D-SafeLogger 0.2.0__tar.gz → 0.2.1__tar.gz

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/CHANGELOG.md

@@ -5,6 +5,22 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.2.1] - 2026-05-20
+
+### Fixed
+- Apply resolved `sens_kws` / `sens_kws_replace` to diagnostic local-variable snapshots across sync text, sync structured, async snapshot, and multiprocess Writer formatter paths.
+- Preserve async producer-side diagnostic frame snapshots in text diagnostic output.
+- Align daily routing examples with the implemented `YYYYMMDD` filename suffix.
+
+### Changed
+- Updated example tests to match the documented runnable examples for migration, web API, long-running service, compliance audit, production debugging, async performance, multiprocess logging, CLI operations, OpenTelemetry, structlog, and container collector coexistence scenarios.
+- Updated public docs, API docs, and test/coverage baselines for the 0.2.1 patch release.
+
+### Verified
+- Full local validation on Python 3.14.3 / Windows: `658 passed, 3 skipped` (`661` collected).
+- Coverage validation regenerated `coverage.xml`: terminal total `87%`, line-rate `88.97%`, branch-rate `81.46%`.
+- API docs, design-doc sync, and whitespace checks pass; `git diff --check` reports only CRLF/LF warnings.
+
 ## [0.2.0] - 2026-05-05
 
 ### Added

{d_safelogger-0.2.0/src/D_SafeLogger.egg-info → d_safelogger-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: D-SafeLogger
-Version: 0.2.0
+Version: 0.2.1
 Summary: Zero-dependency, thread-safe, append-only logging library for Python with 3-layer config pipeline
 Author: D
 License-Expression: Apache-2.0
@@ -28,8 +28,8 @@ Dynamic: license-file
 # D-SafeLogger
 
 [![CI](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml/badge.svg)](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml)
-[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
-[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
+[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
+[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
 [![Zero Dependencies](https://img.shields.io/badge/dependencies-zero-brightgreen.svg)](#main-features)
 
@@ -66,7 +66,7 @@ logger = GetLogger(__name__)
 logger.info("Application started")
 ```
 
-`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_2026-04-03.log`.
+`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_20260403.log`.
 
 Typical text output:
 
@@ -122,7 +122,7 @@ The "Safe" in the name is a design stance that runs across several dimensions of
 
 - **Startup safety:** invalid settings, inconsistent options, and unwritable destinations fail during setup. D-SafeLogger stops a broken logging configuration before the application starts doing real work, instead of silently degrading later.
 - **File safety:** the routing layer opens the next destination instead of renaming or truncating the active log file, which avoids the common Windows failure mode where active log files cannot be renamed. It also avoids the POSIX case where a successful rename leaves the writer appending to the previous file. Routed files can be paired with SHA-256 sidecars and an optional manifest, so log content is verifiable after the fact.
-- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Sensitive-keyword masking is applied on the Writer side using the keyword set established at configure time.
+- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Diagnostic local-variable snapshots and Writer-side formatting use the sensitive-keyword set established at configure time.
 - **Operational control:** environment variables provide explicit runtime overrides for diagnostics, routing, hashing, log levels, and queue/timeout behavior without rebuilding or editing application code.
 - **Concurrency and multiprocess safety:** multiprocess workers do not open the shared log files themselves. A parent-side Writer owns the sinks and accepts records over IPC, with bounded queues and explicit timeouts that keep the host process from unbounded waits.
 - **Failure observability:** when records cannot be delivered, the runtime classifies the outcome where it can: `KnownRejected`, `KnownDropped`, or `UnexplainedLost`. Counters and shutdown summaries make abnormal scenarios describable rather than silent.

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/README.md

@@ -1,8 +1,8 @@
 # D-SafeLogger
 
 [![CI](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml/badge.svg)](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml)
-[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
-[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
+[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
+[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
 [![Zero Dependencies](https://img.shields.io/badge/dependencies-zero-brightgreen.svg)](#main-features)
 
@@ -39,7 +39,7 @@ logger = GetLogger(__name__)
 logger.info("Application started")
 ```
 
-`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_2026-04-03.log`.
+`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_20260403.log`.
 
 Typical text output:
 
@@ -95,7 +95,7 @@ The "Safe" in the name is a design stance that runs across several dimensions of
 
 - **Startup safety:** invalid settings, inconsistent options, and unwritable destinations fail during setup. D-SafeLogger stops a broken logging configuration before the application starts doing real work, instead of silently degrading later.
 - **File safety:** the routing layer opens the next destination instead of renaming or truncating the active log file, which avoids the common Windows failure mode where active log files cannot be renamed. It also avoids the POSIX case where a successful rename leaves the writer appending to the previous file. Routed files can be paired with SHA-256 sidecars and an optional manifest, so log content is verifiable after the fact.
-- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Sensitive-keyword masking is applied on the Writer side using the keyword set established at configure time.
+- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Diagnostic local-variable snapshots and Writer-side formatting use the sensitive-keyword set established at configure time.
 - **Operational control:** environment variables provide explicit runtime overrides for diagnostics, routing, hashing, log levels, and queue/timeout behavior without rebuilding or editing application code.
 - **Concurrency and multiprocess safety:** multiprocess workers do not open the shared log files themselves. A parent-side Writer owns the sinks and accepts records over IPC, with bounded queues and explicit timeouts that keep the host process from unbounded waits.
 - **Failure observability:** when records cannot be delivered, the runtime classifies the outcome where it can: `KnownRejected`, `KnownDropped`, or `UnexplainedLost`. Counters and shutdown summaries make abnormal scenarios describable rather than silent.

