[FLINK-11595] Gelly's addEdge() in certain circumstances still includes duplicate vertices.

• Type: Bug
• Status: Closed
• Priority: Not a Priority
• Resolution: Won't Fix
• Affects Version/s: 1.7.1
• Component/s: Library / Graph Processing

Environment

MacOS, intelliJ

• Assignee: Unassigned
• Reporter: Calvin Han
• Created: 13/Feb/19 22:18
• Updated: 19/Oct/22 11:48
• Resolved: 19/Oct/22 11:48

Description

Assuming a base graph constructed by:

```

public class GraphCorn {

public static Graph<String, VertexLabel, EdgeLabel> gc;

public GraphCorn(String filename) throws Exception {
ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();

DataSet<Tuple6<String, String, String, String, String, String>> csvInput = env.readCsvFile(filename)
.types(String.class, String.class, String.class, String.class, String.class, String.class);

DataSet<Vertex<String, VertexLabel>> srcTuples = csvInput.project(0, 2)
.map(new MapFunction<Tuple, Vertex<String, VertexLabel>>() {
@Override
public Vertex<String, VertexLabel> map(Tuple tuple) throws Exception

{ VertexLabel lb = new VertexLabel(Util.hash(tuple.getField(1))); return new Vertex<>(tuple.getField(0), lb); }
}).returns(new TypeHint<Vertex<String, VertexLabel>>(){});

DataSet<Vertex<String, VertexLabel>> dstTuples = csvInput.project(1, 3)
.map(new MapFunction<Tuple, Vertex<String, VertexLabel>>() {
@Override
public Vertex<String, VertexLabel> map(Tuple tuple) throws Exception { VertexLabel lb = new VertexLabel(Util.hash(tuple.getField(1))); return new Vertex<>(tuple.getField(0), lb); }

}).returns(new TypeHint<Vertex<String, VertexLabel>>(){});

DataSet<Vertex<String, VertexLabel>> vertexTuples = srcTuples.union(dstTuples).distinct(0);

DataSet<Edge<String, EdgeLabel>> edgeTuples = csvInput.project(0, 1, 4, 5)
.map(new MapFunction<Tuple, Edge<String, EdgeLabel>>() {
@Override
public Edge<String, EdgeLabel> map(Tuple tuple) throws Exception

{ EdgeLabel lb = new EdgeLabel(Util.hash(tuple.getField(2)), Long.parseLong(tuple.getField(3))); return new Edge<>(tuple.getField(0), tuple.getField(1), lb); }

}).returns(new TypeHint<Edge<String, EdgeLabel>>(){});

this.gc = Graph.fromDataSet(vertexTuples, edgeTuples, env);
}

}

```

Base graph CSV:

```

0,1,a,b,c,0
0,2,a,d,e,1
1,2,b,d,f,2

```

Attempt to add edges using the following function:

```

try(BufferedReader br = new BufferedReader(new FileReader(this.fileName))) {
for(String line; (line = br.readLine()) != null; )

{ String[] attributes = line.split(","); assert(attributes.length == 6); String srcID = attributes[0]; String dstID = attributes[1]; String srcLb = attributes[2]; String dstLb = attributes[3]; String edgeLb = attributes[4]; String ts = attributes[5]; Vertex<String, VertexLabel> src = new Vertex<>(srcID, new VertexLabel(Util.hash(srcLb))); Vertex<String, VertexLabel> dst = new Vertex<>(dstID, new VertexLabel(Util.hash(dstLb))); EdgeLabel edge = new EdgeLabel(Util.hash(edgeLb), Long.parseLong(ts)); GraphCorn.gc = GraphCorn.gc.addEdge(src, dst, edge); }

} catch (Exception e)

{ System.err.println(e.getMessage()); }

```

The graph components to add is:

```

0,4,a,d,k,3
1,3,b,a,g,3
2,3,d,a,h,4

```

GraphCorn.gc will contain duplicate node 0, 1, and 2 (those that exist in base graph), which should not be the case acceding to the documentation.