SSE Event Parser

Parse server-sent event (SSE) chunks into complete data payloads.

Requirements

• Define parse_sse_events(chunks).
• chunks is a list of already-decoded strings from a streaming response, possibly including empty chunks.
• Events are separated by a blank line.
• Lines may end with LF or CRLF; strip a trailing \r from each completed line before field handling.
• Lines beginning with data: add payload text. After the colon, if the next character is a single space, remove that one space and keep any further spaces (WHATWG SSE rule).
• Multiple data: lines in one event join with \n.
• An event with data: but no payload emits "". A block with no data: lines emits nothing.
• Lines beginning with : are comments and should be ignored.
• Ignore other fields such as event: and id:.
• Return a list of completed event data strings.
• Ignore an incomplete final event without a trailing blank line.

Example

chunks = ["data: hello\n", "data: world\n\n"]
assert parse_sse_events(chunks) == ["hello\nworld"]

Constraints

• Chunk boundaries may split lines anywhere.
• CRLF may split between chunks; it must still count as one line ending.
• Use an explicit state machine or careful line buffering.
• Don't require network or async code.
• This deliberately implements only the stated SSE subset. Bare CR endings, a leading byte order mark, and colonless data fields are outside this exercise. The WHATWG standard covers those cases; production clients also need incremental UTF-8 decoding, event IDs, reconnect handling, and bounded buffering.