Merge pull request #1028 from deepseek-harness/pr/dump-config

feat(cli): dsh --dump-config / --dump-default-config print the composed config tree

packages/ui/app-boot/README.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/ui/app-boot/README.md
-README.md: 54f754842d9a6673ed6791b94656139f0f1be6a3
-README.zh.md: dd56084812e8241f0db24601ce2baeba51252d42
+README.md: 51bc5082512632dd493956b96c605aff47dc872e
+README.zh.md: 644d3a3613a9516cb02881fac8ba7531bffa81eb

packages/ui/app-boot/README.md

@@ -14,6 +14,7 @@ Shared boot glue for the app bins ([`dsh`](../../../apps/cli/README.md), [`dsh-c
 | `loadPersonalPatches(binName, dir?)` | Parse the optional `config.yaml` in the Harness home (default [`resolveDshHome()`](../../util/paths/README.md): `$DSH_HOME`, else `~/.dsh`) — a top-level YAML array of include `PatchOptions` (id-targeted config overrides, `insert` lists, `!!js` allowed); absent file → `undefined`, an unreadable/unparsable/non-array file throws |
 | `loadOverlayPatches(binName, file)` | Parse a required patch-list file with the same shape as personal config; read or parse failures throw a labelled error |
 | `boot(binName, absoluteConfigPath, patches?, prepare?)` | Create the root context, install Loader, run optional host preparation before config-tree entries mount (`prepare` may use Loader and provide launcher-owned context slots such as [`MAIN_SESSION_ID_KEY`](../tui/README.md)), then mount and await the include tree, assert entries loaded and activated, and return the root context |
+| `renderConfigDump(binName, absoluteConfigPath, layers, warn?)` | Compose the base config and labeled overlay layers offline — the include's own parser and patch algorithm (`entryListSchema`/`applyEntryPatches`), so the result equals what `boot()` mounts — and render YAML with `!!js` expressions verbatim; each run of same-provenance rows is preceded by a `# ==` comment naming the contributing file and the layers that patched it, keeping the output one loadable document; a patch matching no row goes to `warn` with its layer label (default: one stderr line), read/parse/shape failures throw |
 | `addHarnessSourceSection(ctx, sourceRoot)` | Add a global `harness:source` prompt section (ordered just after the harness identity, before the persona) telling the agent the on-disk path to its own source checkout; a no-op returning `undefined` when the booted tree has no `systemPrompt` service. The section is registered against that service's fiber, so a dev HMR reload of the system prompt drops it until the next boot |
 | `HARNESS_SOURCE_SECTION` | The `'harness:source'` section name `addHarnessSourceSection` registers under |
 

packages/ui/app-boot/README.zh.md

@@ -14,6 +14,7 @@
 | `loadPersonalPatches(binName, dir?)` | 解析 Harness home 中可选的 `config.yaml`（默认使用 [`resolveDshHome()`](../../util/paths/README.md)：先取 `$DSH_HOME`，否则取 `~/.dsh`）：其顶层是一个 YAML 数组，内容为 include 的 `PatchOptions`（按 id 定位的配置覆盖、`insert` 列表，允许 `!!js`）；文件不存在时返回 `undefined`，文件不可读、不可解析或内容不是数组时抛出异常 |
 | `loadOverlayPatches(binName, file)` | 解析一份必需的 patch 列表文件，其形状与个人配置相同；读取或解析失败时抛出带标签的错误 |
 | `boot(binName, absoluteConfigPath, patches?, prepare?)` | 创建根上下文并安装 Loader，在配置树条目挂载前执行可选的宿主准备操作（`prepare` 可以使用 Loader，也可以提供由启动器拥有的上下文插槽，例如 [`MAIN_SESSION_ID_KEY`](../tui/README.md)），再挂载并等待 include 树结算，断言所有条目均已加载并激活，最后返回根上下文 |
+| `renderConfigDump(binName, absoluteConfigPath, layers, warn?)` | 离线合成基础配置与带标签的覆盖层——使用 include 自己的解析器和补丁算法（`entryListSchema`/`applyEntryPatches`），因此结果与 `boot()` 挂载的内容一致——并渲染为 YAML，`!!js` 表达式原样保留；每段来源相同的连续行之前都有一条 `# ==` 注释，标明贡献该段的文件以及修补过它的层，输出仍是一份可加载的文档；未匹配到行的补丁连同其层标签交给 `warn`（默认：一行 stderr），读取／解析／形状失败则抛出 |
 | `addHarnessSourceSection(ctx, sourceRoot)` | 添加全局 `harness:source` 提示词段落（顺序紧随 harness 身份、位于 persona 之前），告知 agent（智能体）自身源代码 checkout 的磁盘路径；如果已启动树没有此项服务，则不执行操作并返回 `undefined`。这里的服务是 `systemPrompt`；该段落注册到它的 fiber，因此开发环境 HMR（热模块替换）重新加载系统提示词后，它会消失直至下次启动 |
 | `HARNESS_SOURCE_SECTION` | `'harness:source'` 段落名称，供 `addHarnessSourceSection` 注册使用 |
 

packages/ui/app-boot/src/index.ts

@@ -12,7 +12,7 @@ import { basename, dirname, join, resolve } from 'node:path'
 import * as yaml from 'js-yaml'
 import { Context, type FiberState } from 'cordis'
 import Loader from '@cordisjs/plugin-loader'
-import Include, { type PatchOptions } from '@cordisjs/plugin-include'
+import Include, { applyEntryPatches, entryListSchema, type PatchOptions } from '@cordisjs/plugin-include'
 import { resolveDshHome } from '@deepseek-ai/dsh-paths'
 // Side-effect type import: resolves `ctx.get('systemPrompt')` to the service.
 import type {} from '@deepseek-ai/dsh-system-prompt'
@@ -60,16 +60,12 @@ export function loadEnv(
 /** File inside the Harness home holding the personal loader overlay patches. */
 export const PERSONAL_CONFIG_FILENAME = 'config.yaml'
 
-// The include's YAML dialect: `!!js` scalars become expression nodes the
-// Loader interpolates against each entry's context at mount time. Personal
-// patches are parsed with the same schema so they may reference `process.env`.
-// Load-only: this schema never dumps, so no `predicate`/`represent`.
-const jsExprType = new yaml.Type('tag:yaml.org,2002:js', {
-  kind: 'scalar',
-  resolve: data => typeof data === 'string',
-  construct: data => ({ __jsExpr: String(data) }),
-})
-const personalPatchesSchema = yaml.JSON_SCHEMA.extend(jsExprType)
+// The include's YAML dialect (`!!js` scalars become expression nodes the
+// Loader interpolates against each entry's context at mount time), imported
+// from the include itself so patch parsing and config dumping can never drift
+// from what the include mounts. Personal patches share it so they may
+// reference `process.env`.
+const personalPatchesSchema = entryListSchema
 
 /**
  * Load the optional personal overlay patches (`config.yaml` under the Harness
@@ -149,6 +145,141 @@ function parsePatchList(
   return parsed as PatchOptions[]
 }
 
+/** One overlay patch list with the label provenance comments print for it. */
+export interface ConfigDumpLayer {
+  /** Source name shown in provenance comments (a file basename or path). */
+  label: string
+  /** The layer's patches, from {@link loadOverlayPatches} / {@link loadPersonalPatches}. */
+  patches: PatchOptions[]
+}
+
+/**
+ * Compose the effective entry list exactly as `boot()` would mount it: parse
+ * the base config file with the include's entry-list dialect, apply every
+ * layer's patches as ONE flattened list through the include's own patch
+ * algorithm (`applyEntryPatches`) — the same single call `boot()` makes, so
+ * even patch-visibility corner cases (a later layer targeting a group child a
+ * plain config replacement introduced, which the single-pass id index never
+ * sees) compose identically — then render the result as YAML in the same
+ * dialect (`!!js` expressions print verbatim, unevaluated).
+ *
+ * Every run of rows with the same provenance is preceded by a `# ==` comment
+ * naming the file that contributed the rows and any layers that patched them,
+ * so the output stays a loadable YAML document while showing which section
+ * comes from which file. Provenance is derived from single-call prefix
+ * snapshots (base + layers 1..k), diffed positionally: the patch algorithm
+ * only rewrites rows in place or appends, so a top-level index identifies one
+ * row across snapshots, and a layer whose addition changes the row (config
+ * replacement, disable, group insert) is listed as having patched it.
+ *
+ * A patch that matches no row is reported through `warn` with its layer
+ * label, mirroring the Loader's boot-time warning. Earlier layers' patches
+ * see an identical preceding state in every snapshot that includes them, so
+ * each snapshot's warning list extends the previous one and the new tail
+ * belongs to the added layer.
+ * @param binName - the diagnostic prefix on read/parse errors.
+ * @param absoluteConfigPath - the base config file `boot()` would include.
+ * @param layers - overlay layers in application order (later wins).
+ * @param warn - sink for skipped-patch diagnostics; defaults to stderr.
+ * @returns the composed entry list rendered as a YAML document with
+ * provenance comment separators.
+ */
+export function renderConfigDump(
+  binName: string,
+  absoluteConfigPath: string,
+  layers: ConfigDumpLayer[],
+  warn: (line: string) => void = line => void process.stderr.write(`${line}\n`),
+): string {
+  let content: string
+  try {
+    content = readFileSync(absoluteConfigPath, 'utf8')
+  } catch (error) {
+    throw new Error(`${binName}: failed to read config ${absoluteConfigPath}: ${String(error)}`)
+  }
+  let parsed: unknown
+  try {
+    parsed = yaml.load(content, { schema: entryListSchema })
+  } catch (error) {
+    throw new Error(`${binName}: failed to parse config ${absoluteConfigPath}: ${String(error)}`)
+  }
+  if (!Array.isArray(parsed)) {
+    throw new Error(`${binName}: config ${absoluteConfigPath} must be a top-level YAML array of entries`)
+  }
+  const baseLabel = basename(absoluteConfigPath)
+  // The YAML boundary yields untyped rows; the include validates entry shape
+  // at mount, and the dump prints whatever the file holds, so `EntryOptions`
+  // here is structural trust in the same file `boot()` would include.
+  const base = parsed as Parameters<typeof applyEntryPatches>[0]
+  // snapshot_k = ONE application of layers 1..k flattened — boot's exact call
+  // shape for that prefix. snapshot_N is therefore the mounted composition.
+  // The patches are cloned per call: applyEntryPatches detaches the entry
+  // list but pushes `insert` rows by reference from the patch list, so
+  // sharing patch objects across snapshot calls would leak a later
+  // snapshot's mutations into an earlier one's result.
+  const snapshot = (count: number, warnings: string[]): ReturnType<typeof applyEntryPatches> => {
+    const flattened = structuredClone(layers.slice(0, count).flatMap(layer => layer.patches))
+    return applyEntryPatches(base, flattened, (message: string, ...args: unknown[]) => {
+      // The include logs through cordis's printf-style logger (`%C` = code); a
+      // dump has no logger, so substitute inline for a plain line.
+      let index = 0
+      warnings.push(message.replace(/%C/g, () => JSON.stringify(args[index++])))
+    })
+  }
+  let previous = base
+  let previousWarnings: string[] = []
+  const provenance: { origin: string; patchedBy: string[] }[] = base.map(() => ({ origin: baseLabel, patchedBy: [] }))
+  let composed = base
+  for (let count = 1; count <= layers.length; count += 1) {
+    const layer = layers[count - 1]
+    /* v8 ignore next -- count iterates 1..length, so the slot exists */
+    if (layer === undefined) continue
+    const warnings: string[] = []
+    composed = snapshot(count, warnings)
+    for (const line of warnings.slice(previousWarnings.length)) {
+      warn(`${binName}: [${layer.label}] ${line}`)
+    }
+    const before = previous.map(entry => JSON.stringify(entry))
+    for (let index = 0; index < composed.length; index += 1) {
+      if (index >= before.length) provenance.push({ origin: layer.label, patchedBy: [] })
+      else if (JSON.stringify(composed[index]) !== before[index]) provenance[index]?.patchedBy.push(layer.label)
+    }
+    previous = composed
+    previousWarnings = warnings
+  }
+  return groupedDump(composed, provenance)
+}
+
+/** Render the composed rows grouped under one provenance comment per contiguous run. */
+function groupedDump(
+  composed: readonly unknown[],
+  provenance: readonly { origin: string; patchedBy: string[] }[],
+): string {
+  const lines: string[] = []
+  let currentLabel: string | undefined
+  let group: unknown[] = []
+  const flush = (): void => {
+    if (currentLabel === undefined || group.length === 0) return
+    lines.push(`# == ${currentLabel}`)
+    lines.push(yaml.dump(group, { schema: entryListSchema, noRefs: true }).trimEnd())
+    group = []
+  }
+  for (let index = 0; index < composed.length; index += 1) {
+    const record = provenance[index]
+    /* v8 ignore next -- provenance is index-aligned with composed by construction */
+    if (record === undefined) continue
+    const label = record.patchedBy.length === 0
+      ? record.origin
+      : `${record.origin}, patched by ${record.patchedBy.join(', ')}`
+    if (label !== currentLabel) {
+      flush()
+      currentLabel = label
+    }
+    group.push(composed[index])
+  }
+  flush()
+  return lines.join('\n') + '\n'
+}
+
 /**
  * The slice of `process` {@link installFailLoud} needs — injectable so tests
  * exercise the handler without registering on (or exiting) the real process.

packages/ui/app-boot/tests/config-dump.spec.ts

@@ -0,0 +1,187 @@
+/**
+ * `renderConfigDump` behavior: the offline composition must equal what
+ * `boot()` mounts (same parser, same patch algorithm), print `!!js`
+ * expressions verbatim, separate provenance runs with comment lines while
+ * staying one loadable YAML document, and report skipped patches through
+ * `warn` instead of failing — mirroring the Loader's boot-time warning for a
+ * shared overlay whose row exists only on another surface.
+ */
+
+import { mkdtempSync, writeFileSync } from 'node:fs'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+import { describe, expect, it, vi } from 'vitest'
+import * as yaml from 'js-yaml'
+import { entryListSchema } from '@cordisjs/plugin-include'
+import { loadOverlayPatches, renderConfigDump } from '../src/index.ts'
+
+const NAME = 'dsh-test-bin'
+
+const tmp = (): string => mkdtempSync(join(tmpdir(), 'dsh-config-dump-'))
+
+function writeBase(dir: string): string {
+  const base = join(dir, 'base.yml')
+  writeFileSync(base, [
+    '- id: shared',
+    '  name: ./noop.mjs',
+    '  config:',
+    '    value: base',
+    '    key: !!js process.env.DSH_DUMP_SPEC',
+    '- id: untouched',
+    '  name: ./noop.mjs',
+    '',
+  ].join('\n'))
+  return base
+}
+
+describe('renderConfigDump', () => {
+  it('composes overlay layers in order, prints !!js verbatim, and labels each section with its provenance', () => {
+    const dir = tmp()
+    const base = writeBase(dir)
+    const surface = join(dir, 'surface.yml')
+    writeFileSync(surface, [
+      '- id: shared',
+      '  config:',
+      '    value: surface',
+      '    key: !!js process.env.DSH_DUMP_SPEC',
+      '- insert:',
+      '    - id: surface-extra',
+      '      name: ./noop.mjs',
+      '',
+    ].join('\n'))
+    const personal = join(dir, 'personal.yml')
+    writeFileSync(personal, [
+      '- id: surface-extra',
+      '  config:',
+      '    value: personal',
+      '',
+    ].join('\n'))
+
+    const dump = renderConfigDump(NAME, base, [
+      { label: 'surface.yml', patches: loadOverlayPatches(NAME, surface) },
+      { label: 'personal.yml', patches: loadOverlayPatches(NAME, personal) },
+    ], () => {})
+    // Comments do not break loadability: the dump parses as one document
+    // equal to what boot() would mount.
+    const parsed = yaml.load(dump, { schema: entryListSchema }) as {
+      id: string
+      config?: Record<string, unknown>
+    }[]
+    expect(parsed).toEqual([
+      {
+        id: 'shared',
+        name: './noop.mjs',
+        config: { value: 'surface', key: { __jsExpr: 'process.env.DSH_DUMP_SPEC' } },
+      },
+      { id: 'untouched', name: './noop.mjs' },
+      { id: 'surface-extra', name: './noop.mjs', config: { value: 'personal' } },
+    ])
+    // Unevaluated: the expression text round-trips as a !!js scalar.
+    expect(dump).toContain('!!js process.env.DSH_DUMP_SPEC')
+    // Provenance separators: origin file, plus every layer that changed the
+    // row; an inserted row carries the inserting layer as its origin.
+    expect(dump).toContain('# == base.yml, patched by surface.yml')
+    expect(dump).toContain('# == base.yml\n- id: untouched')
+    expect(dump).toContain('# == surface.yml, patched by personal.yml\n- id: surface-extra')
+    expect(dump.indexOf('# == base.yml, patched by surface.yml')).toBeLessThan(dump.indexOf('# == base.yml\n- id: untouched'))
+  })
+
+  it('groups contiguous same-provenance rows under one separator', () => {
+    const dir = tmp()
+    const base = join(dir, 'base.yml')
+    writeFileSync(base, [
+      '- id: a',
+      '  name: ./noop.mjs',
+      '- id: b',
+      '  name: ./noop.mjs',
+      '',
+    ].join('\n'))
+    const dump = renderConfigDump(NAME, base, [], () => {})
