_rating(ocr_result.confidence) lines.append( f"识别置信度: {ocr_result.confidence:.2f} ({confidence_rating['level']})" ) lines.append(f"文本框数量: {len(ocr_result.boxes)}个") return lines def _analyze_text_statistics(self, ocr_result: OCRResult) -> List[str]: """Analyze text statistics.""" lines = [] lines.append("文本统计:") # Character count (excluding whitespace) char_count = len(ocr_result.text.replace(" ", "").replace("\n", "")) lines.append(f"- 总字符数: {char_count}") # Line count line_count = len([line for line in ocr_result.text.split("\n") if line.strip()]) if line_count > 0: lines.append(f"- 文本行数: {line_count}行") # Simple paragraph detection based on empty lines paragraphs_text = [ p.strip() for p in ocr_result.text.split("\n\n") if p.strip() ] if len(paragraphs_text) > 1: lines.append(f"- 段落数量: {len(paragraphs_text)}段（基于空行分析）") elif len(paragraphs_text) == 1 and line_count > 1: lines.append("- 段落数量: 1段") return lines def _analyze_layout_info(self, layout_info: Dict[str, Any]) -> List[str]: """Analyze layout information.""" lines = [] lines.append("版式特征:") # Paragraph count from layout analysis if "paragraph_count" in layout_info: lines.append(f"- 段落数量: {layout_info['paragraph_count']}段（基于位置分析）") # Underline count if "underline_count" in layout_info: underline_count = layout_info["underline_count"] if underline_count > 0: lines.append(f"- 下划线: {underline_count}处") else: lines.append("- 下划线: 无") # Alignment if "alignment" in layout_info: alignment = layout_info["alignment"] alignment_map = { "left": "左对齐", "right": "右对齐", "center": "居中对齐", "justify": "两端对齐", } alignment_text = alignment_map.get(alignment, alignment) lines.append(f"- 主要对齐方式: {alignment_text}") return lines def _generate_quality_assessment(self, ocr_result: OCRResult) -> List[str]: """Generate quality assessment.""" lines = [] lines.append("质量评估:") # Confidence rating rating = self._get_confidence_rating(ocr_result.confidence) lines.append( f"- 置信度评级: {rating['level']} ({rating['threshold']})" ) # Quality description quality_desc = self._get_quality_description(ocr_result) lines.append(f"- 识别质量: {quality_desc}") # Suggestions suggestion = self._get_suggestion(ocr_result) if suggestion: lines.append(f"- 建议: {suggestion}") return lines def _get_confidence_rating(self, confidence: float) -> Dict[str, str]: """Get confidence rating.""" if confidence >= 0.9: return {"level": "优秀", "threshold": "≥0.9"} elif confidence >= 0.7: return {"level": "良好", "threshold": "0.7-0.9"} else: return {"level": "一般", "threshold": "<0.7"} def _get_quality_description(self, ocr_result: OCRResult) -> str: """Get quality description.""" if ocr_result.confidence >= 0.9: return "高，文本清晰可读" elif ocr_result.confidence >= 0.7: return "中等，建议检查关键内容" else: return "较低，建议人工校对" def _get_suggestion(self, ocr_result: OCRResult) -> str: """Get suggestion based on OCR result.""" if ocr_result.confidence >= 0.9: return "识别结果可靠，可直接使用" elif ocr_result.confidence >= 0.7: return "识别结果基本可靠，建议检查重要内容" else: return "识别质量较低，建议人工校对或重新识别"