fix: restore server XLS roster conversion

2026-08-31 16:04:26 +08:00
1 parent 4ba2d43710
commit d7a821da16
@@ -21,6 +21,13 @@ const HTML_WIDTH_ATTRIBUTE_PATTERN = /(\bwidth\s*=\s*["']?)(\d+(?:\.\d+)?)(["']?
 const HTML_WIDTH_STYLE_PATTERN = /(\bwidth\s*:\s*)(\d+(?:\.\d+)?)(pt|px)/gi;

 export type DocumentConversionStatus = 'converted' | 'already_pdf' | 'converted_xlsx' | 'already_xlsx' | 'source_fallback';
+export type DocumentConversionErrorCode =
+  | 'converter_unavailable'
+  | 'converter_timeout'
+  | 'conversion_failed'
+  | 'conversion_output_missing'
+  | 'conversion_output_invalid'
+  | 'converted_file_too_large';

 export interface DocumentConversionInput {
  fileName: string;
@@ -37,7 +44,7 @@ export interface DocumentConversionResult {
  sha256: string;
  converted: boolean;
  conversion_status: DocumentConversionStatus;
-  conversion_error_code?: string;
+  conversion_error_code?: DocumentConversionErrorCode;
 }

 function sourceExtension(fileName: string, contentType: string): string {
@@ -79,7 +86,7 @@ function isXlsx(contentType: string, content: Buffer): boolean {

 function sourceResult(
  input: DocumentConversionInput,
-  errorCode?: string
+  errorCode?: DocumentConversionErrorCode
 ): DocumentConversionResult {
  return {
    fileName: normalizeTaskArtifactFileName(input.fileName) || 'team-file.bin',
@@ -93,7 +100,7 @@ function sourceResult(
  };
 }

-function conversionErrorCode(error: unknown): string {
+function conversionErrorCode(error: unknown): DocumentConversionErrorCode {
  const value = error && typeof error === 'object' ? error as {
    code?: unknown;
    killed?: unknown;
@@ -1,6 +1,7 @@
 import { extname } from 'node:path';
 import ExcelJS from 'exceljs';
 import { fromBufferPromise, type Entry, type ZipFile } from 'yauzl';
+import type { DocumentConversionErrorCode } from './document-converter.js';

 const XLSX_MAGIC = Buffer.from([0x50, 0x4b, 0x03, 0x04]);
 const XLS_MAGIC = Buffer.from([0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1]);
@@ -89,13 +90,18 @@ export class PassengerRosterWorkbookError extends Error {
  readonly code: PassengerRosterWorkbookErrorCode;
  readonly row?: number;
  readonly column?: number;
+  readonly conversionErrorCode?: DocumentConversionErrorCode;

-  constructor(code: PassengerRosterWorkbookErrorCode, location: { row?: number; column?: number } = {}) {
+  constructor(
+    code: PassengerRosterWorkbookErrorCode,
+    location: { row?: number; column?: number; conversionErrorCode?: DocumentConversionErrorCode } = {}
+  ) {
    super(ERROR_MESSAGES[code]);
    this.name = 'PassengerRosterWorkbookError';
    this.code = code;
    this.row = location.row;
    this.column = location.column;
+    this.conversionErrorCode = location.conversionErrorCode;
  }
 }

@@ -132,7 +138,7 @@ type ScalarCellValue = string | number | boolean | Date | null;