+    expect(dump.match(/# == base\.yml/g)).toHaveLength(1)
+    expect(dump).toContain('# == base.yml\n- id: a')
+  })
+
+  it('composes all layers as one flattened patch list, exactly like boot()', () => {
+    // boot() flattens every layer into ONE applyEntryPatches call, whose id
+    // index sees inserted rows but NOT children introduced by a plain group
+    // `config` replacement. A per-layer composition would rebuild the index
+    // between layers and let the second layer patch that child — a tree the
+    // real boot never mounts. Pin the single-call semantics: the child patch
+    // is skipped (with the layer-labeled warning), matching boot.
+    const dir = tmp()
+    const base = join(dir, 'base.yml')
+    writeFileSync(base, [
+      '- id: g',
+      '  name: ./group.mjs',
+      '  group: true',
+      '  config: []',
+      '',
+    ].join('\n'))
+    const warnings: string[] = []
+    const dump = renderConfigDump(NAME, base, [
+      {
+        label: 'a.yml',
+        patches: [{ id: 'g', config: [{ id: 'child', name: './noop.mjs', config: { v: 1 } }] }],
+      },
+      { label: 'b.yml', patches: [{ id: 'child', config: { v: 2 } }] },
+    ], line => void warnings.push(line))
+    expect(warnings).toEqual([`${NAME}: [b.yml] patch: entry "child" not found`])
+    const parsed = yaml.load(dump, { schema: entryListSchema }) as {
+      config?: { config?: { v?: number } }[]
+    }[]
+    expect(parsed[0]?.config?.[0]?.config?.v).toBe(1)
+    // The skipped layer did not change the row, so it is not in provenance.
+    expect(dump).toContain('# == base.yml, patched by a.yml\n- id: g')
+    expect(dump).not.toContain('b.yml\n- id: g')
+  })
+
+  it('reports a patch whose target row is absent through warn with its layer label and keeps composing', () => {
+    const dir = tmp()
+    const base = writeBase(dir)
+    const overlay = join(dir, 'overlay.yml')
+    writeFileSync(overlay, [
+      '- id: only-on-another-surface',
+      '  config:',
+      '    value: ignored',
+      '- id: shared',
+      '  config:',
+      '    value: patched',
+      '',
+    ].join('\n'))
+    const warnings: string[] = []
+    const dump = renderConfigDump(
+      NAME, base,
+      [{ label: 'overlay.yml', patches: loadOverlayPatches(NAME, overlay) }],
+      line => void warnings.push(line),
+    )
+    expect(warnings).toEqual([`${NAME}: [overlay.yml] patch: entry "only-on-another-surface" not found`])
+    const parsed = yaml.load(dump, { schema: entryListSchema }) as { config?: { value?: string } }[]
+    expect(parsed[0]?.config?.value).toBe('patched')
+  })
+
+  it('defaults its warn sink to one stderr line per skipped patch', () => {
+    const dir = tmp()
+    const base = writeBase(dir)
+    const write = vi.spyOn(process.stderr, 'write').mockReturnValue(true)
+    try {
+      renderConfigDump(NAME, base, [{ label: 'x.yml', patches: [{ id: 'absent', config: {} }] }])
+      expect(write).toHaveBeenCalledWith(`${NAME}: [x.yml] patch: entry "absent" not found\n`)
+    } finally {
+      write.mockRestore()
+    }
+  })
+
+  it('fails loud on a missing, unparsable, or non-array base config', () => {
+    const dir = tmp()
+    expect(() => renderConfigDump(NAME, join(dir, 'absent.yml'), [], () => {}))
+      .toThrow(new RegExp(`^${NAME}: failed to read config `))
+    const invalid = join(dir, 'invalid.yml')
+    writeFileSync(invalid, 'invalid: [unclosed\n')
+    expect(() => renderConfigDump(NAME, invalid, [], () => {}))
+      .toThrow(new RegExp(`^${NAME}: failed to parse config `))
+    const scalar = join(dir, 'scalar.yml')
+    writeFileSync(scalar, 'id: not-a-list\n')
+    expect(() => renderConfigDump(NAME, scalar, [], () => {}))
+      .toThrow('must be a top-level YAML array of entries')
+  })
+})