With practical code reuse, the code fragments from developer forums often migrate to applications. Owing to the incomplete nature of such fragments, they often lack the details on exception handling. The adaptation for exception handling to the codebase is not trivial as developers must learn and memorize what API methods could cause exceptions and what exceptions need to be handled. We propose EHBlock, a learning-based exception handling recommender, to accept a given Java code snippet and recommend if a try-catch block is needed. We overcome the limitations of the state-of-the-art information retrieval approaches by our learning exception handling from complete code via Relational Graph Convolutional Network (R-GCN). Via R-GCN, we consider the program dependencies in the surrounding context, which enables EHBlock to learn the identities of the APIs and the relations with the corresponding exception types needed to be handled. Our empirical evaluation shows that EHBlock relatively improves by 12.3% in F-score over the state-of-the-art approach in try-catch block necessity checking.