Move configuration to dataclass

README.md

@@ -1,9 +1,8 @@
 # cihai · [![Python Package](https://img.shields.io/pypi/v/cihai.svg)](https://pypi.org/project/cihai/) [![License](https://img.shields.io/github/license/cihai/cihai.svg)](https://github.com/cihai/cihai/blob/master/LICENSE) [![Code Coverage](https://codecov.io/gh/cihai/cihai/branch/master/graph/badge.svg)](https://codecov.io/gh/cihai/cihai)
 
-Python library for [CJK](https://cihai.git-pull.com/glossary.html#term-cjk) (chinese, japanese,
-korean) data.
-
-This project is under active development. Follow our progress and check back for updates!
+Cihai is an open source Python library for looking up and analyzing Chinese
+characters, Japanese kanji, Korean Hanja, and Vietnamese Chữ Nôm ([CJK]) across various
+datasets.
 
 ## Quickstart
 

examples/dataset.py

@@ -34,7 +34,7 @@ def run() -> None:
     c = Cihai(unihan=False)
 
     c.add_dataset(MyDataset, namespace="moo")
-    my_dataset = MyDataset()
+    my_dataset = MyDataset(cihai=c)
     my_dataset.bootstrap()
 
     print("Definitions exactly for 好", my_dataset.givemedata("好"))

examples/variants.py

@@ -22,7 +22,7 @@ def run(unihan_options: t.Optional[t.Dict[str, object]] = None) -> None:
     """Wrapped so we can test in tests/test_examples.py"""
     print("This example prints variant character data.")
 
-    c = Cihai(config={"unihan_options": unihan_options})
+    c = Cihai(config={"plugins": {"variants": {"options": unihan_options}}})
     if not c.unihan.is_bootstrapped:  # download and install Unihan to db
         c.unihan.bootstrap()
 

pyproject.toml

@@ -1,7 +1,7 @@
 [tool.poetry]
 name = "cihai"
 version = "0.30.0"
-description = "Library for CJK (chinese, japanese, korean) language data."
+description = "Python library designed for CJK (Chinese, Japanese, Korean, Vietnamese) character and word data retrieval and analysis."
 license = "MIT"
 authors = ["Tony Narlock <tony@git-pull.com>"]
 classifiers = [

src/cihai/config.py

@@ -1,23 +1,34 @@
 """Configuration options for Cihai app."""
+import dataclasses
 import os
 import pathlib
 import typing as t
 
-from appdirs import AppDirs
-
-from cihai.constants import app_dirs
+from cihai.constants import DEFAULT_CONFIG, app_dirs
 
 if t.TYPE_CHECKING:
-    from .types import UntypedDict
+    from cihai.constants import Config
+    from cihai.types import UntypedDict
+    from unihan_etl._internal.app_dirs import AppDirs
+
+    C = t.TypeVar("C", Config, UntypedDict)
+
 
+def is_default_option(field_name: str, val: t.Any) -> bool:
+    """Return True if field default option in configuration."""
+    return bool(val == getattr(DEFAULT_CONFIG, field_name, ""))
 
-def expand_config(d: "UntypedDict", dirs: "AppDirs" = app_dirs) -> None:
+
+def expand_config(
+    d: "C",
+    dirs: "AppDirs" = app_dirs,
+) -> "C":
     """
     Expand configuration XDG variables, environmental variables, and tildes.
 
     Parameters
     ----------
-    d : dict
+    d : dict or Options
         config information
     dirs : appdirs.AppDirs
         XDG application mapping
@@ -52,19 +63,56 @@ def expand_config(d: "UntypedDict", dirs: "AppDirs" = app_dirs) -> None:
         "site_data_dir": dirs.site_data_dir,
     }
 
-    if "datasets" in d and "plugins" not in d:
-        d["datasets"] = {}
-
-    for k, v in d.items():
-        if isinstance(v, dict):
-            expand_config(v, dirs)
-        if isinstance(v, str):
-            d[k] = os.path.expanduser(  # NOQA: PTH111
-                os.path.expandvars(v).format(**context),
-            )
-
-            path = pathlib.Path(t.cast(str, d[k]))
-            if path.exists() or any(
-                str(path).startswith(app_dir) for app_dir in context.values()
-            ):
-                d[k] = path
+    if dataclasses.is_dataclass(d):
+        for field in dataclasses.fields(d):
+            if field.name == "dirs":
+                continue
+
+            v = getattr(d, field.name)
+            if dataclasses.is_dataclass(v):
+                setattr(d, field.name, expand_config(getattr(d, field.name)))
+                v = getattr(d, field.name)
+
+            if isinstance(v, dict):
+                setattr(d, field.name, expand_config(v, dirs))
+                v = getattr(d, field.name)
+
+            if isinstance(v, pathlib.Path):
+                setattr(d, field.name, str(v))
+                v = getattr(d, field.name)
+
+            if isinstance(v, str):
+                setattr(
+                    d,
+                    field.name,
+                    os.path.expanduser(  # noqa: PTH111
+                        os.path.expandvars(v).format(**context),
+                    ),
+                )
+
+                path = pathlib.Path(t.cast(str, getattr(d, field.name)))
+                if path.exists() or any(
+                    str(path).startswith(str(app_dir)) for app_dir in context.values()
+                ):
+                    setattr(d, field.name, path)
+    elif isinstance(d, dict):
+        if "datasets" in d and "plugins" not in d:
+            d["datasets"] = {}
+
+        for k, v in d.items():
+            if isinstance(v, dict):
+                v = d[k] = expand_config(v, dirs)
+            if isinstance(v, pathlib.Path):
+                v = d[k] = str(v)
+            if isinstance(v, str):
+                d[k] = os.path.expanduser(  # noqa: PTH111
+                    os.path.expandvars(v).format(**context),
+                )
+
+                path = pathlib.Path(t.cast(str, d[k]))
+                if path.exists() or any(
+                    str(path).startswith(str(app_dir)) for app_dir in context.values()
+                ):
+                    d[k] = path
+
+    return d

