DevConfig-Gen

DevConfig-Gen 是一个开发者工具，用于从结构化输入数据和映射关系生成和校验结构化配置。它围绕一个轻量的 Provider 契约构建，使生成流水线独立于任何特定的配置格式或目标系统。多个输入文档可以在生成前进行合并和覆盖。

A developer tool for generating and validating structured configuration from structured input data and mappings. Built around a small provider contract so the generation pipeline is independent of any particular configuration format or target system. Multiple input documents can be merged and overridden before generation.

内置三个 Provider / Ships with three providers:

• custom — 无 schema 的通用文档，任意 JSON/YAML 层级可增删清空 / schema-free generic document with arbitrary nesting you can add, remove, or clear;
• json — 无依赖的透传/重新序列化 / dependency-free pass-through/re-serialization;
• env — 将嵌套数据扁平化为 UPPER_SNAKE_CASE 的 .env 文件 / flattens nested data into an UPPER_SNAKE_CASE .env file.

它是什么 / What it is

DevConfig-Gen：

• 加载结构化文档（JSON 或 YAML）/ loads a structured document (JSON or YAML);
• 将其规范化为可预测的结构 / normalizes it into a predictable shape;
• 校验并报告每个问题的精确字段路径 / validates and reports every problem with a precise field path;
• 通过 Provider 生成结构化配置文档 / generates a structured configuration document through a provider;
• 将结果写为 JSON 或 YAML / writes the result as JSON or YAML.

刻意保持本地化且无副作用 / Deliberately local and side-effect free. 不会访问远程服务、安装包、管理系统状态 / Does not contact remote services, install packages, or manage system state.

快速开始 / Quick start (3 minutes)

pip install devconfig-gen
devconfig-gen generate \
  --provider custom --input examples/custom.yaml --output-dir generated --format yaml
devconfig-gen validate --provider custom --input examples/custom.yaml

想要引导式流程？运行终端向导或打开本地 Web 工作台 / Prefer a guided flow?

devconfig-gen init --provider custom   # 终端向导 / terminal wizard
devconfig-gen ui                         # Web 工作台 / web studio

安装 / Installation

从 PyPI 安装 / Install from PyPI:

pip install devconfig-gen

YAML 支持无需外部依赖 / YAML support works with no external dependencies.

pip install devconfig-gen[yaml]   # 可选：安装 PyYAML / optional: use PyYAML

开发模式 / Development (editable):

pip install -e ".[yaml]"

命令行 / CLI

列出可用 Provider / List available providers:

devconfig-gen providers
# custom / env / json

生成配置 / Generate configuration:

devconfig-gen generate \
  --provider custom \
  --input examples/custom.yaml \
  --output-dir generated \
  --format yaml

仅校验不写入 / Validate without writing:

devconfig-gen validate --provider custom --input examples/custom.yaml

校验失败时报告字段路径 / Validation failure reports field paths:

$ devconfig-gen validate --provider env --input broken.yaml
invalid: variables must not be empty

结构化诊断输出 / Structured diagnostics:

devconfig-gen validate --provider env --input broken.yaml --json

查看 Provider Schema / Inspect provider schema:

devconfig-gen schema --provider custom

交互式终端向导 / Interactive Terminal Wizard

适用于无头环境、SSH 会话 / For headless servers, SSH sessions:

devconfig-gen init --provider custom

向导按 ProviderField.type 提示每个字段，含默认值、类型校验、选项、边界检查 / Prompts every field by type with defaults, type validation, choices, bounds checking.

========================================================
  DevConfig-Gen Interactive Wizard: 'custom'
  Answer the prompts below. Press Enter to use defaults.
========================================================

--- [1/1] Custom document ---
  自由构建任意嵌套的 JSON/YAML 结构；任意层级都可增删或清空。
? document (load document file or enter entries):
    Path to JSON/YAML file (or press enter for key=value input): examples/custom.yaml
    [✓] Loaded document from examples/custom.yaml

Validating configuration...
[✓] All validations passed!

Select output format (1: YAML [default], 2: JSON): 1
Writing configuration to '.'...
[✓] Generated artifact: .../custom.yaml

Web 可视化工作台 / Configuration Studio WebUI

启动本地 Web 工作台 / Launch local web studio:

devconfig-gen ui
devconfig-gen ui --workspace ~/projects/my-app   # 绑定项目目录 / bind to project dir

浏览器打开 http://127.0.0.1:8848 / Opens at http://127.0.0.1:8848.

