import { readFile, stat } from "node:fs/promises"; import path from "node:path"; import type { TaskStatus } from "./contracts.ts"; export type MdtodoReportSource = { href: string; sourcePath: string; body: string; modifiedAt: string }; export type MdtodoImportPayload = { sourcePath: string; sourceModifiedAt: string; markdown: string; reports: MdtodoReportSource[]; groupName?: string; dryRun?: boolean; }; export type MdtodoImportTask = { sourceId: string; parentSourceId: string | null; title: string; description: string; status: TaskStatus; startAt: string; dueAt: string; reports: Array<{ title: string; body: string; status: string; createdAt: string }>; }; export type MdtodoImportPlan = { sourcePath: string; groupName: string; groupDescription: string; parserMode: "rxx" | "legacy-checkbox"; tasks: MdtodoImportTask[]; summary: { taskCount: number; subtaskCount: number; subsubtaskCount: number; reportCount: number; missingReportCount: number; flattenedTaskCount: number; warningCount: number; }; warnings: Array<{ code: string; message: string; sourceId?: string; href?: string }>; }; const rxxHeadingPattern = /^(#{2,})\s+(R\d+(?:\.\d+)*)(?:\s+(.*?))?\s*$/iu; const checkboxPattern = /^(\s*)(?:[-*+]|\d+[.)])\s+\[([ xX-])\]\s+(.*)$/u; const markdownLinkPattern = /\[([^\]]*)\]\(([^)]+)\)/gu; export async function loadMdtodoImportPayload(filePath: string, options: { groupName?: string; dryRun?: boolean } = {}): Promise { const sourcePath = path.resolve(filePath); const [markdown, sourceInfo] = await Promise.all([readFile(sourcePath, "utf8"), stat(sourcePath)]); const reports: MdtodoReportSource[] = []; const seen = new Set(); for (const link of reportLinks(markdown)) { const reportPath = path.resolve(path.dirname(sourcePath), link.href); if (seen.has(reportPath)) continue; seen.add(reportPath); try { const [body, reportInfo] = await Promise.all([readFile(reportPath, "utf8"), stat(reportPath)]); reports.push({ href: link.href, sourcePath: reportPath, body, modifiedAt: reportInfo.mtime.toISOString() }); } catch { // Missing report files are reported by the pure parser so dry-run and import agree. } } return { sourcePath, sourceModifiedAt: sourceInfo.mtime.toISOString(), markdown, reports, groupName: options.groupName, dryRun: options.dryRun }; } export function parseMdtodoImport(payload: MdtodoImportPayload): MdtodoImportPlan { const lines = String(payload.markdown ?? "").split(/\r?\n/u); const rxxBlocks = collectRxxBlocks(lines); const parsed = rxxBlocks.length ? parseRxx(lines, rxxBlocks, payload) : parseCheckboxes(lines, payload); if (!parsed.tasks.length) throw codedError("mdtodo_tasks_not_found", "MDTODO file does not contain Rxx headings or checkbox tasks"); const reportCount = parsed.tasks.reduce((count, task) => count + task.reports.length, 0); const missingReportCount = parsed.warnings.filter((warning) => warning.code === "report_file_missing").length; const flattenedTaskCount = parsed.warnings.filter((warning) => warning.code === "nested_task_flattened").length; const tasksBySourceId = new Map(parsed.tasks.map((task) => [task.sourceId, task])); return { sourcePath: payload.sourcePath, groupName: cleanTitle(payload.groupName || documentTitle(lines, payload.sourcePath)), groupDescription: `Imported from MDTODO: ${payload.sourcePath}`, parserMode: rxxBlocks.length ? "rxx" : "legacy-checkbox", tasks: parsed.tasks, summary: { taskCount: parsed.tasks.filter((task) => !task.parentSourceId).length, subtaskCount: parsed.tasks.filter((task) => taskHierarchyDepth(task, tasksBySourceId) === 1).length, subsubtaskCount: parsed.tasks.filter((task) => taskHierarchyDepth(task, tasksBySourceId) === 2).length, reportCount, missingReportCount, flattenedTaskCount, warningCount: parsed.warnings.length }, warnings: parsed.warnings.slice(0, 100) }; } function parseRxx(lines: string[], blocks: RxxBlock[], payload: MdtodoImportPayload) { const tasks: MdtodoImportTask[] = []; const warnings: MdtodoImportPlan["warnings"] = []; const sourceIds = new Set(blocks.map((block) => block.sourceId)); for (const block of blocks) { const depth = rxxDepth(block.sourceId); const supportedSourceId = depth > 2 ? block.sourceId.split(".").slice(0, 3).join(".") : block.sourceId; const parentSourceId = nearestRxxParent(supportedSourceId, sourceIds); if (depth > 2) warnings.push({ code: "nested_task_flattened", sourceId: block.sourceId, message: `${block.sourceId} was flattened to the supported third level under ${parentSourceId ?? "the taskgroup"}` }); if (depth > 0 && !parentSourceId) warnings.push({ code: "parent_task_missing", sourceId: block.sourceId, message: `${block.sourceId} has no importable parent task and was imported as a task` }); const description = lines.slice(block.lineIndex + 1, block.endLine).join("\n").trim(); const reports = reportsForTask(description, payload, block.sourceId, warnings); tasks.push({ sourceId: block.sourceId, parentSourceId, title: cleanTitle(`${block.sourceId} ${block.title || conciseBodyTitle(description)}`), description, status: block.status, ...importedTaskRange(payload.sourceModifiedAt, reports), reports }); } expandParentRanges(tasks); return { tasks, warnings }; } function parseCheckboxes(lines: string[], payload: MdtodoImportPayload) { const tasks: MdtodoImportTask[] = []; const warnings: MdtodoImportPlan["warnings"] = []; const parentsByDepth: string[] = []; for (let index = 0; index < lines.length; index += 1) { const match = checkboxPattern.exec(lines[index]); if (!match) continue; const depth = Math.max(0, Math.floor(match[1].replace(/\t/gu, " ").length / 2)); const sourceId = `L${index + 1}`; const supportedDepth = Math.min(depth, 2); const parentSourceId = supportedDepth > 0 ? parentsByDepth[supportedDepth - 1] ?? null : null; if (depth > 2) warnings.push({ code: "nested_task_flattened", sourceId, message: `${sourceId} was flattened to the supported third checkbox level` }); tasks.push({ sourceId, parentSourceId, title: cleanTitle(match[3]), description: "", status: match[2].toLowerCase() === "x" ? "completed" : match[2] === "-" ? "blocked" : "pending", ...importedTaskRange(payload.sourceModifiedAt, []), reports: [] }); parentsByDepth[supportedDepth] = sourceId; parentsByDepth.length = supportedDepth + 1; } expandParentRanges(tasks); return { tasks, warnings }; } function reportsForTask(body: string, payload: MdtodoImportPayload, sourceId: string, warnings: MdtodoImportPlan["warnings"]) { const reports: MdtodoImportTask["reports"] = []; for (const link of reportLinks(body)) { const resolved = path.resolve(path.dirname(payload.sourcePath), link.href); const report = payload.reports.find((candidate) => path.resolve(candidate.sourcePath) === resolved); if (!report) { warnings.push({ code: "report_file_missing", sourceId, href: link.href, message: `Report file could not be read: ${link.href}` }); continue; } reports.push({ title: documentTitle(report.body.split(/\r?\n/u), report.sourcePath), body: report.body, status: "imported", createdAt: requiredIsoDate(report.modifiedAt, "report.modifiedAt") }); } return reports; } function importedTaskRange(sourceModifiedAt: string, reports: MdtodoImportTask["reports"]) { const sourceTime = new Date(requiredIsoDate(sourceModifiedAt, "sourceModifiedAt")).getTime(); const reportTimes = reports.map((report) => new Date(report.createdAt).getTime()); const dueTime = reportTimes.length ? Math.max(...reportTimes) : sourceTime; return { startAt: new Date(dueTime - 86400000).toISOString(), dueAt: new Date(dueTime).toISOString() }; } function expandParentRanges(tasks: MdtodoImportTask[]) { const tasksBySourceId = new Map(tasks.map((task) => [task.sourceId, task])); const deepestFirst = [...tasks].sort((left, right) => taskHierarchyDepth(right, tasksBySourceId) - taskHierarchyDepth(left, tasksBySourceId)); for (const child of deepestFirst) { if (!child.parentSourceId) continue; const parent = tasksBySourceId.get(child.parentSourceId); if (!parent) continue; const startAt = Math.min(new Date(parent.startAt).getTime(), new Date(child.startAt).getTime()); const dueAt = Math.max(new Date(parent.dueAt).getTime(), new Date(child.dueAt).getTime()); parent.startAt = new Date(startAt).toISOString(); parent.dueAt = new Date(dueAt).toISOString(); } } function nearestRxxParent(sourceId: string, sourceIds: Set) { const parts = sourceId.split("."); while (parts.length > 1) { parts.pop(); const candidate = parts.join("."); if (sourceIds.has(candidate)) return candidate; } return null; } function taskHierarchyDepth(task: MdtodoImportTask, tasksBySourceId: Map) { let depth = 0; let parentSourceId = task.parentSourceId; while (parentSourceId && depth < 3) { depth += 1; parentSourceId = tasksBySourceId.get(parentSourceId)?.parentSourceId ?? null; } return depth; } function reportLinks(markdown: string) { const links: Array<{ label: string; href: string }> = []; for (const match of String(markdown ?? "").matchAll(markdownLinkPattern)) { const label = String(match[1] ?? "").trim(); const href = String(match[2] ?? "").trim().replace(/^<|>$/gu, "").split(/\s+/u)[0] ?? ""; if (!href || !/\.md(?:#.*)?$/iu.test(href) || /^(?:[a-z][a-z0-9+.-]*:|\/|[A-Za-z]:\/)/iu.test(href)) continue; if (!/(?:任务报告|task[_ -]?report|report)/iu.test(`${label} ${href}`)) continue; links.push({ label, href: href.split("#")[0] }); } return links; } type RxxBlock = { sourceId: string; title: string; status: TaskStatus; lineIndex: number; endLine: number }; function collectRxxBlocks(lines: string[]): RxxBlock[] { const blocks: Omit[] = []; for (let index = 0; index < lines.length; index += 1) { const match = rxxHeadingPattern.exec(lines[index]); if (!match) continue; const rest = String(match[3] ?? "").trim(); blocks.push({ sourceId: match[2].replace(/^r/iu, "R"), title: stripStatus(rest), status: statusFrom(rest), lineIndex: index }); } return blocks.map((block, index) => ({ ...block, endLine: blocks[index + 1]?.lineIndex ?? lines.length })); } function statusFrom(value: string): TaskStatus { const markers = [...value.matchAll(/\[([^\]]+)\]/gu)].map((match) => String(match[1]).trim().toLowerCase().replace(/[\s-]+/gu, "_")); if (markers.some((marker) => marker === "blocked")) return "blocked"; if (markers.some((marker) => ["completed", "finished", "done"].includes(marker))) return "completed"; if (markers.some((marker) => ["in_progress", "processing"].includes(marker))) return "in_progress"; return "pending"; } function stripStatus(value: string) { return value.replace(/\s*\[(?:open|todo|blocked|in[_ -]?progress|processing|completed|finished|done)\]\s*/giu, " ").trim(); } function rxxDepth(sourceId: string) { return Math.max(0, sourceId.split(".").length - 1); } function documentTitle(lines: string[], sourcePath: string) { return cleanTitle(lines.find((line) => /^#{1,6}\s+\S/u.test(line))?.replace(/^#{1,6}\s+/u, "") || path.basename(sourcePath, path.extname(sourcePath))); } function firstBodyLine(body: string) { return body.split(/\r?\n/u).map((line) => line.trim()).find(Boolean) || "Untitled task"; } function conciseBodyTitle(body: string) { return firstBodyLine(body); } function cleanTitle(value: string) { return String(value ?? "").replace(/\s+\s*$/u, "").trim().slice(0, 500) || "Untitled task"; } function requiredIsoDate(value: unknown, field: string) { const date = new Date(String(value ?? "")); if (Number.isNaN(date.valueOf())) throw codedError("invalid_mdtodo_file_time", `${field} must be an ISO date`); return date.toISOString(); } function codedError(code: string, message: string) { return Object.assign(new Error(message), { code }); }