import { existsSync, mkdirSync, readdirSync, statSync, writeFileSync } from "node:fs"; import path from "node:path"; import { expect, test } from "../helpers/fixtures"; import { EMPLOYEE, login } from "../helpers/auth"; import { createSafeLiveInvoiceOcrEvidence, openInvoicePool, resolveLiveOcrUser, uploadLiveInvoiceAndWait, type SafeLiveInvoiceOcrEvidence, } from "../helpers/liveInvoiceOcr"; type BatchSample = { filePath: string; alias: string; }; const rootDir = path.resolve(process.cwd(), ".."); const reportDir = process.env.PLAYWRIGHT_OCR_BATCH_REPORT_DIR?.trim() || path.join(rootDir, "artifacts/qa-flow", `live-ocr-batch-${new Date().toISOString().replace(/[-:]/g, "").replace(/\.\d{3}Z$/, "Z")}`); const markdownReportPath = path.join(reportDir, "summary.md"); const jsonReportPath = path.join(reportDir, "summary.json"); function resolveInvoiceFiles() { const files = new Set(); const directFile = process.env.PLAYWRIGHT_OCR_LIVE_FILE?.trim(); const batchFiles = process.env.PLAYWRIGHT_OCR_LIVE_FILES?.split(/[\n,]/).map((item) => item.trim()).filter(Boolean) ?? []; const batchDir = process.env.PLAYWRIGHT_OCR_LIVE_DIR?.trim() ?? ""; if (directFile && /\.pdf$/i.test(directFile)) { files.add(path.resolve(directFile)); } for (const file of batchFiles) { if (!/\.pdf$/i.test(file)) continue; files.add(path.resolve(file)); } if (batchDir && existsSync(batchDir) && statSync(batchDir).isDirectory()) { for (const entry of readdirSync(batchDir).sort((left, right) => left.localeCompare(right, "zh-CN"))) { const fullPath = path.join(batchDir, entry); if (!statSync(fullPath).isFile()) continue; if (!/\.pdf$/i.test(entry)) continue; files.add(path.resolve(fullPath)); } } return [...files] .filter((filePath) => existsSync(filePath)) .sort((left, right) => left.localeCompare(right, "zh-CN")) .map((filePath, index): BatchSample => ({ filePath, alias: `SAMPLE-PDF-${String(index + 1).padStart(3, "0")}`, })); } function writeReport(results: SafeLiveInvoiceOcrEvidence[]) { mkdirSync(reportDir, { recursive: true }); const counts = { total: results.length, available: results.filter((item) => item.status === "available").length, ocrFailed: results.filter((item) => item.status === "ocr_failed").length, duplicateOrSameInvoice: results.filter((item) => item.reasonCategory === "duplicate_or_same_invoice").length, alreadyUsed: results.filter((item) => item.reasonCategory === "already_used").length, timeoutOrPending: results.filter((item) => item.reasonCategory === "timeout_or_pending").length, unknown: results.filter((item) => item.reasonCategory === "unknown").length, }; const fieldHitCounts = { invoiceId: results.filter((item) => item.fieldHits.invoiceId).length, invoiceCodeOrNumber: results.filter((item) => item.fieldHits.invoiceCodeOrNumber).length, sellerName: results.filter((item) => item.fieldHits.sellerName).length, totalAmount: results.filter((item) => item.fieldHits.totalAmount).length, invoiceDate: results.filter((item) => item.fieldHits.invoiceDate).length, }; const summary = { mode: "live-ocr-batch-redacted", checkedAt: new Date().toISOString(), counts, fieldHitCounts, privacy: "redacted aliases and boolean field hits only; no invoice number, seller, amount, full filename, full path, raw OCR payload, invoice image, token, or password", results, }; const lines = [ "# 批量 live OCR 脱敏辅助复核报告", "", `生成时间:${new Date().toISOString()}`, "", "## 结果", "", `- 总文件数:${counts.total}`, `- 可用:${counts.available}`, `- 识别失败:${counts.ocrFailed}`, `- 重复或相同发票:${counts.duplicateOrSameInvoice}`, `- 已用于报销/已使用:${counts.alreadyUsed}`, `- 超时或仍在处理中:${counts.timeoutOrPending}`, `- 未分类:${counts.unknown}`, "", "## 明细", "", "| 样本 | 状态 | 耗时(ms) | 原因分类 | 票据ID命中 | 发票号命中 | 销售方命中 | 金额命中 | 日期命中 |", "| --- | --- | ---: | --- | --- | --- | --- | --- | --- |", ...results.map((item) => ( `| ${item.alias} | ${item.status} | ${item.durationMs} | ${item.reasonCategory} | ${item.fieldHits.invoiceId ? "Y" : "N"} | ${item.fieldHits.invoiceCodeOrNumber ? "Y" : "N"} | ${item.fieldHits.sellerName ? "Y" : "N"} | ${item.fieldHits.totalAmount ? "Y" : "N"} | ${item.fieldHits.invoiceDate ? "Y" : "N"} |` )), "", "## 说明", "", "- 本报告用于真实腾讯 OCR 的受控辅助复核,不进入常规本地回归。", "- 本报告只记录样本别名、状态、耗时、字段命中布尔值和原因分类。", "- 本报告不得记录发票号码、销售方、金额、完整文件名、完整路径、原始 OCR payload 或票面原图。", "- 如某张票失败,会继续处理后续文件,并在本报告统一汇总。", ]; writeFileSync(markdownReportPath, `${lines.join("\n")}\n`, "utf-8"); writeFileSync(jsonReportPath, `${JSON.stringify(summary, null, 2)}\n`, "utf-8"); } test("手工 live 腾讯 OCR 批量辅助复核:真实票样逐张上传并输出报告", async ({ page }) => { test.skip(process.env.PLAYWRIGHT_MANUAL !== "1", "仅在显式手工 Playwright 模式下执行"); test.skip(process.env.PLAYWRIGHT_OCR_LIVE !== "1", "仅在显式 live OCR 辅助复核模式下执行"); const invoiceFiles = resolveInvoiceFiles(); test.skip(!invoiceFiles.length, "需要提供 PDF 格式的 PLAYWRIGHT_OCR_LIVE_FILE / PLAYWRIGHT_OCR_LIVE_FILES / PLAYWRIGHT_OCR_LIVE_DIR"); const results: SafeLiveInvoiceOcrEvidence[] = []; const liveOcrUser = resolveLiveOcrUser(EMPLOYEE); await login(page, liveOcrUser); try { await openInvoicePool(page); for (const sample of invoiceFiles) { await openInvoicePool(page); const result = await uploadLiveInvoiceAndWait(page, sample.filePath); results.push(createSafeLiveInvoiceOcrEvidence(result, sample.alias)); } } finally { writeReport(results); await page.context().clearCookies().catch(() => undefined); } const allowDuplicate = process.env.PLAYWRIGHT_OCR_ALLOW_DUPLICATE === "1"; const blockingFailures = results.filter((item) => { if (item.status !== "ocr_failed") return false; return !(allowDuplicate && ["duplicate_or_same_invoice", "already_used"].includes(item.reasonCategory)); }); expect(blockingFailures, JSON.stringify(blockingFailures, null, 2)).toHaveLength(0); });