Abstract: Gravitational waves (GWs) from binary neutron stars (BNSs) offer valuable understanding of the nature of compact objects and hadronic matter. However, the analyses accompanied require massive computational power due to the difficulties in Bayesian stochastic sampling. The third-generation (3G) GW detectors are expected to detect BNS signals with significantly extended signal duration, detection rate, and loudness, the analyses of which would become a major computational burden in the 3G era. We present novel data analysis methods for BNS long signals, including semi-analytical Bayesian approach and machine-learning-based techniques, enabling source pre-merger localization, full parameter estimation and constraint on equations of state (EOSs) for hours-long BNS signals in seconds at minimal hardware cost. Some of these tasks would be prohibitively slow under traditional analysis frameworks. We additionally estimate the computational costs of analyzing BNS signals in the 3G era, showing that the lightweight methods will be crucial for catalog-level analysis in the future.