import { simpleParser } from 'mailparser' import { richEmailBody } from './rich-body.js' import type { AddressEntry, EmailAttachmentMeta, ReadMessageBody } from './types.js' export function flattenAddresses(input: unknown): AddressEntry[] { if (input === null || input === undefined) return [] const list = Array.isArray(input) ? input : (input as { value?: unknown }).value if (!Array.isArray(list)) return [] return list .filter(entry => entry !== null && typeof entry === 'object') .map(entry => { const { name, address } = entry as { name?: string; address?: string } const out: AddressEntry = {} if (typeof name === 'string' && name !== '') out.name = name if (typeof address === 'string' && address !== '') out.address = address return out }) .filter(entry => entry.address !== undefined || entry.name !== undefined) } /** Minimal, dependency-free HTML-to-text: block tags become newlines, tags are dropped, common entities decoded. */ export function stripHtml(html: string): string { let out = html .replace(/<(script|style|head|title)[\s\S]*?<\/\1\s*>/gi, ' ') .replace(/<\/(p|div|tr|li|h[1-6]|table|blockquote|ul|ol|section|article|header|footer)[^>]*>/gi, '\n') .replace(//gi, '\n') .replace(/<[^>]+>/g, ' ') .replace(/ /gi, ' ') .replace(/&/gi, '&') .replace(/</gi, '<') .replace(/>/gi, '>') .replace(/"/gi, '"') .replace(/'|'/gi, "'") out = out.replace(/[ \t]+/g, ' ').replace(/ *\n */g, '\n').replace(/\n{3,}/g, '\n\n') return out.trim() } export function truncateText(text: string, maxChars: number): { text: string; truncated: boolean } { if (text.length <= maxChars) return { text, truncated: false } const cut = text.slice(0, maxChars) const lastBreak = Math.max(cut.lastIndexOf('\n'), cut.lastIndexOf(' ')) // 窗口内没有空格/换行时必须硬切:否则 lastBreak 为 0,正文会被整段丢成空串。 const head = lastBreak > 0 ? cut.slice(0, lastBreak) : cut return { text: head + '\n\n…[正文过长,已截断,共 ' + text.length + ' 字符]', truncated: true } } /** * Turn an untrusted attachment filename into a safe basename: no directory * separators, no traversal, no control characters, bounded length. */ export function sanitizeFilename(raw: unknown, fallback = 'attachment.bin'): string { let name = String(raw ?? '').replace(/\\/g, '/').split('/').pop() ?? '' name = name.replace(/[\u0000-\u001f\u007f]/g, '').replace(/[<>:"|?*]/g, '_').trim().replace(/[. ]+$/g, '') if (name === '' || name === '.' || name === '..' || /^(?:con|prn|aux|nul|com[1-9]|lpt[1-9])(?:\.|$)/i.test(name)) name = fallback if (name.length > 120) { const dot = name.lastIndexOf('.') const ext = dot > 0 && dot >= name.length - 12 ? name.slice(dot) : '' name = name.slice(0, 120 - ext.length) + ext } return name } /** Parse a raw RFC822 message source into the read-result body. */ export async function parseRawMessage(source: Buffer, maxBodyChars: number): Promise { const parsed = await simpleParser(source, { skipImageLinks: true, keepCidLinks: true, skipHtmlToText: false }) let text = parsed.text ?? '' if (text.trim() === '' && typeof parsed.html === 'string' && parsed.html.trim() !== '') { text = stripHtml(parsed.html) } const limited = truncateText(text, maxBodyChars) const attachments: EmailAttachmentMeta[] = (parsed.attachments ?? []).map((att, index) => ({ filename: att.filename ?? '(unnamed)', contentType: typeof att.contentType === 'string' ? att.contentType : 'application/octet-stream', size: Number.isFinite(att.size) && att.size >= 0 ? att.size : 0, part: 'attachment-' + index, ...(att.contentId ? { contentId: att.contentId } : {}), })) const rich = richEmailBody(parsed.html, attachments) return { date: parsed.date instanceof Date && Number.isFinite(parsed.date.getTime()) ? parsed.date.toISOString() : '', from: flattenAddresses(parsed.from), to: flattenAddresses(parsed.to), cc: flattenAddresses(parsed.cc), subject: parsed.subject ?? '', text: limited.text, attachments, ...(rich ? { rich } : {}), truncated: limited.truncated, } }