logging_config.py
python
sha256:7afc59b2ee9c70d18b748d269602c588385d6e936cc22b056fd775ab9f2fd499
fix: real CloudWatch alerting on both envs, add environment…
Sonnet 5
minor
⚠ breaking
1 day ago
| 1 | """Structured JSON logging for MuseHub. |
| 2 | |
| 3 | Replaces plain-text ``basicConfig`` with a JSON formatter that emits one |
| 4 | compact JSON object per log line. Every record includes: |
| 5 | |
| 6 | timestamp — ISO-8601 UTC |
| 7 | level — DEBUG / INFO / WARNING / ERROR / CRITICAL |
| 8 | logger — dotted module name |
| 9 | message — formatted message (PII scrubbed) |
| 10 | request_id — injected by AccessLogMiddleware (empty string outside requests) |
| 11 | user_id — injected by AccessLogMiddleware (empty string outside requests) |
| 12 | |
| 13 | Per-request access records also carry: |
| 14 | method / path / status / duration_ms |
| 15 | |
| 16 | Usage:: |
| 17 | |
| 18 | from musehub.logging_config import configure_logging |
| 19 | configure_logging(debug=settings.debug) |
| 20 | """ |
| 21 | |
| 22 | import json |
| 23 | import logging |
| 24 | import re |
| 25 | from contextvars import ContextVar |
| 26 | from datetime import datetime, timezone |
| 27 | from musehub.types.json_types import JSONObject |
| 28 | |
| 29 | # ── Contextvars ─────────────────────────────────────────────────────────────── |
| 30 | # Set by AccessLogMiddleware at the start of every HTTP request. |
| 31 | # Default to empty string so non-request log lines still produce valid JSON. |
| 32 | |
| 33 | request_id_var: ContextVar[str] = ContextVar("request_id", default="") |
| 34 | user_id_var: ContextVar[str] = ContextVar("user_id", default="") |
| 35 | |
| 36 | # Set once by configure_logging() — module-level rather than threaded through |
| 37 | # every log call, and deliberately not imported from musehub.config to avoid |
| 38 | # any import-order coupling between logging setup and settings loading. |
| 39 | _environment: str = "" |
| 40 | _release_version: str = "" |
| 41 | |
| 42 | # ── PII / secret scrubbing filter ───────────────────────────────────────────── |
| 43 | |
| 44 | # Pattern pairs: (compiled_regex, replacement_string) |
| 45 | # Applied in order to the fully-formatted message string. |
| 46 | _SCRUB_PATTERNS: list[tuple[re.Pattern[str], str]] = [ |
| 47 | # Authorization: Bearer <token> or bearer=<token> |
| 48 | (re.compile(r"(Bearer\s+)[A-Za-z0-9._\-/+=]{8,}", re.IGNORECASE), r"\1***"), |
| 49 | # token=<value> or token: <value> (query-string or log kv) |
| 50 | (re.compile(r"((?:api_)?token[=:\s]+)[^\s&,\"'<>]+", re.IGNORECASE), r"\1***"), |
| 51 | # password=<value> or password: <value> |
| 52 | (re.compile(r"(password[=:\s]+)[^\s&,\"'<>]+", re.IGNORECASE), r"\1***"), |
| 53 | # secret=<value> |
| 54 | (re.compile(r"(secret[=:\s]+)[^\s&,\"'<>]+", re.IGNORECASE), r"\1***"), |
| 55 | ] |
| 56 | |
| 57 | class PiiFilter(logging.Filter): |
| 58 | """Scrub secrets and tokens from the formatted log message. |
| 59 | |
| 60 | Operates on the fully-formatted message string so it catches values |
| 61 | interpolated from args as well as literal strings. |
| 62 | """ |
| 63 | |
| 64 | def filter(self, record: logging.LogRecord) -> bool: |
| 65 | # Format args into msg so we operate on the final string. |
| 66 | try: |
| 67 | msg = record.getMessage() |
| 68 | except Exception: |
| 69 | msg = str(record.msg) |
| 70 | |
| 71 | for pattern, replacement in _SCRUB_PATTERNS: |
| 72 | msg = pattern.sub(replacement, msg) |
| 73 | |
| 74 | record.msg = msg |
| 75 | record.args = () # already expanded — prevent double-formatting |
| 76 | return True |
| 77 | |
| 78 | # ── JSON formatter ───────────────────────────────────────────────────────────── |
| 79 | |
| 80 | # Standard LogRecord attributes — exclude from the "extra" pass-through so we |
| 81 | # don't double-emit them. |
| 82 | _STANDARD_ATTRS = frozenset( |
| 83 | logging.LogRecord("", 0, "", 0, "", (), None).__dict__.keys() |
| 84 | | { |
| 85 | "message", |
| 86 | "asctime", |
| 87 | "exc_text", |
| 88 | "stack_info", |
| 89 | "taskName", |
| 90 | } |
| 91 | ) |
| 92 | |
| 93 | # Per-request fields emitted by AccessLogMiddleware via `extra=`. |
| 94 | _ACCESS_FIELDS = ("method", "path", "status", "duration_ms") |
| 95 | |
| 96 | class JsonFormatter(logging.Formatter): |
| 97 | """Emit one compact JSON object per log record. |
| 98 | |
| 99 | The PiiFilter must be installed on the handler *before* this formatter |
| 100 | runs so that ``record.getMessage()`` returns a scrubbed string. |
| 101 | """ |
| 102 | |
| 103 | def format(self, record: logging.LogRecord) -> str: |
| 104 | # Let the base class populate record.exc_text from exc_info. |
| 105 | super().format(record) |
| 106 | |
| 107 | doc: JSONObject = { |
| 108 | "timestamp": datetime.fromtimestamp(record.created, tz=timezone.utc).isoformat(), |
| 109 | "level": record.levelname, |
| 110 | "logger": record.name, |
| 111 | "message": record.getMessage(), |
| 112 | "request_id": request_id_var.get(), |
| 113 | "user_id": user_id_var.get(), |
| 114 | "environment": _environment, |
| 115 | "release_version": _release_version, |
| 116 | } |
| 117 | |
| 118 | # Optional per-request fields (present on access log records). |
| 119 | for field in _ACCESS_FIELDS: |
| 120 | val = record.__dict__.get(field) |
| 121 | if val is not None: |
| 122 | doc[field] = val |
| 123 | |
| 124 | if record.exc_text: |
| 125 | doc["exc_info"] = record.exc_text |
| 126 | |
| 127 | return json.dumps(doc, ensure_ascii=False) |
| 128 | |
| 129 | # ── Public API ───────────────────────────────────────────────────────────────── |
| 130 | |
| 131 | def configure_logging(debug: bool = False, environment: str = "", release_version: str = "") -> None: |
| 132 | """Install JSON formatter + PII filter on the root logger. |
| 133 | |
| 134 | Safe to call multiple times — clears existing handlers first so there is |
| 135 | no duplication if an earlier ``basicConfig`` call ran before this one. |
| 136 | |
| 137 | ``environment`` (e.g. "staging"/"production") and ``release_version`` |
| 138 | (the deployed image tag) are stamped onto every log line so a single |
| 139 | CloudWatch dashboard/query can distinguish records from either |
| 140 | environment or correlate errors with a specific deploy. |
| 141 | """ |
| 142 | global _environment, _release_version |
| 143 | _environment = environment |
| 144 | _release_version = release_version |
| 145 | |
| 146 | root = logging.getLogger() |
| 147 | root.setLevel(logging.DEBUG if debug else logging.INFO) |
| 148 | |
| 149 | # Remove handlers installed by any earlier basicConfig / configure_logging. |
| 150 | for handler in list(root.handlers): |
| 151 | root.removeHandler(handler) |
| 152 | handler.close() |
| 153 | |
| 154 | handler = logging.StreamHandler() |
| 155 | handler.setFormatter(JsonFormatter()) |
| 156 | handler.addFilter(PiiFilter()) |
| 157 | root.addHandler(handler) |
File History
1 commit
sha256:7afc59b2ee9c70d18b748d269602c588385d6e936cc22b056fd775ab9f2fd499
fix: real CloudWatch alerting on both envs, add environment…
Sonnet 5
minor
⚠
1 day ago