fix file remove

周伟奇
Showing 1 changed file with 26 additions and 6 deletions
src/apps/doc/management/commands/ocr_process.py
--- a/src/apps/doc/management/commands/ocr_process.py
View file @4076848
+++ b/src/apps/doc/management/commands/ocr_process.py
View file @4076848
 import os
 import time
 import json
+import shutil
 import base64
 import signal
 import asyncio
@@ -591,21 +592,40 @@ class Command(BaseCommand, LoggerMixin):
                    doc, business_type = self.get_doc_object(task_str)
                    doc_data_path = os.path.join(self.data_dir, business_type, consts.TMP_DIR_NAME, str(doc.id))
                    excel_path = os.path.join(doc_data_path, '{0}.xlsx'.format(doc.id))
-                    img_save_path = os.path.join(doc_data_path, 'img')
+                    src_excel_path = os.path.join(doc_data_path, 'src.xlsx')
-                    # wb.save(src_excel_path)
+                    wb.save(src_excel_path)
                    count_list = wb.rebuild(merged_bs_summary, license_summary, res_list, doc.document_scheme)
                    wb.save(excel_path)
                except Exception as e:
                    with lock:
                        if task_str in res_dict:
                            del res_dict[task_str]
-                    doc, _ = self.get_doc_object(task_str)
+                    doc, business_type = self.get_doc_object(task_str)
                    doc.status = DocStatus.PROCESS_FAILED.value
                    doc.save()
                    self.cronjob_log.error('{0} [process failed (res to wb)] [task={1}] [error={2}]'.format(
                        self.log_base, task_str, traceback.format_exc()))
+                    try:
+                        doc_data_path = os.path.join(self.data_dir, business_type, consts.TMP_DIR_NAME, str(doc.id))
+                        img_save_path = os.path.join(doc_data_path, 'img')
+                        shutil.rmtree(img_save_path, ignore_errors=True)
+                        pdf_path = os.path.join(doc_data_path, '{0}.pdf'.format(doc.id))
+                        os.remove(pdf_path)
+                    except Exception as e:
+                        self.cronjob_log.error('{0} [file remove failed] [task={1}] [error={2}]'.format(
+                            self.log_base, task_str, traceback.format_exc()))
                else:
                    try:
+                        img_save_path = os.path.join(doc_data_path, 'img')
+                        write_zip_file(img_save_path, os.path.join(doc_data_path, '{0}_img.zip'.format(doc.id)))
+                        shutil.rmtree(img_save_path, ignore_errors=True)
+                        # pdf_path = os.path.join(doc_data_path, '{0}.pdf'.format(doc.id))
+                        # os.remove(pdf_path)
+                        # os.remove(src_excel_path)
+                    except Exception as e:
+                        self.cronjob_log.error('{0} [file remove failed] [task={1}] [error={2}]'.format(
+                            self.log_base, task_str, traceback.format_exc()))
+                    try:
                        # 5.上传至EDMS
                        for times in range(consts.RETRY_TIMES):
                            try:
@@ -629,8 +649,6 @@ class Command(BaseCommand, LoggerMixin):
                        doc.save()
                        self.cronjob_log.error('{0} [process failed (edms upload)] [task={1}] [error={2}]'.format(
                            self.log_base, task_str, traceback.format_exc()))
-                        write_zip_file(img_save_path, os.path.join(doc_data_path, '{0}_img.zip'.format(doc.id)))
                    else:
                        doc.status = DocStatus.COMPLETE.value
                        doc.end_time = timezone.now()
@@ -640,7 +658,9 @@ class Command(BaseCommand, LoggerMixin):
                                setattr(doc, field, count)
                        doc.save()
                        self.cronjob_log.info('{0} [process complete] [task={1}]'.format(self.log_base, task_str))
-                        write_zip_file(img_save_path, os.path.join(doc_data_path, '{0}_img.zip'.format(doc.id)))
+                        # os.remove(excel_path)
    # TODO 细化文件状态，不同异常状态，归还队列，重试时采取不同的处理
    # TODO 异常邮件通知