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/README_ja.md

@@ -39,7 +39,7 @@ logger = GetLogger(__name__)
 logger.info("Application started")
 ```
 
-`pg_name` はルーティング後のログファイル名の接頭辞として使われるアプリケーション名です。例: `MyApp_2026-04-03.log`。
+`pg_name` はルーティング後のログファイル名の接頭辞として使われるアプリケーション名です。例: `MyApp_20260403.log`。
 
 標準的なテキスト出力:
 
@@ -95,7 +95,7 @@ D-SafeLogger は、active file を後から動かす代わりに、書き込み
 
 - **起動時の安全性:** 不正な設定値、矛盾したオプション、書き込めない出力先は初期化時に失敗させます。壊れた logging 構成のまま本体処理を進め、あとから気づきにくい形で動作が崩れることを防ぎます。
 - **ファイルの安全性:** ルーティング層は使用中のログファイルを rename / truncate せず、境界到達時に次の出力先を開きます。これにより active file の rename に起因する Windows 特有の失敗を避けられます。また、POSIX 系で rename が成功しても writer が古い file descriptor へ書き続ける、という世代切り替えのずれも避けられます。ルーティング済みファイルには SHA-256 sidecar と、任意で manifest を付けられ、後からログ内容を検証できます。
-- **レコードとコンテキストの安全性:** request ID、user ID、job ID といったコンテキストは producer 側で snapshot を取り、listener / Writer 側では live な `contextvars` を参照しません。機微情報のマスキングは Writer 側で、初期化時に確定したキーワード集合を使って適用されます。
+- **レコードとコンテキストの安全性:** request ID、user ID、job ID といったコンテキストは producer 側で snapshot を取り、listener / Writer 側では live な `contextvars` を参照しません。diagnostic のローカル変数 snapshot と Writer 側 formatting では、初期化時に確定した機微情報キーワード集合を使います。
 - **運用時の制御:** 環境変数により、コードを変更せずにログレベル、診断出力、モジュール別ルーティング、ハッシュ出力、キュー / timeout 設定を切り替えられます。
 - **並行性とマルチプロセスの安全性:** マルチプロセスでは worker が共有のログファイルを直接開かず、親プロセス側の Writer が出力先を所有して IPC でレコードを受け付けます。上限つきのキューと明示的な timeout により、本体プロセスが無制限に待ち続けることを避けます。
 - **失敗の可観測性:** 配送できなかったレコードは、可能な範囲で `KnownRejected` / `KnownDropped` / `UnexplainedLost` のいずれかに分類されます。カウンタと shutdown summary により、異常時のふるまいが見落とされず、説明できる形で残ります。

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/TESTING.md

@@ -29,8 +29,8 @@ uv run pytest tests -v
 Current v23j local validation on Python 3.14.3 / Windows:
 
 ```text
-651 passed, 3 skipped
-654 collected
+658 passed, 3 skipped
+661 collected
 ```
 
 The collected test count is the baseline. The number of skipped tests is platform-dependent because fork E2E tests are POSIX-only and Windows spawn E2E tests are Windows-only.
@@ -38,9 +38,9 @@ The collected test count is the baseline. The number of skipped tests is platfor
 Latest coverage validation generated `coverage.xml` on Python 3.14.3 / Windows:
 
 ```text
-TOTAL coverage: 86%
-XML line-rate: 88.49%
-XML branch-rate: 80.53%
+TOTAL coverage: 87%
+XML line-rate: 88.97%
+XML branch-rate: 81.46%
 ```
 
 `coverage.xml`, `.coverage`, and `htmlcov/` are local/CI artifacts and are ignored by `.gitignore`; keep coverage numbers in documentation, not the generated files themselves.
@@ -112,6 +112,11 @@ uv run pytest tests -x -v
 | Third-party coexistence | `tests/test_opentelemetry.py`, `tests/test_structlog.py` |
 | Branch coverage edge cases | `tests/test_coverage_boost.py` |
 
+Diagnostic masking coverage must include both component-level checks and
+configuration-to-output integration checks. In particular, `sens_kws` and
+`sens_kws_replace` are validated through `ConfigureLogger(...)` into sync text,
+sync structured, async snapshot, and multiprocess Writer formatter paths.
+
 ## Multiprocess Test Policy
 
 `dsafelogger.mp` is a formal feature, so multiprocess tests are part of the standard quality gate.

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/api/dsafelogger__async.md

@@ -33,5 +33,6 @@ Public methods:
 
 | Name | Type | Value |
 |---|---|---|
+| `MASK_STRING` | `str` | `'*** MASKED ***'` |
 | `QUEUE_DRAIN_TIMEOUT_SEC` | `float` | `10.0` |
 | `WORKER_JOIN_TIMEOUT_SEC` | `float` | `5.0` |

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/api/dsafelogger__pipeline.md

@@ -41,6 +41,6 @@ Public methods:
 
 - `build(self, config: 'ResolvedConfig') -> 'Pipeline'`
 
-### `ResolvedConfig(pg_name: 'str', log_dir: 'Path', file_fmt: 'str | FormatterConfigDict | logging.Formatter', console_fmt: 'str | FormatterConfigDict | logging.Formatter', routing_mode: 'str', routing_kwargs: 'dict', backup_count: 'int', archive_mode: 'bool', enable_hash: 'bool', manifest_path: 'Path | None', encoding: 'str', diagnose: 'bool', max_level: 'str', console: 'bool', is_async: 'bool', queue_size: 'int', log_level: 'str', color_stream: 'bool', module_configs: 'dict[str, dict]', color_overrides: 'dict[str, str]') -> None`
+### `ResolvedConfig(pg_name: 'str', log_dir: 'Path', file_fmt: 'str | FormatterConfigDict | logging.Formatter', console_fmt: 'str | FormatterConfigDict | logging.Formatter', routing_mode: 'str', routing_kwargs: 'dict', backup_count: 'int', archive_mode: 'bool', enable_hash: 'bool', manifest_path: 'Path | None', encoding: 'str', diagnose: 'bool', max_level: 'str', console: 'bool', is_async: 'bool', queue_size: 'int', log_level: 'str', color_stream: 'bool', module_configs: 'dict[str, dict]', color_overrides: 'dict[str, str]', sensitive_keywords: 'frozenset[str]' = <factory>) -> None`
 
 Resolved configuration holding 3-layer merged parameters.

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D-SafeLogger_DetailedDesign_v23j.md

@@ -3328,10 +3328,10 @@ def write_sidecar(file_path: Path, hash_value: str | None = None) -> None:
 #### 出力例
 
 ```
-a1b2c3d4e5f6789...（64文字の16進SHA-256ハッシュ）  MyApp_2026-03-28.log
+a1b2c3d4e5f6789...（64文字の16進SHA-256ハッシュ）  MyApp_20260328.log
 ```
 
-検証: `sha256sum -c MyApp_2026-03-28.log.sha256`
+検証: `sha256sum -c MyApp_20260328.log.sha256`
 
 ### 15.4. マニフェスト追記
 
@@ -3365,8 +3365,8 @@ def append_manifest(file_path: Path, manifest_path: Path, hash_value: str | None
 #### マニフェストフォーマット
 
