The analyzer currently implements a getJSPathFromSourcePath() function using ts.getOutputFileNames().

However, just because we can determine that src/foo.ts gets output by tsc to build/foo.js doesn't mean that's where the file is ultimately importable from (which is what jsPath intends to capture), since additional transforms outside of ts can (and often do) further process that into another location that is ultimately published (i.e. we output to ./development/*.js and process that further with rollup to ./*.js).

This is something we're going to hit pretty quickly when we start testing against real-world packages, and will likely need to solve by allowing the caller to the analyzer to effectively configure the getJSPathFromSourcePath() function.