 function fail(
  code: PassengerRosterWorkbookErrorCode,
-  location: { row?: number; column?: number } = {}
+  location: { row?: number; column?: number; conversionErrorCode?: DocumentConversionErrorCode } = {}
 ): never {
  throw new PassengerRosterWorkbookError(code, location);
 }
@@ -553,7 +559,8 @@ export async function normalizePassengerRosterWorkbook(
    if (!options.convertLegacyXls) fail('roster_workbook_converter_unavailable');
    try {
      xlsxContent = await options.convertLegacyXls(input);
-    } catch {
+    } catch (error) {
+      if (error instanceof PassengerRosterWorkbookError) throw error;
      fail('roster_workbook_conversion_failed');
    }
    if (
@@ -2774,7 +2774,9 @@ export class TaskService {
          maxBytes: this.config.ARTIFACT_MAX_BYTES
        });
        if (!converted.converted || converted.conversion_status !== 'converted_xlsx') {
-          throw new Error(converted.conversion_error_code || 'legacy_xls_conversion_failed');
+          throw new PassengerRosterWorkbookError('roster_workbook_conversion_failed', {
+            conversionErrorCode: converted.conversion_error_code || 'conversion_failed'
+          });
        }
        return converted.content;
      }
@@ -2800,13 +2802,18 @@ export class TaskService {
        WHERE task_id = $1 AND purpose = 'passenger_list' AND sha256 = $2`,
      [target.rowId, digest]
    );
-    if (!prior.rowCount && target.status !== 'awaiting_attachment') {
+    // A rejection is not a permanent verdict on the same bytes: converter or
+    // runtime fixes can make a previously rejected workbook valid. A replay of
+    // the same AgentBus message is still stopped by its idempotency key below.
+    const shouldValidateAttachment = !prior.rowCount
+      || text((prior.rows[0] as Record<string, unknown> | undefined)?.status) === 'rejected';
+    if (shouldValidateAttachment && target.status !== 'awaiting_attachment') {
      throw new TaskError('invalid_transition', `任务当前状态为 ${target.status}，不能接收名单附件。`);
    }

    let normalized: NormalizedPassengerRosterWorkbook | null = null;
    let rejection: { code: string; details: Record<string, unknown> } | null = null;
-    if (!prior.rowCount) {
+    if (shouldValidateAttachment) {
      try {
        normalized = await this.normalizePassengerRosterAttachment(attachment);
      } catch (error) {
@@ -2815,7 +2822,10 @@ export class TaskService {
            code: error.code,
            details: {
              ...(error.row == null ? {} : { row: error.row }),
-              ...(error.column == null ? {} : { column: error.column })
+              ...(error.column == null ? {} : { column: error.column }),
+              ...(error.conversionErrorCode == null
+                ? {}
+                : { conversion_error_code: error.conversionErrorCode })
            }
          };
        } else {
@@ -2877,14 +2887,16 @@ export class TaskService {
      }

