Auto-generate Pydantic models from JSON Schemas

.github/workflows/test.yaml

@@ -43,6 +43,7 @@ jobs:
       run: |
         source .venv/bin/activate
         python tests/run_validation_tests.py schema
+        python tests/run_serialization_tests.py
 
     # - name: Upload test results
     #   uses: actions/upload-artifact@v2

.gitignore

@@ -2,4 +2,5 @@
 .DS_Store
 *.pdf
 *.png
-__pycache__
+__pycache__
+build/

docs/differences-from-notion.md

@@ -0,0 +1,4 @@
+- Certain block types are not supported.
+  - TBD: List which ones
+- Certain fields can be omitted, which would fall back to default values.
+- Metadata field is present in blocks.

docs/notes-on-python-implementation.md

@@ -0,0 +1,45 @@
+# JSON-DOC Python Implementation
+
+JSON Schemas are the single source of truth for page and block types.
+
+We use [datamodel-code-generator](https://koxudaxi.github.io/datamodel-code-generator) to generate Pydantic `BaseModel`s from JSON-DOC JSON schemas. This saves us from writing the Pydantic models by hand.
+
+The JSON Schemas as refer to each other, and there are circular dependencies in some cases, like blocks that have children blocks whose types are also the general block under `/block/block_schema.json`.
+
+We make use of datamodel-code-generator conventions of `customBasePath` and `customTypePath` to rewire inheritance from base classes, so that blocks, rich text objects and file objects (e.g. for image block) can be deserialized elegantly. `jsondoc.serialize` contains logic that initializes Pydantic models from JSON-DOC files or dicts.
+
+- `customBasePath` is used to override the base class of a model, instead of `BaseModel`. This is used to inherit from the correct base class, e.g. `BlockBase` for blocks, `RichTextBase` for rich text objects, etc.
+- `customTypePath` is used to override the type of a field in a model, which will be imported from the given path. If this is given, there won't be a definition for that object in the generated code.
+
+> [!WARNING]
+>
+> This stuff is pretty mind-bending, it probably needs a diagram. It is also not as hacky as it looks.
+
+During model generation, instead of simply substituting the `$ref`s with the actual types, we create dummy types and make use of `customTypePath` to refer to the actual type:
+
+```python
+# We use the existence of title as a control variable to determine if we should substitute the $ref with the actual type, or inherit from the referenced customTypePath.
+title = ref_obj.get("title")
+if title is None:
+    return ref_obj
+
+# We set customTypePath wherever necessary.
+# If it's not found, we construct it from the ref.
+if "customTypePath" in ref_obj:
+    customTypePath = ref_obj["customTypePath"]
+else:
+    ref_tokens = ref.split("/")
+    ref_tokens = ["jsondoc", "models"] + ref_tokens[:-1] + [title]
+    customTypePath = ".".join(ref_tokens)
+
+ret = {
+    "type": "object",
+    "title": title,
+    "customTypePath": customTypePath,
+    "properties": {},
+}
+```
+
+See [autogen_pydantic.py](/scripts/autogen_pydantic.py) for the full implementation.
+
+The module path `jsondoc.models` contains auto-generated models and any edits will be overwritten.

docs/roadmap.md

@@ -116,3 +116,10 @@ These are not "official" blocks, but exist under the `rich_text` key in some blo
 ### Deprecated blocks
 
 - ~~`type: template`~~
+
+
+## Miscellaneous tasks
+
+- [ ] Make non-essential fields optional with default values to make JSON files smaller. Start with rich text fields.
+- [ ] Reserve jsondoc PyPI package name.
+- [ ] Buy a JSON-DOC domain. json-doc.org and json-doc.com are available.

jsondoc/models/block/__init__.py

@@ -0,0 +1,34 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from enum import Enum
+
+from jsondoc.models.block.base import BlockBase
+
+
+class Type(Enum):
+    paragraph = 'paragraph'
+    to_do = 'to_do'
+    bulleted_list_item = 'bulleted_list_item'
+    numbered_list_item = 'numbered_list_item'
+    code = 'code'
+    column = 'column'
+    column_list = 'column_list'
+    divider = 'divider'
+    equation = 'equation'
+    heading_1 = 'heading_1'
+    heading_2 = 'heading_2'
+    heading_3 = 'heading_3'
+    image = 'image'
+    quote = 'quote'
+    equation_1 = 'equation'
+    table = 'table'
+    table_row = 'table_row'
+    toggle = 'toggle'
+
+
+class Block(BlockBase):
+    type: Type

jsondoc/models/block/base/__init__.py

@@ -0,0 +1,50 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from typing import Any, Dict, Optional
+
+from pydantic import AwareDatetime, BaseModel, ConfigDict
+from typing_extensions import Literal
+
+
+class Parent(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    type: str
+    block_id: Optional[str] = None
+    page_id: Optional[str] = None
+
+
+class CreatedBy(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    object: Literal['user']
+    id: str
+
+
+class LastEditedBy(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    object: Literal['user']
+    id: str
+
+
+class BlockBase(BaseModel):
+    object: Literal['block']
+    id: str
+    parent: Optional[Parent] = None
+    type: str
+    created_time: AwareDatetime
+    created_by: Optional[CreatedBy] = None
+    last_edited_time: Optional[AwareDatetime] = None
+    last_edited_by: Optional[LastEditedBy] = None
+    archived: bool
+    in_trash: Optional[bool] = None
+    has_children: bool
+    metadata: Optional[Dict[str, Any]] = None

