Newline-Split Stream Reader

Given a class with a `next()` method that returns arbitrary string chunks containing zero or more embedded newlines, write a wrapper that yields one full line per call. Chunks can split a line mid-word; partial lines from one chunk must be joined with the start of the next. Requirements The given class exposes next() -> string, where each call returns one chunk of the underlying stream. Chunks may contain zero, one, or many \n characters and may end mid-line. Example chunk sequence (literal \n shown for clarity): "one\ntw", "o\n", "three\nfour", "\nfi", "ve" The expected output, one full line per emission: "one", "two", "three", "four", "five" Implement a wrapper class whose own nextLine() (or generator) consumes chunks as needed and emits one complete line per call. Notes Maintain a buffer string of "text seen but not yet emitted". On each nextLine call: scan the buffer for \n; if found, slice and emit the prefix; else call the underlying next() and append, repeat. End-of-stream needs an explicit sentinel — define what next() returns when exhausted and emit any leftover buffered text as the final line. The most common bug: language split quirks. Java's String.split("\n") drops trailing empty strings by default and treats a leading \n as a leading empty match; pass a negative limit (split("\n", -1)) to preserve them. Python's str.split("\n") keeps empties. Edge cases the interviewer will probe: a chunk that is exactly "\n" (emit one empty line and continue), consecutive newlines "a\n\nb" (emit "a", then "", then continue buffering "b"), and a stream that never terminates with a newline (emit the tail on EOF only). Preparation Practice the buffer-scan + lazy-pull pattern in 10 minutes on a small example before adding any language-specific helpers. Pre-rehearse the split quirks for the language you plan to use — losing 15 minutes to a split edge case is the single most-cited reason this round goes sideways. Walk the example chunk sequence by hand on a whiteboard before writing code; the chunk boundaries are deliberately placed to expose all the edge cases at once.