airbyte-internal-ops 0.2.4__py3-none-any.whl → 0.3.1__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -11,9 +11,8 @@ import logging
11
11
  import subprocess
12
12
  from typing import TYPE_CHECKING, Any, Callable
13
13
 
14
- from google.cloud import logging as gcloud_logging
15
-
16
14
  from airbyte_ops_mcp.constants import GCP_PROJECT_NAME
15
+ from airbyte_ops_mcp.gcp_auth import get_logging_client
17
16
 
18
17
  if TYPE_CHECKING:
19
18
  from airbyte_ops_mcp.connection_config_retriever.retrieval import (
@@ -23,21 +22,18 @@ if TYPE_CHECKING:
23
22
  LOGGER = logging.getLogger(__name__)
24
23
 
25
24
  # Lazy-initialized to avoid import-time GCP calls
26
- _logging_client: gcloud_logging.Client | None = None
27
25
  _airbyte_gcloud_logger: Any = None
28
26
 
29
27
 
30
28
  def _get_logger() -> Any:
31
29
  """Get the GCP Cloud Logger, initializing lazily on first use."""
32
- global _logging_client, _airbyte_gcloud_logger
30
+ global _airbyte_gcloud_logger
33
31
 
34
32
  if _airbyte_gcloud_logger is not None:
35
33
  return _airbyte_gcloud_logger
36
34
 
37
- _logging_client = gcloud_logging.Client(project=GCP_PROJECT_NAME)
38
- _airbyte_gcloud_logger = _logging_client.logger(
39
- "airbyte-cloud-connection-retriever"
40
- )
35
+ logging_client = get_logging_client(GCP_PROJECT_NAME)
36
+ _airbyte_gcloud_logger = logging_client.logger("airbyte-cloud-connection-retriever")
41
37
  return _airbyte_gcloud_logger
42
38
 
43
39
 
@@ -10,6 +10,9 @@ from airbyte.exceptions import PyAirbyteInputError
10
10
  MCP_SERVER_NAME = "airbyte-internal-ops"
11
11
  """The name of the MCP server."""
12
12
 
13
+ USER_AGENT = "Airbyte-Internal-Ops Python client"
14
+ """User-Agent string for HTTP requests to Airbyte Cloud APIs."""
15
+
13
16
  # Environment variable names for internal admin authentication
14
17
  ENV_AIRBYTE_INTERNAL_ADMIN_FLAG = "AIRBYTE_INTERNAL_ADMIN_FLAG"
15
18
  ENV_AIRBYTE_INTERNAL_ADMIN_USER = "AIRBYTE_INTERNAL_ADMIN_USER"
@@ -6,94 +6,187 @@ the airbyte-ops-mcp codebase. It supports both standard Application Default
6
6
  Credentials (ADC) and the GCP_PROD_DB_ACCESS_CREDENTIALS environment variable
7
7
  used internally at Airbyte.
8
8
 
9
+ The preferred approach is to pass credentials directly to GCP client constructors
10
+ rather than relying on file-based ADC discovery. This module provides helpers
11
+ that construct credentials from JSON content in environment variables.
12
+
9
13
  Usage:
10
- from airbyte_ops_mcp.gcp_auth import get_secret_manager_client
14
+ from airbyte_ops_mcp.gcp_auth import get_gcp_credentials, get_secret_manager_client
15
+
16
+ # Get credentials object to pass to any GCP client
17
+ credentials = get_gcp_credentials()
18
+ client = logging.Client(project="my-project", credentials=credentials)
11
19
 
12
- # Get a properly authenticated Secret Manager client
20
+ # Or use the convenience helper for Secret Manager
13
21
  client = get_secret_manager_client()
14
22
  """
15
23
 
16
24
  from __future__ import annotations
17
25
 
26
+ import json
18
27
  import logging
19
28
  import os
20
- import tempfile
21
- from pathlib import Path
29
+ import sys
30
+ import threading
22
31
 
32
+ import google.auth
33
+ from google.cloud import logging as gcp_logging
23
34
  from google.cloud import secretmanager
35
+ from google.oauth2 import service_account
24
36
 
25
37
  from airbyte_ops_mcp.constants import ENV_GCP_PROD_DB_ACCESS_CREDENTIALS
26
38
 
27
39
  logger = logging.getLogger(__name__)
28
40
 
29
- # Environment variable name (internal to GCP libraries)
30
- ENV_GOOGLE_APPLICATION_CREDENTIALS = "GOOGLE_APPLICATION_CREDENTIALS"
31
41
 
32
- # Module-level cache for the credentials file path
33
- _credentials_file_path: str | None = None
42
+ def _get_identity_from_service_account_info(info: dict) -> str | None:
43
+ """Extract service account identity from parsed JSON info.
44
+
45
+ Only accesses the 'client_email' key to avoid any risk of leaking
46
+ other credential material.
34
47
 
48
+ Args:
49
+ info: Parsed service account JSON as a dict.
35
50
 
36
- def ensure_adc_credentials() -> str | None:
37
- """Ensure GCP Application Default Credentials are available.
51
+ Returns:
52
+ The client_email if present and a string, otherwise None.
53
+ """
54
+ client_email = info.get("client_email")
55
+ if isinstance(client_email, str):
56
+ return client_email
57
+ return None
38
58
 
39
- If GOOGLE_APPLICATION_CREDENTIALS is not set but GCP_PROD_DB_ACCESS_CREDENTIALS is,
40
- write the JSON credentials to a temp file and set GOOGLE_APPLICATION_CREDENTIALS
41
- to point to that file. This provides a fallback for internal employees who use
42
- GCP_PROD_DB_ACCESS_CREDENTIALS as their standard credential source.
43
59
 
44
- Note: GOOGLE_APPLICATION_CREDENTIALS must be a file path, not JSON content.
45
- The GCP_PROD_DB_ACCESS_CREDENTIALS env var contains the JSON content directly,
46
- so we write it to a temp file first.
60
+ def _get_identity_from_credentials(
61
+ credentials: google.auth.credentials.Credentials,
62
+ ) -> str | None:
63
+ """Extract identity from a credentials object using safe attribute access.
47
64
 
48
- This function is idempotent and safe to call multiple times.
65
+ Only accesses known-safe attributes that don't trigger network calls
66
+ or token refresh.
67
+
68
+ Args:
69
+ credentials: A GCP credentials object.
49
70
 
50
71
  Returns:
51
- The path to the credentials file if one was created, or None if
52
- GOOGLE_APPLICATION_CREDENTIALS was already set.
72
+ The service account email if available, otherwise None.
53
73
  """
54
- global _credentials_file_path
74
+ # Try service_account_email first (most common for service accounts)
75
+ identity = getattr(credentials, "service_account_email", None)
76
+ if isinstance(identity, str):
77
+ return identity
78
+
79
+ # Try signer_email as fallback (sometimes present on impersonated creds)
80
+ identity = getattr(credentials, "signer_email", None)
81
+ if isinstance(identity, str):
82
+ return identity
83
+
84
+ return None
85
+
55
86
 
56
- # If GOOGLE_APPLICATION_CREDENTIALS is already set, nothing to do
57
- if ENV_GOOGLE_APPLICATION_CREDENTIALS in os.environ:
58
- return None
87
+ # Default scopes for GCP services used by this module
88
+ DEFAULT_GCP_SCOPES = ["https://www.googleapis.com/auth/cloud-platform"]
59
89
 
60
- # Check if we have the fallback credentials
61
- gsm_creds = os.getenv(ENV_GCP_PROD_DB_ACCESS_CREDENTIALS)
62
- if not gsm_creds:
63
- return None
90
+ # Module-level cache for credentials (thread-safe)
91
+ _cached_credentials: google.auth.credentials.Credentials | None = None
92
+ _credentials_lock = threading.Lock()
64
93
 
65
- # Reuse the same file path if we've already written credentials and file still exists
66
- if _credentials_file_path is not None and Path(_credentials_file_path).exists():
67
- os.environ[ENV_GOOGLE_APPLICATION_CREDENTIALS] = _credentials_file_path
68
- return _credentials_file_path
69
94
 
70
- # Write credentials to a temp file
71
- # Use a unique filename based on PID to avoid collisions between processes
72
- creds_file = Path(tempfile.gettempdir()) / f"gcp_prod_db_creds_{os.getpid()}.json"
73
- creds_file.write_text(gsm_creds)
95
+ def get_gcp_credentials() -> google.auth.credentials.Credentials:
96
+ """Get GCP credentials, preferring direct JSON parsing over file-based ADC.
74
97
 
75
- # Set restrictive permissions (owner read/write only)
76
- creds_file.chmod(0o600)
98
+ This function resolves credentials in the following order:
99
+ 1. GCP_PROD_DB_ACCESS_CREDENTIALS env var (JSON content) - parsed directly
100
+ 2. Standard ADC discovery (workload identity, gcloud auth, GOOGLE_APPLICATION_CREDENTIALS)
77
101
 
78
- _credentials_file_path = str(creds_file)
79
- os.environ[ENV_GOOGLE_APPLICATION_CREDENTIALS] = _credentials_file_path
102
+ The credentials are cached after first resolution for efficiency.
103
+ Uses the cloud-platform scope which provides access to all GCP services.
80
104
 
81
- logger.debug(
82
- f"Wrote {ENV_GCP_PROD_DB_ACCESS_CREDENTIALS} to {creds_file} and set "
83
- f"{ENV_GOOGLE_APPLICATION_CREDENTIALS}"
84
- )
105
+ Returns:
106
+ A Credentials object that can be passed to any GCP client constructor.
85
107
 
86
- return _credentials_file_path
108
+ Raises:
109
+ google.auth.exceptions.DefaultCredentialsError: If no credentials can be found.
110
+ """
111
+ global _cached_credentials
112
+
113
+ # Return cached credentials if available (fast path without lock)
114
+ if _cached_credentials is not None:
115
+ return _cached_credentials
116
+
117
+ # Acquire lock for thread-safe credential initialization
118
+ with _credentials_lock:
119
+ # Double-check after acquiring lock (another thread may have initialized)
120
+ if _cached_credentials is not None:
121
+ return _cached_credentials
122
+
123
+ # Try GCP_PROD_DB_ACCESS_CREDENTIALS first (JSON content in env var)
124
+ creds_json = os.getenv(ENV_GCP_PROD_DB_ACCESS_CREDENTIALS)
125
+ if creds_json:
126
+ try:
127
+ creds_dict = json.loads(creds_json)
128
+ credentials = service_account.Credentials.from_service_account_info(
129
+ creds_dict,
130
+ scopes=DEFAULT_GCP_SCOPES,
131
+ )
132
+ # Extract identity safely (only after successful credential creation)
133
+ identity = _get_identity_from_service_account_info(creds_dict)
134
+ identity_str = f" (identity: {identity})" if identity else ""
135
+ print(
136
+ f"GCP credentials loaded from {ENV_GCP_PROD_DB_ACCESS_CREDENTIALS}{identity_str}",
137
+ file=sys.stderr,
138
+ )
139
+ logger.debug(
140
+ f"Loaded GCP credentials from {ENV_GCP_PROD_DB_ACCESS_CREDENTIALS} env var"
141
+ )
142
+ _cached_credentials = credentials
143
+ return credentials
144
+ except (json.JSONDecodeError, ValueError) as e:
145
+ # Log only exception type to avoid any risk of leaking credential content
146
+ logger.warning(
147
+ f"Failed to parse {ENV_GCP_PROD_DB_ACCESS_CREDENTIALS}: "
148
+ f"{type(e).__name__}. Falling back to ADC discovery."
149
+ )
150
+
151
+ # Fall back to standard ADC discovery
152
+ credentials, project = google.auth.default(scopes=DEFAULT_GCP_SCOPES)
153
+ # Extract identity safely from ADC credentials
154
+ identity = _get_identity_from_credentials(credentials)
155
+ identity_str = f" (identity: {identity})" if identity else ""
156
+ project_str = f" (project: {project})" if project else ""
157
+ print(
158
+ f"GCP credentials loaded via ADC{project_str}{identity_str}",
159
+ file=sys.stderr,
160
+ )
161
+ logger.debug(f"Loaded GCP credentials via ADC discovery (project: {project})")
162
+ _cached_credentials = credentials
163
+ return credentials
87
164
 
88
165
 
89
166
  def get_secret_manager_client() -> secretmanager.SecretManagerServiceClient:
90
167
  """Get a Secret Manager client with proper credential handling.
91
168
 
92
- This function ensures GCP credentials are available (supporting the
93
- GCP_PROD_DB_ACCESS_CREDENTIALS fallback) before creating the client.
169
+ This function uses get_gcp_credentials() to resolve credentials and passes
170
+ them directly to the client constructor.
94
171
 
95
172
  Returns:
96
173
  A configured SecretManagerServiceClient instance.
97
174
  """
98
- ensure_adc_credentials()
99
- return secretmanager.SecretManagerServiceClient()
175
+ credentials = get_gcp_credentials()
176
+ return secretmanager.SecretManagerServiceClient(credentials=credentials)
177
+
178
+
179
+ def get_logging_client(project: str) -> gcp_logging.Client:
180
+ """Get a Cloud Logging client with proper credential handling.
181
+
182
+ This function uses get_gcp_credentials() to resolve credentials and passes
183
+ them directly to the client constructor.
184
+
185
+ Args:
186
+ project: The GCP project ID to use for logging operations.
187
+
188
+ Returns:
189
+ A configured Cloud Logging Client instance.
190
+ """
191
+ credentials = get_gcp_credentials()
192
+ return gcp_logging.Client(project=project, credentials=credentials)
@@ -0,0 +1,18 @@
1
+ # Copyright (c) 2025 Airbyte, Inc., all rights reserved.
2
+ """GCP Cloud Logging utilities for fetching error details by error ID."""
3
+
4
+ from airbyte_ops_mcp.gcp_logs.error_lookup import (
5
+ GCPLogEntry,
6
+ GCPLogPayload,
7
+ GCPLogSearchResult,
8
+ GCPSeverity,
9
+ fetch_error_logs,
10
+ )
11
+
12
+ __all__ = [
13
+ "GCPLogEntry",
14
+ "GCPLogPayload",
15
+ "GCPLogSearchResult",
16
+ "GCPSeverity",
17
+ "fetch_error_logs",
18
+ ]
@@ -0,0 +1,384 @@
1
+ # Copyright (c) 2025 Airbyte, Inc., all rights reserved.
2
+ """Fetch full stack traces from Google Cloud Logs by error ID.
3
+
4
+ This module provides functionality to look up error details from GCP Cloud Logging
5
+ using an error ID (UUID). This is useful for debugging API errors that return
6
+ only an error ID in the response.
7
+
8
+ Example:
9
+ from airbyte_ops_mcp.gcp_logs import fetch_error_logs
10
+
11
+ result = fetch_error_logs(
12
+ error_id="3173452e-8f22-4286-a1ec-b0f16c1e078a",
13
+ project="prod-ab-cloud-proj",
14
+ lookback_days=7,
15
+ )
16
+ for entry in result.entries:
17
+ print(entry.message)
18
+ """
19
+
20
+ from __future__ import annotations
21
+
22
+ import re
23
+ from datetime import UTC, datetime, timedelta
24
+ from enum import StrEnum
25
+ from typing import Any
26
+
27
+ from google.cloud import logging as gcp_logging
28
+ from google.cloud.logging_v2 import entries
29
+ from pydantic import BaseModel, Field
30
+
31
+ from airbyte_ops_mcp.gcp_auth import get_logging_client
32
+
33
+ # Default GCP project for Airbyte Cloud
34
+ DEFAULT_GCP_PROJECT = "prod-ab-cloud-proj"
35
+
36
+
37
+ class GCPSeverity(StrEnum):
38
+ """Valid GCP Cloud Logging severity levels."""
39
+
40
+ DEBUG = "DEBUG"
41
+ INFO = "INFO"
42
+ NOTICE = "NOTICE"
43
+ WARNING = "WARNING"
44
+ ERROR = "ERROR"
45
+ CRITICAL = "CRITICAL"
46
+ ALERT = "ALERT"
47
+ EMERGENCY = "EMERGENCY"
48
+
49
+
50
+ class GCPLogResourceLabels(BaseModel):
51
+ """Resource labels from a GCP log entry."""
52
+
53
+ pod_name: str | None = Field(default=None, description="Kubernetes pod name")
54
+ container_name: str | None = Field(
55
+ default=None, description="Container name within the pod"
56
+ )
57
+ namespace_name: str | None = Field(default=None, description="Kubernetes namespace")
58
+ cluster_name: str | None = Field(default=None, description="GKE cluster name")
59
+
60
+
61
+ class GCPLogResource(BaseModel):
62
+ """Resource information from a GCP log entry."""
63
+
64
+ type: str | None = Field(default=None, description="Resource type")
65
+ labels: GCPLogResourceLabels = Field(
66
+ default_factory=GCPLogResourceLabels, description="Resource labels"
67
+ )
68
+
69
+
70
+ class GCPLogSourceLocation(BaseModel):
71
+ """Source location information from a GCP log entry."""
72
+
73
+ file: str | None = Field(default=None, description="Source file path")
74
+ line: int | None = Field(default=None, description="Line number")
75
+ function: str | None = Field(default=None, description="Function name")
76
+
77
+
78
+ class GCPLogEntry(BaseModel):
79
+ """A single log entry from GCP Cloud Logging."""
80
+
81
+ timestamp: datetime | None = Field(
82
+ default=None, description="When the log entry was created"
83
+ )
84
+ severity: str | None = Field(
85
+ default=None, description="Log severity (DEBUG, INFO, WARNING, ERROR, etc.)"
86
+ )
87
+ log_name: str | None = Field(default=None, description="Full log name path")
88
+ insert_id: str | None = Field(
89
+ default=None, description="Unique identifier for the log entry"
90
+ )
91
+ trace: str | None = Field(
92
+ default=None, description="Trace ID for distributed tracing"
93
+ )
94
+ span_id: str | None = Field(default=None, description="Span ID within the trace")
95
+ payload: Any = Field(default=None, description="Log entry payload (text or struct)")
96
+ payload_type: str | None = Field(
97
+ default=None, description="Type of payload (text, struct, protobuf)"
98
+ )
99
+ resource: GCPLogResource = Field(
100
+ default_factory=GCPLogResource, description="Resource information"
101
+ )
102
+ source_location: GCPLogSourceLocation | None = Field(
103
+ default=None, description="Source code location"
104
+ )
105
+ labels: dict[str, str] = Field(
106
+ default_factory=dict, description="User-defined labels"
107
+ )
108
+
109
+
110
+ class GCPLogPayload(BaseModel):
111
+ """Extracted and combined payload from grouped log entries."""
112
+
113
+ timestamp: datetime | None = Field(
114
+ default=None, description="Timestamp of the first entry in the group"
115
+ )
116
+ severity: str | None = Field(default=None, description="Severity of the log group")
117
+ resource: GCPLogResource = Field(
118
+ default_factory=GCPLogResource, description="Resource information"
119
+ )
120
+ num_log_lines: int = Field(
121
+ default=0, description="Number of log lines combined into this payload"
122
+ )
123
+ message: str = Field(default="", description="Combined message from all log lines")
124
+
125
+
126
+ class GCPLogSearchResult(BaseModel):
127
+ """Result of searching GCP Cloud Logging for an error ID."""
128
+
129
+ error_id: str = Field(description="The error ID that was searched for")
130
+ project: str = Field(description="GCP project that was searched")
131
+ lookback_days_searched: int = Field(
132
+ description="Number of lookback days that were searched"
133
+ )
134
+ total_entries_found: int = Field(
135
+ description="Total number of log entries found (including related entries)"
136
+ )
137
+ entries: list[GCPLogEntry] = Field(
138
+ default_factory=list, description="Raw log entries found"
139
+ )
140
+ payloads: list[GCPLogPayload] = Field(
141
+ default_factory=list,
142
+ description="Extracted and grouped payloads (reconstructed stack traces)",
143
+ )
144
+
145
+
146
+ def _build_filter(
147
+ error_id: str,
148
+ lookback_days: int,
149
+ min_severity_filter: GCPSeverity | None,
150
+ ) -> str:
151
+ """Build the Cloud Logging filter query."""
152
+ filter_parts = [f'"{error_id}"']
153
+
154
+ start_time = datetime.now(UTC) - timedelta(days=lookback_days)
155
+ filter_parts.append(f'timestamp >= "{start_time.isoformat()}"')
156
+
157
+ if min_severity_filter:
158
+ filter_parts.append(f"severity>={min_severity_filter}")
159
+
160
+ return " AND ".join(filter_parts)
161
+
162
+
163
+ def _entry_to_model(
164
+ entry: entries.StructEntry | entries.TextEntry | entries.ProtobufEntry,
165
+ ) -> GCPLogEntry:
166
+ """Convert a GCP log entry to a Pydantic model."""
167
+ resource_labels = {}
168
+ if entry.resource and entry.resource.labels:
169
+ resource_labels = dict(entry.resource.labels)
170
+
171
+ resource = GCPLogResource(
172
+ type=entry.resource.type if entry.resource else None,
173
+ labels=GCPLogResourceLabels(
174
+ pod_name=resource_labels.get("pod_name"),
175
+ container_name=resource_labels.get("container_name"),
176
+ namespace_name=resource_labels.get("namespace_name"),
177
+ cluster_name=resource_labels.get("cluster_name"),
178
+ ),
179
+ )
180
+
181
+ source_location = None
182
+ if entry.source_location:
183
+ source_location = GCPLogSourceLocation(
184
+ file=entry.source_location.get("file"),
185
+ line=entry.source_location.get("line"),
186
+ function=entry.source_location.get("function"),
187
+ )
188
+
189
+ payload: Any = None
190
+ payload_type = "unknown"
191
+ if isinstance(entry, entries.StructEntry):
192
+ payload = entry.payload
193
+ payload_type = "struct"
194
+ elif isinstance(entry, entries.TextEntry):
195
+ payload = entry.payload
196
+ payload_type = "text"
197
+ elif isinstance(entry, entries.ProtobufEntry):
198
+ payload = str(entry.payload)
199
+ payload_type = "protobuf"
200
+
201
+ return GCPLogEntry(
202
+ timestamp=entry.timestamp,
203
+ severity=entry.severity,
204
+ log_name=entry.log_name,
205
+ insert_id=entry.insert_id,
206
+ trace=entry.trace,
207
+ span_id=entry.span_id,
208
+ payload=payload,
209
+ payload_type=payload_type,
210
+ resource=resource,
211
+ source_location=source_location,
212
+ labels=dict(entry.labels) if entry.labels else {},
213
+ )
214
+
215
+
216
+ def _group_entries_by_occurrence(
217
+ log_entries: list[GCPLogEntry],
218
+ ) -> list[list[GCPLogEntry]]:
219
+ """Group log entries by occurrence (timestamp clusters within 1 second)."""
220
+ if not log_entries:
221
+ return []
222
+
223
+ sorted_entries = sorted(
224
+ log_entries, key=lambda x: x.timestamp or datetime.min.replace(tzinfo=UTC)
225
+ )
226
+
227
+ groups: list[list[GCPLogEntry]] = []
228
+ current_group = [sorted_entries[0]]
229
+ current_timestamp = sorted_entries[0].timestamp or datetime.min.replace(tzinfo=UTC)
230
+
231
+ for entry in sorted_entries[1:]:
232
+ entry_timestamp = entry.timestamp or datetime.min.replace(tzinfo=UTC)
233
+ time_diff = abs((entry_timestamp - current_timestamp).total_seconds())
234
+
235
+ current_pod = current_group[0].resource.labels.pod_name
236
+ entry_pod = entry.resource.labels.pod_name
237
+
238
+ if time_diff <= 1 and entry_pod == current_pod:
239
+ current_group.append(entry)
240
+ else:
241
+ groups.append(current_group)
242
+ current_group = [entry]
243
+ current_timestamp = entry_timestamp
244
+
245
+ if current_group:
246
+ groups.append(current_group)
247
+
248
+ return groups
249
+
250
+
251
+ def _extract_payloads(log_entries: list[GCPLogEntry]) -> list[GCPLogPayload]:
252
+ """Extract and group payloads by occurrence."""
253
+ if not log_entries:
254
+ return []
255
+
256
+ grouped = _group_entries_by_occurrence(log_entries)
257
+
258
+ results = []
259
+ for group in grouped:
260
+ payloads = []
261
+ for entry in group:
262
+ if entry.payload:
263
+ payload_text = str(entry.payload)
264
+ payload_text = re.sub(r"\x1b\[[0-9;]*m", "", payload_text)
265
+ payloads.append(payload_text)
266
+
267
+ combined_message = "\n".join(payloads)
268
+
269
+ first_entry = group[0]
270
+ result = GCPLogPayload(
271
+ timestamp=first_entry.timestamp,
272
+ severity=first_entry.severity,
273
+ resource=first_entry.resource,
274
+ num_log_lines=len(group),
275
+ message=combined_message,
276
+ )
277
+ results.append(result)
278
+
279
+ return results
280
+
281
+
282
+ def fetch_error_logs(
283
+ error_id: str,
284
+ project: str = DEFAULT_GCP_PROJECT,
285
+ lookback_days: int = 7,
286
+ min_severity_filter: GCPSeverity | None = None,
287
+ include_log_envelope_seconds: float = 1.0,
288
+ max_log_entries: int | None = None,
289
+ ) -> GCPLogSearchResult:
290
+ """Fetch logs from Google Cloud Logging by error ID.
291
+
292
+ This function searches GCP Cloud Logging for log entries containing the
293
+ specified error ID, then fetches related log entries (multi-line stack traces)
294
+ from the same timestamp and resource.
295
+ """
296
+ client = get_logging_client(project)
297
+
298
+ filter_str = _build_filter(error_id, lookback_days, min_severity_filter)
299
+
300
+ entries_iterator = client.list_entries(
301
+ filter_=filter_str,
302
+ order_by=gcp_logging.DESCENDING,
303
+ )
304
+
305
+ initial_matches = list(entries_iterator)
306
+
307
+ if not initial_matches:
308
+ return GCPLogSearchResult(
309
+ error_id=error_id,
310
+ project=project,
311
+ lookback_days_searched=lookback_days,
312
+ total_entries_found=0,
313
+ entries=[],
314
+ payloads=[],
315
+ )
316
+
317
+ all_results: list[GCPLogEntry] = []
318
+ seen_insert_ids: set[str] = set()
319
+
320
+ for match in initial_matches:
321
+ timestamp = match.timestamp
322
+ resource_type_val = match.resource.type if match.resource else None
323
+ resource_labels = (
324
+ dict(match.resource.labels)
325
+ if match.resource and match.resource.labels
326
+ else {}
327
+ )
328
+ log_name = match.log_name
329
+
330
+ start_time = timestamp - timedelta(seconds=include_log_envelope_seconds)
331
+ end_time = timestamp + timedelta(seconds=include_log_envelope_seconds)
332
+
333
+ related_filter_parts = [
334
+ f'timestamp >= "{start_time.isoformat()}"',
335
+ f'timestamp <= "{end_time.isoformat()}"',
336
+ ]
337
+
338
+ if log_name:
339
+ related_filter_parts.append(f'logName="{log_name}"')
340
+
341
+ if resource_type_val:
342
+ related_filter_parts.append(f'resource.type="{resource_type_val}"')
343
+
344
+ if "pod_name" in resource_labels:
345
+ related_filter_parts.append(
346
+ f'resource.labels.pod_name="{resource_labels["pod_name"]}"'
347
+ )
348
+ if "container_name" in resource_labels:
349
+ related_filter_parts.append(
350
+ f'resource.labels.container_name="{resource_labels["container_name"]}"'
351
+ )
352
+
353
+ # Note: resource_type_val is extracted from the matched entry, and
354
+ # min_severity_filter is already applied in the initial search filter
355
+
356
+ related_filter = " AND ".join(related_filter_parts)
357
+
358
+ related_entries = client.list_entries(
359
+ filter_=related_filter,
360
+ order_by=gcp_logging.ASCENDING,
361
+ )
362
+
363
+ for entry in related_entries:
364
+ if entry.insert_id and entry.insert_id not in seen_insert_ids:
365
+ seen_insert_ids.add(entry.insert_id)
366
+ all_results.append(_entry_to_model(entry))
367
+
368
+ all_results.sort(
369
+ key=lambda x: x.timestamp or datetime.min.replace(tzinfo=UTC), reverse=True
370
+ )
371
+
372
+ if max_log_entries:
373
+ all_results = all_results[:max_log_entries]
374
+
375
+ payloads = _extract_payloads(all_results)
376
+
377
+ return GCPLogSearchResult(
378
+ error_id=error_id,
379
+ project=project,
380
+ lookback_days_searched=lookback_days,
381
+ total_entries_found=len(all_results),
382
+ entries=all_results,
383
+ payloads=payloads,
384
+ )