mirror of
https://github.com/larksuite/cli.git
synced 2026-07-03 14:02:43 +08:00
main
10 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
dc-bytedance
|
bdffffb368 | feat: interactive upgrade prompt for bare lark-cli (#1498) | ||
|
mew
|
fb042758db | feat: add whoami command showing effective identity (#1666) | ||
|
liangshuo-1
|
ebb0b6fe73 | feat(affordance): per-command usage guidance system (markdown source) (#1565) | ||
|
hanshaoshuai
|
c61acb5264 | feat: add ci quality gate | ||
|
liangshuo-1
|
9d845442ce | feat: add skills command to read embedded skill content (#1318) | ||
|
xiongyuanwen-byted
|
b07a6003f9 |
feat(sheets): spec-driven shortcut refactor with backward-compatible package (#1220)
* refactor(sheets): rebuild lark-sheets on sheet-skill-spec canonical + One-OpenAPI
Restart lark-sheets as a spec-driven downstream. Skill content (SKILL.md
and 16 references covering 13 operations skills + 3 workflow skills,
including the standalone filter-view skill) is mirrored from the
sheet-skill-spec canonical-spec; do not hand-edit, change upstream and
rerun npm run sync:consumers.
Drop the 11 legacy shortcut sources (spreadsheet / sheet management,
cell ops, dropdown, filter-view, float image, etc.) and 10 associated
tests. Wire up the new sheet_ai/v2 One-OpenAPI single entry that
dispatches by tool_name with JSON-string input/output, and land the
first canonical shortcut +workbook-info as a template that exercises
the public token XOR pair, Risk tiering, and zero-side-effect DryRun.
sheet_ai_api.go provides callTool / invokeToolDryRun and bypasses
runtime.CallAPI's silent swallowing of non-envelope responses so
gateway and business errors from the new endpoint surface precisely.
The remaining 55 shortcuts will be designed and landed separately,
canonical skill by canonical skill.
* feat(sheets): implement lark_sheet_workbook shortcuts (B1)
Land the 8 modify_workbook_structure shortcuts that round out the
lark_sheet_workbook canonical skill alongside the existing +workbook-info:
+sheet-create / +sheet-delete / +sheet-rename / +sheet-move / +sheet-copy
/ +sheet-hide / +sheet-unhide / +sheet-set-tab-color. All eight call
modify_workbook_structure via the One-OpenAPI invoke_write endpoint,
dispatched by the `operation` enum.
Helpers in helpers.go grow publicSheetFlags() / resolveSheetSelector() /
sheetSelectorForToolInput() / sheetSelectorPlaceholder() so future
sheet-level shortcuts share the public --sheet-id / --sheet-name XOR
treatment. +sheet-create intentionally drops the sheet selector pair since
create has no existing-sheet anchor (matches the spec fix in
tool-shortcut-map.json).
+sheet-delete is the first high-risk-write shortcut in the canonical
package; the framework requires --yes (exit code 10 otherwise).
+sheet-move's tool requires source_index in addition to target_index. The
CLI accepts an optional --source-index override and falls back to a
single get_workbook_structure read to derive it (and to resolve sheet_id
from --sheet-name). DryRun stays network-free by rendering <resolve>
placeholders for any field that would need that read.
* feat(sheets): implement lark_sheet_sheet_structure shortcuts (B2)
Add 8 shortcuts under the lark_sheet_sheet_structure canonical skill:
+sheet-info (get_sheet_structure) plus +dim-insert / +dim-delete /
+dim-hide / +dim-unhide / +dim-freeze / +dim-group / +dim-ungroup
(modify_sheet_structure, dispatched by operation enum).
Two reusable conversion helpers cover the impedance mismatch between
the CLI surface and the tool input:
- dimRange / dimPosition translate the CLI's 0-based exclusive-end
range into the tool's 1-based A1 notation. row 5..8 becomes
position "6" + count 3 (insert) or range "6:8" (range ops); column
26..29 becomes "AA:AC".
- infoTypeFromInclude maps the fine-grained --include vocabulary
(row_heights / col_widths / merges / hidden_rows / hidden_cols /
groups / frozen) to the coarse info_type enum the tool accepts;
mixed categories collapse to "all".
+dim-delete is high-risk-write (irreversible row/column removal).
+dim-freeze --count 0 auto-dispatches to operation=unfreeze. +dim-group
accepts --depth for forward-compat with a future server-side nested
group endpoint but does not pass it through today.
* feat(sheets): implement read_data / search_replace / write_cells shortcuts (B3)
Land 11 shortcuts across three canonical skills:
- lark_sheet_read_data (3): +cells-get / +csv-get / +dropdown-get
- lark_sheet_search_replace (2): +cells-search / +cells-replace
- lark_sheet_write_cells (6): +cells-set / +cells-set-style / +csv-put
/ +dropdown-set / +dropdown-update / +dropdown-delete
+dropdown-get reads the data_validation field via get_cell_ranges with
the range carrying its own sheet prefix (no --sheet-id needed). The
fine-grained --include vocabulary (value / formula / style / comment /
data_validation) maps to the tool's coarse include_styles bool plus
value_render_option enum. +csv-get's --include-row-prefix=false strips
the [row=N] prefix client-side because the tool only emits the
annotated form.
+cells-search / +cells-replace flatten the tool's options sub-object
into four independent flags (--match-case / --match-entire-cell /
--regex / --include-formulas) per the flat-flag rule, then repack them on the way
in.
+cells-set takes a raw --data JSON body whose `cells` array must match
the --range dimensions. +cells-set-style fans a single --style block
out to every cell in the range via a new fillCellsMatrix helper; the
range parser (rangeDimensions / splitCellRef / letterToColumnIndex)
only accepts rectangular A1:B2 forms — whole-column / whole-row need
sheet totals and are deferred.
+dropdown-set fans the validation block out to one range; +dropdown-
update / +dropdown-delete iterate sheet-prefixed --ranges and call
set_cell_range sequentially (partial failure leaves earlier ranges
already mutated; the Tip calls this out). +dropdown-delete is
high-risk-write and requires --yes.
+cells-set-image stays deferred to the cli-only batch (needs the
shared local-file upload helper alongside +workbook-create / +dim-move
/ +workbook-export).
* refactor(sheets): move +dropdown-update / +dropdown-delete to lark_sheet_batch_update
Follow-up to B3 after the spec re-mapped these two shortcuts to the
batch_update tool (atomic multi-range CRUD) instead of fan-out via
set_cell_range. Drop their Go implementations + helper validateDropdownRanges
+ splitSheetPrefixedRange from lark_sheet_write_cells.go and remove the
registrations from Shortcuts(); the shortcuts will reappear under
lark_sheet_batch_update during B7.
Also pull in the re-rendered reference docs:
- skills/lark-sheets/references/lark-sheets-write-cells.md
- skills/lark-sheets/references/lark-sheets-batch-update.md
* feat(sheets): implement lark_sheet_range_operations shortcuts (B4)
Land 8 shortcuts across four canonical tools:
- clear_cell_range → +cells-clear (high-risk-write)
- merge_cells → +cells-merge / +cells-unmerge
- resize_range → +dim-resize
- transform_range → +range-move / +range-copy / +range-fill / +range-sort
Three CLI↔tool vocabulary bridges live in this file:
- +cells-clear: --scope content normalizes to the tool's clear_type
"contents" (singular/plural spec mismatch is absorbed in the CLI).
- +dim-resize: --size <px> wraps as resize_{height,width}:{value:N};
--reset wraps as {reset:true}. The two flags are mutually exclusive
and at least one is required.
- +range-fill: CLI's five-valued --series-type collapses to the tool's
binary fill_type — `copy` → "copyCells", anything else → "fillSeries"
(the actual series progression is inferred server-side from the
seed cells in --source-range).
- +range-copy: --paste-type {values, formulas, formats} maps to the
tool's {value_only, formula_only, format_only}; "all" omits the
field entirely so the server applies its default.
+cells-clear is the second high-risk-write shortcut in the package;
the framework enforces --yes with exit code 10 as usual.
* feat(sheets): implement object-list shortcuts (B5)
Land 7 read shortcuts, one per object skill — chart / pivot table /
conditional format / filter / filter view / sparkline / float image. All
share the same shape (public sheet selector + optional <obj>-id filter)
so they're declared via newObjectListShortcut + an objectListSpec.
Notes:
- +cond-format-list exposes --rule-id, which is renamed to
conditional_format_id on the wire (the tool's full field name).
- +sparkline-list exposes --group-id (the higher-level handle); the
tool also accepts sparkline_id, intentionally not surfaced.
- +filter-list takes no id filter — at most one sheet-level filter
per sheet, so the listing is already unique.
- +filter-view-list is `cli_status: cli-only` but get_filter_view_objects
is in mcp-tools.json and dispatches through the same One-OpenAPI
endpoint; no special path required.
* feat(sheets): implement object CRUD shortcuts (B6)
Land 21 shortcuts — three (create / update / delete) per object skill —
backed by the manage_<obj>_object tools dispatched on the operation
enum. Five standard objects (chart / cond-format / sparkline /
float-image / filter-view) share an objectCRUDSpec factory; pivot and
filter are special-cased.
Shared wire contract:
excel_id + sheet_id|sheet_name + operation + [<obj>_id] + [properties]
CLI --data is passed through as the tool's `properties` field as-is, so
callers shape it per each object's spec doc.
Special cases:
- pivot adds optional --target-sheet-id / --target-position on create
(siblings of properties, not inside it).
- cond-format exposes --rule-id (short CLI name) wired to the tool's
conditional_format_id on the wire.
- sparkline uses --group-id (higher-level object handle) instead of
sparkline_id.
- filter has no separate id flag — at most one filter per sheet, so
filter_id is implicit. +filter-create promotes --range to a first-
class flag (instead of burying it inside --data).
- filter-view CRUD are `cli_status: cli-only` but
manage_filter_view_object is in mcp-tools.json, so they go through
callTool / One-OpenAPI alongside everything else.
All delete shortcuts are high-risk-write and require --yes.
* feat(sheets): implement lark_sheet_batch_update shortcuts (B7)
Land 4 shortcuts that all funnel through the batch_update tool's atomic
operations array:
- +batch-update raw passthrough; --data carries the full
{ operations: [{tool, params}, ...] } payload
plus optional continue_on_error. high-risk-write
since the caller may stuff anything inside.
- +cells-batch-set-style --data is [{ranges, style}, ...]; CLI flattens
each (entry × range) pair into a set_cell_range
op with a fan-out cells matrix carrying
cell_styles + border_styles.
- +dropdown-update --ranges + --options (+ --colors / --multiple /
--highlight) — installs/replaces one dropdown
across many ranges, each becoming a separate
set_cell_range op with data_validation in cells.
- +dropdown-delete --ranges — clears data_validation across many
ranges (high-risk-write).
Default is strict transaction: if any sub-tool fails the whole batch rolls
back. +batch-update exposes --continue-on-error to flip the policy; the
three fan-out shortcuts leave it strict (they're meant to be all-or-nothing).
Reinstates validateDropdownRanges + splitSheetPrefixedRange that were
removed during B3 → B7 relocation.
* feat(sheets): implement cli-only shortcuts (B8) — 70/70 complete
Land the four cli-only shortcuts that can't route through the One-OpenAPI
dispatcher (their backing capabilities aren't in mcp-tools.json):
- +workbook-create POST /open-apis/sheets/v3/spreadsheets
+ optional set_cell_range follow-up that zips
--headers and --data into the first sheet starting
at A1.
- +workbook-export POST /open-apis/drive/v1/export_tasks (type=sheet)
→ poll /export_tasks/:ticket up to ~30s
→ optional GET /export_tasks/file/:file_token/download.
CSV mode requires --sheet-id (single sheet export).