 ```
-[2026-03-28T23:59:59.123] a1b2c3d4e5f6789...  MyApp_2026-03-28.log
-[2026-03-29T23:59:59.456] b2c3d4e5f6789a1...  MyApp_2026-03-29.log
+[2026-03-28T23:59:59.123] a1b2c3d4e5f6789...  MyApp_20260328.log
+[2026-03-29T23:59:59.456] b2c3d4e5f6789a1...  MyApp_20260329.log
 ```
 
 各行の構成: `[ISO8601タイムスタンプ] SHA-256ハッシュ  ファイル名`

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D-SafeLogger_TestDesign_v23j.md

@@ -9,7 +9,7 @@
 
 v23j は v23h 完了時点の OSS 公開前指摘に対するユーザー判断に基づく修正版である。v23h の sink 分類 / bounded shutdown 契約を維持した上で、公開前のテスト契約・MP E2E・benchmark 成果物管理を確定する。
 
-現行 v23j のローカル検証 baseline は Python 3.14.3 / Windows 上の結果であり、`651 passed, 3 skipped`（`654` collected）である。fork E2E は POSIX-only、Windows spawn E2E は Windows-only であるため、OS によって skipped 数は変動し得る。
+現行 v23j のローカル検証 baseline は Python 3.14.3 / Windows 上の結果であり、`658 passed, 3 skipped`（`661` collected）である。fork E2E は POSIX-only、Windows spawn E2E は Windows-only であるため、OS によって skipped 数は変動し得る。
 
 v23h から継続する動作変更は以下:
 

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D-SafeLogger_v23j_WhitePaper.md

@@ -16,7 +16,7 @@ SPDX-FileCopyrightText: 2026 D-SafeLogger contributors
 | 文書バージョン | 1.0 |
 | 発行日 | 2026-05-09 |
 | 対象ライブラリ | **D-SafeLogger v23j** |
-| pyproject バージョン | `0.2.0` |
+| pyproject バージョン | `0.2.1` |
 | import name | `dsafelogger` |
 | distribution name | `d-safelogger` |
 | ライセンス | Apache License 2.0 |
@@ -879,7 +879,7 @@ class DSafeFormatter(logging.Formatter):
 
 - `is_async=True` 時、producer thread 側で:
   - `contextualize()` 情報を `LogRecord` の private 属性 (`_ds_context`) へスナップショット
-  - `diagnose=True` かつ `exc_info` ありの場合のみ `f_locals` を repr 済みスナップショットに変換 (`_ds_diag_frames`)
+  - `diagnose=True` かつ `exc_info` ありの場合のみ `f_locals` をマスク済み repr スナップショットに変換 (`_ds_diag_frames`)
 - consumer thread 側では live `contextvars` を参照せず、producer 側 snapshot を優先使用。
 - multiprocess 版で `is_async=True` を併用すると **process-local async queue + multiprocess log queue + Writer dispatch の二重キューイング**になる。通常は `is_async=False` で十分。
 
@@ -1390,7 +1390,7 @@ bounded wait → visible reject/drop → process survives
 
 free-threaded build では実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。
 
-- queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全な repr 済みスナップショット**に変換。
+- queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全なマスク済み repr スナップショット**に変換。
 - consumer thread 側では live 参照を行わない。
 
 フォールバック規則:
@@ -2370,7 +2370,7 @@ auth, credential, private, cert
 
 設計書 §9.4:
 
-> free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全な repr 済みスナップショット**に変換し、consumer thread 側では live 参照を行わない。
+> free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全なマスク済み repr スナップショット**に変換し、consumer thread 側では live 参照を行わない。
 > （`docs/design/D_SafeLogger_Specification_v23j_full.md` §9.4）
 
 これは security ではなく correctness の話であるが、free-threaded 環境での「他スレッドの内部状態を意図せず参照してしまう」事故を排除する点で、データ漏洩リスクの低減にもつながる。
@@ -3467,7 +3467,7 @@ class QueueTransport(Transport):
 producer thread 側の責務:
 
 - `contextualize()` 情報を `LogRecord` の private 属性 `_ds_context` へ snapshot
-- `diagnose=True` かつ `exc_info` ありの場合のみ `f_locals` を repr 済みスナップショットに変換 (`_ds_diag_frames`)
+- `diagnose=True` かつ `exc_info` ありの場合のみ `f_locals` をマスク済み repr スナップショットに変換 (`_ds_diag_frames`)
 - 通常ログでは copy + context snapshot の軽量 hand-off
 
 #### 5.9.3 安全な終了の保証レベル
@@ -4099,7 +4099,7 @@ dsafelogger init | less
 
 設計書 §9.4:
 
-> free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全な repr 済みスナップショット**に変換し、consumer thread 側では live 参照を行わない。
+> free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を**安全なマスク済み repr スナップショット**に変換し、consumer thread 側では live 参照を行わない。
 > （`docs/design/D_SafeLogger_Specification_v23j_full.md` §9.4）
 
 #### 5.19.4 thread 境界の意味論
@@ -5103,7 +5103,7 @@ stdlib 公式が明文化している Windows rename 不能制約に対し、Pyt
 - 設計書 §1 が **Python 3.13 以上の free-threaded build を設計対象として明示**
 - 設計書 §2 が「`_configure_state`、`_active_pipeline`、`_active_workers`、`_custom_levels` 等の共有状態は GIL の存在を前提にせず、明示ロックにより保護する。`list` / `dict` の実装依存の原子性には依存しない」と規定
 - v21 改訂で `ConfigureLogger` 全体を `_lifecycle_lock` 保持下で実行
-- `f_locals` を producer thread で repr 済みスナップショットに変換（cross-thread 安全性、§9.4）
+- `f_locals` を producer thread でマスク済み repr スナップショットに変換（cross-thread 安全性、§9.4）
 - 内部 thread を空 `Context` で開始（§9.5）
 - ピュア Python 実装のため `Py_mod_gil` 対応（C 拡張側）の必要がない
 - `TESTING.md` に手動テスト手順掲載: `PYTHON_GIL=0 uvx --python cpython-3.13+freethreaded --from pytest pytest tests -v`
@@ -5338,9 +5338,9 @@ OSS 公開時には国内・海外双方からの評価軸が想定されるが
 #### 7.10.3 品質ゲート
 
 `TESTING.md` および公開検証手順:
-- v23j ローカル検証（Python 3.14.3 / Windows）: **651 passed, 3 skipped**（収集 654、`uv run pytest tests -v`）
+- v23j ローカル検証（Python 3.14.3 / Windows）: **658 passed, 3 skipped**（収集 661、`uv run pytest tests -v`）
 - skipped 数は OS 依存。fork E2E は POSIX-only、Windows spawn E2E は Windows-only であるため、CI matrix では OS により skipped 数が変動し得る。
-- カバレッジ: terminal total **86%**, XML line-rate **88.49%**, branch-rate **80.53%**
+- カバレッジ: terminal total **87%**, XML line-rate **88.97%**, branch-rate **81.46%**
 - multiprocess tests / OTel・structlog coexistence tests は公式品質ゲートに含まれる
 - free-threaded build テスト: `PYTHON_GIL=0 uvx --python cpython-3.13+freethreaded --from pytest pytest tests -v`
 
@@ -5355,7 +5355,7 @@ OSS 公開時には国内・海外双方からの評価軸が想定されるが
 #### 7.10.5 公開時のステータス
 
 公開前レビュー記録（2026-05-07 時点）:
-- 現在の公開対象バージョン: `0.2.0`
+- 現在の公開対象バージョン: `0.2.1`
 - 最新公開前レビュー結果: **GO-with-fixes**
 - 公開前必須修正項目（release blocker）が列挙されている
 
@@ -5429,7 +5429,7 @@ OSS 公開時には国内・海外双方からの評価軸が想定されるが
 7. **海外特有の対応軸**: PEP 703 先行採用傾向 / Executive Order 14028 等のサプライチェーン規制要件化 / 大規模 microservice 構成。examples / README 英語、Apache 2.0、依存ゼロが対応軸。
 8. **設計目的が「広範な普及」ではないことが明示されている**: 設計書 §1 は「広く普及させる目的よりも、D エコシステムの共通基盤として運用することを最優先とする」と宣言しており、これは「特定の運用要件を持つ subset 組織への対応」を優先する設計姿勢として観測される。
 9. **失敗の境界を能動的に明示するドキュメント運用**: `examples/08_compliance_audit.md` の Threat Model / `examples/12_multiprocess_logging.md` の Writer does not guarantee / `BENCHMARK.md` の What Not To Claim / 設計書 §7.6.7 の HMAC スコープ外宣言 / 設計書 §11.2 の remote aggregation スコープ外宣言。これらは「過剰な期待による誤用を防ぐ」運用姿勢として一貫している。
-10. **品質ゲートの透明性**: Python 3.14.3 / Windows で 651 passed / 3 skipped（収集 654）、カバレッジ 86% terminal、free-threaded build テスト手順、`scripts/check_design_docs_sync.py` および `scripts/generate_api_docs.py --check` による内部同期検証、`benchmarks/summary/manifest.json` による benchmark セッション固定。skipped 数は OS 依存で変動し得る。これらは導入候補ライブラリの評価時に観測可能な品質指標として記録される。
+10. **品質ゲートの透明性**: Python 3.14.3 / Windows で 658 passed / 3 skipped（収集 661）、カバレッジ 87% terminal、free-threaded build テスト手順、`scripts/check_design_docs_sync.py` および `scripts/generate_api_docs.py --check` による内部同期検証、`benchmarks/summary/manifest.json` による benchmark セッション固定。skipped 数は OS 依存で変動し得る。これらは導入候補ライブラリの評価時に観測可能な品質指標として記録される。
