#!/usr/bin/env python
# coding: utf-8

# # Biolink Metamodel Test Notebook

# In[20]:


get_ipython().system('pip install yamlmagic')
get_ipython().run_line_magic('reload_ext', 'yamlmagic')


# In[21]:


from IPython.core.display import display, HTML
from types import ModuleType
from json import loads, dumps
from jsonasobj import JsonObj, as_json_object
from rdflib import Graph

from biolinkml.meta import SchemaDefinition
from biolinkml.utils.schemaloader import SchemaLoader
from biolinkml.utils.yamlutils import DupCheckYamlLoader, as_json_object as yaml_to_json

from biolinkml.generators.shexgen import ShExGenerator
from biolinkml.generators.pythongen import PythonGenerator
from biolinkml.generators.yumlgen import YumlGenerator
from biolinkml.generators.jsonldcontextgen import ContextGenerator


# ## Basic model structure
# A biolink model consists of:
# * a name
# * a uri
# * type definitions
# * slot definitions
# * class definitions
# * subset definitions
# 
# As an example, the model below defines:

# In[22]:


get_ipython().run_cell_magic('yaml', '--loader DupCheckYamlLoader yaml', 'id: http://example.org/sample/example1\nname: synopsis2\nprefixes:\n    foaf: http://xmlns.com/foaf/0.1/\n    samp: http://example.org/model/\n    xsd: http://www.w3.org/2001/XMLSchema#\n    \ndefault_prefix: samp\n\ndefault_curi_maps:\n    - semweb_context\n    \ndefault_range: string\n\ntypes:\n    string:\n        base: str\n        uri: xsd:string\n    int:\n        base: int\n        uri: xsd:integer\n    boolean:\n        base: Bool\n        uri: xsd:boolean\n        \n\nclasses:\n    person:\n        description: A person, living or dead\n        slots:\n            - id\n            - first name\n            - last name\n            - age\n            - living\n            - knows\n                  \n    friendly_person:\n        description: Any person that knows someone\n        is_a: person\n        slot_usage:\n            knows:\n                required: True\n\nslots:\n    id:\n        description: Unique identifier of a person\n        identifier: true\n\n    first name:\n        description: The first name of a person\n        slot_uri: foaf:firstName\n        multivalued: true\n        \n    last name:\n        description: The last name of a person\n        slot_uri: foaf:lastName\n        required: true\n        \n    living:\n        description: Whether the person is alive\n        range: boolean\n        comments:\n            - unspecified means unknown\n        \n    age:\n        description: The age of a person if living or age of death if not\n        range: int\n        slot_uri: foaf:age\n        \n    knows:\n        description: A person known by this person (indicating some level of reciprocated interaction between the parties).\n        range: person\n        slot_uri: foaf:knows\n        multivalued: true\n')


# ### We can emit this model as a Python class

# In[23]:


print(PythonGenerator(yaml).serialize())


# ### Compile the python into a module

# In[24]:


spec = compile(PythonGenerator(yaml).serialize(), 'test', 'exec')
module = ModuleType('test')
exec(spec, module.__dict__)


# ### We can emit a UML rendering of  this model

# In[25]:


display(HTML(f'<img src="{YumlGenerator(yaml).serialize()}"/>'))


# ### We can emit a JSON-LD context for the model:

# In[26]:


cntxt = loads(ContextGenerator(yaml).serialize(base="http://example.org/people/"))
print(dumps(cntxt, indent="  "))


# ### The python model can be used to create classes

# In[27]:


# Generate a person
joe_smith = module.Person(id="42", last_name="smith", first_name=['Joe', 'Bob'], age=43)
print(joe_smith)


# ### and can be combined w/ the JSON-LD Context to generate RDF

# In[28]:


# Add the context and turn it into RDF
jsonld = as_json_object(yaml_to_json(joe_smith, cntxt))
print(jsonld)
g = Graph()
g.parse(data=jsonld, format="json-ld")
print(g.serialize(format="turtle").decode())


# ### The model can be turned into ShEx

# In[29]:


shex = ShExGenerator(yaml).serialize(collections=False)
print(shex)


# ### The ShEx can then be used to validate RDF

# In[30]:


from pyshex.evaluate import evaluate
r = evaluate(g, shex, 
             start="http://example.org/model/Person", 
             focus="http://example.org/people/42")
print("Conforms" if r[0] else r[1])


# In[31]:


r = evaluate(g, shex, 
             start="http://example.org/model/FriendlyPerson", 
             focus="http://example.org/people/42")
print("Conforms" if r[0] else r[1])


# In[ ]: