instruktai-python-logger 0.5.0__tar.gz → 0.5.1__tar.gz

{instruktai_python_logger-0.5.0 → instruktai_python_logger-0.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: instruktai-python-logger
-Version: 0.5.0
+Version: 0.5.1
 Summary: Centralized logging utilities for Python services.
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown

{instruktai_python_logger-0.5.0 → instruktai_python_logger-0.5.1}/instrukt_ai_logging/logging.py

@@ -548,6 +548,46 @@ def iter_recent_log_lines(log_file: Path, since: timedelta) -> list[str]:
     return lines
 
 
+# Initial backward-probe size for locating the start of the --since window in a
+# large file. A typical multi-minute window fits in one or two probes; the probe
+# doubles when the window is larger, so worst case is still O(window), not O(file).
+_WINDOW_PROBE_BYTES = 262144  # 256 KiB
+
+
+def _window_start_offset(path: Path, cutoff: datetime) -> int:
+    """Return a byte offset at or before the first line newer than `cutoff`.
+
+    A log file is appended in timestamp order, so the window's start can be found
+    by probing backward from EOF in exponentially growing chunks instead of
+    scanning from byte 0: once a probe's earliest timestamped line is already
+    older than the cutoff, the window begins within that probe and everything
+    before it is skippable. Returns 0 when the window spans the whole file (or the
+    file is smaller than a single probe), which reproduces a full forward scan.
+    """
+    try:
+        size = path.stat().st_size
+    except OSError:
+        return 0
+    if size <= _WINDOW_PROBE_BYTES:
+        return 0
+
+    probe = _WINDOW_PROBE_BYTES
+    with path.open("rb") as fb:
+        while probe < size:
+            start = size - probe
+            fb.seek(start)
+            fb.readline()  # discard the partial line spanning the probe boundary
+            first_ts: datetime | None = None
+            for raw in fb:
+                first_ts = parse_log_timestamp(raw.decode("utf-8", errors="replace"))
+                if first_ts is not None:
+                    break
+            if first_ts is not None and first_ts < cutoff:
+                return start
+            probe *= 2
+    return 0
+
+
 def iter_recent_log_lines_merged(files: Iterable[Path], since: timedelta) -> Iterator[str]:
     """Yield lines newer than now-`since` merged across multiple files by timestamp.
 
@@ -567,12 +607,17 @@ def iter_recent_log_lines_merged(files: Iterable[Path], since: timedelta) -> Ite
 
     def _file_stream(path: Path) -> Iterator[tuple[datetime, str]]:
         try:
-            f = path.open("r", encoding="utf-8", errors="replace")
+            fb = path.open("rb")
         except FileNotFoundError:
             return
         try:
+            offset = _window_start_offset(path, cutoff)
+            if offset:
+                fb.seek(offset)
+                fb.readline()  # discard the partial line at the seek boundary
             last_ts: datetime | None = None
-            for line in f:
+            for raw in fb:
+                line = raw.decode("utf-8", errors="replace")
                 ts = parse_log_timestamp(line)
                 if ts is None:
                     if last_ts is None or last_ts < cutoff:
@@ -584,7 +629,7 @@ def iter_recent_log_lines_merged(files: Iterable[Path], since: timedelta) -> Ite
                     continue
                 yield (ts, line)
         finally:
-            f.close()
+            fb.close()
 
     eligible: list[Path] = []
     for path in files:

{instruktai_python_logger-0.5.0 → instruktai_python_logger-0.5.1}/instruktai_python_logger.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: instruktai-python-logger
-Version: 0.5.0
+Version: 0.5.1
 Summary: Centralized logging utilities for Python services.
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown

{instruktai_python_logger-0.5.0 → instruktai_python_logger-0.5.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "instruktai-python-logger"
-version = "0.5.0"
+version = "0.5.1"
 description = "Centralized logging utilities for Python services."
 readme = "README.md"
 requires-python = ">=3.11"

{instruktai_python_logger-0.5.0 → instruktai_python_logger-0.5.1}/tests/test_log_aggregation.py

@@ -34,6 +34,9 @@ _MSG_GIT_ERROR = "git-error"
 _MSG_INFO = "info-line"
 _MSG_CRON = "cron-line"
 _MSG_DAEMON = "daemon-line"
+_MSG_OLD_PREFIX = "msg=old-"
+_MSG_LINE_PREFIX = "msg=line-"
+_LARGE_LINE_COUNT = 20000
 
 
 @pytest.fixture()
@@ -176,6 +179,59 @@ def test_iter_recent_log_lines_merged_keeps_continuation_lines_with_parent(
     assert _MSG_AFTER in lines[3]
 
 
+def test_iter_recent_log_lines_merged_reads_tail_window_of_large_file(
+    app_log_dir: Path,
+) -> None:
+    # A large body of old lines (well past the 256 KiB backward probe) followed by
+    # a recent window that includes a multi-line entry. The reader must return only
+    # the recent lines — proving it locates the window near EOF rather than scanning
+    # the whole file — and must keep the continuation line with its parent across
+    # the seek boundary.
+    daemon = app_log_dir / "demo-app.log"
+    old_block = "".join(
+        f"{_now_iso(-30)} level=INFO logger=demo.daemon {_MSG_OLD_PREFIX}{i}\n" for i in range(_LARGE_LINE_COUNT)
+    )
+    recent_block = (
+        f"{_now_iso(-1)} level=ERROR logger=demo.daemon msg={_MSG_BOOM}\n"
+        f"{_TRACEBACK_LINE}\n"
+        f"{_now_iso(0)} level=INFO logger=demo.daemon msg={_MSG_AFTER}\n"
+    )
+    daemon.write_text(old_block + recent_block, encoding="utf-8")
+    assert daemon.stat().st_size > 262144  # forces the backward-probe path (offset > 0)
+
+    files = resolve_log_files("demo-app")
+    lines = list(iter_recent_log_lines_merged(files, since=timedelta(minutes=5)))
+    assert len(lines) == 3
+    assert _MSG_BOOM in lines[0]
+    assert lines[1].startswith(_TRACEBACK_LINE)
+    assert _MSG_AFTER in lines[2]
+    assert all(_MSG_OLD_PREFIX not in line for line in lines)
+
+
+def test_iter_recent_log_lines_merged_returns_whole_large_file_within_window(
+    app_log_dir: Path,
+) -> None:
+    # Every line is inside the window but the file exceeds one probe: the backward
+    # probe must grow to the file start (offset 0) and return all lines, not just
+    # the final probe's worth.
+    daemon = app_log_dir / "demo-app.log"
+    daemon.write_text(
+        "".join(
+            f"{_now_iso(-1)} level=INFO logger=demo.daemon {_MSG_LINE_PREFIX}{i}\n" for i in range(_LARGE_LINE_COUNT)
+        ),
+        encoding="utf-8",
+    )
+    assert daemon.stat().st_size > 262144
+
+    files = resolve_log_files("demo-app")
+    lines = list(iter_recent_log_lines_merged(files, since=timedelta(minutes=5)))
+    assert len(lines) == _LARGE_LINE_COUNT
+    first_marker = f"{_MSG_LINE_PREFIX}0"
+    last_marker = f"{_MSG_LINE_PREFIX}{_LARGE_LINE_COUNT - 1}"
+    assert first_marker in lines[0]
+    assert last_marker in lines[-1]
+
+
 def test_cli_exclude_drops_matching_lines(
     app_log_dir: Path, monkeypatch: pytest.MonkeyPatch, capsys: pytest.CaptureFixture[str]
 ) -> None: