update

output/output_article_distribution_0917.py

@@ -134,6 +134,7 @@ time.sleep(10)
 second_demands_tag_count = {}
 projects_demands_tag_count = {}
 total_tag_count = {}
+total_tag_count_pro = {}
 for redis_count,spark_res in enumerate(sql_res):
     # if redis_count >= 50:break
     second_demands = []
@@ -175,39 +176,43 @@ for redis_count,spark_res in enumerate(sql_res):
         bulk_dict[0][1] += 1
         if second_demands:
             for tag in second_demands:
-                if tag in second_demands_tag_count:
+                if tag in total_tag_count:
                     total_tag_count[tag] += 1
                 else:
                     total_tag_count[tag] = 1
         if projects:
             for tag in projects:
-                if tag in projects_demands_tag_count:
-                    total_tag_count[tag] += 1
+                if tag in total_tag_count_pro:
+                    total_tag_count_pro[tag] += 1
                 else:
-                    total_tag_count[tag] = 1
-    elif 10 <= total_tractate_content_num < 50:
-        bulk_dict[10][1] += 1
-    elif 50 <= total_tractate_content_num < 100:
-        bulk_dict[50][1] += 1
-    elif 100 <= total_tractate_content_num < 200:
-        bulk_dict[100][1] += 1
-    elif 200 <= total_tractate_content_num < 500:
-        bulk_dict[200][1] += 1
-    elif 500 <= total_tractate_content_num < 1000:
-        bulk_dict[500][1] += 1
-    else:
-        bulk_dict[1000][1] += 1
+                    total_tag_count_pro[tag] = 1
+        else:
+            continue
+    # elif 10 <= total_tractate_content_num < 50:
+    #     bulk_dict[10][1] += 1
+    # elif 50 <= total_tractate_content_num < 100:
+    #     bulk_dict[50][1] += 1
+    # elif 100 <= total_tractate_content_num < 200:
+    #     bulk_dict[100][1] += 1
+    # elif 200 <= total_tractate_content_num < 500:
+    #     bulk_dict[200][1] += 1
+    # elif 500 <= total_tractate_content_num < 1000:
+    #     bulk_dict[500][1] += 1
+    # else:
+    #     bulk_dict[1000][1] += 1
 
     if redis_count % 5000 == 0:
         print(redis_count,bulk_dict)
         # print(second_demands_tag_count,projects_demands_tag_count)
 print(total_tag_count)
+print(total_tag_count_pro)
 # print(bulk_dict)
 #
 # print("second_demands_tag_count",second_demands_tag_count)
 # print("projects_demands_tag_count",projects_demands_tag_count)
 with open("log.log","w",encoding='utf-8') as f:
     f.write(total_tag_count)
+    f.write(total_tag_count_pro)
     # f.write(second_demands_tag_count)
     # f.write(projects_demands_tag_count)