The deployment of caches in the Internet has grown significantly in the last decade, thus enabling the vision of Content-Centric Networks (CCNs). The caching policy employed at these routers has significant impact on the potential gains in network performance. Policies that adapt to changes in content popularities are of special interest. In this paper, we propose a novel caching policy called Data Stream Caching Algorithm (DSCA) with the goal of maximizing cache hit rate of CCN routers by incorporating content popularity in caching decisions. In contrast to existing popularitybased caching policies, DSCA copes with dynamics in content popularity while operating under the memory and high processing rate constraints of CCN network routers. DSCA achieves the above objectives using a data streaming algorithm that identifies the most popular contents adapted to work in a windowed manner. We analyze the performance and robustness of the proposed caching policy through simulations. Evaluations on synthetic data and real-world traces show that DSCA outperforms LRU and other caching policies evaluated in this work.