revision/convert.py

import json
import xml.etree.ElementTree as ET
from xml.dom import minidom
import re
from datetime import datetime

def parse_knowledge_graph_txt(txt_file):
    """
    解析知识图谱txt文件，提取节点和边信息
    
    参数:
    - txt_file: txt文件路径
    
    返回:
    - nodes: 节点列表 [(id, label, degree, importance)]
    - edges: 边列表 [(source, target, relation, weight)]
    """
    nodes = []
    edges = []
    
    with open(txt_file, 'r', encoding='utf-8') as f:
        content = f.read()
    
    # 解析节点信息
    node_section_match = re.search(r'【节点信息】.*?格式: 节点名称.*?\n-{40}\n(.*?)\n={80}', content, re.DOTALL)
    if node_section_match:
        node_lines = node_section_match.group(1).strip().split('\n')
        for line in node_lines:
            if line.strip():
                # 解析格式: "  1. 节点名称 | 度数: X | 重要性: 级别"
                match = re.match(r'\s*(\d+)\.\s+(.*?)\s+\|\s+度数:\s+(\d+)\s+\|\s+重要性:\s+(\S+)', line)
                if match:
                    node_id = int(match.group(1)) - 1  # 从0开始的ID
                    label = match.group(2).strip()
                    degree = int(match.group(3))
                    importance = match.group(4)
                    nodes.append((node_id, label, degree, importance))
    
    # 解析边关系信息
    edge_section_match = re.search(r'【边关系信息】.*?格式: 节点1.*?\n-{60}\n(.*?)\n={80}', content, re.DOTALL)
    if edge_section_match:
        edge_lines = edge_section_match.group(1).strip().split('\n')
        for line in edge_lines:
            if line.strip():
                # 解析格式: "  1. 节点1 <--[关系]--> 节点2 | 重要性分数: X"
                match = re.match(r'\s*(\d+)\.\s+(.*?)\s+<--\[(.*?)\]-->\s+(.*?)\s+\|\s+重要性分数:\s+(\d+)', line)
                if match:
                    edge_id = int(match.group(1)) - 1
                    source_label = match.group(2).strip()
                    relation = match.group(3).strip()
                    target_label = match.group(4).strip()
                    weight = int(match.group(5))
                    
                    # 找到对应的节点ID
                    source_id = None
                    target_id = None
                    for node_id, label, _, _ in nodes:
                        if label == source_label:
                            source_id = node_id
                        if label == target_label:
                            target_id = node_id
                    
                    if source_id is not None and target_id is not None:
                        edges.append((source_id, target_id, relation, weight))
    
    return nodes, edges

def create_gexf_from_txt(txt_file, output_file='knowledge_graph.gexf', 
                        graph_title="Knowledge Graph", creator="Knowledge Graph Generator"):
    """
    将txt格式的知识图谱转换为GEXF格式
    
    参数:
    - txt_file: 输入的txt文件路径
    - output_file: 输出的gexf文件路径
    - graph_title: 图的标题
    - creator: 创建者信息
    """
    
    # 解析txt文件
    nodes, edges = parse_knowledge_graph_txt(txt_file)
    
    if not nodes:
        print("警告：未找到节点数据")
        return
    
    # 创建GEXF根元素
    gexf = ET.Element('gexf')
    gexf.set('xmlns', 'http://www.gexf.net/1.2draft')
    gexf.set('xmlns:viz', 'http://www.gexf.net/1.2draft/viz')
    gexf.set('version', '1.2')
    
    # 添加meta信息
    meta = ET.SubElement(gexf, 'meta')
    meta.set('lastmodifieddate', datetime.now().strftime('%Y-%m-%d'))
    
    creator_elem = ET.SubElement(meta, 'creator')
    creator_elem.text = creator
    
    description = ET.SubElement(meta, 'description')
    description.text = f"{graph_title} - Generated from knowledge graph data"
    
    # 创建graph元素
    graph = ET.SubElement(gexf, 'graph')
    graph.set('defaultedgetype', 'undirected')
    graph.set('mode', 'static')
    
    # 定义节点属性
    attributes = ET.SubElement(graph, 'attributes')
    attributes.set('class', 'node')
    
    degree_attr = ET.SubElement(attributes, 'attribute')
    degree_attr.set('id', 'degree')
    degree_attr.set('title', 'Degree')
    degree_attr.set('type', 'integer')
    
    importance_attr = ET.SubElement(attributes, 'attribute')
    importance_attr.set('id', 'importance')
    importance_attr.set('title', 'Importance')
    importance_attr.set('type', 'string')
    
    # 定义边属性 - 同时添加label和relationship
    edge_attributes = ET.SubElement(graph, 'attributes')
    edge_attributes.set('class', 'edge')
    
    label_attr = ET.SubElement(edge_attributes, 'attribute')
    label_attr.set('id', 'label')
    label_attr.set('title', 'Label')
    label_attr.set('type', 'string')
    
    relationship_attr = ET.SubElement(edge_attributes, 'attribute')
    relationship_attr.set('id', 'relationship')
    relationship_attr.set('title', 'Relationship')
    relationship_attr.set('type', 'string')
    
    # 添加节点
    nodes_elem = ET.SubElement(graph, 'nodes')
    nodes_elem.set('count', str(len(nodes)))
    
    # 计算节点的可视化属性
    max_degree = max(degree for _, _, degree, _ in nodes) if nodes else 1
    min_degree = min(degree for _, _, degree, _ in nodes) if nodes else 1
    
    for node_id, label, degree, importance in nodes:
        node = ET.SubElement(nodes_elem, 'node')
        node.set('id', str(node_id))
        node.set('label', label)
        
        # 添加节点属性值
        attvalues = ET.SubElement(node, 'attvalues')
        
        degree_val = ET.SubElement(attvalues, 'attvalue')
        degree_val.set('for', 'degree')
        degree_val.set('value', str(degree))
        
        importance_val = ET.SubElement(attvalues, 'attvalue')
        importance_val.set('for', 'importance')
        importance_val.set('value', importance)
        
        # 添加可视化属性
        viz = ET.SubElement(node, 'viz:size')
        # 根据度数计算节点大小（5-30之间）
        if max_degree > min_degree:
            size = 5 + 25 * (degree - min_degree) / (max_degree - min_degree)
        else:
            size = 15
        viz.set('value', f'{size:.1f}')
        
        # 根据重要性设置颜色
        color = ET.SubElement(node, 'viz:color')
        if importance == '高':
            color.set('r', '46')    # 深绿色
            color.set('g', '139')
            color.set('b', '87')
        elif importance == '中':
            color.set('r', '144')   # 浅绿色
            color.set('g', '238')
            color.set('b', '144')
        else:  # 低
            color.set('r', '224')   # 很浅的绿色
            color.set('g', '255')
            color.set('b', '224')
        color.set('a', '0.8')
    
    # 添加边
    edges_elem = ET.SubElement(graph, 'edges')
    edges_elem.set('count', str(len(edges)))
    
    for edge_id, (source_id, target_id, relation, weight) in enumerate(edges):
        edge = ET.SubElement(edges_elem, 'edge')
        edge.set('id', str(edge_id))
        edge.set('source', str(source_id))
        edge.set('target', str(target_id))
        edge.set('weight', str(weight))
        edge.set('label', relation)  # 直接设置边的label属性
        
        # 添加边属性值 - 同时设置label和relationship
        attvalues = ET.SubElement(edge, 'attvalues')
        
        label_val = ET.SubElement(attvalues, 'attvalue')
        label_val.set('for', 'label')
        label_val.set('value', relation)
        
        relationship_val = ET.SubElement(attvalues, 'attvalue')
        relationship_val.set('for', 'relationship')
        relationship_val.set('value', relation)  # 与label值完全一致
        
        # 添加边的可视化属性
        edge_color = ET.SubElement(edge, 'viz:color')
        edge_color.set('r', '128')
        edge_color.set('g', '128')
        edge_color.set('b', '128')
        edge_color.set('a', '0.5')
        
        # 根据权重设置边的粗细
        max_weight = max(w for _, _, _, w in edges) if edges else 1
        min_weight = min(w for _, _, _, w in edges) if edges else 1
        if max_weight > min_weight:
            thickness = 1 + 4 * (weight - min_weight) / (max_weight - min_weight)
        else:
            thickness = 2
        
        edge_thickness = ET.SubElement(edge, 'viz:thickness')
        edge_thickness.set('value', f'{thickness:.1f}')
    
    # 格式化XML并保存
    rough_string = ET.tostring(gexf, encoding='unicode')
    reparsed = minidom.parseString(rough_string)
    pretty_xml = reparsed.toprettyxml(indent="  ", encoding=None)
    
    # 移除空行
    pretty_lines = [line for line in pretty_xml.split('\n') if line.strip()]
    pretty_xml = '\n'.join(pretty_lines)
    
    with open(output_file, 'w', encoding='utf-8') as f:
        f.write(pretty_xml)
    
    print(f"GEXF文件已生成: {output_file}")
    print(f"节点数: {len(nodes)}, 边数: {len(edges)}")
    print("边属性: 同时包含label和relationship属性，值完全一致")
    
    return output_file

def create_enhanced_gexf_from_data(nodes_data, edges_data, output_file='enhanced_knowledge_graph.gexf',
                                  graph_title="Enhanced Knowledge Graph"):
    """
    直接从节点和边数据创建增强的GEXF文件
    
    参数:
    - nodes_data: 节点数据列表 [(label, degree, importance)]
    - edges_data: 边数据列表 [(source_label, target_label, relation, weight)]
    - output_file: 输出文件路径
    - graph_title: 图标题
    """
    
    # 创建节点ID映射
    label_to_id = {label: idx for idx, (label, _, _) in enumerate(nodes_data)}
    
    # 创建GEXF根元素
    gexf = ET.Element('gexf')
    gexf.set('xmlns', 'http://www.gexf.net/1.2draft')
    gexf.set('xmlns:viz', 'http://www.gexf.net/1.2draft/viz')
    gexf.set('version', '1.2')
    
    # 添加meta信息
    meta = ET.SubElement(gexf, 'meta')
    meta.set('lastmodifieddate', datetime.now().strftime('%Y-%m-%d+%H:%M'))
    
    creator_elem = ET.SubElement(meta, 'creator')
    creator_elem.text = "Knowledge Graph GEXF Generator v1.0"
    
    description = ET.SubElement(meta, 'description')
    description.text = f"{graph_title} - 地质知识图谱可视化数据"
    
    keywords = ET.SubElement(meta, 'keywords')
    keywords.text = "知识图谱,地质学,三元组,网络分析"
    
    # 创建graph元素
    graph = ET.SubElement(gexf, 'graph')
    graph.set('defaultedgetype', 'undirected')
    graph.set('mode', 'static')
    graph.set('timeformat', 'date')
    
    # 定义节点属性
    attributes = ET.SubElement(graph, 'attributes')
    attributes.set('class', 'node')
    
    degree_attr = ET.SubElement(attributes, 'attribute')
    degree_attr.set('id', '0')
    degree_attr.set('title', 'degree')
    degree_attr.set('type', 'integer')
    
    importance_attr = ET.SubElement(attributes, 'attribute')
    importance_attr.set('id', '1')
    importance_attr.set('title', 'importance')
    importance_attr.set('type', 'string')
    
    category_attr = ET.SubElement(attributes, 'attribute')
    category_attr.set('id', '2')
    category_attr.set('title', 'category')
    category_attr.set('type', 'string')
    
    # 定义边属性 - 同时添加label和relationship
    edge_attributes = ET.SubElement(graph, 'attributes')
    edge_attributes.set('class', 'edge')
    
    label_attr = ET.SubElement(edge_attributes, 'attribute')
    label_attr.set('id', '0')
    label_attr.set('title', 'label')
    label_attr.set('type', 'string')
    
    relationship_attr = ET.SubElement(edge_attributes, 'attribute')
    relationship_attr.set('id', '1')
    relationship_attr.set('title', 'relationship')
    relationship_attr.set('type', 'string')
    
    relation_type_attr = ET.SubElement(edge_attributes, 'attribute')
    relation_type_attr.set('id', '2')
    relation_type_attr.set('title', 'relation_type')
    relation_type_attr.set('type', 'string')
    
