We present a framework for analyzing weak gravitational lensing survey data, including lensing and source-density observables, plus spectroscopic redshift calibration data. All two-point observables are predicted in terms of parameters of a perturbed Robertson-Walker metric, making the framework independent of the models for gravity, dark energy, or galaxy properties. For Gaussian fluctuations, the two-point model determines the survey likelihood function and allows Fisher matrix forecasting. The framework includes nuisance terms for the major systematic errors: shear measurement errors, magnification bias and redshift calibration errors, intrinsic galaxy alignments, and inaccurate theoretical predictions. We propose flexible parameterizations of the many nuisance parameters related to galaxy bias and intrinsic alignment. For the first time, we can integrate many different observables and systematic errors into a single analysis. As a first application of this framework, we demonstrate that: uncertainties in power-spectrum theory cause very minor degradation to cosmological information content; nearly all useful information (excepting baryon oscillations) is extracted with 3 bins per decade of angular scale; and the rate at which galaxy bias varies with redshift, substantially influences the strength of cosmological inference. The framework will permit careful study of the interplay between numerous observables, systematic errors, and spectroscopic calibration data for large weak lensing surveys.