Skip to content

errors.md

Latest commit

 

History

History
92 lines (72 loc) · 2.64 KB

errors.md

File metadata and controls

92 lines (72 loc) · 2.64 KB

Errors

Every error raised by the SDK subclasses hirebase.HirebaseError, so a single except can catch anything from the library.

import hirebase

try:
    client.jobs.search(query)
except hirebase.HirebaseError as e:
    print("Hirebase call failed:", e)

Exception hierarchy

HirebaseError
├── ConfigurationError        # no/invalid client config (e.g. missing API key)
├── APIError                  # any non-2xx HTTP response
│   ├── AuthenticationError   # 401 — missing/invalid API key
│   ├── PaymentRequiredError  # 402 — plan or credits required
│   ├── PermissionError_      # 403 — key valid but not allowed
│   ├── NotFoundError         # 404
│   ├── RateLimitError        # 429
│   └── ServerError           # 5xx
└── TaskError
    ├── TaskFailed            # task ended failed/canceled
    └── TaskTimeout           # task didn't finish within poll timeout

APIError attributes

APIError (and its subclasses) carry the HTTP context:

try:
    client.jobs.get("does-not-exist")
except hirebase.APIError as e:
    print(e.status_code)   # e.g. 404
    print(e.message)       # human-readable detail from the API
    print(e.body)          # decoded response body (dict or str), if any

FastAPI validation errors (422) are flattened into a readable message, e.g. body.limit: Limit cannot be greater than 100.

Handling specific cases

import time
import hirebase

def search_with_retry(client, query, retries=3):
    for attempt in range(retries):
        try:
            return client.jobs.search(query)
        except hirebase.RateLimitError:
            time.sleep(2 ** attempt)        # back off and retry
        except hirebase.AuthenticationError:
            raise                            # bad key — don't retry
    raise RuntimeError("exhausted retries")

Task errors

tasks.poll returns (False, task) for a failed/canceled task rather than raising, so you can inspect task.error. It does raise TaskTimeout if the task doesn't finish within timeout:

try:
    success, result = client.tasks.poll(task, timeout=120)
except hirebase.TaskTimeout as e:
    print("Still running:", e.task.state, e.task.progress)

Configuration errors

Raised before any network call when the client can't be built:

try:
    client = hirebase.Client()      # no api_key arg and no HIREBASE_API_KEY env
except hirebase.ConfigurationError as e:
    print(e)

See also