Incorrect YAML validation: when is a string a map?

[setjmp]

In the example below, the pajv certifies that the YAML file contains a list of strings. However, the Python yaml module parses the data into a map.

The schema (example_schema.json):

{
	"description": "Test case",
	"title": "Stuff",
	"type": "array",
	"additionalProperties": false,
	"items": {
		"type": "array",
		"properties": {
		  "items": {
			"type": "string"
		  }
		}
	}
}

The data file (example.yaml):

---
- [this is a test]
- [this:{foo,bar}]

The validation:

$ pajv -s ./example_schema.json -d example.yaml
example.yaml valid

Python test code (example.py):

import yaml

with open("example.yaml") as f:
    yaml_list = yaml.safe_load(f.read())
for item in yaml_list:
    print(item)

The outcome from the Python code:

$ python example.py 
['this is a test']
[{'this': {'foo': None, 'bar': None}}]

Either Python's built in YAML parser is wrong about this or pajv is. I am not an expert, but....

[setjmp]

It looks like the schema had a problem where the "properties" portion needs to be eliminated. I discovered this by testing the example using ajv which told me that the schema had problems worth digging into:

$ ajv  -s ./example_schema.json -d example.yaml 
strict mode: missing type "object" for keyword "additionalProperties" at "#" (strictTypes)
strict mode: missing type "object" for keyword "properties" at "#/items" (strictTypes)
example.yaml valid