aice-data-quality 0.1.0__tar.gz

aice_data_quality-0.1.0/.gitignore

@@ -0,0 +1,4 @@
+.venv/
+.dq-spool/
+__pycache__/
+.pytest_cache/

aice_data_quality-0.1.0/CHANGELOG.md

@@ -0,0 +1,22 @@
+# Changelog
+
+SDK versions are independent of contract versions. Each release states the contract it produces.
+
+## 0.1.0 (unreleased)
+
+**Produces contract:** `1.0` (`contract-v1.0.0`, pending sign-off)
+
+- Core: `QualityClient`, `Run`, `Report`, `Dataset`; contract models in pydantic.
+- Sources: `DataFrameSource`, `CsvSource`, `ExcelSource`, `SqlSource`.
+- Checks: `schema`, `not_null`, `parses_as`, `accepted_values`, `unique`, `row_count` (including `vs_previous`).
+- YAML suites validated with pydantic; `dq suite validate`.
+- Sinks: `LogSink`, `HttpSink` (client of the server); local spool with ordered replay. The SDK stores nothing and ships no results DDL.
+- `dq dbt publish`; dbt artifacts parsed with pydantic models.
+- Sample policy (`keys`, `none`), HMAC key hashing for sensitive datasets, OpenTelemetry spans.
+
+Differences from the specification, to confirm:
+
+- No `PostgresSink` and no results DDL (spec P2 dropped): the server is the only results store. Acceptance criteria revised in `docs/phase1-acceptance.md`.
+- `vs_previous` history (served by the server) counts `pass` **and `skipped`** results; counting only passing results can never bootstrap.
+- Constructor/suite default severity is `flag`; the spec did not state a default.
+- `HttpSink` and a proof-of-concept server are part of phase 1 (spec said phase 3).

aice_data_quality-0.1.0/PKG-INFO

@@ -0,0 +1,79 @@
+Metadata-Version: 2.5
+Name: aice-data-quality
+Version: 0.1.0
+Summary: Data quality SDK: checks, one result contract, one destination
+Author-email: Adanian Labs <engineering@adanianlabs.io>
+License: MIT
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3
+Classifier: Typing :: Typed
+Requires-Python: <3.14,>=3.10
+Requires-Dist: pydantic>=2
+Requires-Dist: pyyaml>=6
+Provides-Extra: dbt
+Provides-Extra: dev
+Requires-Dist: httpx>=0.27; extra == 'dev'
+Requires-Dist: jsonschema[format]>=4.21; extra == 'dev'
+Requires-Dist: openpyxl>=3.1; extra == 'dev'
+Requires-Dist: opentelemetry-api>=1.20; extra == 'dev'
+Requires-Dist: opentelemetry-sdk>=1.20; extra == 'dev'
+Requires-Dist: pandas>=2; extra == 'dev'
+Requires-Dist: psycopg[binary]>=3.1; extra == 'dev'
+Requires-Dist: pytest>=8.3; extra == 'dev'
+Requires-Dist: sqlalchemy>=2; extra == 'dev'
+Provides-Extra: excel
+Requires-Dist: openpyxl>=3.1; extra == 'excel'
+Requires-Dist: pandas>=2; extra == 'excel'
+Provides-Extra: http
+Requires-Dist: httpx>=0.27; extra == 'http'
+Provides-Extra: otel
+Requires-Dist: opentelemetry-api>=1.20; extra == 'otel'
+Provides-Extra: pandas
+Requires-Dist: pandas>=2; extra == 'pandas'
+Provides-Extra: sql
+Requires-Dist: psycopg[binary]>=3.1; extra == 'sql'
+Requires-Dist: sqlalchemy>=2; extra == 'sql'
+Description-Content-Type: text/markdown
+
+# aice-data-quality (SDK)
+
+Check data, record the outcome in the result contract, send it to one destination.
+Import name: `dq`. Produces records of contract `1.0` ([`contract/`](contract/README.md)).
+
+```python
+from dq import QualityClient, Dataset, Severity, checks as c
+from dq.sources import DataFrameSource
+
+client = QualityClient(project="my-project", environment="production")   # or DQ_* env vars
+with client.run(pipeline="uploads") as run:
+    report = run.validate(DataFrameSource(df), dataset=Dataset("finance_uploads", "statement", "bronze"),
+                          owner="finance",
+                          checks=[c.not_null("amount", name="amount_set", severity=Severity.BLOCK)])
+if report.blocked:
+    raise UploadRejected(report.messages())
+```
+
+- Full guide: [`docs/user-guide.md`](docs/user-guide.md)
+- Runnable example: [`examples/dataframe_example.py`](examples/dataframe_example.py)
+- Specification (shared with the server): [`../README.md`](../README.md)
+
+## What belongs here, and what does not
+
+| In the SDK (runs inside each project) | In the server (`../server/`, never here) |
+|---|---|
+| Sources, checks, suites, privacy rules | Storing results: schema, upserts, retention |
+| Contract models, run outcome, dbt mapping | Authentication and token scopes |
+| Sinks: `LogSink`, `HttpSink` (the server's client), the local spool | Queries, history for `vs_previous`, dataset and check listings |
+| `SqlSource`, which reads the project's own data tables | Alert routing, UI |
+
+The SDK has no results database and no results DDL. `SqlSource` is the only code that talks to a database,
+and only to read data being checked.
+
+## Develop
+
+```bash
+docker compose up -d --wait      # sample Postgres on :55432 for SqlSource tests only
+uv sync --extra dev
+uv run pytest                    # SqlSource tests skip themselves if the database is down
+```

aice_data_quality-0.1.0/README.md

@@ -0,0 +1,41 @@
+# aice-data-quality (SDK)
+
+Check data, record the outcome in the result contract, send it to one destination.
+Import name: `dq`. Produces records of contract `1.0` ([`contract/`](contract/README.md)).
+
+```python
+from dq import QualityClient, Dataset, Severity, checks as c
+from dq.sources import DataFrameSource
+
+client = QualityClient(project="my-project", environment="production")   # or DQ_* env vars
+with client.run(pipeline="uploads") as run:
+    report = run.validate(DataFrameSource(df), dataset=Dataset("finance_uploads", "statement", "bronze"),
+                          owner="finance",
+                          checks=[c.not_null("amount", name="amount_set", severity=Severity.BLOCK)])
+if report.blocked:
+    raise UploadRejected(report.messages())
+```
+
+- Full guide: [`docs/user-guide.md`](docs/user-guide.md)
+- Runnable example: [`examples/dataframe_example.py`](examples/dataframe_example.py)
+- Specification (shared with the server): [`../README.md`](../README.md)
+
+## What belongs here, and what does not
+
+| In the SDK (runs inside each project) | In the server (`../server/`, never here) |
+|---|---|
+| Sources, checks, suites, privacy rules | Storing results: schema, upserts, retention |
+| Contract models, run outcome, dbt mapping | Authentication and token scopes |
+| Sinks: `LogSink`, `HttpSink` (the server's client), the local spool | Queries, history for `vs_previous`, dataset and check listings |
+| `SqlSource`, which reads the project's own data tables | Alert routing, UI |
+
+The SDK has no results database and no results DDL. `SqlSource` is the only code that talks to a database,
+and only to read data being checked.
+
+## Develop
+
+```bash
+docker compose up -d --wait      # sample Postgres on :55432 for SqlSource tests only
+uv sync --extra dev
+uv run pytest                    # SqlSource tests skip themselves if the database is down
+```

aice_data_quality-0.1.0/contract/CHANGELOG.md

@@ -0,0 +1,15 @@
+# Contract changelog
+
+Versioning rules: see [README.md](README.md#versioning).
+
+## 1.0.0 (unreleased, pending sign-off and tag `contract-v1.0.0`)
+
+First published version.
+
+- `run.schema.json` and `result.schema.json` (JSON Schema 2020-12).
+- `origin` includes `manual` (decision C8) so reconciliation results use the same records.
+- Timestamps are UTC with `Z` only (decision C9), enforced by pattern.
+- `owner` required and non-empty on every result (C1).
+- `check_id` pattern `<source>.<name>:<check_name>` (C6); at most 100 `failing_keys` (C4).
+- Valid and invalid examples, validation test in `tests/test_contract.py`.
+

aice_data_quality-0.1.0/contract/DECISIONS.md

@@ -0,0 +1,44 @@
+# Phase 0 decision log
+
+**Status: drafted, not yet reviewed.** Every outcome below is the *proposed
+default* from the phase plan. "Decided by" stays `pending` until the review
+session (day 2) and is filled in by the SDK owner. Nothing here has been agreed
+by anyone yet.
+
+Timebox rule: anything not agreed by day 5 is left out of 1.0 and recorded under
+"Proposals for a later minor version".
+
+| # | Question | Outcome (proposed) | Decided by | Reflected in |
+|---|---|---|---|---|
+| C1 | Is `owner` required, and what is the fallback? | Required, non-empty; fallback `data-platform` | pending | `result.schema.json` (`required`, `minLength`) |
+| C2 | Is the `layer` list complete? | `raw`, `bronze`, `silver`, `gold`, `app`, `other` | pending | `dataset.layer` enum |
+| C3 | Is the `dimension` list complete; is `conformity` separate from `validity`? | Ten dimensions as in the specification; `conformity` stays separate | pending | `dimension` enum |
+| C4 | Maximum `failing_keys` | 100 | pending | `maxItems: 100`; `too_many_keys.json` |
+| C5 | Which datasets are sensitive; is sensitivity a result field? | SDK configuration, not a field. Default: all employee-level and finance transaction datasets (spec Q6) | pending | README (privacy); no schema field |
+| C6 | `check_id` format | `<source>.<name>:<check_name>`, lowercase snake case | pending | `check_id` pattern; `bad_check_id.json` |
+| C7 | dbt `warn` mapping | `status: fail` with `severity: warn` | pending | README (check status); `dbt_warn.json` |
+| C8 | Is `accuracy` recorded with the same record, with `origin: manual`? | Yes | pending | `origin` enum; `manual_accuracy.json` |
+| C9 | Timestamp format | UTC with `Z` only | pending | `utc_timestamp` pattern; `local_timestamp.json` |
+
+## Drafting choices to confirm in review
+
+Small tightenings made while turning the draft into schemas. Raise any you disagree with.
+
+1. `owner`, `check_type`, `sdk_version` and `external_run_id` have `minLength: 1`, so an empty string cannot satisfy a required field (otherwise C1 is meaningless).
+2. Timestamps use both `format: date-time` and a `Z`-only `pattern`. The pattern is what enforces C9, because `format` is advisory in many validators.
+3. The timestamp rule is shared through `$defs/utc_timestamp` in each schema (schemas are self-contained; no cross-file references).
+4. `observed_value` allows `null` (as in the specification draft).
+5. Not enforced by schema, by design: `check_id` matching `dataset` + `check_name`, and `failing_count` vs `failing_keys` length. Both belong to SDK tests in phase 1.
+6. v1 uses `additionalProperties: false`, which makes "ignore unknown optional fields" (spec section 14) a consumer-side duty rather than something the schema allows. Confirm this is the intended reading.
+
+## Sign-off
+
+| Role | Name | Date | Signed off |
+|---|---|---|---|
+| SDK owner | TBC (spec Q1 is still open) | | no |
+| Data Platform | | | no |
+| Second project representative | | | no |
+
+## Proposals for a later minor version
+
+None yet. Items not agreed by day 5 are recorded here.

aice_data_quality-0.1.0/contract/README.md

@@ -0,0 +1,106 @@
+# Data quality result contract (v1)
+
+The format every project uses to record data quality results, and that every
+destination (warehouse table, later a server) accepts.
+
+**Source of truth:** the JSON Schemas in [`v1/`](v1/). This document covers only
+what a schema cannot express. Background and rationale: [`../../README.md`](../../README.md).
+
+| File | Describes |
+|---|---|
+| [`v1/run.schema.json`](v1/run.schema.json) | One execution of a pipeline step |
+| [`v1/result.schema.json`](v1/result.schema.json) | The outcome of one check in one run |
+| [`v1/examples/valid/`](v1/examples/valid/) | Records that must pass (files named `run_*` are runs, the rest results) |
+| [`v1/examples/invalid/`](v1/examples/invalid/) | Records that must be rejected |
+
+Validate with `uv run --extra dev pytest` (see `tests/test_contract.py`).
+
+## Enumerations
+
+| Field | Values |
+|---|---|
+| `environment` | `development`, `staging`, `production` |
+| `origin` (run) | `sdk`, `dbt`, `manual` (reconciliation recorded by hand) |
+| `layer` | `raw`, `bronze`, `silver`, `gold`, `app`, `other` |
+| `dimension` | `completeness`, `validity`, `uniqueness`, `consistency`, `referential_integrity`, `timeliness`, `volume`, `distribution`, `conformity`, `accuracy` |
+| `severity` | `block`, `flag`, `warn` |
+| `status` | `pass`, `fail`, `error`, `skipped` |
+| `outcome` (run) | `passed`, `flagged`, `blocked`, `error` |
+
+## Rules the schema cannot express
+
+These are enforced by the SDK (phase 1), not by the schemas.
+
+### Severity
+
+| Severity | Effect when the check has status `fail` or `error` |
+|---|---|
+| `block` | The run is `blocked`. The caller must stop (reject the upload, skip downstream steps). The SDK never exits the process |
+| `flag` | Data is kept; the run is `flagged`; the owner is notified |
+| `warn` | Recorded only |
+
+Automated sources are never rejected: at ingestion they use `flag` at most. Only
+human uploads, which can be fixed and resubmitted, use `block`.
+
+### Run outcome
+
+Computed when the run closes, first match wins:
+
+1. `blocked`: any `block` result has status `fail` or `error`
+2. `flagged`: any `flag` result has status `fail` or `error`
+3. `passed`: otherwise (`warn` failures and `skipped` results included)
+
+`error` is set instead when the run itself could not complete (for example the
+source could not be read). An open run has neither `finished_at` nor `outcome`;
+`outcome` requires `finished_at` (the schema enforces this).
+
+### Check status
+
+- `pass`: rule satisfied.
+- `fail`: rule not satisfied.
+- `error`: the check could not run. It counts as a failure for severity and outcome. A broken check is never a pass.
+- `skipped`: deliberately not run (for example no history to compare against). Has no effect on the outcome.
+- dbt `warn` is recorded as `status: fail` with `severity: warn`, so it is visible but never blocks.
+
+### Idempotency
+
+A result is unique on (`run_id`, `check_id`). Sending it again replaces the
+earlier copy; it is never counted twice. A run is unique on `run_id`; closing a
+run replaces the open record.
+
+### Dataset naming and `check_id`
+
+- `dataset.source`, `dataset.name`: lowercase snake case (`^[a-z0-9_]+$`). `source` is the system or channel (`finance_uploads`, `bd_sheet`, `warehouse`); `name` is the entity, file type or table.
+- dbt models use `source: warehouse` and `name` = model name.
+- `check_id` = `<source>.<name>:<check_name>`, all lowercase snake case, and must be consistent with `dataset` and `check_name`. The schema checks the pattern only; the SDK checks the match.
+- Names must stay stable: renaming a dataset or check breaks its history.
+
+### Timestamps
+
+UTC with a `Z` suffix only, for example `2027-01-12T07:41:03Z` (fractional
+seconds allowed). Offsets such as `+03:00` are rejected; convert before sending.
+
+### Owner
+
+`owner` is required on every result. When none is given, the SDK and the dbt
+publisher use `data-platform`. `owner` is a team or role, never a person's email.
+
+### Sample policy and privacy (`failing_keys`, `message`, `error_detail`)
+
+| Policy | `failing_keys` contains | Allowed in |
+|---|---|---|
+| `none` | Nothing | All environments |
+| `keys` (default) | Row identifiers only: a primary key, or `row:<n>` for files | All environments |
+| `rows` | Full failing rows | `development` only; the SDK refuses it in `staging` and `production` |
+
+1. `failing_keys` holds at most 100 entries (the schema enforces this). `failing_count` carries the true count.
+2. `message` and `error_detail` describe the failure (column, rule, count). They never contain data values from personal or financial fields.
+3. Keys of datasets marked sensitive are hashed before leaving the process. Sensitivity is SDK configuration, not a field on the result.
+4. Credentials, connection strings and tokens never appear in any field.
+
+## Versioning
+
+1. `spec_version` is on every record. This contract is `1.0`.
+2. **Minor** (`1.x`): new optional fields, new check types, new enum values for `layer`. Consumers must ignore unknown optional fields. (Note that the v1 schemas set `additionalProperties: false`, so a producer may not send a field until its minor version is published here, and validators must be updated with it.)
+3. **Breaking** (`2.0`): renamed or removed fields, changed meanings, new required fields. Sinks and the server accept both versions for at least one release cycle. Schemas live in a new `contract/v2/` folder.
+4. Every change goes through a pull request to this folder, with a [`CHANGELOG.md`](CHANGELOG.md) entry and a version bump, before any code changes. Release tags are `contract-vX.Y.Z`.

