In [138]:

from pathlib import Path
import json

In [139]:

schemas = {}

for p in Path('schema').glob('*.schema.json'):
    schemas[p.name] = json.loads(p.read_text())

In [140]:

defs = {}
i = 0
for root, val in schemas.items():
    vx = {(root,name):v for (name,v) in val['$defs'].items()}
    i += len(vx)
    defs.update(vx)
    
print(len(defs.keys()), i)

60 60

In [141]:

def _parse(path):
    root, local = path.split('#/')
    keys = local.split('/')
    return root, keys

def resolve(schemas, path, current_root='?'):
    root,keys = _parse(path)
    if root == '':
        root = current_root
    s = schemas[root]
    for k in keys:
        s = s[k]
    return k, s

Out[141]:

('StaticPhrasingContent',
 {'anyOf': [{'$ref': 'commonmark.schema.json#/$defs/Text'},
   {'$ref': 'commonmark.schema.json#/$defs/HTML'},
   {'$ref': 'commonmark.schema.json#/$defs/EmphasisStatic'},
   {'$ref': 'commonmark.schema.json#/$defs/StrongStatic'},
   {'$ref': 'commonmark.schema.json#/$defs/InlineCode'},
   {'$ref': 'commonmark.schema.json#/$defs/Break'},
   {'$ref': 'commonmark.schema.json#/$defs/Image'},
   {'$ref': 'commonmark.schema.json#/$defs/ImageReference'},
   {'$ref': 'roles.schema.json#/$defs/Role'},
   {'$ref': 'styles.schema.json#/$defs/SubscriptStatic'},
   {'$ref': 'styles.schema.json#/$defs/SuperscriptStatic'},
   {'$ref': 'styles.schema.json#/$defs/UnderlineStatic'},
   {'$ref': 'math.schema.json#/$defs/InlineMath'}]})

In [173]:

defs = {}
i = 0
for root, val in schemas.items():
    vx = {(root,name):v for (name,v) in val['$defs'].items()}
    i += len(vx)
    defs.update(vx)
    
print(len(defs.keys()), i)


def _handle_prop(name, prop, current_root):
    
    prop = {k:v for k,v in prop.items()}
    if 'const' in prop:
        assert set(prop.keys()) == {'const'}
        return f'{name} = ' + repr(prop['const'])
    if not prop:
        return f"{name} : Any"
    if 'type' not in prop and 'anyOf' in prop:
        print(prop)
        return f"{name}: TBD"# "{_union_value({'anyOf':prop['anyOf']}, current_root)}"
    if 'type' not in prop :
        return '!!!'+str(prop)
    if prop['type'] == 'integer':
        prop.pop('type')
        prop.pop('minimum', None)
        prop.pop('maximum', None)
        prop.pop('description', None)
        assert not prop, prop
        return f'{name} : int'
    if prop['type'] == 'boolean':
        prop.pop('type')
        prop.pop('description', None)
        assert not prop, prop
        return f'{name}  : bool'
    if prop['type'] == 'string':
        prop.pop('type')
        prop.pop('description', None)
        prop.pop('enum', None)
        assert not prop, prop
        return f'{name}  : str'
    
    if prop['type'] == 'array':
        prop.pop('type')
        items = prop.pop('items')
        prop.pop('description', None)
        assert not prop, prop
        if list(items.keys()) == ['$ref']:
            nx, sch = resolve(schemas, items['$ref'], current_root)
            return f'{name} : List[{nx}]'
            
        return f'{name} : List[{items}]'

    
    
    return f"?{prop.get('type', prop)}"


def _union_value(v, root):
    assert list(v.keys()) == ['anyOf']
    subs = ','.join([resolve(schemas, x['$ref'], root)[0] for x in v['anyOf']])
    return f'Union[{subs}]'
        
for (root,k),v in defs.items():
    print()
    if v.get('type',None) != 'object':
        print('We likely have a union')
        print(f'{k} = {_union_value(v, root)}')
        continue
    print(f'class {k}:')
    #v.pop('type')
    v.pop('description', None)
    properties = {}
    allOf = v.get('allOf', [])
    for it in allOf :
        kys = set(it.keys()) - {'properties', 'additionalProperties', '$ref'}
        if kys:
            print('Refs ?',kys)
        if it.get('$ref'):
            pass
            #print('Allof Ref:', it['$ref'], resolve(schemas, it['$ref'], root)[1])
        properties.update(it.get('properties',{}))
    properties.update(v.get('properties',{}))
    for p_name,p_val in properties.items():
            print('   ', _handle_prop(p_name, p_val, root))
            #print('    # additionalProperties: ', it.get('additionalProperties', None))
            #print('    # required: ', it.get('required', None))
            

60 60

class Math:
    type = 'math'
    enumerated  : bool
    enumerator  : str
    identifier : Any
    label : Any
    value : Any
    position : Any
    data : Any

class InlineMath:
    type = 'inlineMath'
    value : Any
    position : Any
    data : Any

class Abbreviation:
    type = 'abbreviation'
    children : List[StaticPhrasingContent]
    title  : str
    position : Any
    data : Any

class Directive:
Refs ? {'required'}
    type = 'mystDirective'
    name  : str
    args  : str
    ?object
    value  : str
    children : List[{'anyOf': [{'$ref': 'myst.schema.json#/$defs/FlowContent'}, {'$ref': 'myst.schema.json#/$defs/PhrasingContent'}]}]
    position : Any
    data : Any

class Comment:
    type = 'mystComment'
    value : Any
    position : Any
    data : Any

class Paragraph:
    type = 'paragraph'
    children : List[PhrasingContent]
    position : Any
    data : Any

class Heading:
Refs ? {'required'}
    type = 'heading'
    depth : int
    enumerated  : bool
    enumerator  : str
    children : List[PhrasingContent]
    identifier : Any
    label : Any
    position : Any
    data : Any

class ThematicBreak:
    type = 'thematicBreak'
    position : Any
    data : Any

class Blockquote:
    type = 'blockquote'
    children : List[FlowContent]
    position : Any
    data : Any

class List:
    type = 'list'
    ordered  : bool
    start : int
    spread  : bool
    children : List[ListContent]
    position : Any
    data : Any

class ListItem:
    type = 'listItem'
    spread  : bool
    children : List[{'anyOf': [{'$ref': 'myst.schema.json#/$defs/PhrasingContent'}, {'$ref': 'myst.schema.json#/$defs/FlowContent'}]}]
    position : Any
    data : Any

class HTML:
    type = 'html'
    value : Any
    position : Any
    data : Any

class Code:
Refs ? {'if', 'then', 'else'}
Refs ? {'if', 'then', 'else'}
    type = 'code'
    lang  : str
    meta  : str
    class  : str
    showLineNumbers  : bool
    startingLineNumber : int
    emphasizeLines : List[{'type': 'integer', 'minimum': 1}]
    identifier : Any
    label : Any
    value : Any
    position : Any
    data : Any

class Definition:
    type = 'definition'
    identifier : Any
    label : Any
    url : Any
    title : Any
    position : Any
    data : Any

class Text:
    type = 'text'
    value : Any
    position : Any
    data : Any

class Emphasis:
    type = 'emphasis'
    children : List[PhrasingContent]
    position : Any
    data : Any

class EmphasisStatic:
    type = 'emphasis'
    children : List[StaticPhrasingContent]
    position : Any
    data : Any

class Strong:
    type = 'strong'
    children : List[PhrasingContent]
    position : Any
    data : Any

class StrongStatic:
    type = 'strong'
    children : List[StaticPhrasingContent]
    position : Any
    data : Any

class InlineCode:
    type = 'inlineCode'
    value : Any
    position : Any
    data : Any

class Break:
    type = 'break'
    position : Any
    data : Any

class Link:
    type = 'link'
    children : List[StaticPhrasingContent]
    url : Any
    title : Any
    position : Any
    data : Any

class Image:
    type = 'image'
    class  : str
    width  : str
    align  : str
    url : Any
    title : Any
    alt : Any
    position : Any
    data : Any

class LinkReference:
    type = 'linkReference'
    children : List[StaticPhrasingContent]
    referenceType : Any
    identifier : Any
    label : Any
    position : Any
    data : Any

class ImageReference:
    type = 'imageReference'
    referenceType : Any
    identifier : Any
    label : Any
    alt : Any
    position : Any
    data : Any

class Resource:
    url  : str
    title  : str

class OptionalAssociation:
    identifier  : str
    label  : str

class Association:
Refs ? {'required'}
    identifier : Any

class Alternative:
    alt  : str

class Reference:
    referenceType  : str

class Target:
    type = 'mystTarget'
    label  : str
    position : Any
    data : Any

class CrossReference:
    type = 'crossReference'
    kind  : str
    children : List[StaticPhrasingContent]
    identifier : Any
    label : Any
    position : Any
    data : Any

We likely have a union
FlowContent = Union[Paragraph,Definition,Heading,ThematicBreak,Blockquote,List,HTML,Code,Comment,Target,Directive,Admonition,Container,Math,Table,FootnoteDefinition]

We likely have a union
ListContent = Union[ListItem]

