KCIDB I/O Library

141 verified Mon Apr 27 auth: no python

I/O library for KCIDB, the Linux Kernel CI reporting system. Handles schema validation, data loading, submission, and querying against Google Cloud Platform BigQuery. Current version 141 (2025), requires Python >=3.10, major schema v4 with renamed 'revisions' to 'checkouts'.

pip install kcidb-io

Common errors

error AttributeError: module 'kcidb' has no attribute 'load' ↓

cause load() is in kcidb.io submodule, not top-level.

fix

Use 'from kcidb.io import load'.

error google.api_core.exceptions.Forbidden: 403 Access denied ↓

cause Client lacks BigQuery permissions on the project/dataset or GOOGLE_APPLICATION_CREDENTIALS not set.

fix

Ensure service account has BigQuery Data Editor role. Set GOOGLE_APPLICATION_CREDENTIALS or provide credentials parameter.

error TypeError: expected dictionary, got str ↓

cause submit() expects a dict, not a JSON string.

fix

Pass a Python dictionary: client.submit({'version': {'major':4, 'minor':0}, 'checkouts': []}).

Warnings

breaking Schema v4 renamed 'revisions' to 'checkouts'. Data using v3 field names will fail validation. ↓

fix Update all data to use 'checkouts' instead of 'revisions'. kcidb-io v141 (and its schema) no longer accepts 'revisions'.

deprecated Environment variable KCIDB_GS_KEYFILE is deprecated. Use GOOGLE_APPLICATION_CREDENTIALS or explicit credentials parameter. ↓

fix Unset KCIDB_GS_KEYFILE and export GOOGLE_APPLICATION_CREDENTIALS='/path/to/key.json'.

gotcha Validation of internal consistency is disabled by default. To enable, set KCIDB_VALIDATE_INTERNAL=1. ↓

fix Export KCIDB_VALIDATE_INTERNAL=1 in environment before importing.

Imports

Client
```
from kcidb import Client
```
Top-level package, no submodule 'io'. Client is for BigQuery I/O.
load
wrong
```
from kcidb import load
```
correct
```
from kcidb.io import load
```
load() lives in kcidb.io submodule, not top-level.
schema
```
from kcidb import schema
```
Schema module is at package root.

Quickstart

Submit an empty v4 data structure to BigQuery dataset. Requires GCP credentials.

import os
from kcidb import Client
client = Client(project=os.environ.get('GCP_PROJECT', 'my-project'), dataset='kernelci04', credentials=os.environ.get('GOOGLE_APPLICATION_CREDENTIALS', ''))
client.submit({'version': {'major': 4, 'minor': 0}, 'checkouts': []})