aice_data_quality-0.1.0/contract/v1/examples/invalid/bad_check_id.json

@@ -0,0 +1,18 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "Finance-Uploads/date parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "block",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 1
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/extra_field.json

@@ -0,0 +1,23 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:date_parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "block",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 1,
+  "sample_rows": [
+    {
+      "amount": 100
+    }
+  ]
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/local_timestamp.json

@@ -0,0 +1,18 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:date_parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "block",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T10:41:03+03:00",
+  "failing_count": 1
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/missing_owner.json

@@ -0,0 +1,17 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:date_parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "block",
+  "status": "fail",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 1
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/run_outcome_without_finished.json

@@ -0,0 +1,11 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "project": "organization-data-model",
+  "environment": "production",
+  "pipeline": "finance-uploads",
+  "origin": "sdk",
+  "started_at": "2027-01-12T07:40:58Z",
+  "outcome": "blocked",
+  "sdk_version": "0.1.0"
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/too_many_keys.json

@@ -0,0 +1,121 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:date_parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "block",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 101,
+  "failing_keys": [
+    "row:1",
+    "row:2",
+    "row:3",
+    "row:4",
+    "row:5",
+    "row:6",
+    "row:7",
+    "row:8",
+    "row:9",
+    "row:10",
+    "row:11",
+    "row:12",
+    "row:13",
+    "row:14",
+    "row:15",
+    "row:16",
+    "row:17",
+    "row:18",
+    "row:19",
+    "row:20",
+    "row:21",
+    "row:22",
+    "row:23",
+    "row:24",
+    "row:25",
+    "row:26",
+    "row:27",
+    "row:28",
+    "row:29",
+    "row:30",
+    "row:31",
+    "row:32",
+    "row:33",
+    "row:34",
+    "row:35",
+    "row:36",
+    "row:37",
+    "row:38",
+    "row:39",
+    "row:40",
+    "row:41",
+    "row:42",
+    "row:43",
+    "row:44",
+    "row:45",
+    "row:46",
+    "row:47",
+    "row:48",
+    "row:49",
+    "row:50",
+    "row:51",
+    "row:52",
+    "row:53",
+    "row:54",
+    "row:55",
+    "row:56",
+    "row:57",
+    "row:58",
+    "row:59",
+    "row:60",
+    "row:61",
+    "row:62",
+    "row:63",
+    "row:64",
+    "row:65",
+    "row:66",
+    "row:67",
+    "row:68",
+    "row:69",
+    "row:70",
+    "row:71",
+    "row:72",
+    "row:73",
+    "row:74",
+    "row:75",
+    "row:76",
+    "row:77",
+    "row:78",
+    "row:79",
+    "row:80",
+    "row:81",
+    "row:82",
+    "row:83",
+    "row:84",
+    "row:85",
+    "row:86",
+    "row:87",
+    "row:88",
+    "row:89",
+    "row:90",
+    "row:91",
+    "row:92",
+    "row:93",
+    "row:94",
+    "row:95",
+    "row:96",
+    "row:97",
+    "row:98",
+    "row:99",
+    "row:100",
+    "row:101"
+  ]
+}

aice_data_quality-0.1.0/contract/v1/examples/invalid/unknown_severity.json

@@ -0,0 +1,18 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:date_parses",
+  "check_name": "date_parses",
+  "check_type": "parses_as",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "validity",
+  "severity": "critical",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 1
+}

aice_data_quality-0.1.0/contract/v1/examples/valid/check_error.json

@@ -0,0 +1,19 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:amount_not_null",
+  "check_name": "amount_not_null",
+  "check_type": "not_null",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "completeness",
+  "severity": "block",
+  "status": "error",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "error_detail": "Column 'amount' not found in the uploaded sheet.",
+  "message": "Check could not run: required column is missing."
+}

aice_data_quality-0.1.0/contract/v1/examples/valid/dbt_block_pass.json

@@ -0,0 +1,18 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "warehouse.fin_cash_activity:unique_fin_cash_activity_id",
+  "check_name": "unique_fin_cash_activity_id",
+  "check_type": "dbt:unique",
+  "dataset": {
+    "source": "warehouse",
+    "name": "fin_cash_activity",
+    "layer": "gold"
+  },
+  "dimension": "uniqueness",
+  "severity": "block",
+  "status": "pass",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 0
+}

aice_data_quality-0.1.0/contract/v1/examples/valid/dbt_warn.json

@@ -0,0 +1,18 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "warehouse.fin_cash_activity:not_null_fin_cash_activity_amount",
+  "check_name": "not_null_fin_cash_activity_amount",
+  "check_type": "dbt:not_null",
+  "dataset": {
+    "source": "warehouse",
+    "name": "fin_cash_activity",
+    "layer": "gold"
+  },
+  "dimension": "completeness",
+  "severity": "warn",
+  "status": "fail",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "failing_count": 7
+}

aice_data_quality-0.1.0/contract/v1/examples/valid/history_skipped.json

@@ -0,0 +1,23 @@
+{
+  "spec_version": "1.0",
+  "run_id": "0b6c2f5e-9a51-4d4f-8a7e-3f1d2b9c7a10",
+  "check_id": "finance_uploads.bank_statement_kcb:row_count_vs_previous",
+  "check_name": "row_count_vs_previous",
+  "check_type": "row_count",
+  "dataset": {
+    "source": "finance_uploads",
+    "name": "bank_statement_kcb",
+    "layer": "bronze"
+  },
+  "dimension": "volume",
+  "severity": "flag",
+  "status": "skipped",
+  "owner": "finance",
+  "evaluated_at": "2027-01-12T07:41:03Z",
+  "expected": {
+    "vs_previous": {
+      "tolerance_pct": 30
+    }
+  },
+  "message": "No previous run to compare against."
+}