We likely have a union
PhrasingContent = Union[StaticPhrasingContent,Emphasis,Strong,Link,LinkReference,Subscript,Superscript,Underline,Abbreviation,CrossReference,FootnoteReference]

We likely have a union
StaticPhrasingContent = Union[Text,HTML,EmphasisStatic,StrongStatic,InlineCode,Break,Image,ImageReference,Role,SubscriptStatic,SuperscriptStatic,UnderlineStatic,InlineMath]

class Role:
Refs ? {'required'}
    type = 'mystRole'
    name  : str
    value  : str
    children : List[PhrasingContent]
    position : Any
    data : Any

class Admonition:
    type = 'admonition'
    kind  : str
    class  : str
    children : List[{'anyOf': [{'$ref': '#/$defs/AdmonitionTitle'}, {'$ref': 'myst.schema.json#/$defs/FlowContent'}]}]
    position : Any
    data : Any

class AdmonitionTitle:
    type = 'admonitionTitle'
    children : List[PhrasingContent]
    position : Any
    data : Any

class FootnoteReference:
    type = 'footnoteReference'
    identifier : Any
    label : Any
    position : Any
    data : Any

class FootnoteDefinition:
    type = 'footnoteDefinition'
    children : List[FlowContent]
    identifier : Any
    label : Any
    position : Any
    data : Any

class Table:
    type = 'table'
    align  : str
    children : List[TableRow]
    position : Any
    data : Any

class TableRow:
    type = 'tableRow'
    children : List[TableCell]
    position : Any
    data : Any

class TableCell:
    type = 'tableCell'
    header  : bool
    align  : str
    children : List[PhrasingContent]
    position : Any
    data : Any

class BlockBreak:
    type = 'blockBreak'
    meta  : str
    position : Any
    data : Any

class Block:
    type = 'block'
    meta  : str
{'description': 'Top-level children of myst document', 'anyOf': [{'type': 'array', 'items': {'$ref': 'myst.schema.json#/$defs/FlowContent'}}, {'type': 'array', 'items': {'$ref': 'myst.schema.json#/$defs/ListContent'}}, {'type': 'array', 'items': {'$ref': 'myst.schema.json#/$defs/PhrasingContent'}}]}
    children: TBD
    position : Any
    data : Any

class Subscript:
    type = 'subscript'
    children : List[PhrasingContent]
    position : Any
    data : Any

class SubscriptStatic:
    type = 'subscript'
    children : List[StaticPhrasingContent]
    position : Any
    data : Any

class Superscript:
    type = 'superscript'
    children : List[PhrasingContent]
    position : Any
    data : Any

class SuperscriptStatic:
    type = 'superscript'
    children : List[StaticPhrasingContent]
    position : Any
    data : Any

class Underline:
    type = 'underline'
    children : List[PhrasingContent]
    position : Any
    data : Any

class UnderlineStatic:
    type = 'underline'
    children : List[StaticPhrasingContent]
    position : Any
    data : Any

class Container:
Refs ? {'required'}
Refs ? {'if', 'then', 'else'}
    type = 'container'
    kind  : str
    class  : str
    enumerated  : bool
    enumerator  : str
    children : List[{'anyOf': [{'$ref': '#/$defs/Caption'}, {'$ref': '#/$defs/Legend'}, {'$ref': 'commonmark.schema.json#/$defs/Image'}, {'$ref': 'tables.schema.json#/$defs/Table'}]}]
    identifier : Any
    label : Any
    position : Any
    data : Any

class Caption:
    type = 'caption'
    children : List[FlowContent]
    position : Any
    data : Any

class Legend:
    type = 'legend'
    children : List[FlowContent]
    position : Any
    data : Any

class Node:
    type  : str
    ?object
    !!!{'description': 'location of node in source file; must not be present for generated nodes', '$ref': '#/$defs/Position'}

class Literal:
Refs ? {'required'}
    value  : str

class Parent:
Refs ? {'required'}
    children : List[Node]

class Point:
    line : int
    column : int
    offset : int

class Position:
    !!!{'description': 'place of first character of parsed source region', '$ref': '#/$defs/Point'}
    !!!{'description': 'place of first character after parsed source region, whether it exists or not', '$ref': '#/$defs/Point'}
    indent : List[{'type': 'integer', 'minimum': 0}]

In [104]:

Out[104]:

{'required': ['referenceType'],
 'properties': {'referenceType': {'description': 'explicitness of the reference:\n`shortcut` - reference is implicit, identifier inferred\n`collapsed` - reference explicit, identifier inferred\n`full` - reference explicit, identifier explicit',
   'type': 'string',
   'enum': ['shortcut', 'collapsed', 'full']}}}

In [1]:

!open .

In [ ]: