/** * 设定集多格式导入导出工具 * 支持 JSON / TXT / Markdown / DOCX / PDF 格式的互相转换 */ // ==================== 字段映射表 ==================== // 中文标签 → 字段 key 的映射（用于导入时自动匹配） const FIELD_ALIASES = { character: { _nameAliases: ['姓名', '名字', '名称', '角色名', '人名'], role: ['角色', '角色类型', '身份', '定位'], gender: ['性别'], age: ['年龄', '岁数'], appearance: ['外貌', '外表', '外观', '容貌', '长相', '相貌', '外貌描写'], personality: ['性格', '个性', '人格', '性格特征', '性格特点'], background: ['背景', '身世', '经历', '来历', '背景故事', '过往'], motivation: ['动机', '目标', '追求', '愿望', '目的', '理想'], skills: ['能力', '技能', '特长', '武功', '法术', '本领', '擅长'], speechStyle: ['说话风格', '口头禅', '语气', '话风', '说话方式', '对白风格'], relationships: ['关系', '人际关系', '社交', '人物关系', '人际'], arc: ['成长', '发展', '弧线', '成长弧线', '角色发展', '变化'], notes: ['备注', '其他', '注释', '补充', '其他信息'], }, location: { _nameAliases: ['地点名', '地名', '场景名', '名称'], description: ['描述', '简介', '介绍', '说明', '概述'], slugline: ['场景标题', '标题行'], sensoryVisual: ['视觉', '视觉描写', '画面', '光线'], sensoryAudio: ['听觉', '听觉描写', '声音', '音效'], sensorySmell: ['嗅觉', '触觉', '嗅觉描写', '气味', '温度'], mood: ['氛围', '情绪', '基调', '氛围基调'], dangerLevel: ['危险等级', '危险', '安全等级'], notes: ['备注', '其他', '注释'], }, object: { _nameAliases: ['物品名', '道具名', '名称'], description: ['描述', '简介', '介绍', '说明'], objectType: ['类型', '分类', '物品类型', '道具分类'], rank: ['品阶', '等级', '品级', '稀有度'], currentHolder: ['持有者', '拥有者', '持有人', '归属'], numericStats: ['数值', '属性', '数值属性', '面板'], symbolism: ['象征', '象征意义', '寓意', '隐喻'], notes: ['备注', '其他', '注释'], }, world: { _nameAliases: ['设定名', '名称'], description: ['描述', '简介', '介绍', '内容', '说明', '详情'], notes: ['备注', '其他', '注释'], }, plot: { _nameAliases: ['标题', '章节名', '名称'], status: ['状态', '进度'], description: ['描述', '简介', '剧情', '内容', '说明'], notes: ['备注', '其他', '注释'], }, rules: { _nameAliases: ['规则名', '名称'], description: ['描述', '内容', '规则', '说明', '详情'], notes: ['备注', '其他'], }, }; // 字段 key → 中文显示名（用于导出） const FIELD_LABELS = { character: { role: '角色', gender: '性别', age: '年龄', appearance: '外貌', personality: '性格', background: '背景故事', motivation: '动机', skills: '能力', speechStyle: '说话风格', relationships: '人物关系', arc: '成长弧线', notes: '备注', }, location: { description: '描述', slugline: '场景标题', sensoryVisual: '视觉描写', sensoryAudio: '听觉描写', sensorySmell: '嗅觉/触觉', mood: '氛围基调', dangerLevel: '危险等级', notes: '备注', }, object: { description: '描述', objectType: '物品类型', rank: '品阶', currentHolder: '持有者', numericStats: '数值属性', symbolism: '象征意义', notes: '备注', }, world: { description: '描述', notes: '备注' }, plot: { status: '状态', description: '描述', notes: '备注' }, rules: { description: '描述', notes: '备注' }, }; // 分类中文名 const CATEGORY_NAMES = { character: '人物设定', location: '空间/地点', object: '物品/道具', world: '世界观/设定', plot: '大纲', rules: '写作规则', }; // 分类检测关键词 const CATEGORY_KEYWORDS = { character: ['性别', '年龄', '性格', '外貌', '人物', '角色', '长相', '口头禅', '身世', '人际关系'], location: ['地点', '地理', '场景', '地图', '位置', '区域', '氛围', '视觉描写'], object: ['物品', '道具', '装备', '武器', '法宝', '品阶', '持有者'], world: ['世界', '设定', '体系', '历史', '势力', '社会', '文化', '世界观'], plot: ['大纲', '剧情', '章节', '伏笔', '结局', '冲突', '支线'], rules: ['规则', '禁忌', '风格', '禁止', '必须'], }; // 中文分类名 → category key 映射（用于解析节头） const CATEGORY_NAME_MAP = { '人物设定': 'character', '人物': 'character', '空间/地点': 'location', '地点': 'location', '场景': 'location', '世界观/设定': 'world', '世界观': 'world', '世界': 'world', '物品/道具': 'object', '物品': 'object', '道具': 'object', '大纲': 'plot', '剧情': 'plot', '写作规则': 'rules', '规则': 'rules', }; // ==================== 分类检测 ==================== export function detectCategory(text) { const scores = {}; for (const [cat, keywords] of Object.entries(CATEGORY_KEYWORDS)) { scores[cat] = 0; for (const kw of keywords) { if (text.includes(kw)) scores[cat]++; } } let best = 'character'; let bestScore = 0; for (const [cat, score] of Object.entries(scores)) { if (score > bestScore) { best = cat; bestScore = score; } } return best; } // ==================== 结构化文本解析（导入） ==================== /** * 检测文本是否为结构化格式（带有分类节头或【】条目标记 + 字段） */ export function isStructuredText(text) { // TXT 导出格式: │ 分类名 或 ┌─── if (/^[│┌└]/m.test(text)) return true; // 有多个 【name】 单独一行 const entryHeaders = text.match(/^【.+?】\s*$/gm); if (entryHeaders && entryHeaders.length >= 2) return true; // Markdown 格式: ## 分类 或 ### 条目 const mdHeaders = text.match(/^#{2,3}\s+.+$/gm); if (mdHeaders && mdHeaders.length >= 2) return true; return false; } /** * 预处理 PDF 提取的纯文本 — 恢复分类和条目结构 * 优先检测 ■ / ◆ 标记（我们的 PDF 导出格式）， * 否则用启发式规则（行级分析） */ export function preprocessPdfText(text) { const lines = text.split('\n'); // 判断是否为字段行（短标签 + 冒号 + 内容） const isFieldLine = (s) => { if (!s) return false; return /^[^\s：:【】。！？■◆]{1,6}[：:]\s*/.test(s.trim()); }; const SENTENCE_ENDINGS = /[。！？）】」、；\.\!\?\)\]]$/; const categoryIndices = new Set(); const entryIndices = new Set(); // 检测是否有 ■ / ◆ 标记 const hasMarkers = lines.some(l => /^[■◆]\s/.test(l.trim())); if (hasMarkers) { // ===== 标记模式：直接使用 ■/◆ ===== for (let i = 0; i < lines.length; i++) { const trimmed = lines[i].trim(); if (/^■\s/.test(trimmed)) { categoryIndices.add(i); } else if (/^◆\s/.test(trimmed)) { entryIndices.add(i); } } } else { // ===== 启发式模式：用于非本应用导出的 PDF ===== for (let i = 0; i < lines.length; i++) { const trimmed = lines[i].trim(); if (!trimmed) continue; if (/—\s*设定集\s*$/.test(trimmed)) continue; if (SENTENCE_ENDINGS.test(trimmed)) continue; if (isFieldLine(trimmed)) continue; if (trimmed.length > 50) continue; const nextTrimmed = (lines[i + 1] || '').trim(); if (isFieldLine(nextTrimmed)) { // 下一行是字段 → 本行是条目名 entryIndices.add(i); } } // 检测分类：非字段、非条目的短行，后面的非空行是一个条目 for (let i = 0; i < lines.length; i++) { const trimmed = lines[i].trim(); if (!trimmed) continue; if (entryIndices.has(i)) continue; if (isFieldLine(trimmed)) continue; if (SENTENCE_ENDINGS.test(trimmed)) continue; if (trimmed.length > 30) continue; if (/—\s*设定集\s*$/.test(trimmed)) continue; // 向前找下一个非空行，看是否是条目 for (let j = i + 1; j < lines.length; j++) { const nextT = lines[j].trim(); if (!nextT) continue; if (entryIndices.has(j)) { categoryIndices.add(i); } break; } } } console.log('[preprocessPdfText] mode:', hasMarkers ? 'markers' : 'heuristic'); console.log('[preprocessPdfText] categories:', [...categoryIndices].map(i => `L${i}: "${lines[i].trim()}"`)); console.log('[preprocessPdfText] entries:', [...entryIndices].map(i => `L${i}: "${lines[i].trim()}"`)); // 生成结果：去掉 ■/◆ 前缀，加上 ##/### 标记 const result = []; for (let i = 0; i < lines.length; i++) { if (categoryIndices.has(i)) { result.push(`## ${lines[i].trim().replace(/^■\s*/, '')}`); } else if (entryIndices.has(i)) { result.push(`### ${lines[i].trim().replace(/^◆\s*/, '')}`); } else { result.push(lines[i]); } } return result.join('\n'); } /** * 结构化文本解析 — 正确处理分类节头 + 【条目】 + 字段：值 格式 * 支持 TXT 导出格式和 Markdown 导出格式 * @returns {Array<{name: string, category: string, fields: Object}>} */ export function parseStructuredText(text) { const entries = []; let currentCategory = 'character'; let currentEntryName = null; let currentFields = {}; let currentFieldLabel = null; let currentFieldLines = []; const flushField = () => { if (currentFieldLabel) { const val = currentFieldLines.join('\n').trim(); if (val) currentFields[currentFieldLabel] = val; } currentFieldLabel = null; currentFieldLines = []; }; const flushEntry = () => { flushField(); if (currentEntryName && Object.keys(currentFields).length > 0) { entries.push({ name: currentEntryName, category: currentCategory, fields: { ...currentFields }, }); } currentEntryName = null; currentFields = {}; }; const lines = text.split('\n'); for (let i = 0; i < lines.length; i++) { const line = lines[i]; const trimmed = line.trim(); // 跳过装饰线（═══ ┌── └── ───等） if (/^[═─┌└┐┘]+\s*$/.test(trimmed)) continue; // 跳过空行 if (!trimmed) continue; // 跳过顶部标题 (xxx — 设定集) if (/—\s*设定集\s*$/.test(trimmed)) continue; // === TXT 分类节头: │ 人物设定 === const catBoxMatch = trimmed.match(/^│\s*(.+)$/); if (catBoxMatch) { flushEntry(); const catName = catBoxMatch[1].trim(); currentCategory = CATEGORY_NAME_MAP[catName] || detectCategory(catName); continue; } // === Markdown 二级标题作为分类: ## 人物设定 === const mdH2 = trimmed.match(/^##\s+(.+)$/); if (mdH2 && !trimmed.startsWith('###')) { const catName = mdH2[1].replace(/\s*—\s*设定集.*$/, '').trim(); flushEntry(); currentCategory = CATEGORY_NAME_MAP[catName] || detectCategory(catName); continue; } // === 条目标记: 【name】 单独一行 === const entryBracket = trimmed.match(/^【(.+?)】\s*$/); if (entryBracket) { flushEntry(); currentEntryName = entryBracket[1].trim(); continue; } // === Markdown 三级标题作为条目: ### name === const entryMd = trimmed.match(/^###\s+(.+)$/); if (entryMd) { flushEntry(); currentEntryName = entryMd[1].trim(); continue; } // 只有在有条目名的情况下才解析字段 if (!currentEntryName) continue; // === Markdown 加粗标签: **label**：value === const mdBold = trimmed.match(/^\*\*(.+?)\*\*\s*[：:]\s*(.*)/); if (mdBold) { flushField(); currentFieldLabel = mdBold[1].trim(); if (mdBold[2].trim()) currentFieldLines.push(mdBold[2].trim()); continue; } // === 冒号格式: label：value === const colonMatch = trimmed.match(/^([^\s：:\[\]【】]{1,15})[：:]\s*(.*)/); if (colonMatch) { flushField(); currentFieldLabel = colonMatch[1].trim(); if (colonMatch[2].trim()) currentFieldLines.push(colonMatch[2].trim()); continue; } // === 普通内容行（追加到当前字段） === if (currentFieldLabel) { currentFieldLines.push(line); } } flushEntry(); return entries; } /** * 简单文本解析（用于没有明确结构的纯文本，单个条目） */ export function parseTextToFields(text) { const fields = {}; const lines = text.split('\n'); let currentLabel = null; let currentLines = []; const flush = () => { if (currentLabel) { const val = currentLines.join('\n').trim(); if (val) fields[currentLabel] = val; } currentLabel = null; currentLines = []; }; for (const line of lines) { const trimmed = line.trim(); if (!trimmed) continue; const mdMatch = trimmed.match(/^#{1,3}\s+(.+)$/); if (mdMatch) { flush(); currentLabel = mdMatch[1].trim(); continue; } const bracketMatch = trimmed.match(/^【(.+?)】\s*(.*)/); if (bracketMatch) { flush(); currentLabel = bracketMatch[1].trim(); if (bracketMatch[2].trim()) currentLines.push(bracketMatch[2].trim()); continue; } const colonMatch = trimmed.match(/^([^\s：:]{1,10})[：:]\s*(.*)/); if (colonMatch) { flush(); currentLabel = colonMatch[1].trim(); if (colonMatch[2].trim()) currentLines.push(colonMatch[2].trim()); continue; } currentLines.push(line); } flush(); return fields; } /** * 将解析出的 {中文标签: 内容} 映射到 content 字段 * @returns {{ name, content, category }} */ export function mapFieldsToContent(parsedFields, category) { const aliases = FIELD_ALIASES[category] || {}; const content = {}; let name = ''; for (const [label, value] of Object.entries(parsedFields)) { let matched = false; if (aliases._nameAliases?.some(a => label.includes(a))) { name = value; matched = true; continue; } for (const [fieldKey, fieldAliases] of Object.entries(aliases)) { if (fieldKey === '_nameAliases') continue; if (fieldAliases.some(a => label.includes(a) || a.includes(label))) { content[fieldKey] = value; matched = true; break; } } if (!matched) { content[label] = value; } } return { name, content, category }; } /** * 解析多个条目（用于无结构的纯文本，用 --- 或连续空行分隔） */ export function parseMultipleEntries(text) { const blocks = text.split(/(?:\n\s*---+\s*\n|\n{3,})/).filter(b => b.trim()); if (blocks.length <= 1) return [text]; return blocks; } // ==================== 文本导出 ==================== /** * 将节点数组导出为 TXT 文本 */ export function exportNodesToTxt(nodes) { const workNode = nodes.find(n => n.type === 'work'); let output = ''; if (workNode) { output += `═══════════════════════════\n`; output += ` ${workNode.name} — 设定集\n`; output += `═══════════════════════════\n\n`; } const categories = ['character', 'location', 'world', 'object', 'plot', 'rules']; for (const cat of categories) { const items = nodes.filter(n => n.type === 'item' && n.category === cat); if (items.length === 0) continue; const catName = CATEGORY_NAMES[cat] || cat; output += `┌──────────────────────────\n`; output += `│ ${catName}\n`; output += `└──────────────────────────\n\n`; const labels = FIELD_LABELS[cat] || {}; for (const item of items) { output += `【${item.name}】\n`; const content = item.content || {}; for (const [key, label] of Object.entries(labels)) { if (content[key]) { output += `${label}：${content[key]}\n`; } } const knownKeys = new Set(Object.keys(labels)); for (const [key, val] of Object.entries(content)) { if (!knownKeys.has(key) && val) { output += `${key}：${val}\n`; } } output += '\n'; } } return output.trimEnd() + '\n'; } /** * 将节点数组导出为 Markdown */ export function exportNodesToMarkdown(nodes) { const workNode = nodes.find(n => n.type === 'work'); let output = ''; if (workNode) { output += `# ${workNode.name} — 设定集\n\n`; } const categories = ['character', 'location', 'world', 'object', 'plot', 'rules']; for (const cat of categories) { const items = nodes.filter(n => n.type === 'item' && n.category === cat); if (items.length === 0) continue; const catName = CATEGORY_NAMES[cat] || cat; output += `## ${catName}\n\n`; const labels = FIELD_LABELS[cat] || {}; for (const item of items) { output += `### ${item.name}\n\n`; const content = item.content || {}; for (const [key, label] of Object.entries(labels)) { if (content[key]) { output += `**${label}**：${content[key]}\n\n`; } } const knownKeys = new Set(Object.keys(labels)); for (const [key, val] of Object.entries(content)) { if (!knownKeys.has(key) && val) { output += `**${key}**：${val}\n\n`; } } } } return output.trimEnd() + '\n'; } // ==================== DOCX 导出 ==================== /** * 将节点数组导出为 DOCX 文件（返回 Blob） */ export async function exportNodesToDocx(nodes) { const docx = await import('docx'); const { Document, Paragraph, TextRun, HeadingLevel, Packer, AlignmentType } = docx; const workNode = nodes.find(n => n.type === 'work'); const children = []; children.push(new Paragraph({ text: `${workNode?.name || '设定集'} — 设定集`, heading: HeadingLevel.TITLE, spacing: { after: 300 }, })); const categories = ['character', 'location', 'world', 'object', 'plot', 'rules']; for (const cat of categories) { const items = nodes.filter(n => n.type === 'item' && n.category === cat); if (items.length === 0) continue; const catName = CATEGORY_NAMES[cat] || cat; children.push(new Paragraph({ text: catName, heading: HeadingLevel.HEADING_1, spacing: { before: 400, after: 200 }, })); const labels = FIELD_LABELS[cat] || {}; for (const item of items) { children.push(new Paragraph({ text: item.name, heading: HeadingLevel.HEADING_2, spacing: { before: 300, after: 150 }, })); const content = item.content || {}; for (const [key, label] of Object.entries(labels)) { if (content[key]) { children.push(new Paragraph({ children: [ new TextRun({ text: `${label}：`, bold: true, size: 24, font: '宋体' }), new TextRun({ text: content[key], size: 24, font: '宋体' }), ], spacing: { after: 100, line: 360 }, alignment: AlignmentType.LEFT, })); } } const knownKeys = new Set(Object.keys(labels)); for (const [key, val] of Object.entries(content)) { if (!knownKeys.has(key) && val) { children.push(new Paragraph({ children: [ new TextRun({ text: `${key}：`, bold: true, size: 24, font: '宋体' }), new TextRun({ text: val, size: 24, font: '宋体' }), ], spacing: { after: 100, line: 360 }, alignment: AlignmentType.LEFT, })); } } } } const doc = new Document({ styles: { default: { document: { run: { size: 24, font: '宋体' }, paragraph: { alignment: AlignmentType.LEFT, spacing: { line: 360 } }, }, }, }, sections: [{ children }], }); return await Packer.toBlob(doc); } // ==================== DOCX / PDF 导入 ==================== /** * 从 DOCX 文件中提取结构化文本（保留标题层级） * H1 → ## (分类) H2 → ### (条目名) → 保留标签格式 */ export async function parseDocxToText(file) { const mammoth = await import('mammoth'); const arrayBuf = await file.arrayBuffer(); const result = await mammoth.convertToHtml({ arrayBuffer: arrayBuf }); const html = result.value || ''; if (!html.trim()) throw new Error('文件内容为空'); const text = html // 标题转 Markdown 格式（保留结构） .replace(/]*>([\s\S]*?)<\/h1>/gi, (_, c) => `\n## ${c.replace(/<[^>]*>/g, '').trim()}\n`) .replace(/]*>([\s\S]*?)<\/h2>/gi, (_, c) => `\n### ${c.replace(/<[^>]*>/g, '').trim()}\n`) .replace(/]*>([\s\S]*?)<\/h[3-6]>/gi, (_, c) => `\n#### ${c.replace(/<[^>]*>/g, '').trim()}\n`) // 加粗文本保留为标签格式 .replace(/([\s\S]*?)<\/strong>/gi, (_, c) => c.replace(/<[^>]*>/g, '')) .replace(/([\s\S]*?)<\/b>/gi, (_, c) => c.replace(/<[^>]*>/g, '')) // 段落和换行 .replace(/<\/(?:p|li|div)>/gi, '\n') .replace(//gi, '\n') // 去除剩余 HTML 标签 .replace(/<[^>]*>/g, '') // HTML 实体 .replace(/ /g, ' ') .replace(/&/g, '&') .replace(/</g, '<') .replace(/>/g, '>') // 清理多余空行 .replace(/\n{3,}/g, '\n\n') .trim(); if (!text) throw new Error('文件内容为空'); return text; } /** * 从 PDF 文件中提取纯文本（通过服务端 API） */ export async function parsePdfToText(file) { const formData = new FormData(); formData.append('file', file); const response = await fetch('/api/parse-file', { method: 'POST', body: formData, }); if (!response.ok) { const data = await response.json().catch(() => ({})); throw new Error(data.error || `PDF 解析失败 (${response.status})`); } const data = await response.json(); if (!data.text?.trim()) { throw new Error(data.warning || 'PDF 内容为空，可能是扫描件或图片PDF'); } return data.text; } // ==================== PDF 导出 ==================== /** * 将节点数组导出为 PDF（通过浏览器打印） * 条目名前加 ◆ 标记、分类名前加 ■ 标记，便于导入时识别 */ export function exportSettingsAsPdf(nodes) { const workNode = nodes.find(n => n.type === 'work'); const title = `${workNode?.name || '设定集'} — 设定集`; let html = ''; const categories = ['character', 'location', 'world', 'object', 'plot', 'rules']; for (const cat of categories) { const items = nodes.filter(n => n.type === 'item' && n.category === cat); if (items.length === 0) continue; const catName = CATEGORY_NAMES[cat] || cat; html += `

■ ${catName}

`; const labels = FIELD_LABELS[cat] || {}; for (const item of items) { html += `

◆ ${item.name}

`; const content = item.content || {}; for (const [key, label] of Object.entries(labels)) { if (content[key]) { html += `

${label}：${content[key]}

`; } } const knownKeys = new Set(Object.keys(labels)); for (const [key, val] of Object.entries(content)) { if (!knownKeys.has(key) && val) { html += `

${key}：${val}

`; } } } } const fullHtml = `${title}

${title}

${html}`; const win = window.open('', '_blank'); if (win) { win.document.write(fullHtml); win.document.close(); setTimeout(() => win.print(), 500); } }