特性 / Features:

• 中英双语界面，一键切换，偏好本地保存 / Bilingual UI (中文/English) with one-click toggle and saved preference;
• Apple 原生排版，亮色/暗色主题 / Apple-native typography, light/dark theme;
• 分步表单向导，内联校验 / Step-by-step wizard with inline validation;
• custom Provider 递归树编辑器：任意层级增删字段/项、切换类型、逐层清空 / recursive tree editor for the custom provider: add/remove fields or items at any depth, switch types, clear per node;
• json Provider 文档上传与内联编辑器（拖拽 .json/.yaml 反向解析）/ document drop-zone and inline editor for the json provider (drag .json/.yaml to backfill);
• 顶部「全部清空」一键重置当前 Provider 内容 / header "Clear All" resets the current provider;
• 双栏实时预览 / Dual-pane live preview;
• 模板预设、文件上传、草稿保存、磁盘导出 / Template presets, file upload, auto-save, disk export;
• 输出格式开关（YAML/JSON，XML 及未来格式已预留）/ output format switch (YAML/JSON; XML and future formats reserved);
• 零外部依赖 / Zero external build dependencies.

Provider 的步骤与字段元数据自带 i18n 翻译（内置 Provider 已提供中文）/ Provider step and field metadata carry optional i18n translations (the built-in providers ship Chinese).

仅绑定本地回环 / Binds to loopback only. 磁盘导出沙箱限制在工作空间内 / Disk export sandboxed to workspace root.

退出码 / Exit codes: 0 成功/success, 1 校验失败/validation failure, 2 输入错误/input error.

多源合并与覆盖 / Multi-source input and overrides

--input 可重复，文档从左到右深度合并 / --input may be repeated, deep-merged left to right:

devconfig-gen generate \
  --provider custom \
  --input configs/base.yaml \
  --input configs/prod.json \
  --set app.port=9090 \
  --set app.environment=production \
  --output-dir dist --format yaml

--set 值自动解析 JSON 类型 / --set values parsed as JSON when possible (true→bool, 42→int, null→null).

生成任意结构的自定义文档 / Generating an arbitrary custom document

custom Provider 不限定 schema，任意 JSON/YAML 层级都能生成 / The custom provider imposes no schema; any JSON/YAML nesting is generated as-is:

devconfig-gen generate --provider custom --input configs/anything.yaml --output-dir dist --format yaml

from devconfig_gen import GenerationRequest, generate

result = generate(
    "custom",
    GenerationRequest(
        context={"document": {"app": {"name": "web", "limits": {"cpu": "500m"}}}},
        options={"format": "yaml"},
    ),
)

在 Web 工作台中选择 custom，它提供两种编辑模式 / In the studio, custom offers two editing modes:

• 结构模式：递归树编辑器，任意层级增删字段/项、切换类型（string/number/boolean/object/array/null）、逐层清空 / Tree mode: a recursive editor to add/remove fields or items at any depth, switch types, and clear per node;
• 文本模式：直接输入 JSON/YAML，支持列表 [1,2,3,4,5]、嵌套 {"1":{"2":{}}}、YAML 缩进与 0: [1,2,3,4,5] 等任意混合 / Text mode: type JSON/YAML directly — lists [1,2,3,4,5], nesting {"1":{"2":{}}}, YAML indentation, and 0: [1,2,3,4,5], freely mixed.

顶层可以是映射、列表或标量 / The root may be a mapping, a list, or a scalar.

顶部「全部清空」一键重置 / The header "Clear All" resets the current provider.

生成 .env 文件 / Generating a .env file

devconfig-gen generate --provider env --input examples/vars.yaml --output-dir dist
# generated dist/.env

DATABASE_HOST=localhost
DATABASE_PORT=5432
DATABASE_NAME=appdb
DEBUG=true
LOG_LEVEL=info

Python API

from devconfig_gen import GenerationRequest, generate, generate_from_file

# 内存上下文 / In-memory context
result = generate(
    "custom",
    GenerationRequest(
        context={"document": {"app": {"name": "checkout-api", "port": 8080}}},
        options={"format": "yaml"},
    ),
)
artifact = result.artifacts[0]
print(artifact.name, artifact.media_type)   # custom.yaml application/yaml

# 文件到文件，与 CLI 相同流水线 / File to file, same pipeline as CLI
generate_from_file(
    "custom",
    "examples/custom.yaml",
    output_dir="generated",
    output_format="yaml",
)

直接加载和序列化 / Load and serialize directly:

from devconfig_gen import load_file, loads, dumps

data = load_file("examples/custom.yaml")
text = dumps(data, "json")

结构化诊断与 Schema 元数据 / Diagnostics and schema metadata:

from devconfig_gen import diagnose_request, describe_provider

for d in diagnose_request("env", context={}):
    print(d.field, "->", d.message, f"({d.severity})")

for step in describe_provider("custom"):
    print(step.id, step.title, [f.name for f in step.fields])

示例输入 / Example input

examples/custom.yaml:

app:
  name: checkout-api
  version: "2.4.0"
  port: 8080
  environment: production
  replicas: 3
  labels:
    team: payments
    tier: backend
  health_check:
    path: /healthz
    interval_seconds: 15
    timeout_seconds: 5

examples/custom.json 是等效 JSON 文档 / is the equivalent JSON document.