      const existingAttachment = await client.query(
-        `SELECT status, row_count, error_code
+        `SELECT id, status, row_count, error_code
           FROM task_input_attachments
          WHERE task_id = $1 AND purpose = 'passenger_list' AND sha256 = $2
          FOR UPDATE`,
        [target.rowId, digest]
      );
-      if (existingAttachment.rowCount) {
-        const previous = existingAttachment.rows[0] as Record<string, unknown>;
+      const previousAttachment = existingAttachment.rowCount
+        ? existingAttachment.rows[0] as Record<string, unknown>
+        : null;
+      if (previousAttachment && text(previousAttachment.status) !== 'rejected') {
        if (scopedIdempotencyKey) {
          await client.query(
            `INSERT INTO idempotency_keys (organization_id, scope, idempotency_key, request_hash, task_id)
@@ -2903,17 +2915,13 @@ export class TaskService {
        return {
          row: taskRow,
          event: null,
-          inputAttachment: text(previous.status) === 'rejected'
-            ? {
-                status: 'rejected' as const,
-                error_code: text(previous.error_code) || 'roster_workbook_rejected',
-                message: `该名单附件此前已校验失败【${text(previous.error_code) || 'roster_workbook_rejected'}】，请修正后重新发送。`
-              }
-            : {
-                status: 'duplicate' as const,
-                ...(Number(previous.row_count) > 0 ? { row_count: Number(previous.row_count) } : {}),
-                message: '该名单附件已经接收，无需重复发送。'
-              }
+          inputAttachment: {
+            status: 'duplicate' as const,
+            ...(Number(previousAttachment.row_count) > 0
+              ? { row_count: Number(previousAttachment.row_count) }
+              : {}),
+            message: '该名单附件已经接收，无需重复发送。'
+          }
        };
      }
      if (text(taskRow.status) !== 'awaiting_attachment') {
@@ -2930,26 +2938,48 @@ export class TaskService {
      let inputAttachment: NonNullable<TaskMessageResult['input_attachment']>;
      let event: TaskEvent;
      if (normalized) {
-        await client.query(
-          `INSERT INTO task_input_attachments
-            (organization_id, task_id, purpose, source, file_name_ciphertext,
-             content_type, byte_size, sha256, status, normalizer_version,
-             normalized_text_ciphertext, row_count)
-           VALUES ($1, $2, 'passenger_list', $3, $4, $5, $6, $7,
-                   'normalized', $8, $9, $10)`,
-          [
-            context.organizationId,
-            target.rowId,
-            attachment.source,
-            encryptedFileName,
-            attachment.contentType,
-            attachment.content.byteLength,
-            digest,
-            PASSENGER_ROSTER_WORKBOOK_VERSION,
-            encryptText(this.config, normalized.canonicalTsv),
-            normalized.rowCount
-          ]
-        );
+        const encryptedCanonicalTsv = encryptText(this.config, normalized.canonicalTsv);
+        if (previousAttachment) {
+          await client.query(
+            `UPDATE task_input_attachments
+                SET source = $2, file_name_ciphertext = $3, content_type = $4,
+                    byte_size = $5, status = 'normalized', normalizer_version = $6,
+                    normalized_text_ciphertext = $7, row_count = $8,
+                    error_code = NULL, error_details = '{}'::jsonb, updated_at = now()
+              WHERE id = $1 AND status = 'rejected'`,
+            [
+              previousAttachment.id,
+              attachment.source,
+              encryptedFileName,
+              attachment.contentType,
+              attachment.content.byteLength,
+              PASSENGER_ROSTER_WORKBOOK_VERSION,
+              encryptedCanonicalTsv,
+              normalized.rowCount
+            ]
+          );
+        } else {
+          await client.query(
+            `INSERT INTO task_input_attachments
+              (organization_id, task_id, purpose, source, file_name_ciphertext,
+               content_type, byte_size, sha256, status, normalizer_version,
+               normalized_text_ciphertext, row_count)
+             VALUES ($1, $2, 'passenger_list', $3, $4, $5, $6, $7,
+                     'normalized', $8, $9, $10)`,
+            [
+              context.organizationId,
+              target.rowId,
+              attachment.source,
+              encryptedFileName,
+              attachment.contentType,
+              attachment.content.byteLength,
+              digest,
+              PASSENGER_ROSTER_WORKBOOK_VERSION,
+              encryptedCanonicalTsv,
+              normalized.rowCount
+            ]
+          );
+        }
        const updated = await client.query(
          `UPDATE tasks
              SET status = 'parse_queued', stage = 'parse',
@@ -2984,24 +3014,44 @@ export class TaskService {
        });
      } else {
        const safeRejection = rejection || { code: 'roster_workbook_processing_failed', details: {} };
-        await client.query(
-          `INSERT INTO task_input_attachments
-            (organization_id, task_id, purpose, source, file_name_ciphertext,
-             content_type, byte_size, sha256, status, error_code, error_details)
-           VALUES ($1, $2, 'passenger_list', $3, $4, $5, $6, $7,
-                   'rejected', $8, $9::jsonb)`,
-          [
-            context.organizationId,
-            target.rowId,
-            attachment.source,
-            encryptedFileName,
-            attachment.contentType,
-            attachment.content.byteLength,
-            digest,
-            safeRejection.code,
-            safeRejection.details
-          ]
-        );
+        if (previousAttachment) {
+          await client.query(
+            `UPDATE task_input_attachments
+                SET source = $2, file_name_ciphertext = $3, content_type = $4,
+                    byte_size = $5, status = 'rejected', normalizer_version = NULL,
+                    normalized_text_ciphertext = NULL, row_count = NULL,
+                    error_code = $6, error_details = $7::jsonb, updated_at = now()
+              WHERE id = $1 AND status = 'rejected'`,
+            [
+              previousAttachment.id,
+              attachment.source,
+              encryptedFileName,
+              attachment.contentType,
+              attachment.content.byteLength,
+              safeRejection.code,
+              safeRejection.details
+            ]
+          );
+        } else {
+          await client.query(
+            `INSERT INTO task_input_attachments
+              (organization_id, task_id, purpose, source, file_name_ciphertext,
+               content_type, byte_size, sha256, status, error_code, error_details)
+             VALUES ($1, $2, 'passenger_list', $3, $4, $5, $6, $7,
+                     'rejected', $8, $9::jsonb)`,
+            [
+              context.organizationId,
+              target.rowId,
+              attachment.source,
+              encryptedFileName,
+              attachment.contentType,
+              attachment.content.byteLength,
+              digest,
+              safeRejection.code,
+              safeRejection.details
+            ]
+          );
+        }
        const updated = await client.query(
          `UPDATE tasks
              SET message = $2, error = '', updated_at = now()