mirror of
https://github.com/nesquena/hermes-webui.git
synced 2026-07-11 18:20:30 +00:00
nesquena-hermes
5296b15956
Codex reproduced a SILENT regression: the PR's 4096-BYTE read could miss the continuation marker the old read_text()[:4096] (4096 CHARS) found. Session.save() writes compression_anchor_summary (ensure_ascii=False) BEFORE parent_session_id, so a multi-byte emoji summary puts the marker at char offset ~1230 but byte offset ~4900 — within the old 4096-char window, past a 4096-byte one → continuation child no longer detected. Fix: read 16384 bytes (4096 chars worst-case for 4-byte UTF-8) then slice [:4096] chars, keeping the same needle + bounded I/O. Multibyte regression test added + proven non-vacuous; updated the read-size assertion to <=16384. Co-authored-by: starship-s <starship-s@users.noreply.github.com>
120 lines
4.6 KiB
Python
120 lines
4.6 KiB
Python
import api.models as models
|
|
|
|
|
|
def test_compression_continuation_fallback_reads_only_file_head(monkeypatch, tmp_path):
|
|
session_dir = tmp_path / "sessions"
|
|
session_dir.mkdir()
|
|
index_file = tmp_path / "_index.json"
|
|
index_file.write_text("[]", encoding="utf-8")
|
|
|
|
monkeypatch.setattr(models, "SESSION_DIR", session_dir)
|
|
monkeypatch.setattr(models, "SESSION_INDEX_FILE", index_file)
|
|
models.SESSIONS.clear()
|
|
|
|
parent_sid = "parent"
|
|
candidate = session_dir / "child.json"
|
|
candidate.write_text("x" * 8192 + f'"parent_session_id": "{parent_sid}"', encoding="utf-8")
|
|
|
|
opened_for_read = []
|
|
read_text_calls = []
|
|
|
|
original_path_open = models.Path.open
|
|
original_path_read_text = models.Path.read_text
|
|
|
|
class _TrackingHandle:
|
|
def __init__(self, path, inner):
|
|
self._path = str(path)
|
|
self._inner = inner
|
|
|
|
def read(self, size=-1):
|
|
opened_for_read.append((self._path, size))
|
|
return self._inner.read(size)
|
|
|
|
def __enter__(self):
|
|
self._inner.__enter__()
|
|
return self
|
|
|
|
def __exit__(self, exc_type, exc, tb):
|
|
return self._inner.__exit__(exc_type, exc, tb)
|
|
|
|
def __getattr__(self, name):
|
|
return getattr(self._inner, name)
|
|
|
|
def tracking_open(path, *args, **kwargs):
|
|
return _TrackingHandle(path, original_path_open(path, *args, **kwargs))
|
|
|
|
def tracking_read_text(self, *args, **kwargs):
|
|
read_text_calls.append(str(self))
|
|
return original_path_read_text(self, *args, **kwargs)
|
|
|
|
monkeypatch.setattr(models.Path, "open", tracking_open)
|
|
monkeypatch.setattr(models.Path, "read_text", tracking_read_text)
|
|
|
|
session = models.Session(session_id=parent_sid)
|
|
|
|
assert models._has_compression_continuation(session) is False
|
|
|
|
assert str(index_file) in read_text_calls
|
|
assert all(path != str(candidate) for path in read_text_calls)
|
|
|
|
candidate_reads = [size for path, size in opened_for_read if path == str(candidate)]
|
|
assert candidate_reads, "fallback should read the candidate sidecar"
|
|
# Bounded read: at most 16384 bytes (== 4096 UTF-8 chars worst case), never
|
|
# an unbounded -1 read of the full sidecar.
|
|
assert all(size != -1 and size <= 16384 for size in candidate_reads)
|
|
|
|
|
|
def test_compression_continuation_prefix_match_stays_true(monkeypatch, tmp_path):
|
|
session_dir = tmp_path / "sessions"
|
|
session_dir.mkdir()
|
|
index_file = tmp_path / "_index.json"
|
|
index_file.write_text("[]", encoding="utf-8")
|
|
|
|
monkeypatch.setattr(models, "SESSION_DIR", session_dir)
|
|
monkeypatch.setattr(models, "SESSION_INDEX_FILE", index_file)
|
|
models.SESSIONS.clear()
|
|
|
|
parent_sid = "parent"
|
|
candidate = session_dir / "child.json"
|
|
candidate.write_text('{"session_id":"child", "parent_session_id": "parent",', encoding="utf-8")
|
|
|
|
session = models.Session(session_id=parent_sid)
|
|
|
|
assert models._has_compression_continuation(session) is True
|
|
|
|
|
|
def test_compression_continuation_multibyte_summary_before_marker(monkeypatch, tmp_path):
|
|
"""A multi-byte (emoji) compression summary written BEFORE parent_session_id
|
|
can push the marker past a 4096-BYTE cutoff while still inside the old
|
|
4096-CHARACTER prefix. The bounded read must preserve the char-prefix
|
|
semantics (read 16384 bytes, slice 4096 chars) so the continuation is still
|
|
detected. Regression for the gate-found byte-vs-char truncation."""
|
|
session_dir = tmp_path / "sessions"
|
|
session_dir.mkdir()
|
|
index_file = tmp_path / "_index.json"
|
|
index_file.write_text("[]", encoding="utf-8")
|
|
|
|
monkeypatch.setattr(models, "SESSION_DIR", session_dir)
|
|
monkeypatch.setattr(models, "SESSION_INDEX_FILE", index_file)
|
|
models.SESSIONS.clear()
|
|
|
|
parent_sid = "parent"
|
|
candidate = session_dir / "child.json"
|
|
# 1200 emoji (4 bytes each in UTF-8) → marker at char offset ~1230 but byte
|
|
# offset ~4900: within the old 4096-char window, PAST a 4096-byte window.
|
|
emoji_summary = "😀" * 1200
|
|
candidate.write_text(
|
|
'{"session_id":"child", "compression_anchor_summary": "' + emoji_summary
|
|
+ '", "parent_session_id": "parent",',
|
|
encoding="utf-8",
|
|
)
|
|
|
|
# sanity: the marker is past 4096 bytes but within 4096 chars
|
|
text = candidate.read_text(encoding="utf-8")
|
|
needle = '"parent_session_id": "parent"'
|
|
assert text.encode("utf-8").index(needle.encode()) > 4096, "test setup: needle must be past 4096 bytes"
|
|
assert text.index(needle) < 4096, "test setup: needle must be within 4096 chars"
|
|
|
|
session = models.Session(session_id=parent_sid)
|
|
assert models._has_compression_continuation(session) is True
|