Generate JSON schema from documentation

Using the structured information extracted from the documentation (see
previous commit), build and write a JSON schema. This also introduces a
new utility library to convert build the schema description from the set
of objects and attributes.

_extensions/cps.py

@@ -1,3 +1,4 @@
+import os
 import re
 from dataclasses import dataclass
 from typing import cast
@@ -6,6 +7,8 @@
 from docutils.parsers.rst import Directive, directives, roles
 from docutils.transforms import Transform
 
+from jsb import JsonSchema
+
 from sphinx import addnodes, domains
 from sphinx.util import logging
 from sphinx.util.nodes import clean_astext
@@ -346,9 +349,53 @@ def add_role(self, name, styles=None, parent=roles.generic_custom_role):
     def add_code_role(self, name, styles=None, parent=roles.code_role):
         self.add_role(name, styles, parent)
 
+
+# =============================================================================
+def write_schema(app, exception):
+    if exception is not None:
+        return
+
+    config = app.env.config
+    title = f'{config.project} v{config.version}'
+
+    domain = cast(CpsDomain, app.env.get_domain('cps'))
+    schema = JsonSchema(title, config.schema_id)
+
+    object_attributes = {}
+    for attribute_set in domain.attributes.values():
+        for i, attribute in enumerate(attribute_set.instances):
+            schema.add_attribute(
+                attribute_set.name, i,
+                attribute.typedesc,
+                attribute.description,
+            )
+
+        for context, attribute_ref in attribute_set.context.items():
+            attribute = (
+                attribute_set.name,
+                attribute_ref[0],
+                attribute_set.instances[attribute_ref[0]].required
+            )
+            if context in object_attributes:
+                object_attributes[context].append(attribute)
+            else:
+                object_attributes[context] = [attribute]
+
+    for name, description in domain.objects.items():
+        schema.add_object_type(name, description, object_attributes[name])
+
+    output_path = os.path.join(app.outdir, config.schema_filename)
+    schema.write(config.schema_root_object, output_path)
+
 # =============================================================================
 def setup(app):
     app.add_domain(CpsDomain)
+    app.add_config_value('schema_id', '', '', [str])
+    app.add_config_value('schema_filename', 'schema.json', '', [str])
+    app.add_config_value('schema_root_object', None, '', [str])
 
     # Add custom transform to resolve cross-file references
     app.add_transform(InternalizeLinks)
+
+    # Add hook to write machine-readable schema description on completion
+    app.connect('build-finished', write_schema)

_packages/jsb.py

@@ -0,0 +1,102 @@
+import json
+import re
+
+# =============================================================================
+class JsonSchema:
+    # -------------------------------------------------------------------------
+    def __init__(self, title, uri):
+        self.meta = {
+            '$schema': 'http://json-schema.org/draft-07/schema',
+            '$id': uri,
+            'title': title,
+        }
+        self.types = {}
+        self.object_types = {}
+        self.attributes = {}
+
+    # -------------------------------------------------------------------------
+    def add_type(self, typedesc):
+        if typedesc in self.types:
+            # Type already defined; nothing to do
+            return
+
+        if '|' in typedesc:
+            types = typedesc.split('|')
+            for t in types:
+                self.add_type(t)
+
+            self.types[typedesc] = {
+                'oneOf': [
+                    {'$ref': f'#/definitions/types/{t}'} for t in types
+                ],
+            }
+
+            return
+
+        m = re.match(r'^(list|map)[(](.*)[)]$', typedesc)
+        if m:
+            outer, inner = m.groups()
+            self.add_type(inner)
+
+            if outer == 'list':
+                self.types[typedesc] = {
+                    'type': 'array',
+                    'items': {'$ref': f'#/definitions/types/{inner}'},
+                }
+
+            else:
+                self.types[typedesc] = {
+                    'type': 'object',
+                    'patternProperties': {
+                        '': {'$ref': f'#/definitions/types/{inner}'},
+                    },
+                }
+
+        elif typedesc in {'string'}:
+            # Handle simple (non-compound) types
+            self.types[typedesc] = {'type': typedesc}
+
+        else:
+            # Anything unknown is assumed to be an object type, which must be
+            # defined via add_object_type
+            pass
+
+    # -------------------------------------------------------------------------
+    def add_attribute(self, name, instance, typedesc, description):
+        self.attributes[f'{name}@{instance}'] = {
+            'description': description,
+            '$ref': f'#/definitions/types/{typedesc}',
+        }
+
+        self.add_type(typedesc)
+
+    # -------------------------------------------------------------------------
+    def add_object_type(self, name, description, attributes, strict=False):
+        self.object_types[name] = {
+            'type': 'object',
+            'description': description,
+            'properties': {
+                n: {'$ref': f'#/definitions/attributes/{n}@{i}'}
+                for n, i, r in attributes
+            },
+            'required': [n for n, i, r in attributes if r],
+            'additionalProperties': not strict,
+        }
+
+    # -------------------------------------------------------------------------
+    def _build_schema(self, root_object):
+        all_types = self.types
+        all_types.update(self.object_types)
+
+        schema = self.meta
+        schema['definitions'] = {
+            'types': all_types,
+            'attributes': self.attributes,
+        }
+        schema['$ref'] = f'#/definitions/types/{root_object}'
+
+        return schema
+
+    # -------------------------------------------------------------------------
+    def write(self, root_object, path):
+        json.dump(self._build_schema(root_object), open(path, 'wt'))

conf.py

@@ -28,6 +28,11 @@
 highlight_language = 'none'
 pygments_style = 'sphinx'
 
+# -- Options for JSON schema output ---------------------------------------
+schema_filename = 'cps.schema.json'
+schema_root_object = 'package'
+schema_id = 'https://cps-org.github.io/cps/' + schema_filename
+
 # -- Options for HTML output ----------------------------------------------
 html_style = 'cps.css'
 html_theme = 'default'

pyproject.toml

@@ -9,8 +9,9 @@ description = ""
 authors = ["The CPS Project <@>"]
 readme = "readme.md"
 packages = [
+  { include = "jsb.py", from = "_packages" },
   { include = "cps.py", from = "_extensions" },
-  { include = "autosectionlabel.py", from = "_extensions" }
+  { include = "autosectionlabel.py", from = "_extensions" },
 ]
 
 [tool.poetry.dependencies]