infinity-data 2.2.0__tar.gz → 3.1.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {infinity_data-2.2.0/src/infinity_data.egg-info → infinity_data-3.1.1}/PKG-INFO +1 -1
- {infinity_data-2.2.0 → infinity_data-3.1.1}/pyproject.toml +2 -2
- infinity_data-3.1.1/src/infinity_data/emit/__init__.py +35 -0
- infinity_data-3.1.1/src/infinity_data/emit/config.py +37 -0
- infinity_data-3.1.1/src/infinity_data/emit/converter.py +258 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/path.py +17 -1
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/parser/__init__.py +2 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/parser/diagnostics.py +5 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/parser/models.py +78 -38
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/parser/parser.py +65 -29
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/pipeline.py +13 -6
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/sandbox/mediator.py +14 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/__init__.py +4 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/builder/builder.py +146 -100
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/constraints.py +3 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/diagnostics.py +60 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/executor/executor.py +23 -2
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/registry/__init__.py +37 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/registry/_core.py +31 -5
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/registry/general.py +16 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/registry/logic.py +14 -3
- infinity_data-3.1.1/src/infinity_data/semantic/registry/types.py +321 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/resolver/imports.py +1 -1
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/std.py +135 -5
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/diagnostics.py +5 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/finalizer.py +26 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/models/raw_tokens.py +3 -2
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/models/tokens.py +11 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/tokenizer.py +57 -17
- {infinity_data-2.2.0 → infinity_data-3.1.1/src/infinity_data.egg-info}/PKG-INFO +1 -1
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data.egg-info/SOURCES.txt +1 -0
- infinity_data-2.2.0/src/infinity_data/emit/__init__.py +0 -9
- infinity_data-2.2.0/src/infinity_data/emit/converter.py +0 -56
- infinity_data-2.2.0/src/infinity_data/semantic/registry/types.py +0 -136
- {infinity_data-2.2.0 → infinity_data-3.1.1}/LICENSE +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/README.md +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/setup.cfg +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/frontend.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/diagnostics.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/file.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/ll1_stream.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/infra/location.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/parser/token_stream.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/py.typed +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/sandbox/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/sandbox/config.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/sandbox/errors.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/sandbox/schema.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/builder/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/builder/models.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/executor/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/jsonpath.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/registry/dict_constraints.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/resolver/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/resolver/identity.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/resolver/models.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/semantic/resolver/resolver.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/char_stream.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data/tokenizer/models/__init__.py +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data.egg-info/dependency_links.txt +0 -0
- {infinity_data-2.2.0 → infinity_data-3.1.1}/src/infinity_data.egg-info/top_level.txt +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
[project]
|
|
2
2
|
name = "infinity-data"
|
|
3
|
-
version = "
|
|
3
|
+
version = "3.1.1"
|
|
4
4
|
description = "InfinityData:声明式配置语言(.infd/.inft)的 Python 编译器库"
|
|
5
5
|
readme = "README.md"
|
|
6
6
|
license = { text = "MIT" }
|
|
@@ -95,6 +95,6 @@ venvPath = "."
|
|
|
95
95
|
venv = ".venv"
|
|
96
96
|
|
|
97
97
|
[tool.interrogate]
|
|
98
|
-
fail-under =
|
|
98
|
+
fail-under = 60
|
|
99
99
|
# 测试函数名自解释,docstring 可选;产品代码(src)不豁免
|
|
100
100
|
ignore-regex = ["^test_"]
|
|
@@ -0,0 +1,35 @@
|
|
|
1
|
+
"""产物发射层:Python / std → 输出格式(JSON / YAML / TOML)。
|
|
2
|
+
|
|
3
|
+
与 semantic 层边界:
|
|
4
|
+
- **内部表示(忠实双向)** → :mod:`infinity_data.semantic.std` 的
|
|
5
|
+
``python_to_std`` / ``std_to_python``(path → PurePosixPath、float → Decimal、
|
|
6
|
+
noexist → :data:`~infinity_data.semantic.std.NOEXIST` 哨兵)
|
|
7
|
+
- **本层(emit,有损投影)** → :func:`to_json` / :func:`to_yaml` / :func:`to_toml`,
|
|
8
|
+
按 :class:`EmitConfig` 控制 Decimal 编码与 noexist 标记;:func:`project_output`
|
|
9
|
+
为纯 Python 值的轻量投影;:func:`restore_python` 反向**尝试还原**加载的原始
|
|
10
|
+
dict 为 :data:`~infinity_data.semantic.std.StdPythonValue`
|
|
11
|
+
|
|
12
|
+
**emit 不应被流水线内部使用**——内部表示一律走 semantic 的忠实转换。
|
|
13
|
+
"""
|
|
14
|
+
|
|
15
|
+
from infinity_data.emit.config import EmitConfig
|
|
16
|
+
from infinity_data.emit.converter import (
|
|
17
|
+
dump_to_json,
|
|
18
|
+
load_from_json,
|
|
19
|
+
project_output,
|
|
20
|
+
restore_python,
|
|
21
|
+
to_json,
|
|
22
|
+
to_toml,
|
|
23
|
+
to_yaml,
|
|
24
|
+
)
|
|
25
|
+
|
|
26
|
+
__all__ = [
|
|
27
|
+
'EmitConfig',
|
|
28
|
+
'dump_to_json',
|
|
29
|
+
'load_from_json',
|
|
30
|
+
'project_output',
|
|
31
|
+
'restore_python',
|
|
32
|
+
'to_json',
|
|
33
|
+
'to_toml',
|
|
34
|
+
'to_yaml',
|
|
35
|
+
]
|
|
@@ -0,0 +1,37 @@
|
|
|
1
|
+
"""输出转换配置(emit 层):控制 std/python → 输出格式的有损投影语义。"""
|
|
2
|
+
|
|
3
|
+
from __future__ import annotations
|
|
4
|
+
|
|
5
|
+
from dataclasses import dataclass
|
|
6
|
+
|
|
7
|
+
__all__ = ['EmitConfig']
|
|
8
|
+
|
|
9
|
+
|
|
10
|
+
@dataclass(frozen=True)
|
|
11
|
+
class EmitConfig:
|
|
12
|
+
"""emit 层输出转换配置。
|
|
13
|
+
|
|
14
|
+
转换语义:
|
|
15
|
+
- ``full_float``:``float`` 值(:class:`decimal.Decimal`)以自描述标记
|
|
16
|
+
``{"__type__": "decimal", "num": "<十进制字符串>"}`` 编码(无损);False 时
|
|
17
|
+
有限值转普通 JSON 数字,NaN / ±Infinity 无法用 JSON 数字表达 → 仍以标记编码
|
|
18
|
+
- ``full_path``:路径值(:class:`PurePosixPath`)以自描述标记
|
|
19
|
+
``{"__type__": "path", "path": "<POSIX 字符串>"}`` 编码(可辨识「这是路径」);
|
|
20
|
+
False 时投影为普通 POSIX 字符串
|
|
21
|
+
- ``keep_noexist``:``noexist`` 字段保留为 ``{"__type__": "noexist"}`` 标记
|
|
22
|
+
(依赖 :data:`~infinity_data.semantic.std.NOEXIST` 哨兵保留三态);默认丢弃
|
|
23
|
+
(键不出现,§1.6)
|
|
24
|
+
|
|
25
|
+
序列化选项:
|
|
26
|
+
- ``indent`` / ``sort_keys`` / ``ensure_ascii``:JSON 缩进 / 排序 / ASCII 转义
|
|
27
|
+
"""
|
|
28
|
+
|
|
29
|
+
# ── 转换语义 ──────────────────────────────────────
|
|
30
|
+
full_float: bool = False
|
|
31
|
+
full_path: bool = False
|
|
32
|
+
keep_noexist: bool = False
|
|
33
|
+
|
|
34
|
+
# ── 序列化选项 ────────────────────────────────────
|
|
35
|
+
indent: int | None = 2
|
|
36
|
+
sort_keys: bool = False
|
|
37
|
+
ensure_ascii: bool = False
|
|
@@ -0,0 +1,258 @@
|
|
|
1
|
+
"""产物发射层:Python / std → 输出格式(JSON / YAML / TOML)。
|
|
2
|
+
|
|
3
|
+
职责边界(与 semantic 层严格分离):
|
|
4
|
+
|
|
5
|
+
- **内部表示**(忠实双向,语义最小丢失)::func:`~infinity_data.semantic.std.python_to_std`
|
|
6
|
+
与 :func:`~infinity_data.semantic.std.std_to_python`——path 保持
|
|
7
|
+
:class:`PurePosixPath`、float 保持 :class:`decimal.Decimal`、noexist 经
|
|
8
|
+
:data:`~infinity_data.semantic.std.NOEXIST` 哨兵保留三态。
|
|
9
|
+
流水线内部(如 :class:`~infinity_data.pipeline.CompilationResult.value`)一律走该层。
|
|
10
|
+
- **本层(emit)**:做**有损投影**供输出格式消费,无需担心完整性——
|
|
11
|
+
path → POSIX 字符串;Decimal 的 NaN / ±Infinity 等特殊编码由本层按
|
|
12
|
+
:class:`EmitConfig` 决定(``full_float`` 标记编码)。
|
|
13
|
+
|
|
14
|
+
入口:
|
|
15
|
+
- :func:`to_json` / :func:`to_yaml` / :func:`to_toml`:输出文本(接受 std 或 Python 值)
|
|
16
|
+
- :func:`project_output`:纯 Python 值的轻量有损投影(path → 字符串)
|
|
17
|
+
"""
|
|
18
|
+
|
|
19
|
+
from __future__ import annotations
|
|
20
|
+
|
|
21
|
+
import importlib
|
|
22
|
+
import json
|
|
23
|
+
from decimal import Decimal, InvalidOperation
|
|
24
|
+
from pathlib import PurePath, PurePosixPath
|
|
25
|
+
from typing import Any, cast
|
|
26
|
+
|
|
27
|
+
from infinity_data.emit.config import EmitConfig
|
|
28
|
+
from infinity_data.semantic.std import NOEXIST, StdPythonValue, is_std_value, std_to_python
|
|
29
|
+
|
|
30
|
+
__all__ = [
|
|
31
|
+
'dump_to_json',
|
|
32
|
+
'load_from_json',
|
|
33
|
+
'project_output',
|
|
34
|
+
'restore_python',
|
|
35
|
+
'to_json',
|
|
36
|
+
'to_yaml',
|
|
37
|
+
'to_toml',
|
|
38
|
+
]
|
|
39
|
+
|
|
40
|
+
# ── 自描述标记(输出格式无法原生表达的值的编码) ──────
|
|
41
|
+
|
|
42
|
+
|
|
43
|
+
def _decimal_marker(dec: Any) -> dict[str, str]:
|
|
44
|
+
"""Decimal 自描述标记:``{"__type__": "decimal", "num": "<十进制字符串>"}``。"""
|
|
45
|
+
return {'__type__': 'decimal', 'num': str(dec)}
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
def _path_marker(s: str) -> dict[str, str]:
|
|
49
|
+
"""路径自描述标记:``{"__type__": "path", "path": "<POSIX 字符串>"}``。"""
|
|
50
|
+
return {'__type__': 'path', 'path': s}
|
|
51
|
+
|
|
52
|
+
|
|
53
|
+
def _noexist_marker() -> dict[str, str]:
|
|
54
|
+
"""noexist 自描述标记:``{"__type__": "noexist"}``。"""
|
|
55
|
+
return {'__type__': 'noexist'}
|
|
56
|
+
|
|
57
|
+
|
|
58
|
+
# ── 标记还原(反向:加载的原始 dict → StdPythonValue) ────
|
|
59
|
+
|
|
60
|
+
|
|
61
|
+
def _restore_marker(value: dict[Any, Any]) -> Any | None:
|
|
62
|
+
"""若 dict 是 emit 自描述标记则还原为 Python 值,否则 None(非标记)。
|
|
63
|
+
|
|
64
|
+
仅识别 emit 层产出的**精确标记形状**;结构不符(额外键 / 负载非字符串 /
|
|
65
|
+
数值无法解析)一律视为普通 dict(返回 None,尽力而为)。
|
|
66
|
+
"""
|
|
67
|
+
typ = value.get('__type__')
|
|
68
|
+
if typ == 'decimal':
|
|
69
|
+
if set(value) == {'__type__', 'num'} and isinstance(value.get('num'), str):
|
|
70
|
+
try:
|
|
71
|
+
return Decimal(value['num'])
|
|
72
|
+
except (InvalidOperation, ValueError, TypeError):
|
|
73
|
+
return None
|
|
74
|
+
return None
|
|
75
|
+
if typ == 'path':
|
|
76
|
+
if set(value) == {'__type__', 'path'} and isinstance(value.get('path'), str):
|
|
77
|
+
return PurePosixPath(value['path'])
|
|
78
|
+
return None
|
|
79
|
+
if typ == 'noexist':
|
|
80
|
+
if set(value) == {'__type__'}:
|
|
81
|
+
return NOEXIST
|
|
82
|
+
return None
|
|
83
|
+
return None
|
|
84
|
+
|
|
85
|
+
|
|
86
|
+
def restore_python(value: Any) -> StdPythonValue:
|
|
87
|
+
"""从加载的原始 dict(含可选 emit 自描述标记)**尝试**还原为 StdPythonValue。
|
|
88
|
+
|
|
89
|
+
- ``{"__type__": "decimal", "num": "..."}`` → :class:`decimal.Decimal`
|
|
90
|
+
- ``{"__type__": "path", "path": "..."}`` → :class:`PurePosixPath`
|
|
91
|
+
- ``{"__type__": "noexist"}`` → :data:`NOEXIST`
|
|
92
|
+
- 普通 ``float`` → :class:`decimal.Decimal`(忠实形式,与
|
|
93
|
+
:func:`~infinity_data.semantic.std.python_to_std` 规范一致)
|
|
94
|
+
- int / bool / str / None 原样;dict / list 递归
|
|
95
|
+
- 识别不了的形状(含疑似标记但结构不符)按普通 dict / list 保留(尽力而为)
|
|
96
|
+
|
|
97
|
+
配合 :func:`to_json` 等 + ``EmitConfig(full_float=True, full_path=True,
|
|
98
|
+
keep_noexist=True)`` 可实现「std → 输出文本 → 加载 → StdPythonValue」
|
|
99
|
+
无损闭环(再经 :func:`~infinity_data.semantic.std.python_to_std` 回 std)。
|
|
100
|
+
"""
|
|
101
|
+
if isinstance(value, dict):
|
|
102
|
+
obj = cast(dict[Any, Any], value)
|
|
103
|
+
marker = _restore_marker(obj)
|
|
104
|
+
if marker is not None:
|
|
105
|
+
return marker
|
|
106
|
+
return {str(k): restore_python(v) for k, v in obj.items()}
|
|
107
|
+
if isinstance(value, list):
|
|
108
|
+
arr = cast(list[Any], value)
|
|
109
|
+
return [restore_python(v) for v in arr]
|
|
110
|
+
if isinstance(value, bool):
|
|
111
|
+
return value
|
|
112
|
+
if isinstance(value, int):
|
|
113
|
+
return value
|
|
114
|
+
if isinstance(value, float):
|
|
115
|
+
return Decimal(str(value))
|
|
116
|
+
return value
|
|
117
|
+
|
|
118
|
+
|
|
119
|
+
# ── 序列化核心 ──────────────────────────────────────
|
|
120
|
+
|
|
121
|
+
|
|
122
|
+
def _to_serializable(value: Any, config: EmitConfig) -> Any:
|
|
123
|
+
"""std 或 Python 值 → 可序列化结构(按 config 做有损投影)。"""
|
|
124
|
+
if is_std_value(value):
|
|
125
|
+
# std 输入:忠实转 Python(noexist 按需保留为 NOEXIST 哨兵),再投影
|
|
126
|
+
py = std_to_python(value, keep_noexist=config.keep_noexist)
|
|
127
|
+
return _serializable(py, config)
|
|
128
|
+
# Python 输入(如 CompilationResult.value):直接投影
|
|
129
|
+
return _serializable(value, config)
|
|
130
|
+
|
|
131
|
+
|
|
132
|
+
def _serializable(value: Any, config: EmitConfig) -> Any:
|
|
133
|
+
"""Python 值 → 可序列化结构(递归有损投影)。
|
|
134
|
+
|
|
135
|
+
- :data:`NOEXIST` 哨兵 → ``{"__type__": "noexist"}``
|
|
136
|
+
- :class:`PurePath` → POSIX 字符串
|
|
137
|
+
- :class:`decimal.Decimal` → 按 ``full_float`` 编码(有限值默认 JSON 数字)
|
|
138
|
+
- dict / list 递归;其余(int / bool / str / None 等)原样
|
|
139
|
+
"""
|
|
140
|
+
if value is NOEXIST:
|
|
141
|
+
return _noexist_marker()
|
|
142
|
+
if isinstance(value, PurePath):
|
|
143
|
+
s = value.as_posix()
|
|
144
|
+
return _path_marker(s) if config.full_path else s
|
|
145
|
+
if isinstance(value, dict):
|
|
146
|
+
obj = cast(dict[Any, Any], value)
|
|
147
|
+
return {k: _serializable(v, config) for k, v in obj.items()}
|
|
148
|
+
if isinstance(value, list):
|
|
149
|
+
arr = cast(list[Any], value)
|
|
150
|
+
return [_serializable(v, config) for v in arr]
|
|
151
|
+
if isinstance(value, Decimal):
|
|
152
|
+
return _decimal_serializable(value, config)
|
|
153
|
+
return value
|
|
154
|
+
|
|
155
|
+
|
|
156
|
+
def _decimal_serializable(dec: Any, config: EmitConfig) -> Any:
|
|
157
|
+
"""Decimal → 可序列化形式。
|
|
158
|
+
|
|
159
|
+
- ``full_float`` → 自描述标记(无损)
|
|
160
|
+
- 否则:有限值 → JSON 数字(整值转 int,避免 ``1.0``;溢出回退标记);
|
|
161
|
+
NaN / ±Infinity 无法用 JSON 数字表达 → 自描述标记
|
|
162
|
+
"""
|
|
163
|
+
if config.full_float:
|
|
164
|
+
return _decimal_marker(dec)
|
|
165
|
+
if dec.is_nan() or dec.is_infinite():
|
|
166
|
+
return _decimal_marker(dec)
|
|
167
|
+
try:
|
|
168
|
+
if dec == dec.to_integral_value():
|
|
169
|
+
return int(dec)
|
|
170
|
+
return float(dec)
|
|
171
|
+
except (OverflowError, ValueError):
|
|
172
|
+
return _decimal_marker(dec)
|
|
173
|
+
|
|
174
|
+
|
|
175
|
+
# ── 输出入口 ────────────────────────────────────────
|
|
176
|
+
|
|
177
|
+
|
|
178
|
+
def to_json(value: Any, *, config: EmitConfig | None = None) -> str:
|
|
179
|
+
"""std 或 Python 值 → JSON 文本。
|
|
180
|
+
|
|
181
|
+
``value`` 可为 :class:`~infinity_data.semantic.std.StdValue`(如
|
|
182
|
+
``CompilationResult.document.root``,支持完整选项)或 Python 值
|
|
183
|
+
(如 ``CompilationResult.value``,noexist 信息已在转换时丢失)。
|
|
184
|
+
"""
|
|
185
|
+
cfg = config or EmitConfig()
|
|
186
|
+
data = _to_serializable(value, cfg)
|
|
187
|
+
return json.dumps(data, indent=cfg.indent, sort_keys=cfg.sort_keys, ensure_ascii=cfg.ensure_ascii)
|
|
188
|
+
|
|
189
|
+
|
|
190
|
+
def to_yaml(value: Any, *, config: EmitConfig | None = None) -> str:
|
|
191
|
+
"""std 或 Python 值 → YAML 文本(需安装 PyYAML)。"""
|
|
192
|
+
cfg = config or EmitConfig()
|
|
193
|
+
data = _to_serializable(value, cfg)
|
|
194
|
+
try:
|
|
195
|
+
yaml = importlib.import_module('yaml')
|
|
196
|
+
except ImportError as exc: # pragma: no cover - 依赖缺失路径
|
|
197
|
+
raise ImportError('to_yaml 需要 PyYAML(pip install pyyaml)') from exc
|
|
198
|
+
return yaml.safe_dump(data, allow_unicode=not cfg.ensure_ascii, sort_keys=cfg.sort_keys)
|
|
199
|
+
|
|
200
|
+
|
|
201
|
+
def to_toml(value: Any, *, config: EmitConfig | None = None) -> str:
|
|
202
|
+
"""std 或 Python 值 → TOML 文本(需安装 tomli-w)。"""
|
|
203
|
+
cfg = config or EmitConfig()
|
|
204
|
+
data = _to_serializable(value, cfg)
|
|
205
|
+
try:
|
|
206
|
+
tomli_w = importlib.import_module('tomli_w')
|
|
207
|
+
except ImportError as exc: # pragma: no cover - 依赖缺失路径
|
|
208
|
+
raise NotImplementedError('to_toml 需要 tomli-w(pip install tomli-w);当前未安装') from exc
|
|
209
|
+
return tomli_w.dumps(data)
|
|
210
|
+
|
|
211
|
+
|
|
212
|
+
# ── 便捷自由函数(默认尝试保留一切语义) ────────────
|
|
213
|
+
|
|
214
|
+
|
|
215
|
+
def dump_to_json(value: Any, *, config: EmitConfig | None = None) -> str:
|
|
216
|
+
"""std / Python 值 → JSON 文本(便捷入口,**默认保留一切语义**)。
|
|
217
|
+
|
|
218
|
+
默认等价 ``EmitConfig(full_float=True, full_path=True, keep_noexist=True)``——
|
|
219
|
+
Decimal / PurePosixPath / noexist 以 ``{"__type__": ...}`` 标记编码;
|
|
220
|
+
传入 ``config`` 可覆盖(关闭标记、改缩进等,见 :class:`EmitConfig`)。
|
|
221
|
+
|
|
222
|
+
与 :func:`load_from_json` 配对可实现「std → JSON → StdPythonValue」无损闭环。
|
|
223
|
+
"""
|
|
224
|
+
cfg = config or EmitConfig(full_float=True, full_path=True, keep_noexist=True)
|
|
225
|
+
return to_json(value, config=cfg)
|
|
226
|
+
|
|
227
|
+
|
|
228
|
+
def load_from_json(text: str) -> StdPythonValue:
|
|
229
|
+
"""JSON 文本 → StdPythonValue(便捷入口,**默认尝试还原全部语义**)。
|
|
230
|
+
|
|
231
|
+
解析后经 :func:`restore_python`:``{"__type__": ...}`` 标记还原为
|
|
232
|
+
Decimal / PurePosixPath / NOEXIST,普通 float 归一到 Decimal;
|
|
233
|
+
识别不了的形状原样保留(尽力而为)。
|
|
234
|
+
"""
|
|
235
|
+
return restore_python(json.loads(text))
|
|
236
|
+
|
|
237
|
+
|
|
238
|
+
# ── 轻量有损投影(纯 Python 值) ─────────────────────
|
|
239
|
+
|
|
240
|
+
|
|
241
|
+
def project_output(value: Any) -> Any:
|
|
242
|
+
"""Python 值 → 输出友好投影(递归,有损,无配置)。
|
|
243
|
+
|
|
244
|
+
- :class:`PurePath`(含 :class:`PurePosixPath`)→ POSIX 字符串
|
|
245
|
+
- dict / list 递归投影;其余(int / Decimal / bool / str / None 等)原样保留
|
|
246
|
+
|
|
247
|
+
完整选项(Decimal 编码 / noexist 标记)请用 :func:`to_json` 等带
|
|
248
|
+
:class:`EmitConfig` 的入口。
|
|
249
|
+
"""
|
|
250
|
+
if isinstance(value, PurePath):
|
|
251
|
+
return value.as_posix()
|
|
252
|
+
if isinstance(value, dict):
|
|
253
|
+
obj = cast(dict[Any, Any], value)
|
|
254
|
+
return {k: project_output(v) for k, v in obj.items()}
|
|
255
|
+
if isinstance(value, list):
|
|
256
|
+
arr = cast(list[Any], value)
|
|
257
|
+
return [project_output(v) for v in arr]
|
|
258
|
+
return value
|
|
@@ -14,7 +14,23 @@ from __future__ import annotations
|
|
|
14
14
|
import sys
|
|
15
15
|
from pathlib import Path, PurePosixPath, PureWindowsPath
|
|
16
16
|
|
|
17
|
-
__all__ = ['posix_to_native', 'native_to_posix']
|
|
17
|
+
__all__ = ['posix_to_native', 'native_to_posix', 'is_valid_posix_path']
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
def is_valid_posix_path(s: str) -> bool:
|
|
21
|
+
"""语言内 POSIX 路径合法性(纯语法校验,无文件系统访问)。
|
|
22
|
+
|
|
23
|
+
- 非空
|
|
24
|
+
- 无 NUL 字符(:class:`PurePosixPath` 拒绝)
|
|
25
|
+
- 可被 :class:`PurePosixPath` 解析
|
|
26
|
+
"""
|
|
27
|
+
if not s or '\x00' in s:
|
|
28
|
+
return False
|
|
29
|
+
try:
|
|
30
|
+
PurePosixPath(s)
|
|
31
|
+
return True
|
|
32
|
+
except ValueError:
|
|
33
|
+
return False
|
|
18
34
|
|
|
19
35
|
|
|
20
36
|
def posix_to_native(path: str, *, platform: str | None = None) -> Path:
|
|
@@ -34,6 +34,7 @@ from infinity_data.parser.models import (
|
|
|
34
34
|
JsonPathKey,
|
|
35
35
|
JsonPathSegment,
|
|
36
36
|
LiteralValue,
|
|
37
|
+
NamedArg,
|
|
37
38
|
Statement,
|
|
38
39
|
TemplateCallValue,
|
|
39
40
|
TemplateConfig,
|
|
@@ -72,6 +73,7 @@ __all__ = [
|
|
|
72
73
|
'JsonPathKey',
|
|
73
74
|
'JsonPathSegment',
|
|
74
75
|
'LiteralValue',
|
|
76
|
+
'NamedArg',
|
|
75
77
|
'Parser',
|
|
76
78
|
'Statement',
|
|
77
79
|
'TemplateCallValue',
|
|
@@ -63,6 +63,11 @@ register_diagnostic_define(
|
|
|
63
63
|
'[{location}] 导入语句必须以换行结尾(同一行逗号后不能再接其他语句)',
|
|
64
64
|
en='[{location}] import statement must end with a newline (cannot be followed by more tokens on the same line)',
|
|
65
65
|
),
|
|
66
|
+
diagnostic_define(
|
|
67
|
+
'parse.import_path_required',
|
|
68
|
+
'[{location}] 导入路径必须用 p"..." 路径字面量(§1.5 / §3.2),已按字符串容错取用',
|
|
69
|
+
en='[{location}] import path must use a p"..." path literal (§1.5 / §3.2); recovered as string',
|
|
70
|
+
),
|
|
66
71
|
diagnostic_define(
|
|
67
72
|
'parse.value_field',
|
|
68
73
|
'[{location}] 值位置出现字段定义 {name}(外层数组/对象未闭合)',
|
|
@@ -14,6 +14,7 @@ from infinity_data.tokenizer.models.tokens import (
|
|
|
14
14
|
IntegerToken,
|
|
15
15
|
NoexistToken,
|
|
16
16
|
NullToken,
|
|
17
|
+
PathToken,
|
|
17
18
|
StringToken,
|
|
18
19
|
)
|
|
19
20
|
|
|
@@ -27,6 +28,11 @@ def _canonical_str(s: str) -> str:
|
|
|
27
28
|
return json.dumps(s, ensure_ascii=False)
|
|
28
29
|
|
|
29
30
|
|
|
31
|
+
def _canonical_path(s: str) -> str:
|
|
32
|
+
"""路径 → 标准 infd 路径字面量(``p"..."``,§1.5)。"""
|
|
33
|
+
return 'p' + json.dumps(s, ensure_ascii=False)
|
|
34
|
+
|
|
35
|
+
|
|
30
36
|
def _canonical_constraint_list(constraints: Iterable['Constraint']) -> str:
|
|
31
37
|
"""约束列表 → infd 类型标注:单约束省略尖括号,多约束 ``<a, b>``。"""
|
|
32
38
|
items = list(constraints)
|
|
@@ -37,6 +43,16 @@ def _canonical_constraint_list(constraints: Iterable['Constraint']) -> str:
|
|
|
37
43
|
return '<' + ', '.join(c.canonical() for c in items) + '>'
|
|
38
44
|
|
|
39
45
|
|
|
46
|
+
def _canonical_error(message: str) -> str:
|
|
47
|
+
"""错误节点容错序列化:``error("...")`` 占位符。
|
|
48
|
+
|
|
49
|
+
错误节点是**解析失败恢复**的产物,对应的源码本就无效,无法还原为等价 AST——
|
|
50
|
+
canonical 仅作**容错**(模板真名哈希 / 调试 / LSP 提示不因错误节点崩溃),
|
|
51
|
+
不承诺 round-trip(``error(...)`` 是显示占位符,不是可解析的模板调用语法)。
|
|
52
|
+
"""
|
|
53
|
+
return f'error({_canonical_str(message)})'
|
|
54
|
+
|
|
55
|
+
|
|
40
56
|
# ═══════════════════════════════════════════════════════════
|
|
41
57
|
# 节点
|
|
42
58
|
# ═══════════════════════════════════════════════════════════
|
|
@@ -106,7 +122,7 @@ class TemplateImportItem(AstNode):
|
|
|
106
122
|
|
|
107
123
|
@dataclass
|
|
108
124
|
class TemplateImportStmt(AstNode):
|
|
109
|
-
"""模板导入: !from "path" import Name1, Name2 as N2"""
|
|
125
|
+
"""模板导入: !from p"path" import Name1, Name2 as N2"""
|
|
110
126
|
|
|
111
127
|
from_path: str # 文件路径(unix 风格)
|
|
112
128
|
items: list[TemplateImportItem] # 导入项列表
|
|
@@ -116,7 +132,7 @@ class TemplateImportStmt(AstNode):
|
|
|
116
132
|
|
|
117
133
|
def canonical(self) -> str:
|
|
118
134
|
items = ', '.join(i.canonical() for i in self.items)
|
|
119
|
-
return f'!from {
|
|
135
|
+
return f'!from {_canonical_path(self.from_path)} import {items}'
|
|
120
136
|
|
|
121
137
|
|
|
122
138
|
@dataclass
|
|
@@ -189,7 +205,7 @@ class FileImportItem(AstNode):
|
|
|
189
205
|
|
|
190
206
|
@dataclass
|
|
191
207
|
class FileImportStmt(AstNode):
|
|
192
|
-
"""配置文件导入: !file "path" as <format> import .path.to.key as alias, ..."""
|
|
208
|
+
"""配置文件导入: !file p"path" as <format> import .path.to.key as alias, ..."""
|
|
193
209
|
|
|
194
210
|
file_path: str # 文件路径
|
|
195
211
|
format: str | None # 文件格式: "yaml", "json", "toml", "raw" 或 None(自动检测后缀)
|
|
@@ -201,7 +217,7 @@ class FileImportStmt(AstNode):
|
|
|
201
217
|
def canonical(self) -> str:
|
|
202
218
|
fmt = f' as {self.format}' if self.format else ''
|
|
203
219
|
items = ', '.join(i.canonical() for i in self.imports)
|
|
204
|
-
return f'!file {
|
|
220
|
+
return f'!file {_canonical_path(self.file_path)}{fmt} import {items}'
|
|
205
221
|
|
|
206
222
|
|
|
207
223
|
@dataclass
|
|
@@ -418,7 +434,7 @@ class Constraints(AstNode):
|
|
|
418
434
|
class LiteralValue(AstNode):
|
|
419
435
|
"""字面量值"""
|
|
420
436
|
|
|
421
|
-
value: FloatToken | IntegerToken | BoolToken | NullToken | NoexistToken | StringToken
|
|
437
|
+
value: FloatToken | IntegerToken | BoolToken | NullToken | NoexistToken | StringToken | PathToken
|
|
422
438
|
|
|
423
439
|
def canonical(self) -> str:
|
|
424
440
|
return self.value.canonical()
|
|
@@ -432,7 +448,7 @@ class DollarValue(AstNode):
|
|
|
432
448
|
"""
|
|
433
449
|
|
|
434
450
|
name: str # 变量名
|
|
435
|
-
type_cast: Literal['int', 'float', 'bool', 'str', None] # 可选类型转换
|
|
451
|
+
type_cast: Literal['int', 'float', 'bool', 'str', 'path', None] # 可选类型转换
|
|
436
452
|
|
|
437
453
|
def canonical(self) -> str:
|
|
438
454
|
if self.type_cast is None:
|
|
@@ -457,24 +473,41 @@ class UnpackValue(AstNode):
|
|
|
457
473
|
return ('**' if self.double else '*') + self.value.canonical()
|
|
458
474
|
|
|
459
475
|
|
|
476
|
+
@dataclass
|
|
477
|
+
class NamedArg(AstNode):
|
|
478
|
+
"""模板调用命名参数槽: name = value。
|
|
479
|
+
|
|
480
|
+
与 ``**expr`` 解包项(:class:`UnpackValue`)在 :attr:`TemplateCallValue.named_args`
|
|
481
|
+
中按**源码顺序**混排——键冲突「保留先到者」(disjoint merge,§2.7)与书写顺序一致。
|
|
482
|
+
"""
|
|
483
|
+
|
|
484
|
+
name: str
|
|
485
|
+
value: Value
|
|
486
|
+
|
|
487
|
+
def children(self) -> Iterable[AstNode]:
|
|
488
|
+
return [self.value]
|
|
489
|
+
|
|
490
|
+
def canonical(self) -> str:
|
|
491
|
+
return f'{self.name} = {self.value.canonical()}'
|
|
492
|
+
|
|
493
|
+
|
|
460
494
|
@dataclass
|
|
461
495
|
class DictValue(AstNode):
|
|
462
496
|
"""对象值: { ... }
|
|
463
497
|
|
|
464
498
|
``: <constraint, ...>`` 结构级约束作用于该字面量 dict 的整体。
|
|
465
|
-
``
|
|
499
|
+
``items``:字段(:class:`Field`)与 ``**expr`` 解包项(:class:`UnpackValue`)按
|
|
500
|
+
**源码顺序**混排——键冲突「保留先到者」(disjoint merge,§2.7)与书写顺序一致。
|
|
466
501
|
"""
|
|
467
502
|
|
|
468
|
-
|
|
503
|
+
items: list[Field | UnpackValue]
|
|
469
504
|
constraints: list[Constraint] = field(default_factory=lambda: [])
|
|
470
|
-
unpacks: list[UnpackValue] = field(default_factory=lambda: [])
|
|
471
505
|
|
|
472
506
|
def children(self) -> Iterable[AstNode]:
|
|
473
|
-
return [*self.
|
|
507
|
+
return [*self.items, *self.constraints]
|
|
474
508
|
|
|
475
509
|
def canonical(self) -> str:
|
|
476
|
-
inner: list[str] = [
|
|
477
|
-
inner.extend(u.canonical() for u in self.unpacks)
|
|
510
|
+
inner: list[str] = [i.canonical() for i in self.items]
|
|
478
511
|
cs = _canonical_constraint_list(self.constraints)
|
|
479
512
|
if cs:
|
|
480
513
|
inner.append(': ' + cs)
|
|
@@ -501,18 +534,21 @@ class TemplateCallValue(AstNode):
|
|
|
501
534
|
展开(§2.8):参数值后缀 ``...`` = 展开轴;调用级(``)`` 后)``...`` = 展开传播
|
|
502
535
|
(本调用的展开结果作为包围模板调用的轴,整体模式逐元素重复)、``^`` = 笛卡尔积
|
|
503
536
|
(多轴全组合,首轴最慢变化;与 ``propagate`` 正交,``^...`` 可叠加)。
|
|
504
|
-
- ``
|
|
505
|
-
|
|
506
|
-
- ``
|
|
537
|
+
- ``positional_args``:位置参数槽,**保持源码顺序**;``*expr`` 解包项
|
|
538
|
+
(:class:`UnpackValue`,double=False)也在其中,实例化时展开为多个位置参数(§2.7)
|
|
539
|
+
- ``named_args``:命名参数槽,**保持源码顺序**;显式 ``name = value``(:class:`NamedArg`)
|
|
540
|
+
与 ``**expr`` 解包项(:class:`UnpackValue`,double=True)混排——键冲突「保留先到者」
|
|
541
|
+
与书写顺序一致(§2.7 disjoint merge)
|
|
542
|
+
- ``axis_positional``:位置参数槽中带 ``...`` 的索引(``*expr`` 解包项不是轴)
|
|
543
|
+
- ``axis_named``:命名参数(:class:`NamedArg`)中带 ``...`` 的键
|
|
544
|
+
- ``axis_unpack_kwargs``:``**expr`` 解包项(named_args 槽索引)中带 ``...`` 的索引
|
|
507
545
|
- ``propagate``:调用级 ``...``(展开传播;无 = 展开止于本调用)
|
|
508
546
|
- ``cartesian``:调用级 ``^``(笛卡尔积展开;无 = zip 配对)
|
|
509
547
|
"""
|
|
510
548
|
|
|
511
549
|
template_name: str
|
|
512
|
-
positional_args: list[Value]
|
|
513
|
-
named_args:
|
|
514
|
-
unpack_args: list[UnpackValue] = field(default_factory=lambda: []) # *expr(list → 位置参数)
|
|
515
|
-
unpack_kwargs: list[UnpackValue] = field(default_factory=lambda: []) # **expr(dict → 命名参数)
|
|
550
|
+
positional_args: list[Value | UnpackValue]
|
|
551
|
+
named_args: list[NamedArg | UnpackValue]
|
|
516
552
|
axis_positional: frozenset[int] = field(default_factory=frozenset[int])
|
|
517
553
|
axis_named: frozenset[str] = field(default_factory=frozenset[str])
|
|
518
554
|
axis_unpack_kwargs: frozenset[int] = field(default_factory=frozenset[int])
|
|
@@ -522,27 +558,22 @@ class TemplateCallValue(AstNode):
|
|
|
522
558
|
def children(self) -> Iterable[AstNode]:
|
|
523
559
|
return [
|
|
524
560
|
*self.positional_args,
|
|
525
|
-
*self.named_args
|
|
526
|
-
*self.unpack_args,
|
|
527
|
-
*self.unpack_kwargs,
|
|
561
|
+
*self.named_args,
|
|
528
562
|
]
|
|
529
563
|
|
|
530
564
|
def canonical(self) -> str:
|
|
531
565
|
args: list[str] = []
|
|
532
566
|
for i, a in enumerate(self.positional_args):
|
|
533
567
|
s = a.canonical()
|
|
534
|
-
if i in self.axis_positional:
|
|
535
|
-
s += '...'
|
|
536
|
-
args.append(s)
|
|
537
|
-
args.extend(u.canonical() for u in self.unpack_args)
|
|
538
|
-
for k, v in sorted(self.named_args.items()):
|
|
539
|
-
s = f'{k} = {v.canonical()}'
|
|
540
|
-
if k in self.axis_named:
|
|
568
|
+
if not isinstance(a, UnpackValue) and i in self.axis_positional:
|
|
541
569
|
s += '...'
|
|
542
570
|
args.append(s)
|
|
543
|
-
for
|
|
544
|
-
s =
|
|
545
|
-
if
|
|
571
|
+
for i, n in enumerate(self.named_args):
|
|
572
|
+
s = n.canonical()
|
|
573
|
+
if isinstance(n, UnpackValue):
|
|
574
|
+
if i in self.axis_unpack_kwargs:
|
|
575
|
+
s += '...'
|
|
576
|
+
elif n.name in self.axis_named:
|
|
546
577
|
s += '...'
|
|
547
578
|
args.append(s)
|
|
548
579
|
out = f'{self.template_name}({", ".join(args)})'
|
|
@@ -560,32 +591,41 @@ class TemplateCallValue(AstNode):
|
|
|
560
591
|
|
|
561
592
|
@dataclass
|
|
562
593
|
class ErrorStatement(AstNode):
|
|
563
|
-
"""解析失败的语句。用于错误恢复。
|
|
594
|
+
"""解析失败的语句。用于错误恢复。
|
|
595
|
+
|
|
596
|
+
canonical 容错输出 ``error("message")``(不抛异常),供哈希 / LSP 提示。
|
|
597
|
+
"""
|
|
564
598
|
|
|
565
599
|
message: str
|
|
566
600
|
|
|
567
601
|
def canonical(self) -> str:
|
|
568
|
-
|
|
602
|
+
return _canonical_error(self.message)
|
|
569
603
|
|
|
570
604
|
|
|
571
605
|
@dataclass
|
|
572
606
|
class ErrorValue(AstNode):
|
|
573
|
-
"""解析失败的值。用于错误恢复。
|
|
607
|
+
"""解析失败的值。用于错误恢复。
|
|
608
|
+
|
|
609
|
+
canonical 容错输出 ``error("message")``(不抛异常),供哈希 / LSP 提示。
|
|
610
|
+
"""
|
|
574
611
|
|
|
575
612
|
message: str
|
|
576
613
|
|
|
577
614
|
def canonical(self) -> str:
|
|
578
|
-
|
|
615
|
+
return _canonical_error(self.message)
|
|
579
616
|
|
|
580
617
|
|
|
581
618
|
@dataclass
|
|
582
619
|
class ErrorConstraint(AstNode):
|
|
583
|
-
"""解析失败的约束。用于错误恢复。
|
|
620
|
+
"""解析失败的约束。用于错误恢复。
|
|
621
|
+
|
|
622
|
+
canonical 容错输出 ``error("message")``(不抛异常),供哈希 / LSP 提示。
|
|
623
|
+
"""
|
|
584
624
|
|
|
585
625
|
message: str
|
|
586
626
|
|
|
587
627
|
def canonical(self) -> str:
|
|
588
|
-
|
|
628
|
+
return _canonical_error(self.message)
|
|
589
629
|
|
|
590
630
|
|
|
591
631
|
# ═══════════════════════════════════════════════════════════
|