jsondoc/models/block/types/bulleted_list_item/__init__.py

@@ -0,0 +1,31 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from typing import List, Optional
+
+from jsondoc.models.block.base import BlockBase
+from jsondoc.models.block.types.rich_text.base import RichTextBase
+from jsondoc.models.shared_definitions import Color
+from pydantic import BaseModel, ConfigDict
+from typing_extensions import Literal
+
+
+class BulletedListItem(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+        arbitrary_types_allowed=True,
+    )
+    rich_text: List[RichTextBase]
+    color: Optional[Color] = None
+
+
+class BulletedListItemBlock(BlockBase):
+    model_config = ConfigDict(
+        arbitrary_types_allowed=True,
+    )
+    type: Literal['bulleted_list_item']
+    bulleted_list_item: BulletedListItem
+    children: Optional[List[BlockBase]] = None

jsondoc/models/block/types/code/__init__.py

@@ -0,0 +1,107 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from enum import Enum
+from typing import List, Optional
+
+from jsondoc.models.block.base import BlockBase
+from jsondoc.models.block.types.rich_text.base import RichTextBase
+from pydantic import BaseModel, ConfigDict
+from typing_extensions import Literal
+
+
+class Language(Enum):
+    abap = 'abap'
+    arduino = 'arduino'
+    bash = 'bash'
+    basic = 'basic'
+    c = 'c'
+    clojure = 'clojure'
+    coffeescript = 'coffeescript'
+    c__ = 'c++'
+    c_ = 'c#'
+    css = 'css'
+    dart = 'dart'
+    diff = 'diff'
+    docker = 'docker'
+    elixir = 'elixir'
+    elm = 'elm'
+    erlang = 'erlang'
+    flow = 'flow'
+    fortran = 'fortran'
+    f_ = 'f#'
+    gherkin = 'gherkin'
+    glsl = 'glsl'
+    go = 'go'
+    graphql = 'graphql'
+    groovy = 'groovy'
+    haskell = 'haskell'
+    html = 'html'
+    java = 'java'
+    javascript = 'javascript'
+    json = 'json'
+    julia = 'julia'
+    kotlin = 'kotlin'
+    latex = 'latex'
+    less = 'less'
+    lisp = 'lisp'
+    livescript = 'livescript'
+    lua = 'lua'
+    makefile = 'makefile'
+    markdown = 'markdown'
+    markup = 'markup'
+    matlab = 'matlab'
+    mermaid = 'mermaid'
+    nix = 'nix'
+    objective_c = 'objective-c'
+    ocaml = 'ocaml'
+    pascal = 'pascal'
+    perl = 'perl'
+    php = 'php'
+    plain_text = 'plain text'
+    powershell = 'powershell'
+    prolog = 'prolog'
+    protobuf = 'protobuf'
+    python = 'python'
+    r = 'r'
+    reason = 'reason'
+    ruby = 'ruby'
+    rust = 'rust'
+    sass = 'sass'
+    scala = 'scala'
+    scheme = 'scheme'
+    scss = 'scss'
+    shell = 'shell'
+    sql = 'sql'
+    swift = 'swift'
+    typescript = 'typescript'
+    vb_net = 'vb.net'
+    verilog = 'verilog'
+    vhdl = 'vhdl'
+    visual_basic = 'visual basic'
+    webassembly = 'webassembly'
+    xml = 'xml'
+    yaml = 'yaml'
+    java_c_c___c_ = 'java/c/c++/c#'
+
+
+class Code(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+        arbitrary_types_allowed=True,
+    )
+    caption: Optional[List[RichTextBase]] = None
+    rich_text: List[RichTextBase]
+    language: Optional[Language] = None
+
+
+class CodeBlock(BlockBase):
+    model_config = ConfigDict(
+        arbitrary_types_allowed=True,
+    )
+    type: Literal['code']
+    code: Code
+    children: Optional[List[BlockBase]] = None

jsondoc/models/block/types/column/__init__.py

@@ -0,0 +1,20 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional
+
+from jsondoc.models.block.base import BlockBase
+from pydantic import ConfigDict
+from typing_extensions import Literal
+
+
+class ColumnBlock(BlockBase):
+    model_config = ConfigDict(
+        arbitrary_types_allowed=True,
+    )
+    type: Literal['column']
+    column: Dict[str, Any]
+    children: Optional[List[BlockBase]] = None

jsondoc/models/block/types/column_list/__init__.py

@@ -0,0 +1,21 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional
+
+from jsondoc.models.block.base import BlockBase
+from jsondoc.models.block.types.column import ColumnBlock
+from pydantic import ConfigDict
+from typing_extensions import Literal
+
+
+class ColumnListBlock(BlockBase):
+    model_config = ConfigDict(
+        arbitrary_types_allowed=True,
+    )
+    type: Literal['column_list']
+    column_list: Dict[str, Any]
+    children: Optional[List[ColumnBlock]] = None

jsondoc/models/block/types/divider/__init__.py

@@ -0,0 +1,15 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from typing import Any, Dict
+
+from jsondoc.models.block.base import BlockBase
+from typing_extensions import Literal
+
+
+class DividerBlock(BlockBase):
+    type: Literal['divider']
+    divider: Dict[str, Any]

jsondoc/models/block/types/equation/__init__.py

@@ -0,0 +1,21 @@
+# generated by datamodel-codegen:
+#   filename:  example.json
+#   timestamp: 2024-08-21T17:19:54+00:00
+
+from __future__ import annotations
+
+from jsondoc.models.block.base import BlockBase
+from pydantic import BaseModel, ConfigDict
+from typing_extensions import Literal
+
+
+class Equation(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    expression: str
+
+
+class EquationBlock(BlockBase):
+    type: Literal['equation']
+    equation: Equation