    # 添加节点
    nodes_elem = ET.SubElement(graph, 'nodes')
    nodes_elem.set('count', str(len(nodes_data)))
    
    # 计算节点的可视化属性
    degrees = [degree for _, degree, _ in nodes_data]
    max_degree = max(degrees) if degrees else 1
    min_degree = min(degrees) if degrees else 1
    
    for idx, (label, degree, importance) in enumerate(nodes_data):
        node = ET.SubElement(nodes_elem, 'node')
        node.set('id', str(idx))
        node.set('label', label)
        
        # 添加节点属性值
        attvalues = ET.SubElement(node, 'attvalues')
        
        degree_val = ET.SubElement(attvalues, 'attvalue')
        degree_val.set('for', '0')
        degree_val.set('value', str(degree))
        
        importance_val = ET.SubElement(attvalues, 'attvalue')
        importance_val.set('for', '1')
        importance_val.set('value', importance)
        
        # 根据实体名称推断类别
        category = "地质实体"
        if any(keyword in label for keyword in ["组", "段", "层"]):
            category = "地层单位"
        elif any(keyword in label for keyword in ["岩", "石"]):
            category = "岩石类型"
        elif any(keyword in label for keyword in ["构造", "断层", "褶皱"]):
            category = "构造单元"
        elif any(keyword in label for keyword in ["矿物", "成分"]):
            category = "矿物组分"
        
        category_val = ET.SubElement(attvalues, 'attvalue')
        category_val.set('for', '2')
        category_val.set('value', category)
        
        # 添加可视化属性
        viz_size = ET.SubElement(node, 'viz:size')
        # 根据度数计算节点大小（10-50之间）
        if max_degree > min_degree:
            size = 10 + 40 * (degree - min_degree) / (max_degree - min_degree)
        else:
            size = 25
        viz_size.set('value', f'{size:.1f}')
        
        # 根据重要性和类别设置颜色
        color = ET.SubElement(node, 'viz:color')
        if importance == '高':
            if category == "地层单位":
                color.set('r', '34'); color.set('g', '139'); color.set('b', '34')   # 森林绿
            elif category == "岩石类型":
                color.set('r', '139'); color.set('g', '69'); color.set('b', '19')   # 棕色
            elif category == "构造单元":
                color.set('r', '220'); color.set('g', '20'); color.set('b', '60')   # 深红
            else:
                color.set('r', '46'); color.set('g', '139'); color.set('b', '87')   # 默认深绿
        elif importance == '中':
            if category == "地层单位":
                color.set('r', '124'); color.set('g', '252'); color.set('b', '0')   # 草绿
            elif category == "岩石类型":
                color.set('r', '210'); color.set('g', '180'); color.set('b', '140') # 棕褐色
            elif category == "构造单元":
                color.set('r', '255'); color.set('g', '105'); color.set('b', '180') # 粉红
            else:
                color.set('r', '144'); color.set('g', '238'); color.set('b', '144') # 默认浅绿
        else:  # 低
            if category == "地层单位":
                color.set('r', '240'); color.set('g', '255'); color.set('b', '240') # 很浅绿
            elif category == "岩石类型":
                color.set('r', '245'); color.set('g', '222'); color.set('b', '179') # 小麦色
            elif category == "构造单元":
                color.set('r', '255'); color.set('g', '228'); color.set('b', '225') # 很浅粉
            else:
                color.set('r', '224'); color.set('g', '255'); color.set('b', '224') # 默认很浅绿
        color.set('a', '0.9')
        
        # 添加节点位置（可选，用于初始布局）
        position = ET.SubElement(node, 'viz:position')
        # 简单的圆形布局
        import math
        angle = 2 * math.pi * idx / len(nodes_data)
        radius = 100 + degree * 10
        x = radius * math.cos(angle)
        y = radius * math.sin(angle)
        position.set('x', f'{x:.2f}')
        position.set('y', f'{y:.2f}')
        position.set('z', '0.0')
    