JSON/YAML 支持与限制 / Support and limitations

JSON 由标准库处理 / JSON handled by standard library. YAML 在安装 PyYAML 时使用，否则用内置解析器 / YAML uses PyYAML when available, otherwise bundled parser.

支持 / Supported:

• 缩进嵌套的映射和序列 / mappings and sequences by indentation;
• 标量：字符串、整数、浮点、布尔、null / scalars: strings, integers, floats, booleans, null;
• 引号字符串和裸字符串 / quoted and plain strings;
• 流式集合 [a, b] {x: 1} / flow collections;
• 注释和空行 / comments and blank lines;
• 块标量 | > 及 -/+ 修剪 / block scalars with chomping.

不支持 / Not supported:

• 锚点别名 &/*、自定义标签 !tag、合并键 << / anchors, aliases, tags, merge keys;
• 多文档 --- / multiple documents;
• 块标 scalar 内注释可能丢失 / comments inside block scalars may be dropped.

需要完整 YAML 行为请安装 PyYAML / For full YAML, install PyYAML.

env Provider 渲染纯文本 .env（media_type: text/plain）/ renders plain-text .env.

架构与数据流 / Architecture and data flow

输入文件 / 上下文 (input file / context)
        |
        v
  formats.load_*        JSON/YAML 解析 + 格式检测
        |
        v
  Provider.diagnose     规范化 + 结构化诊断
        |
        v
  Provider.generate     结构化配置文档
        |
        v
  engine.generate       产物序列化 + 可选持久化
        |
        v
   JSON / YAML 输出

CLI、Python API、向导、工作台调用相同的 engine 函数 / CLI, API, wizard, and studio all call the same engine functions. CLI 本身不含生成逻辑 / CLI contains no generation logic of its own.

核心模块 / Key modules:

模块 / Module	职责 / Responsibility
formats	JSON/YAML 加载导出、格式检测、媒体类型
validation	路径感知校验辅助工具 / path-aware validation helpers
models	请求、产物、诊断、字段、步骤、Provider 协议
registry	ProviderRegistry 及内置 Provider
engine	编排、诊断、Schema、持久化 / orchestration, diagnostics, persistence
providers	custom、json、env Provider
interactive	init 终端向导（延迟加载）/ terminal wizard (lazy-loaded)
web_ui	ui 本地工作台（延迟加载）/ local studio (lazy-loaded)
cli	仅参数解析 / argument parsing only

编写自定义 Provider / Writing a custom provider (5 minutes)

实现 ConfigProvider 协议 / Implement the ConfigProvider protocol. 只需 name、validate、generate 必需 / Only name, validate, generate required.

from devconfig_gen import (
    Diagnostic, GeneratedArtifact, GenerationResult,
    ProviderField, ProviderStep, ValidationError,
)

class GreetingProvider:
    name = "greeting"
    steps = (
        ProviderStep(
            id="input",
            title="Greeting input",
            i18n={"zh": {"title": "问候输入", "description": "输入要问候的对象。"}},
            fields=(
                ProviderField(
                    "who",
                    type="string",
                    required=True,
                    title="Who",
                    description="Who to greet.",
                    i18n={"zh": {"title": "对象", "description": "要问候的对象。"}},
                ),
            ),
        ),
    )

    def describe_schema(self):
        return self.steps

    def diagnose(self, request):
        if not request.context.get("who"):
            return (Diagnostic("who", "missing required field: 'who'"),)
        return ()

    def validate(self, request):
        return tuple(item.message for item in self.diagnose(request))

    def generate(self, request):
        diagnostics = self.diagnose(request)
        if diagnostics:
            raise ValidationError(diagnostics)
        return GenerationResult(
            provider=self.name,
            artifacts=(
                GeneratedArtifact(
                    name="greeting.json",
                    content={"message": f"hello {request.context['who']}"},
                    media_type="application/json",
                ),
            ),
        )

注册并运行 / Register and run:

from devconfig_gen import GenerationRequest, ProviderRegistry, generate
from devconfig_gen.providers import CustomProvider, JsonProvider

registry = ProviderRegistry((CustomProvider(), JsonProvider(), GreetingProvider()))
result = generate("greeting", GenerationRequest(context={"who": "world"}), registry=registry)

GeneratedArtifact.content 可以是数据结构或预渲染字符串 / may be a data structure or pre-rendered string.

开发与测试 / Development and testing

python -m unittest discover -s tests -v
# 或无需安装 / or without installing:
PYTHONPATH=src python3 -m unittest discover -s tests -v

测试覆盖格式解析、校验、Provider、多源合并、CLI/API 等价性、向导、WebUI / Covers format parsing, validation, providers, merging, CLI/API parity, wizard, WebUI.

许可证 / License

Apache-2.0. 详见 LICENSE / See LICENSE.