完整配置文件

config/config.yaml

@@ -26,7 +26,7 @@ log:
   console: true
 
 sources:
-  - path: /data/y4/tools/data/user-0s.csv
+  - path: /data/y4/tools/data/users-0.csv
     csv:
       delimiter: "|"
       withHeader: true
@@ -86,7 +86,7 @@ sources:
             type: "STRING"
             index: 16
 
-  - path: /data/y4/tools/data/post-0s.csv
+  - path: /data/y4/tools/data/posts-0.csv
     csv:
       delimiter: "|"
       withHeader: true
@@ -140,7 +140,7 @@ sources:
             type: "INT"
             index: 14
 
-  - path: /data/y4/tools/data/group-0s.csv
+  - path: /data/y4/tools/data/groups-0.csv
     csv:
       delimiter: "|"
       withHeader: true
@@ -161,7 +161,7 @@ sources:
             type: "STRING"
             index: 3
 
-  - path: /data/y4/tools/data/organization-0s.csv
+  - path: /data/y4/tools/data/organizations-0.csv
     csv:
       delimiter: "|"
       withHeader: true
@@ -185,7 +185,7 @@ sources:
             type: "STRING"
             index: 4
 
-  - path: /data/y4/tools/data/task-0s.csv
+  - path: /data/y4/tools/data/tasks-0.csv
     csv:
       delimiter: "|"
       withHeader: true
@@ -284,25 +284,25 @@ sources:
             type: "STRING"
             index: 0
 
-#  - path: /data/y4/tools/data/post_repost_post-0.csv
-#    csv:
-#      delimiter: "|"
-#      withHeader: true
-#    edges:
-#      - name: post_repost_post
-#        mode: INSERT
-#        src:
-#          id:
-#            type: "STRING"
-#            index: 2 # src
-#        dst:
-#          id:
-#            type: "STRING"
-#            index: 1 # dst
-#        props:
-#          - name: "extra_prop"
-#            type: "STRING"
-#            index: 0
+  - path: /data/y4/tools/data/post_repost_post-0.csv
+    csv:
+      delimiter: "|"
+      withHeader: true
+    edges:
+      - name: post_repost_post
+        mode: INSERT
+        src:
+          id:
+            type: "STRING"
+            index: 2 # src
+        dst:
+          id:
+            type: "STRING"
+            index: 1 # dst
+        props:
+          - name: "extra_prop"
+            type: "STRING"
+            index: 0
 
   - path: /data/y4/tools/data/post_cite_post-0.csv
     csv:
@@ -324,25 +324,25 @@ sources:
             type: "STRING"
             index: 0
 
-#  - path: /data/y4/tools/data/user_post_post-0.csv
-#    csv:
-#      delimiter: "|"
-#      withHeader: true
-#    edges:
-#      - name: user_post_post
-#        mode: INSERT
-#        src:
-#          id:
-#            type: "STRING"
-#            index: 2 # src
-#        dst:
-#          id:
-#            type: "STRING"
-#            index: 1 # dst
-#        props:
-#          - name: "extra_prop"
-#            type: "STRING"
-#            index: 0
+  - path: /data/y4/tools/data/user_post_post-0.csv
+    csv:
+      delimiter: "|"
+      withHeader: true
+    edges:
+      - name: user_post_post
+        mode: INSERT
+        src:
+          id:
+            type: "STRING"
+            index: 2 # src
+        dst:
+          id:
+            type: "STRING"
+            index: 1 # dst
+        props:
+          - name: "extra_prop"
+            type: "STRING"
+            index: 0
 
   - path: /data/y4/tools/data/user_comment_user-0.csv
     csv:
@@ -544,22 +544,22 @@ sources:
             type: "STRING"
             index: 0
 
-#  - path: /data/y4/tools/data/user_join_organization-0.csv
-#    csv:
-#      delimiter: "|"
-#      withHeader: true
-#    edges:
-#      - name: user_join_organization
-#        mode: INSERT
-#        src:
-#          id:
-#            type: "STRING"
-#            index: 2 # src
-#        dst:
-#          id:
-#            type: "STRING"
-#            index: 1 # dst
-#        props:
-#          - name: "extra_prop"
-#            type: "STRING"
-#            index: 0
+  - path: /data/y4/tools/data/user_join_organization-0.csv
+    csv:
+      delimiter: "|"
+      withHeader: true
+    edges:
+      - name: user_join_organization
+        mode: INSERT
+        src:
+          id:
+            type: "STRING"
+            index: 2 # src
+        dst:
+          id:
+            type: "STRING"
+            index: 1 # dst
+        props:
+          - name: "extra_prop"
+            type: "STRING"
+            index: 0

src/main/java/com/idata/tools/GeneratorTestData.java

@@ -341,15 +341,23 @@ public class GeneratorTestData {
         int i = 0;
         List<String> edgeLists = Arrays.asList("post_in_task", "user_participate_task", "user_join_group", "user_join_organization");
         for (String src : srcList) {
-            i++;
             if (i % 1000 == 0) {
                 log.info("edge_type: {}-{}, {}",edgeType, count, i);
             }
 
             List<String> dsts;
             if (i == 0 && !edgeLists.contains(edgeType)) {
+                // 当索引等于0 时，选择 1000 个目标节点
+                dsts = getRandomSample(dstList, 1000);
+                log.info("super vertex id is {}; dst size is {}", src, dsts.size());
+            } else if (i == 1 && !edgeLists.contains(edgeType)) {
+                // 当索引等于0 时，选择 10001 个目标节点
+                dsts = getRandomSample(dstList, 1500);
+                log.info("ssuper vertex id is {}; dst size is {}", src, dsts.size());
+            } else if (i == 2 && !edgeLists.contains(edgeType)) {
                 // 当索引等于0 时，选择 10001 个目标节点
                 dsts = getRandomSample(dstList, 10001);
+                log.info("sssuper vertex id is {}; dst size is {}", src, dsts.size());
             } else {
                 // 否则选择最多 num 个目标节点
                 dsts = getRandomSample(dstList, Math.min(dstList.size(), num));
@@ -364,6 +372,7 @@ public class GeneratorTestData {
                 edge.put("edge_type", edgeType);
                 edges.add(edge);
             }
+            i++;
         }
         return edges;
     }