    # 添加边
    edges_elem = ET.SubElement(graph, 'edges')
    edges_elem.set('count', str(len(edges_data)))
    
    # 计算边权重范围
    weights = [weight for _, _, _, weight in edges_data]
    max_weight = max(weights) if weights else 1
    min_weight = min(weights) if weights else 1
    
    for edge_id, (source_label, target_label, relation, weight) in enumerate(edges_data):
        if source_label in label_to_id and target_label in label_to_id:
            edge = ET.SubElement(edges_elem, 'edge')
            edge.set('id', str(edge_id))
            edge.set('source', str(label_to_id[source_label]))
            edge.set('target', str(label_to_id[target_label]))
            edge.set('weight', str(weight))
            edge.set('label', relation)  # 直接设置边的label属性
            
            # 添加边属性值 - 同时设置label和relationship
            attvalues = ET.SubElement(edge, 'attvalues')
            
            label_val = ET.SubElement(attvalues, 'attvalue')
            label_val.set('for', '0')
            label_val.set('value', relation)
            
            relationship_val = ET.SubElement(attvalues, 'attvalue')
            relationship_val.set('for', '1')
            relationship_val.set('value', relation)  # 与label值完全一致
            
            # 根据关系推断关系类型
            relation_type = "其他"
            if relation in ["位于", "分布于", "出露于"]:
                relation_type = "空间关系"
            elif relation in ["含有", "包含", "夹有"]:
                relation_type = "组成关系"
            elif relation in ["厚度", "长度", "宽度"]:
                relation_type = "数量关系"
            elif relation in ["整合", "不整合", "断层接触"]:
                relation_type = "接触关系"
            elif relation in ["发育", "具有", "呈现"]:
                relation_type = "特征关系"
            elif relation in ["属于", "归属"]:
                relation_type = "分类关系"
            
            relation_type_val = ET.SubElement(attvalues, 'attvalue')
            relation_type_val.set('for', '2')
            relation_type_val.set('value', relation_type)
            
            # 添加边的可视化属性
            edge_color = ET.SubElement(edge, 'viz:color')
            # 根据关系类型设置颜色
            if relation_type == "空间关系":
                edge_color.set('r', '0'); edge_color.set('g', '0'); edge_color.set('b', '255')     # 蓝色
            elif relation_type == "组成关系":
                edge_color.set('r', '255'); edge_color.set('g', '165'); edge_color.set('b', '0')   # 橙色
            elif relation_type == "数量关系":
                edge_color.set('r', '128'); edge_color.set('g', '0'); edge_color.set('b', '128')   # 紫色
            elif relation_type == "接触关系":
                edge_color.set('r', '255'); edge_color.set('g', '0'); edge_color.set('b', '0')     # 红色
            elif relation_type == "特征关系":
                edge_color.set('r', '0'); edge_color.set('g', '128'); edge_color.set('b', '0')     # 绿色
            elif relation_type == "分类关系":
                edge_color.set('r', '255'); edge_color.set('g', '20'); edge_color.set('b', '147')  # 深粉色
            else:
                edge_color.set('r', '128'); edge_color.set('g', '128'); edge_color.set('b', '128') # 灰色
            edge_color.set('a', '0.7')
            
            # 根据权重设置边的粗细
            if max_weight > min_weight:
                thickness = 1 + 5 * (weight - min_weight) / (max_weight - min_weight)
            else:
                thickness = 2
            
            edge_thickness = ET.SubElement(edge, 'viz:thickness')
            edge_thickness.set('value', f'{thickness:.1f}')
    
    # 格式化XML并保存
    rough_string = ET.tostring(gexf, encoding='unicode')
    reparsed = minidom.parseString(rough_string)
    pretty_xml = reparsed.toprettyxml(indent="  ", encoding=None)
    
    # 移除空行并清理格式
    pretty_lines = [line for line in pretty_xml.split('\n') if line.strip()]
    pretty_xml = '\n'.join(pretty_lines)
    
    with open(output_file, 'w', encoding='utf-8') as f:
        f.write(pretty_xml)
    
    print(f"增强的GEXF文件已生成: {output_file}")
    print(f"节点数: {len(nodes_data)}, 边数: {len(edges_data)}")
    print("可视化特性:")
    print("- 节点大小: 基于度数")
    print("- 节点颜色: 基于重要性和类别")
    print("- 边颜色: 基于关系类型")
    print("- 边粗细: 基于权重")
    print("- 边标签: 同时包含label和relationship属性，值完全一致")
    print("- Gephi可直接识别label属性，relationship属性可用于分类分析")
    
    return output_file

def fix_existing_gexf(input_file, output_file=None):
    """
    修复现有的GEXF文件，将relation属性改为label属性，并添加relationship属性
    
    参数:
    - input_file: 输入的GEXF文件路径
    - output_file: 输出的GEXF文件路径，如果为None则覆盖原文件
    """
    if output_file is None:
        output_file = input_file
    
    try:
        # 读取并解析GEXF文件
        tree = ET.parse(input_file)
        root = tree.getroot()
        
        # 查找并修改边属性定义
        for attributes in root.findall('.//{http://www.gexf.net/1.2draft}attributes[@class="edge"]'):
            # 查找现有的relation属性，改为label
            for attr in attributes.findall('.//{http://www.gexf.net/1.2draft}attribute[@id="relation"]'):
                attr.set('id', 'label')
                attr.set('title', 'Label')
                print("已修改边属性定义: relation -> label")
            
            # 添加relationship属性（如果不存在）
            existing_relationship = attributes.find('.//{http://www.gexf.net/1.2draft}attribute[@id="relationship"]')
            if existing_relationship is None:
                relationship_attr = ET.SubElement(attributes, 'attribute')
                relationship_attr.set('id', 'relationship')
                relationship_attr.set('title', 'Relationship')
                relationship_attr.set('type', 'string')
                print("已添加边属性定义: relationship")
        
        # 查找并修改边属性值
        modified_count = 0
        for attvalue in root.findall('.//{http://www.gexf.net/1.2draft}attvalue[@for="relation"]'):
            relation_value = attvalue.get('value')
            attvalue.set('for', 'label')
            modified_count += 1
            
            # 在同一个edge下添加relationship属性值
            edge_attvalues = attvalue.getparent()
            if edge_attvalues is not None:
                # 检查是否已存在relationship属性值
                existing_relationship_val = edge_attvalues.find('.//{http://www.gexf.net/1.2draft}attvalue[@for="relationship"]')
                if existing_relationship_val is None:
                    relationship_val = ET.SubElement(edge_attvalues, 'attvalue')
                    relationship_val.set('for', 'relationship')
                    relationship_val.set('value', relation_value)
        
        # 为所有边添加label属性（如果还没有的话）
        for edge in root.findall('.//{http://www.gexf.net/1.2draft}edge'):
            # 查找边的关系值
            for attvalue in edge.findall('.//{http://www.gexf.net/1.2draft}attvalue[@for="label"]'):
                relation_value = attvalue.get('value')
                if relation_value:
                    edge.set('label', relation_value)
                    break
        
        # 保存修改后的文件
        rough_string = ET.tostring(root, encoding='unicode')
        reparsed = minidom.parseString(rough_string)
        pretty_xml = reparsed.toprettyxml(indent="  ", encoding=None)
        
        # 移除空行
        pretty_lines = [line for line in pretty_xml.split('\n') if line.strip()]
        pretty_xml = '\n'.join(pretty_lines)
        
        with open(output_file, 'w', encoding='utf-8') as f:
            f.write(pretty_xml)
        
        print(f"GEXF文件修复完成: {output_file}")
        print(f"修改了 {modified_count} 个边属性值")
        print("已同时添加label和relationship属性，值完全一致")
        
    except Exception as e:
        print(f"修复GEXF文件时出错: {e}")

# 使用示例和测试函数
def test_conversion():
    """测试转换功能"""
    # 测试从txt文件转换
    txt_file = 'knowledge_graph_export.txt'
    gexf_file = 'knowledge_graph.gexf'
    
    try:
        create_gexf_from_txt(txt_file, gexf_file)
        print(f"转换成功: {txt_file} -> {gexf_file}")
    except Exception as e:
        print(f"转换失败: {e}")

if __name__ == "__main__":
    test_conversion()