Struct LineBuffer 

pub struct LineBuffer { /* private fields */ }

Accumulates raw stream bytes and yields complete newline-terminated lines.

Streaming responses arrive as arbitrary byte chunks: a single multi-byte UTF-8 character (e.g. é, or an emoji) can be split across two chunks. The naive buffer.push_str(&String::from_utf8_lossy(&chunk)) decodes each raw chunk in isolation, so the leading bytes of a split character decode to U+FFFD (the replacement character) and the trailing bytes decode to another — permanently corrupting the text.

LineBuffer accumulates raw bytes and only decodes complete, newline-terminated lines. A \n (0x0A) byte can never appear inside a multi-byte UTF-8 sequence, so decoding a whole line is always safe even when the underlying chunk boundary fell mid-character.

Consumed bytes are tracked with a cursor rather than drained per line: draining left-shifts every remaining byte, which turns a chunk carrying K lines into O(K·chunk) work. The buffer compacts once the consumed prefix grows past a threshold, so memory stays bounded over a long stream.

Implementations

impl LineBuffer

pub fn new() -> Self

Creates an empty buffer.

pub fn push(&mut self, chunk: &[u8])

Appends a raw byte chunk from the stream.

pub fn next_line(&mut self) -> Option<String>

Removes and returns the next complete line (trailing \n stripped), or None when no complete line is buffered yet.

pub fn take_remaining(&mut self) -> Option<String>

Drains and returns whatever bytes remain after the last complete line.

A stream can end without a trailing newline, leaving a final, otherwise-complete record buffered that next_line will never yield. Call this once at end-of-stream to recover it. Returns None when the buffer is empty or the remainder is only whitespace.

Trait Implementations

impl Default for LineBuffer

fn default() -> LineBuffer

Returns the “default value” for a type.