src/cihai/constants.py

@@ -1,30 +1,98 @@
 """Constants for cihai."""
+import dataclasses
 import pathlib
 import typing as t
 
-from appdirs import AppDirs
+from appdirs import AppDirs as BaseAppDirs
+
+from cihai.__about__ import (
+    __author__,
+    __package_name__,
+)
+from unihan_etl._internal.app_dirs import AppDirs
 
 if t.TYPE_CHECKING:
-    from cihai.types import UntypedDict
+    from typing_extensions import TypeAlias
+
+    from cihai.types import PluginMap
+
 
 #: XDG App directory locations
-app_dirs = AppDirs("cihai", "cihai team")
+_app_dirs = BaseAppDirs(__package_name__, __author__)
+app_dirs = AppDirs(_app_dirs=_app_dirs)
+
 
 #: Default configuration
-DEFAULT_CONFIG: "UntypedDict" = {
-    "debug": False,
-    "database": {"url": "sqlite:///{user_data_dir}/cihai.db"},
-    "dirs": {
-        "cache": pathlib.Path(app_dirs.user_cache_dir),
-        "log": pathlib.Path(app_dirs.user_log_dir),
-        "data": pathlib.Path(app_dirs.user_data_dir),
-    },
-    "datasets": {},
-    "plugins": {},
-}
+# DEFAULT_CONFIG: "UntypedDict" = {}
+
+
+@dataclasses.dataclass
+class Database:
+    """Database configuration for Cihai."""
+
+    url: str = "sqlite:///{user_data_dir}/cihai.db"
+
+
+DatasetName: "TypeAlias" = str
+Dataset: "TypeAlias" = str
+Datasets = t.Dict[DatasetName, Dataset]
+
+
+@dataclasses.dataclass
+class Config:
+    """Cihai configuration."""
+
+    debug: bool = False
+    database: Database = dataclasses.field(default_factory=Database)
+    dirs: AppDirs = dataclasses.field(default_factory=lambda: app_dirs)
+    datasets: Datasets = dataclasses.field(default_factory=dict)
+    plugins: "PluginMap" = dataclasses.field(default_factory=dict)
+
+    def __post_init__(self) -> None:
+        """Resolve variables and paths for Cihai configuration."""
+        if isinstance(self.dirs, BaseAppDirs):
+            self.dirs = AppDirs(_app_dirs=self.dirs)
+        if isinstance(self.dirs, dict):
+            self.dirs = AppDirs(_app_dirs=_app_dirs, **self.dirs)
+            # for k, v in directories.items():
+            #     setattr(
+            #         self.dirs,
+            #         k,
+
+        if isinstance(self.database, dict):
+            database_url = self.database.get("url")
+            if isinstance(database_url, pathlib.Path):
+                database_url = str(database_url)
+            if isinstance(database_url, str):
+                self.database["url"] = database_url.format(
+                    **dataclasses.asdict(self.dirs)
+                )
+
+            self.database = Database(**self.database)
+
+
+DEFAULT_CONFIG = Config()
+
+
+class UnihanConfigDict(t.TypedDict):
+    """Unihan configuration dictionary."""
+
+    datasets: Datasets
+
 
 #: User will be prompted to automatically configure their installation for UNIHAN
-UNIHAN_CONFIG: "UntypedDict" = {
-    "datasets": {"unihan": "cihai.data.unihan.dataset.Unihan"},
-    # Turn off by default for using as a plugin example in examples/
-}
+UNIHAN_CONFIG: "UnihanConfigDict" = UnihanConfigDict(
+    {
+        "datasets": {"unihan": "cihai.data.unihan.dataset.Unihan"},
+        # Turn off by default for using as a plugin example in examples/
+    }
+)
+
+
+@dataclasses.dataclass
+class UnihanConfig:
+    """Unihan configuration."""
+
+    datasets: Datasets = dataclasses.field(
+        default_factory=lambda: UNIHAN_CONFIG["datasets"]
+    )