session_id cache

@@ -2,6 +2,7 @@ import requests
 from zeep import Client
 from settings import conf
 from . import consts
+from common.redis_cache import redis_handler as rh
 
 
 class EDMS:
@@ -15,12 +16,25 @@ class EDMS:
         self.pwd = pwd
         self.session_id = None
 
-    def get_session_id(self):
+    def set_session_id(self):
         self.session_id = self.sm_client.service.StartSession(login=self.user_name,
                                                               password=self.pwd,
                                                               clientType=consts.CUSTOM_CLIENT)
+        rh.set_session_id(self.session_id)
         return self.session_id
 
+    def get_session_id(self):
+        if self.session_id is None:
+            # redis获取session_id
+            self.session_id = rh.get_session_id()
+            if self.session_id is None:
+                return self.set_session_id()
+        # 验证session_id
+        is_session_valid = self.sm_client.service.IsSessionValid(sessionId=self.session_id, keepAlive=False)
+        if is_session_valid is True:
+            return self.session_id
+        return self.set_session_id()
+
     def get_download_token(self, headers, metadata_version_id):
         with self.dm_client.settings(extra_http_headers=headers):
             res = self.dm_client.service.PrepareSingleDocumentToDownload(metadataVersionId=metadata_version_id,

@@ -12,12 +12,11 @@ from zeep import Client
 
 from django.core.management import BaseCommand
 from common.mixins import LoggerMixin
-from common.redis_cache import redis_handler as rh
 from common.tools.file_tools import write_zip_file
 from apps.doc.models import DocStatus, HILDoc, AFCDoc
 from apps.doc import consts
 from settings import conf
-from apps.doc.edms import EDMS
+from apps.doc.edms import EDMS, rh
 
 
 class Command(BaseCommand, LoggerMixin):
@@ -57,7 +56,7 @@ class Command(BaseCommand, LoggerMixin):
         doc_id = int(doc_id_str)
         doc_class = HILDoc if business_type == consts.HIL_PREFIX else AFCDoc
         doc_info = doc_class.objects.filter(id=doc_id, status=DocStatus.INIT.value).values(
-            'id', 'metadata_version_id', 'document_name').first()
+            'id', 'metadata_version_id', 'application_id', 'document_name').first()
         if doc_info is None:
             self.cronjob_log.warn('{0} [get_doc_info] [doc completed] [task_str={1}] [is_priority={2}]'.format(
                 self.log_base, task_str, is_priority))
@@ -125,7 +124,7 @@ class Command(BaseCommand, LoggerMixin):
         s = item[1]  # xref of its /SMask
         is_rgb = True if item[5] == 'DeviceRGB' else False
 
-        # RGB
+        # GRAY/RGB  # TODO 颜色空间不同处理
         if is_rgb:
             if s == 0:
                 return doc.extractImage(x)
@@ -143,7 +142,7 @@ class Command(BaseCommand, LoggerMixin):
             pix1 = pix2 = None  # free temp pixmaps
             return self.getimage(pix)
 
-        # GRAY/CMYK
+        # CMYK
         pix1 = fitz.Pixmap(doc, x)
         pix = fitz.Pixmap(pix1)  # copy of pix1, alpha channel added
 
@@ -159,7 +158,7 @@ class Command(BaseCommand, LoggerMixin):
 
         pix1 = pix2 = None  # free temp pixmaps
 
-        pix = fitz.Pixmap(fitz.csRGB, pix)  # GRAY/CMYK to RGB
+        pix = fitz.Pixmap(fitz.csRGB, pix)  # CMYK to RGB
         return self.getimage(pix)
 
     @staticmethod

@@ -6,22 +6,22 @@ from . import consts
 class DocHandler:
 
     @staticmethod
-    def get_link(doc_id, file='pdf'):
+    def get_link(doc_id, business_type, file='pdf'):
         if file == 'pdf':
-            return '/data/{0}/{0}.pdf'.format(doc_id)
+            return '/data/{1}/{0}/{0}.pdf'.format(doc_id, business_type)
         elif file == 'img':
-            return '/data/{0}/{0}_img.zip'.format(doc_id)
+            return '/data/{1}/{0}/{0}_img.zip'.format(doc_id, business_type)
         else:
-            return '/data/{0}/{0}.xls'.format(doc_id)
+            return '/data/{1}/{0}/{0}.xls'.format(doc_id, business_type)
 
-    def get_doc_list(self, doc_queryset):
+    def get_doc_list(self, doc_queryset, business_type):
         for doc_dict in doc_queryset:
             if doc_dict['status'] != DocStatus.COMPLETE.value:
                 continue
             doc_id = doc_dict.get('id')
-            doc_dict['pdf_link'] = self.get_link(doc_id)
-            doc_dict['img_link'] = self.get_link(doc_id, 'img')
-            doc_dict['excel_link'] = self.get_link(doc_id, 'excel')
+            doc_dict['pdf_link'] = self.get_link(doc_id, business_type)
+            doc_dict['img_link'] = self.get_link(doc_id, business_type, file='img')
+            doc_dict['excel_link'] = self.get_link(doc_id, business_type, file='excel')
         return list(doc_queryset)
 
     @staticmethod

@@ -247,9 +247,9 @@ class DocView(GenericView, DocHandler):
             if create_time_start is not None and create_time_end is not None else Q()
         query = application_id_query & status_query & data_source_query & upload_finish_time_query & create_time_query
         val_tuple = ('id', 'application_id', 'upload_finish_time', 'create_time', 'data_source', 'status')
-        doc_class, _ = self.get_doc_class(business_type)
+        doc_class, prefix = self.get_doc_class(business_type)
         doc_queryset = doc_class.objects.filter(query).values(*val_tuple).order_by('-upload_finish_time')
-        doc_list = self.get_doc_list(doc_queryset)
+        doc_list = self.get_doc_list(doc_queryset, prefix)
 
         total = len(doc_list)
         start_index = page_size * (page - 1)

@@ -35,6 +35,7 @@ class RedisHandler:
         self.prefix = 'bwm_ocr'
         self.common_queue_key = '{0}:common_queue'.format(self.prefix)
         self.priority_queue_key = '{0}:priority_queue'.format(self.prefix)
+        self.session_id_key = '{0}:session_id'.format(self.prefix)
 
     def enqueue(self, tasks, is_priority=False):
         # 1
@@ -50,3 +51,9 @@ class RedisHandler:
             is_priority = False
         return task, is_priority
 
+    def set_session_id(self, session_id):
+        return self.redis.set(self.session_id_key, session_id)
+
+    def get_session_id(self):
+        return self.redis.get(self.session_id_key)
+