fix(sessions): reject session IDs with a trailing newline

[shoemoney]

The bug

_validate_uuid documents a precise contract (src/claude_agent_sdk/_internal/sessions.py:70):

Returns the string if it is a valid UUID, else None.

It accepts one string that is not a valid UUID. Python's $ also matches immediately before a trailing newline, so re.match(r"^...$", s) succeeds on "<uuid>\n" — and the newline is preserved in the returned value:

>>> from claude_agent_sdk._internal.sessions import _validate_uuid
>>> _validate_uuid("3f2504e0-4f89-11d3-9a0c-0305e82c3301")
'3f2504e0-4f89-11d3-9a0c-0305e82c3301'
>>> _validate_uuid("3f2504e0-4f89-11d3-9a0c-0305e82c3301\n")
'3f2504e0-4f89-11d3-9a0c-0305e82c3301\n'

\r\n, leading whitespace and trailing junk are all correctly rejected — it is specifically the single trailing \n that slips through, which is the one a caller is most likely to actually have.

Why it matters

There are nine call sites. The validated value is used to build a filename (sessions.py:781):

file_name = f"{uuid}.jsonl"

3f2504e0-...-0305e82c3301\n.jsonl never exists, so the failure is silent and typed as "not found":

entry point	result with a trailing newline
get_session_info	None
get_session_messages	[]
list_subagents	[]

A caller cannot distinguish "you passed a malformed session ID" from "that session does not exist" — which is the distinction the validator exists to make. Acquiring the stray byte is ordinary: reading an ID out of a file, or capturing it from command output.

The fix

fullmatch anchors both ends with no newline exemption, so the redundant ^/$ come out with it. Two lines.

Verification

Against the real imported function, before and after:

--- before (main) ---
  clean       -> '3f2504e0-4f89-11d3-9a0c-0305e82c3301'
  trailing \n -> '3f2504e0-4f89-11d3-9a0c-0305e82c3301\n'    <- accepted

--- after ---
  clean       -> '3f2504e0-4f89-11d3-9a0c-0305e82c3301'
  trailing \n -> None                                          <- rejected

Full suite and lint:

$ uv run --extra dev pytest -q
1337 passed, 5 skipped in 11.15s

$ uv run --extra dev ruff check src/
All checks passed!

$ uv run --extra dev ruff format --check src/claude_agent_sdk/_internal/sessions.py
1 file already formatted

Not included

I didn't add a regression test, to keep this to the two-line change — but tests/test_sessions.py is the obvious home for one and it's a three-line assert. Happy to push it here if you'd rather have the behaviour pinned.

I also left the other three compiled patterns in this file alone. _SKIP_FIRST_PROMPT_PATTERN.match is a deliberate prefix match, and the other two are used with findall/sub, so none of them have this issue.