Online video streaming has been experiencing great popularity in recent years, and an increasing number of video contents are now delivered via wireless networks, which brings an overwhelming burden towards mobile operators. Femtocells are introduced to improve the area spectral efficiency of wireless video delivery via short link transmission. The backhaul pressure in the presence of dense deployment of femtocells can be reduced by the technologies of femtocaching. In this paper, video clip selection and video clip assignment problems are considered for femtocaching. On one hand, since mobile users always pursue the video clips that are highly popular and can be played smoothly, it is important to ensure the satisfaction of mobile users. On the other hand, the assignment of multiple video clips among multiple femto-caches is also significant as the capacity of each cache is limited. To model the video selection dynamics of mobile users, an evolutionary game is introduced and two algorithms are proposed to result in an evolutionary stable strategies, which correspond to the balance of video popularity and video quality. For the video assignment problem, the auction-based algorithms are proposed with low complexity to approximate the social optimality with the limited caching capacities. As a consequence, all femtocells and mobile video users are satisfied with their utilities respectively.