 11. **配布構造の明確性**: wheel は runtime package files のみを含み、`py.typed` を同梱する。sdist は公開検証・再現性のため docs / examples / tests / benchmark summaries / selected benchmark summaries を含む。private planning materials と一時作業ファイルは含めない。
 12. **competitor との関係は競合ではなく責務分離**: structlog（フロントエンド）/ OTel（emission）/ Loguru（DX 置換）/ picologging（速度差別化）/ logfire（SaaS）/ Eliot（causal）は責務軸が異なり、本ライブラリと共存または並走する。Loguru の 23.9k stars 等の知名度指標は、設計軸の比較とは独立した文脈である。
 
@@ -5706,8 +5706,8 @@ structlog（2 パターンで共存）/ OpenTelemetry Python（`LoggingHandler`
 
 #### 8.7.2 品質ゲート
 
-- 公式テスト baseline: 651 passed / 3 skipped（収集 654、`uv run pytest tests -v`、Python 3.14.3 / Windows）。fork E2E は POSIX-only、Windows spawn E2E は Windows-only のため、OS によって skipped 数は変動し得る。
-- カバレッジ: terminal total 86%, XML line-rate 88.49%, branch-rate 80.53%
+- 公式テスト baseline: 658 passed / 3 skipped（収集 661、`uv run pytest tests -v`、Python 3.14.3 / Windows）。fork E2E は POSIX-only、Windows spawn E2E は Windows-only のため、OS によって skipped 数は変動し得る。
+- カバレッジ: terminal total 87%, XML line-rate 88.97%, branch-rate 81.46%
 - multiprocess tests / OTel・structlog coexistence tests は公式品質ゲートに含まれる
 - free-threaded build テスト手順あり（`PYTHON_GIL=0 uvx ...`）
 - `scripts/check_design_docs_sync.py` および `scripts/generate_api_docs.py --check` で内部同期検証
@@ -5722,7 +5722,7 @@ structlog（2 パターンで共存）/ OpenTelemetry Python（`LoggingHandler`
 - `pyproject.toml` ランタイム依存ゼロ
 - wheel は `src/dsafelogger/` 配下の runtime package files のみを含む
 - sdist は docs / examples / tests / benchmark summaries / selected benchmark summaries を含み、private planning materials と一時作業ファイルは含めない
-- 公開対象バージョン: 0.2.0
+- 公開対象バージョン: 0.2.1
 - 最新公開前レビュー結果: GO-with-fixes（2026-05-07）
 
 #### 8.7.4 ドキュメント運用の特徴
@@ -5760,7 +5760,7 @@ structlog（2 パターンで共存）/ OpenTelemetry Python（`LoggingHandler`
 
 #### 8.8.5 ドキュメント・品質運用
 
-10. **品質ゲートと内部同期検証スクリプトが整備されている**: Python 3.14.3 / Windows で 651 passed / 3 skipped（収集 654）、カバレッジ 86%、`scripts/check_design_docs_sync.py` と `scripts/generate_api_docs.py --check` による内部同期検証、`benchmarks/summary/manifest.json` による公開代表セッション固定。skipped 数は OS 依存で変動し得る。これらは導入候補ライブラリの評価時に観測可能な品質指標として記録される。
+10. **品質ゲートと内部同期検証スクリプトが整備されている**: Python 3.14.3 / Windows で 658 passed / 3 skipped（収集 661）、カバレッジ 87%、`scripts/check_design_docs_sync.py` と `scripts/generate_api_docs.py --check` による内部同期検証、`benchmarks/summary/manifest.json` による公開代表セッション固定。skipped 数は OS 依存で変動し得る。これらは導入候補ライブラリの評価時に観測可能な品質指標として記録される。
 
 ---
 

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D-SafeLogger_v23j_WhitePaper_en.md

@@ -15,7 +15,7 @@ SPDX-FileCopyrightText: 2026 D-SafeLogger contributors
 | Document version | 1.0 |
 | Publication date | 2026-05-09 |
 | Target library | **D-SafeLogger v23j** |
-| pyproject version | `0.2.0` |
+| pyproject version | `0.2.1` |
 | import name | `dsafelogger` |
 | distribution name | `d-safelogger` |
 | License | Apache License 2.0 |
@@ -242,7 +242,7 @@ The `README.md` Overview section organizes Safe into six operational dimensions.
 |---|---|
 | **Startup safety** | Invalid settings, inconsistent options, and non-writable destinations cause **setup to fail** instead of letting the application enter production with broken logging. |
 | **File safety** | The routing layer does not rename or truncate the active log file. It switches output by opening the next append-only destination, avoiding common Windows rename failures and POSIX-style stale-FD failure modes while enabling post-mortem verification with SHA-256 sidecars and manifests. |
-| **Record / context safety** | Request ID, user ID, job ID, and similar context are **snapshotted** at producer-side hand-off. Listeners and writers do not depend on live `contextvars`, and sensitive keyword masking is applied on the Writer side. |
+| **Record / context safety** | Request ID, user ID, job ID, and similar context are **snapshotted** at producer-side hand-off. Listeners and writers do not depend on live `contextvars`; diagnostic snapshots and Writer-side formatting use the sensitive-keyword set established at configure time. |
 | **Operational control** | Diagnostics, routing, hashing, log level, queue/timeout can be overwritten with environment variables without rebuilding or editing. |
 | **Concurrency / multiprocess safety** | Multiprocess workers **do not open shared log files directly**. The parent-side Writer owns the sink and accepts records via IPC. Bounded queues and explicit timeouts prevent the host process from waiting indefinitely. |
 | **Failure observability** | Delivery failures are **classified** as `KnownRejected` / `KnownDropped` / `UnexplainedLost`, making log loss describable as counters and shutdown summaries instead of invisible gaps in a file. |
@@ -696,7 +696,7 @@ Detailed design document §15a stipulates. The Transport layer is an abstraction
 Design documents §9.3 and §11.17:
 - At `is_async=True`, on the producer thread side:
   - Snapshot `contextualize()` information to private attribute (`_ds_context`) of `LogRecord`
-  - Convert `f_locals` to repr-converted snapshot only when `diagnose=True` and `exc_info` are present (`_ds_diag_frames`)
+  - Convert `f_locals` to a masked, repr-converted snapshot only when `diagnose=True` and `exc_info` are present (`_ds_diag_frames`)
 - The consumer thread side does not refer to live `contextvars`, and preferentially uses the producer side snapshot.
 - When used with `is_async=True` in the multiprocess version, it becomes double queuing of **process-local async queue + multiprocess log queue + Writer dispatch**. Usually `is_async=False` is sufficient.
 #### 2.5.3 No-Copy Snapshot (FrozenContext)
@@ -1061,7 +1061,7 @@ The design document clearly states the policy of not relying on the existence of
 - The independent `self._lock` of `AppendOnlyFileHandler` has been abolished in v21 and unified to the `acquire()` / `release()` API of the parent class `logging.Handler` (double lock overhead eliminated).
 #### 2.10.2 cross-thread safety (§9.4)
 In a free-threaded build, a `f_locals` live reference to the frame of another running thread is unsafe.
-- If a cross-queue hand-off occurs, traceback and `f_locals` are converted to a **safe repr-converted snapshot** on the producer thread side.
+- If a cross-queue hand-off occurs, traceback and `f_locals` are converted to a **safe masked, repr-converted snapshot** on the producer thread side.
 - No live references are made on the consumer thread side.
 Fallback rules:
 1. Use queue hand-off diagnostic snapshot if available
@@ -1620,7 +1620,7 @@ These points are revisited in the next chapter, “4. Security,” as safety asp
 |---|---|
 | **Startup safety** | Reject invalid settings and unwritable paths during setup → Structurally eliminate situations where "it appears to be working with broken settings" |
 | **File safety** | Do not rename/truncate → Structurally avoids the problem of not being able to rename Windows active logs + Post-mortem verification possible with SHA-256 sidecar |
-| **Record/context safety** | Eliminate dependence on snapshot → live `contextvars` during hand-off on the producer side. Apply sens_kws masking on Writer side |
+| **Record/context safety** | Eliminate dependence on snapshot → live `contextvars` during hand-off on the producer side. Apply `sens_kws` masking during diagnostic snapshots and Writer-side formatting |
 | **Operational control** | Diagnostics, routing, and hashes can be overwritten with environment variables without rebuilding |
 | **Concurrency/multiprocess safety** | Workers do not directly open shared log files, and the parent Writer owns the sink. Eliminate unlimited waiting for host process with bounded queue + explicit timeout |
 | **Failure observability** | Classify delivery failures by `KnownRejected` / `KnownDropped` / `UnexplainedLost` → silent loss is not allowed structurally |
@@ -1774,7 +1774,7 @@ Design document §9.4:
 This functions as an indirect defense against ``attacker-induced `__repr__` exceptions'' and ``log size attacks using large objects.''
 #### 4.4.8 cross-thread safety
 Design document §9.4:
-> In a free-threaded build, a `f_locals` live reference to frame of another running thread is unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to a **safe repr-converted snapshot** on the producer thread side, and no live reference is made on the consumer thread side.
+> In a free-threaded build, a `f_locals` live reference to frame of another running thread is unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to a **safe masked, repr-converted snapshot** on the producer thread side, and no live reference is made on the consumer thread side.
 > (`docs/design/D_SafeLogger_Specification_v23j_full.md` §9.4)
 Although this is a matter of correctness rather than security, it also helps reduce the risk of data leakage by eliminating accidents in which the internal state of other threads is unintentionally referenced in a free-threaded environment.
 ---
@@ -2537,7 +2537,7 @@ Design document §9.3:
 - Reason: To separate stdlib differences between Python 3.11 / 3.13 / 3.14 from semantics.
 Responsibilities on the producer thread side:
 - `contextualize()` information to private attribute `_ds_context` of `LogRecord` snapshot
-- Convert `f_locals` to repr-converted snapshot only if `diagnose=True` and `exc_info` are present (`_ds_diag_frames`)
+- Convert `f_locals` to a masked, repr-converted snapshot only if `diagnose=True` and `exc_info` are present (`_ds_diag_frames`)
 - Lightweight hand-off of copy + context snapshot in normal logs
 #### 5.9.3 Safe Termination Assurance Level
 Design document §9.3:
@@ -2982,7 +2982,7 @@ Design document §2 / §9.2:
 > (`docs/design/D_SafeLogger_Specification_v23j_full.md` §2)
 #### 5.19.3 cross-thread safety
 Design document §9.4:
-> In a free-threaded build, a `f_locals` live reference to frame of another running thread is unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to a **safe repr-converted snapshot** on the producer thread side, and no live reference is made on the consumer thread side.
+> In a free-threaded build, a `f_locals` live reference to frame of another running thread is unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to a **safe masked, repr-converted snapshot** on the producer thread side, and no live reference is made on the consumer thread side.
 > (`docs/design/D_SafeLogger_Specification_v23j_full.md` §9.4)
 #### 5.19.4 thread boundary semantics
 Design document §9.5:
@@ -2995,7 +2995,7 @@ Design document §2 v21 revised:
 - Eliminate double lock overhead by abolishing the independent `self._lock` of `AppendOnlyFileHandler` and unifying it with the lock API (`self.acquire()/release()`) of the parent class `logging.Handler`
 #### 5.19.6 Functional Observation
 - Structurally eliminates shared state destruction in free-threaded builds by using an explicit locking design that does not assume the existence of GIL.
-- Eliminate the risk of live reference to other thread frames in free-threaded build by making `f_locals` a repr-converted snapshot.
+- Eliminate the risk of live reference to other thread frames in free-threaded build by making `f_locals` a masked, repr-converted snapshot.
 - Eliminate accidents where the user's request context is leaked to the internal thread by starting an empty Context in the internal thread.
 ---
 
@@ -3026,7 +3026,7 @@ formatter falls back in the following order:
 #### 5.20.6 Functional Observation
 - The diagnostic function is positioned as an ``operation tool during production'' rather than a ``useful function during development'' (only environment variables enabled, intentional operation by the operator).
 - repr Defensive implementation against failures and large objects.
-- Maintains safety in a free-threaded environment because hand-off is performed using repr-converted snapshots even via async/multiprocess.
+- Maintains safety in a free-threaded environment because hand-off is performed using masked, repr-converted snapshots even via async/multiprocess.
 ---
 
 ### 5.21 sens_kws masking
@@ -3868,9 +3868,9 @@ Facts that can be confirmed from `pyproject.toml` and `MANIFEST.in`:
 - Operation guide: `TESTING.md` / `BENCHMARK.md` / `CONTRIBUTING.md` / `CHANGELOG.md`
 #### 7.10.3 Quality Gate
 `TESTING.md` and public validation procedures:
-- v23j local validation on Python 3.14.3 / Windows: **651 passed, 3 skipped** (654 collected, `uv run pytest tests -v`)
+- v23j local validation on Python 3.14.3 / Windows: **658 passed, 3 skipped** (661 collected, `uv run pytest tests -v`)
 - The skipped count is platform-dependent because fork E2E tests are POSIX-only and Windows spawn E2E tests are Windows-only.
-- Coverage: terminal total **86%**, XML line-rate **88.49%**, branch-rate **80.53%**
+- Coverage: terminal total **87%**, XML line-rate **88.97%**, branch-rate **81.46%**
 - multiprocess tests / OTel/structlog coexistence tests are included in the official quality gate
 - free-threaded build test: `PYTHON_GIL=0 uvx --python cpython-3.13+freethreaded --from pytest pytest tests -v`
 #### 7.10.4 Release Management
@@ -3881,7 +3881,7 @@ Public validation procedures:
 - `BENCHMARK.md` is an interpretation of manual editing and will not be regenerated from benchmark runner
 #### 7.10.5 Status at time of publication
 Pre-publication review record (as of 2026-05-07):
-- Current release target version: `0.2.0`
+- Current release target version: `0.2.1`
 - Latest pre-release review results: **GO-with-fixes**
 - Required correction items before publication (release blockers) are listed
 Although these indicate the existence of a ``prepared release operation flow,'' this chapter does not consider this to be the ``reason for popularity,'' but only records it as the technical structure of the distribution.
@@ -3935,7 +3935,7 @@ The observed facts referenced in this chapter can be summarized as follows. This
 7. **International-market perspective**: Trend in early adoption of PEP 703 / Supply chain regulatory requirements such as Executive Order 14028 / Large-scale microservice configuration. examples / README English, Apache 2.0, zero dependencies are supported.
 8. **The design purpose is clearly not “wide dissemination”**: §1 of the design document declares that the “top priority is to operate as a common foundation for the D ecosystem rather than to pursue wide adoption.” This reflects a design stance that prioritizes fit for organizations with specific operational requirements.
 9. **Document operations that actively specify failure boundaries**: Threat Model of `examples/08_compliance_audit.md` / Writer does not guarantee of `examples/12_multiprocess_logging.md` / What Not To Claim of `BENCHMARK.md` / HMAC out-of-scope declaration in design document §7.6.7 / remote aggregation out-of-scope declaration in design document §11.2. These are consistent with the operational stance of ``preventing misuse due to excessive expectations.''
-10. **Quality gate transparency**: 651 passed / 3 skipped on Python 3.14.3 / Windows (654 collected), coverage 86% terminal, free-threaded build test procedure, internal synchronization verification by `scripts/check_design_docs_sync.py` and `scripts/generate_api_docs.py --check`, benchmark session fixation by `benchmarks/summary/manifest.json`. The skipped count can vary by OS. These are recorded as observable quality indicators when evaluating candidate libraries for introduction.
+10. **Quality gate transparency**: 658 passed / 3 skipped on Python 3.14.3 / Windows (661 collected), coverage 87% terminal, free-threaded build test procedure, internal synchronization verification by `scripts/check_design_docs_sync.py` and `scripts/generate_api_docs.py --check`, benchmark session fixation by `benchmarks/summary/manifest.json`. The skipped count can vary by OS. These are recorded as observable quality indicators when evaluating candidate libraries for introduction.
 11. **Clear distribution structure**: The wheel contains runtime package files only and includes `py.typed`. The sdist includes docs / examples / tests / benchmark summaries / selected benchmark summaries for public validation and reproducibility. Private planning materials and temporary working files are excluded.
 12. **Relationship with competitors is not competition but separation of responsibilities**: structlog (front end) / OTel (emission) / Loguru (DX replacement) / picologging (speed differentiation) / logfire (SaaS) / Eliot (causal) have different responsibility axes and coexist or run in parallel with this library. Popularity indicators such as Loguru's 23.9k stars are a context independent of comparison of design dimensions.
 ---
@@ -4131,8 +4131,8 @@ Boundaries that public bench analysis actively enumerates:
 | API | `docs/api/dsafelogger*.md` | Automatically generated |
 | Operation | `TESTING.md` / `BENCHMARK.md` / `CONTRIBUTING.md` / `CHANGELOG.md` | — |
 #### 8.7.2 Quality Gate
-- Official test baseline: 651 passed / 3 skipped (654 collected, `uv run pytest tests -v`, Python 3.14.3 / Windows). The skipped count can vary by OS because fork E2E tests are POSIX-only and Windows spawn E2E tests are Windows-only.
-- Coverage: terminal total 86%, XML line-rate 88.49%, branch-rate 80.53%
+- Official test baseline: 658 passed / 3 skipped (661 collected, `uv run pytest tests -v`, Python 3.14.3 / Windows). The skipped count can vary by OS because fork E2E tests are POSIX-only and Windows spawn E2E tests are Windows-only.
+- Coverage: terminal total 87%, XML line-rate 88.97%, branch-rate 81.46%
 - multiprocess tests / OTel/structlog coexistence tests are included in the official quality gate
 - free-threaded build test procedure included (`PYTHON_GIL=0 uvx ...`)
 - Internal synchronization verification on `scripts/check_design_docs_sync.py` and `scripts/generate_api_docs.py --check`
@@ -4144,7 +4144,7 @@ Boundaries that public bench analysis actively enumerates:
 - `py.typed` included
 - `pyproject.toml` Zero runtime dependency
 - Distribution target is only under `src/dsafelogger/`
-- Release target version: 0.2.0
+- Release target version: 0.2.1
 - Latest pre-release review results: GO-with-fixes (2026-05-07)
 #### 8.7.4 Characteristics of document operation
 - Multilingual: README in Japanese and English, design document in Japanese, examples/API/operation guide in English
@@ -4168,7 +4168,7 @@ Based on the observed facts in this report as a whole, we have summarized the ob
 8. **Multiprocess raw throughput is led by stdlib logging**: In `root_p8`, D-SafeLogger reaches 63-75% of stdlib throughput. `BENCHMARK.md` states this clearly as a design tradeoff reflecting fixed costs in the specification (IPC + Writer dispatch). The multiprocess value of this library is not raw throughput, but delivery-state observability.
 9. **Classifies and explains 12/12 rows in the multiprocess resilience profile**: stdlib / loguru rows are marked with `observability_gap`. Delivery-state explainability is recorded as observability specific to this library.
 #### 8.8.5 Documentation/quality operations
-10. **Quality gate and internal synchronization verification scripts are in place**: 651 passed / 3 skipped on Python 3.14.3 / Windows (654 collected), coverage 86%, internal synchronization verification by `scripts/check_design_docs_sync.py` and `scripts/generate_api_docs.py --check`, public representative session fixation by `benchmarks/summary/manifest.json`. The skipped count can vary by OS. These are recorded as observable quality indicators when evaluating candidate libraries for introduction.
+10. **Quality gate and internal synchronization verification scripts are in place**: 658 passed / 3 skipped on Python 3.14.3 / Windows (661 collected), coverage 87%, internal synchronization verification by `scripts/check_design_docs_sync.py` and `scripts/generate_api_docs.py --check`, public representative session fixation by `benchmarks/summary/manifest.json`. The skipped count can vary by OS. These are recorded as observable quality indicators when evaluating candidate libraries for introduction.
