iso27diy-corp/AuditGlue/System alternative/JSON validation for Postgres.md

---
tags:
  - json
  - supabase
---
# JSON Schema validation for Postgres

When using the JSON (or JSONB) datatype, the data needs to be validated to assure database integrity.

pg_jsonschema is a PostgreSQL extension for SupaBase that can validate `json` and `jsonb` data types against a JSON Schema. The extension offers two functions:


```java
-- Validates a json *instance* against a JSON Schema *schema*
json_matches_schema(schema json, instance json) returns bool

-- Validates a jsonb *instance* against a JSON Schema *schema*
jsonb_matches_schema(schema json, instance jsonb) returns bool
```

JSON Schema is a way to define what valid JSON should look like for a particular use case:

- What properties an object should have
- What data types are expected
- Which fields are required vs optional
- Validation constraints (like minimum/maximum values, string patterns, etc.)
- Default values and descriptions

A JSON Schema is itself a JSON document. Here's a simple example of a JSON schema:

```json
{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "properties": {
    "name": {
      "type": "string",
      "description": "Person's full name"
    },
    "age": {
      "type": "integer",
      "minimum": 0,
      "maximum": 150
    },
    "email": {
      "type": "string",
      "format": "email"
    }
  },
  "required": ["name", "email"]
}
```

You can check input against a schema in SQL like this: 

```sql
create table some_table(
id serial primary key, -- db-column `id` column is an auto-incrementing primary key
metadata json not null, -- db-column `metadata` must contain a JSON value and cannot be null
check ( -- table-level check constraint to match the JSON in `metadata` to the schema
	json_matches_schema(
		schema :='{
			"type": "object", -- we require an object ...
			"properties": {
				"foo": { -- with a single string property `"foo"` ...
				"type": "string"
			}
		},
		"required": ["foo"], -- property `"foo"` is required ...
		"additionalProperties": false -- and no additional properties are allowed
	}',
	instance := metadata -- the value of the `metadata` column is passed ...
											-- as the `instance` argument to the `json_matches_schema` function, for each row 
)
)
);


-- Now we can attempt to insert a row into `some_table`,
-- with the `metadata` value provided as `<SQL input>`

insert into some_table(metadata)
values
	(<SQL input>);

-- <SQL input> needs to be replaced with an actual JSON value, e.g. '{"foo": "bar"}'.
-- The insert will only succeed if the contents of `metadata` matches the schema in the check constraint.
```

## Validating for a set of allowed values

Use the `enum` keyword to validate that a value must be one of a specific set of allowed values.

**String values:**

```json
{
  "type": "string",
  "enum": ["red", "green", "blue"]
}
```

**Mixed data types:**

```json
{
  "enum": ["active", "inactive", null, 42]
}
```

**In an object property:**

```json
{
  "type": "object",
  "properties": {
    "status": {
      "type": "string",
      "enum": ["pending", "approved", "rejected"]
    },
    "priority": {
      "type": "integer",
      "enum": [1, 2, 3, 4, 5]
    }
  }
}
```


## Validating for a data range

**Inclusive bounds (default):**

```json
{
  "type": "integer",
  "minimum": 1,
  "maximum": 10
}
```

This allows values from 1 to 10, including 1 and 10.

**Exclusive bounds:**

```json
{
  "type": "number",
  "exclusiveMinimum": 0,
  "exclusiveMaximum": 100
}
```

This allows values greater than 0 and less than 100, but not 0 or 100 themselves.

**Mixed Bounds:**

```json
{
  "type": "number",
  "minimum": 0,
  "exclusiveMaximum": 1
}
```

This allows values from 0 (inclusive) to 1 (exclusive), so 0 ≤ value < 1.

**One-Sided Ranges:**

```json
{
  "type": "integer",
  "minimum": 18
}
```

```json
{
  "type": "number",
  "maximum": 3.14159
}
```

 
**In Object Properties:**

```json
{
  "type": "object",
  "properties": {
    "age": {
      "type": "integer",
      "minimum": 0,
      "maximum": 150
    },
    "temperature": {
      "type": "number",
      "minimum": -273.15,
      "maximum": 1000.0
    }
  }
}
```

 
 **Regex Validation:**

```json
{
  "type": "string",
  "pattern": "^[a-zA-Z0-9]+$"
}
```

**Date Validation:**

JSON Schema supports the ISO 8601 date format:

```json
{
  "type": "string",
  "format": "date"
}
```

`"date"` validates dates like: `2023-12-25`
`"date-time"` validates like: `2023-12-25T10:30:00Z` or `2023-12-25T10:30:00.123Z`
`"time"` validates like: `10:30:00` or `10:30:00.123`

Using the ISO 8601 date format is recommended for interoperability.
Custom date patterns can be validated with Regex.

Ranges can be validated using the `"minimum"` and `"maximum"` keywords like before.

## Documentation**
- [pg_jsonschema](https://github.com/supabase/pg_jsonschema)
- [JSON Schema](https://json-schema.org/)