 ---
 
 ### 8.9 Summary of this chapter
@@ -4336,7 +4336,7 @@ We will organize the D-SafeLogger-specific terms that frequently appear in this
 |---|---|
 | **Startup safety** | Reject invalid settings and unwritable paths during setup |
 | **File safety** | append-only routing without rename/truncate + SHA-256 sidecar |
-| **Record / context safety** | Snapshot at hand-off on the producer side, sens_kws masking on the Writer side |
+| **Record / context safety** | Snapshot at hand-off on the producer side; `sens_kws` masking during diagnostic snapshots and Writer-side formatting |
 | **Operational control** | Overriding without rebuilding with environment variables |
 | **Concurrency / multiprocess safety** | parent-side Writer owns sink, bounded queue + explicit timeout |
 | **Failure observability** | Classification of `KnownRejected` / `KnownDropped` / `UnexplainedLost` |

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D_SafeLogger_Specification_v23j_full.md

@@ -1103,7 +1103,7 @@ INIファイルで指定されたモジュール別の `path` についても同
   * **`sens_kws_replace` による完全置換**: `sens_kws_replace=True` を明示的に指定した場合、ビルトインキーワードは破棄され、`sens_kws` で指定したキーワード**のみ**がマスキング対象となる。これにより「ビルトインの `key` が広すぎてマスキングされすぎる」等の問題をユーザーが完全に制御可能。
   * **マッチング**: 変数名に対する部分一致（大文字小文字不問）で判定する。例えば `password` は `user_password`, `PASSWORD_HASH`, `my_password_field` のいずれにもマッチする。
 * **巨大reprの抑制**: 個々のローカル変数の `repr()` は一定長で打ち切り、巨大オブジェクトや過度に冗長なデータがログを汚染しないようにする。`repr()` 自体に失敗した場合も診断ログ全体を壊さず、失敗した旨をプレースホルダとして出力する。
-* **cross-thread 安全性**: free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を安全な repr 済みスナップショットへ変換し、consumer thread 側では live 参照を行わない。
+* **cross-thread 安全性**: free-threaded build では、実行中の他 thread の frame に対する `f_locals` live 参照は unsafe である。したがって、queue を跨ぐ hand-off が発生する場合は producer thread 側で traceback と `f_locals` を安全なマスク済み repr スナップショットへ変換し、consumer thread 側では live 参照を行わない。
 * **フォールバック規則**: formatter は (1) queue hand-off 済みの診断スナップショットがあればそれを使用、(2) 同一 thread 内で `exc_info` が保持されている場合のみ live 参照を許可、(3) それ以外は standard traceback のみを出力する。
 * **[実装方針]**: diagnose の有効/無効は `ConfigureLogger` 実行時に `{prefix}_DIAGNOSE` から解決し、formatter / queue hand-off / diagnostic snapshot へ一貫して伝搬させる。heavy path は `exc_info` の存在時にのみ通す。
 

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/docs/design/D_SafeLogger_Specification_v23j_full_en.md

@@ -1095,7 +1095,7 @@ If `manifest_path` is specified, the writability of that directory is also verif
   * **Complete replacement by `sens_kws_replace`**: If `sens_kws_replace=True` is explicitly specified, the built-in keyword will be discarded and only the keyword** specified by `sens_kws` will be masked. This gives the user complete control over problems such as "built-in `key` is too wide and is masked too much".
   * **Matching**: Determine by partial match (case does not matter) for the variable name. For example, `password` matches `user_password`, `PASSWORD_HASH`, `my_password_field`.
 * **Suppressing large reprs**: Truncate `repr()` of individual local variables to a fixed length to prevent large objects or excessively redundant data from polluting the log. Even if `repr()` fails, the entire diagnostic log is not destroyed, but the failure is output as a placeholder.
-* **Cross-thread safety**: In a free-threaded build, live `f_locals` references to frames in other running threads are unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to safe repr snapshots on the producer thread side, and live references are not performed on the consumer thread side.
+* **Cross-thread safety**: In a free-threaded build, live `f_locals` references to frames in other running threads are unsafe. Therefore, when a hand-off across queues occurs, the traceback and `f_locals` are converted to safe masked repr snapshots on the producer thread side, and live references are not performed on the consumer thread side.
 * **Fallback rules**: The formatter (1) uses a queue hand-off diagnostic snapshot if available, (2) allows live references only if `exc_info` is held in the same thread, and (3) otherwise outputs only a standard traceback.
 * **[Implementation policy]**: The enable/disable of diagnose is resolved from `{prefix}_DIAGNOSE` during `ConfigureLogger` execution, and consistently propagated to formatter / queue hand-off / diagnostic snapshot. Heavy path is passed only when `exc_info` exists.
 

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/examples/06_web_api_logging.md

@@ -88,7 +88,7 @@ def handle_order(request_id: str, user: str, amount: float):
 
 ## The Output
 
-### Main Log (`./logs/OrderAPI.log`)
+### Main Log (`./logs/OrderAPI_20260403.log`)
 
 Every JSON line includes the context keys set by `contextualize()`:
 
@@ -120,7 +120,7 @@ Because the output is structured JSON, you can filter by any field:
 cat logs/*.log | jq -s 'sort_by(.timestamp) | .[] | select(.request_id == "req-a1b2")'
 
 # Show only warnings and errors
-cat logs/OrderAPI_2026-04-03.log | jq 'select(.level == "WAR" or .level == "ERR")'
+cat logs/OrderAPI_20260403.log | jq 'select(.level == "WAR" or .level == "ERR")'
 
 # List unique request IDs that hit the payment module
 cat logs/payment_alerts.log | jq -r '.request_id' | sort -u

{d_safelogger-0.2.0 → d_safelogger-0.2.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "D-SafeLogger"
-version = "0.2.0"
+version = "0.2.1"
 description = "Zero-dependency, thread-safe, append-only logging library for Python with 3-layer config pipeline"
 readme = "README.md"
 requires-python = ">=3.11"

{d_safelogger-0.2.0 → d_safelogger-0.2.1/src/D_SafeLogger.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: D-SafeLogger
-Version: 0.2.0
+Version: 0.2.1
 Summary: Zero-dependency, thread-safe, append-only logging library for Python with 3-layer config pipeline
 Author: D
 License-Expression: Apache-2.0
@@ -28,8 +28,8 @@ Dynamic: license-file
 # D-SafeLogger
 
 [![CI](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml/badge.svg)](https://github.com/nightmarewalker/D-SafeLogger/actions/workflows/ci.yml)
-[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
-[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg)](https://pypi.org/project/d-safelogger/)
+[![PyPI version](https://img.shields.io/pypi/v/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
+[![Python](https://img.shields.io/pypi/pyversions/d-safelogger.svg?cacheSeconds=3600)](https://pypi.org/project/d-safelogger/)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
 [![Zero Dependencies](https://img.shields.io/badge/dependencies-zero-brightgreen.svg)](#main-features)
 
@@ -66,7 +66,7 @@ logger = GetLogger(__name__)
 logger.info("Application started")
 ```
 
-`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_2026-04-03.log`.
+`pg_name` is the application name used as the prefix of routed log file names, for example `MyApp_20260403.log`.
 
 Typical text output:
 
@@ -122,7 +122,7 @@ The "Safe" in the name is a design stance that runs across several dimensions of
 
 - **Startup safety:** invalid settings, inconsistent options, and unwritable destinations fail during setup. D-SafeLogger stops a broken logging configuration before the application starts doing real work, instead of silently degrading later.
 - **File safety:** the routing layer opens the next destination instead of renaming or truncating the active log file, which avoids the common Windows failure mode where active log files cannot be renamed. It also avoids the POSIX case where a successful rename leaves the writer appending to the previous file. Routed files can be paired with SHA-256 sidecars and an optional manifest, so log content is verifiable after the fact.
-- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Sensitive-keyword masking is applied on the Writer side using the keyword set established at configure time.
+- **Record and context safety:** request IDs, user IDs, job IDs, and other context are snapshotted on the producer side at hand-off, so listeners and Writers do not depend on live `contextvars`. Diagnostic local-variable snapshots and Writer-side formatting use the sensitive-keyword set established at configure time.
 - **Operational control:** environment variables provide explicit runtime overrides for diagnostics, routing, hashing, log levels, and queue/timeout behavior without rebuilding or editing application code.
 - **Concurrency and multiprocess safety:** multiprocess workers do not open the shared log files themselves. A parent-side Writer owns the sinks and accepts records over IPC, with bounded queues and explicit timeouts that keep the host process from unbounded waits.
 - **Failure observability:** when records cannot be delivered, the runtime classifies the outcome where it can: `KnownRejected`, `KnownDropped`, or `UnexplainedLost`. Counters and shutdown summaries make abnormal scenarios describable rather than silent.