- +dim-move POST /open-apis/sheets/v2/spreadsheets/:token
/dimension_range
CLI is 0-indexed inclusive (--start / --end); the v2
endpoint expects half-open [startIndex, endIndex)
so the body uses endIndex = --end + 1. --sheet-name
is resolved client-side to sheet_id via
lookupSheetIndex when needed.
- +cells-set-image common.UploadDriveMediaAll
(parent_type=sheet_image, parent_node=token)
then callTool set_cell_range with cells carrying
rich_text: [{type:"embed-image", attachment_token, attachment_name}].
--range must be exactly one cell.
All four use runtime.CallAPI / DoAPI directly; only +cells-set-image
combines a legacy upload with the new One-OpenAPI for the second step
(set_cell_range is in mcp-tools.json so callTool is the right path).
This closes the migration: 70 shortcuts × 17 canonical skills × matching
the sheet-skill-spec v0.5.0 tool-shortcut-map.
* test(sheets): cover all 70 shortcuts with dry-run + execute-path tests
Twelve _test.go files alongside the implementation, mirroring the legacy
package's coverage style:
- testhelpers_test.go shared rig: TestFactory + Mount + dry-run
capture + JSON-input decode + envelope helpers.
- lark_sheet_*_test.go one test file per implementation file (9
files), table-driven dry-run cases per shortcut
plus targeted validation guards.
- execute_paths_test.go end-to-end execute paths via httpmock stubs.
Covers callTool unwrap, JSON-string output
decoding, two-step lookup (+sheet-move),
batch_update fan-out, dropdown atomic writes,
and the legacy OAPI shortcuts (+workbook-create,
+dim-move) including CLI inclusive → API
half-open index conversion.
Test coverage on the sheets package is 60.5 % of statements with -race
clean, meeting the dev manual's ≥ 60 % patch-coverage gate.
* refactor(sheets): inline cli-only shortcuts into their canonical skill files
Two naming cleanups:
- lark_sheet_cli_only.go is gone. The four shortcuts it grouped
(+workbook-create / +workbook-export / +dim-move / +cells-set-image)
were bundled by their implementation pattern (legacy OAPI direct
calls) rather than by canonical skill. The whole sheets package IS
the CLI implementation, so "cli only" wasn't a meaningful grouping
at the Go layer. Each shortcut now lives next to its skill peers:
+workbook-create / +workbook-export → lark_sheet_workbook.go
+dim-move → lark_sheet_sheet_structure.go
+cells-set-image → lark_sheet_write_cells.go
Per-skill shortcut counts now match tool-shortcut-map.json exactly
(workbook: 11, sheet_structure: 9, write_cells: 5). Helpers
(buildInitialFillInput, pollExportTask, downloadExportFile,
dimMoveBody) move with their shortcuts; nothing else in the package
referenced them.
- testhelpers_test.go → helpers_test.go. The _test.go suffix already
conveys "test"; the leading "test" was redundant. Matches the
helpers.go naming convention.
Behavior unchanged. go test -race -cover stays at 60.5 %.
* refactor(sheets): sync shortcut flags with sheet-skill-spec v0.5.0
Upstream hoisted a batch of high-frequency scalar fields out of --data
into independent flags and renamed several composite-JSON flags to
match their semantic content. CLI catches up.
Renames (drop-in, same payload semantics):
- +cells-replace --replace → --replacement
- +cells-set --data → --cells
- +workbook-create --data → --values
- +batch-update --data → --operations (now a bare array;
still accepts the envelope form for
back-compat with continue_on_error)
Flat-flag hoists out of --style / --data:
- +cells-set-style / +cells-batch-set-style
--style JSON drops; replaced by 11 flat style flags
(--background-color / --font-color / --font-size / --font-style /
--font-weight / --font-line / --horizontal-alignment /
--vertical-alignment / --word-wrap / --number-format) plus
--border-styles for the one field that's still nested. Both
shortcuts share styleFlatFlags() + buildCellStyleFromFlags().
- +cells-batch-set-style also drops the [{ranges, style}] array shape
in favor of one --ranges + the same flat style flags applied to
all of them.
Object CRUD --data → --properties everywhere (chart / pivot / cond-format
/ filter / filter-view / sparkline / float-image). Per-skill scalar
hoists merged into properties via an enhanceCreate/UpdateInput callback:
- +pivot-create adds --source (required), --range
(and continues to expose --target-sheet-id /
--target-position at top level)
- +cond-format-{create,update}
adds --rule-type (enum) + --ranges (JSON array);
merged into properties.rule.type and
properties.ranges respectively
- +filter-view-{create,update}
adds --view-name and --range; both override
their properties.* counterparts
- +filter-update adds first-class --range (was buried in --data)
Float-image is fully hoisted — no --properties flag at all. Ten flat
flags (--image-name / --image-token | --image-uri / --position-row /
--position-col / --size-width / --size-height / --offset-row /
--offset-col / --z-index) compose the properties block. Implemented as
its own factory (newFloatImageWriteShortcut) since it diverges from the
shared CRUD spec.
Tests track every flag renamed and add explicit cases for the new flag
combos. go test -race -cover stays at 60.3 %.
* refactor(sheets): align batch_update + cells-set with synced reference docs
Sync to upstream reference doc updates for 9 skills:
- batch_update sub-ops: rewrite wire fields tool/params -> tool_name/input
in CellsBatchSetStyle and DropdownUpdate/Delete fan-out (the actual
server contract per Schemas section); update --operations flag desc
and tests.
- +cells-set --cells: accept bare 2D matrix [[{cell},...],...] instead
of envelope {"cells":[[...]]}; spec example shows bare-array form.
- sparkline createDataDesc enum: win_loss -> winLoss (camelCase).
All other doc changes (float-image flat flags, cond-format
--rule-type/--ranges, pivot create-only --source/--range, filter /
filter-view extra flags, chart --properties) were already aligned in
commit
|
||
|
sang-neo03
|
33c292c05e |
feat(extension): Plugin / Hook framework with command pruning (#910)
* feat(extension): introduce Plugin / Hook framework with command pruning
Add a single public extension contract under extension/platform: integrators
implement the Plugin interface and register Observers, Wrappers, Lifecycle
handlers, and pruning Rules through the Registrar in one Install call.
Command pruning:
- Rule (Allow / Deny / MaxRisk / Identities) with doublestar globs
- 4-axis AND evaluation, parent-group aggregation, unknown-risk allow
- Sources: Plugin.Restrict (single-rule) and ~/.lark-cli/policy.yml
- Plugin path is fail-closed (envelope on rule error / multiple Restrict);
yaml path is fail-open (warning, CLI continues)
- strict-mode stubs now also write the denial annotation so the hook
layer's denial guard physically isolates Wrap chains on them
- HOME path never leaked through policy_source label
Hook framework:
- Observer (panic-safe, Before/After), Wrapper (middleware, may short-circuit
via AbortError), Lifecycle (Startup + Shutdown only)
- Recover guards every plugin entry point: Capabilities(), Install(),
Wrapper factory composition AND inner Handler, Lifecycle handlers
- namespacedWrap copies AbortError so a plugin's package-level sentinel
is never mutated across concurrent invocations
- Selector unknown-risk uniform: ByExactRisk / ByWrite / ByReadOnly never
match unannotated commands; safety-side hooks opt in via
ByWrite().Or(ByUnknownRisk())
Bootstrap orchestration (cmd/build.go + cmd/policy.go):
- InstallAll uses a staging Registrar + atomic commit
- FailClosed plugin install / Plugin.Restrict conflict / Startup handler
failure each install a structured envelope guard at every dispatch path
- walkGuard neutralises every cobra bypass we know of (PersistentPreRunE
first-wins, ValidateArgs, ParseFlags, legacyArgs, __complete /
__completeNoDesc, non-runnable groups, required-arg subcommands)
- cmd/root.go::Execute calls hook.Emit(Shutdown, runErr) after
rootCmd.Execute; isCompletionCommand skips both __complete and
__completeNoDesc so Tab completion never triggers Shutdown handlers
Capabilities consistency:
- Restricts=true must declare FailurePolicy=FailClosed
- RequiredCLIVersion (semver constraint) is validated against build.Version;
a malformed constraint is treated as untrusted-config and aborts
unconditionally, regardless of FailurePolicy (DEV builds included)
JSON envelope contract:
- error.type closed enum: pruning / strict_mode / hook / plugin_install /
plugin_conflict / plugin_lifecycle
- reason_code closed enums per type, all referenced by structured tests
Bootstrap surfaces (new user commands):
- lark-cli config policy show -- JSON view of the active Rule + source
- lark-cli config policy validate -- parse + schema + glob check, no apply
Coverage:
- extension/platform: every public type has a unit test
- internal/{pruning,hook,platformhost,policydecision,cmdmeta}: full coverage
of denial guard isolation, AbortError sentinel safety, observer panic
safety, lifecycle error/panic typing, staging atomic rollback
- cmd/plugin_integration_test.go: end-to-end through buildInternal with
synthetic and real command trees
- cmd/install_guard_test.go: walkGuard covers auth / config / __complete /
__completeNoDesc / non-runnable parents
* fix(pruning): deny stub must override Args + PersistentPreRunE
The pruning denyStub and the strict-mode stub previously only swapped
RunE plus Hidden + DisableFlagParsing. Cobra's dispatch order means
several pre-RunE gates can fire BEFORE the stub's RunE ever runs:
1. Args validator: shortcut commands often declare cobra.NoArgs.
With DisableFlagParsing=true the user's `--doc xxx --mode append`
looks like positional args, so ValidateArgs surfaces a usage
error instead of the pruning / strict_mode envelope. Observer
hooks also miss the dispatch entirely.
2. Parent PersistentPreRunE: cmd/auth/auth.go declares a
PersistentPreRunE that returns external_provider when env
credentials are set. Cobra's "first PersistentPreRunE wins
walking up from the leaf" then short-circuits with
external_provider instead of the leaf's denial envelope.
Both stubs now also set:
- Args = cobra.ArbitraryArgs (bypass gate 1)
- PersistentPreRunE = no-op leaf hook (bypass gate 2)
- PreRunE / PreRun / PersistentPreRun = nil (defensive)
Effect: dispatch reaches the wrapped RunE, observers fire, the real
pruning / strict_mode envelope is emitted regardless of credential
provider or flag count.
Adds regression tests covering both gates on both stub paths.
* fix(config): policy subcommand bypasses parent's credential check
cmd/config/config.go::NewCmdConfig declares a PersistentPreRunE that
calls f.RequireBuiltinCredentialProvider; with env credentials set,
it returns external_provider for every config subcommand.
`config policy show` and `config policy validate` are READ-ONLY
diagnostic commands -- they inspect or parse the user-layer rule
without touching credentials. They MUST work regardless of which
credential provider is active, otherwise users on env-credential
deployments cannot debug their policy.
Same shape as the codex C11/C13 fix: install a no-op leaf-level
PersistentPreRunE on the `policy` group so cobra's "first walking up
from leaf" rule picks ours over the config parent's.
Regression caught by divergent e2e (F1-F6 all returned external_provider
before this fix; all pass after). Adds a unit test pinning the
PersistentPreRunE override.
* feat(shortcuts): tag service groups with cmdmeta.Domain
RegisterShortcutsWithContext now calls cmdmeta.SetDomain on each
service-level cobra.Command (im, docs, drive, calendar, ...) so the
business-domain axis is actually populated on every shortcut leaf via
parent-chain inheritance.
Before this change, platform.ByDomain("docs") never matched any
command: the domain annotation was unset across the entire shortcut
tree, so the selector's d != "" guard always failed and risk-style
selectors silently degraded to no-op.
The SetDomain call is placed AFTER the create-or-reuse branch so it
fires whether the service command was freshly created here or had
already been added by cmd/service/service.go's OpenAPI auto-
registration (which runs first and creates im, drive, calendar, etc.).
Without this placement only pure-shortcut services like docs would
have been tagged.
Adds a regression test asserting:
- service-group cobra.Command carries the cmdmeta.domain annotation
- leaf shortcuts inherit the domain via parent-chain walk
* feat(diagnostic): add unconditionally allowed command paths for introspection
* feat(plugins): add diagnostic command to inspect installed plugins and their contributions
* fix(cli): surface unknown_subcommand error instead of silent help fallback
When a user passed an unknown subcommand or shortcut (e.g. `lark-cli drive
+bogus`), cobra returned `flag.ErrHelp` for the non-runnable group command,
printed the parent help, and exited 0. AI agents couldn't distinguish a
typo from an intentional help request.
Install a tree-wide guard that attaches a RunE to every group command
without its own Run/RunE. The RunE forwards no-args invocations to help
(preserving prior behavior) and emits a structured unknown_subcommand
ExitError (exit 2) listing available subcommands when args are present.
* refactor(envelope): rename error.type pruning/strict_mode to command_denied
The envelope's `type` field was leaking implementation terms ("pruning",
"strict_mode") that describe enforcement mechanism rather than the user-
facing semantic. It also duplicated `detail.layer`, and forced consumers
to branch on two values for the same conceptual error ("a command was
denied by policy").
Collapse both into a single semantic type "command_denied". The
enforcement layer ("pruning" / "strict_mode") is preserved in
`detail.layer` so debugging and per-layer diagnostics still work.
* feat(platform): fail closed on unannotated/invalid risk when a Rule is active
The pruning engine used to treat any command without a risk annotation as
ALLOW even when a Rule with MaxRisk was set, and would silently skip the
MaxRisk comparison whenever the command's risk string was outside the
closed taxonomy. Both gaps let an unannotated or typo'd write command
slip past an "agent read-only" pruning rule.
Engine now denies before any other axis when a Rule is registered:
- reason_code "risk_not_annotated" for commands with no risk
- reason_code "risk_invalid" for commands whose risk is outside
the read | write | high-risk-write
taxonomy (e.g. typo "wrtie")
Main-flow is preserved: a nil Rule still returns Allowed=true
unconditionally, so a CLI with no pruning plugin behaves identically to
before. ByUnknownRisk() is removed from the public surface since the
Unknown state is no longer reachable through risk-based selectors when
any Rule is active; safety-side widening composition is no longer needed.
* chore(config): hide diagnostic policy/plugins commands from --help
`config policy show`, `config policy validate`, and `config plugins show`
are local-introspection-only commands kept behind the pruning
diagnostic whitelist so operators can always inspect why a command was
denied. They do not need to surface in `--help` for AI agents and were
contributing to help noise.
Hide the `policy` and `plugins` parent groups and both `show` /
`validate` leaves. Commands remain callable by exact name and continue
to bypass user-layer pruning via diagnosticPaths.
* style: gofmt
* fix(platform): nil Selector honours None contract; reject multi-doc policy yaml
- selector.go: And/Or/Not now treat nil Selector as None() per godoc,
preventing runtime panic when composed selectors are invoked.
- schema.go: Parse rejects multi-document YAML input so a stray '---'
separator can't silently drop trailing policy constraints.
* chore: go mod tidy
* feat(extension/platform): plugin SDK with policy engine, hooks, and Builder
Introduces extension/platform — the in-process plugin SDK external
Go forks of lark-cli use to extend or restrict the command surface.
Plugins compile in via blank import; there is no dynamic loading
and no RPC isolation.
Public SDK (extension/platform):
- Plugin interface (Name / Version / Capabilities / Install).
- Registrar verbs: Observe, Wrap, On, Restrict.
- Hook types: Observer (side-effect, panic-safe, fires Before/After
RunE), Wrapper (middleware, may short-circuit via AbortError),
LifecycleHandler (Startup / Shutdown), Selector with nil-safe
And/Or/Not composition.
- Risk / Identity are defined string types with closed taxonomies;
ParseRisk / ParseIdentity convert raw strings with the
absent-vs-invalid distinction the engine relies on.
- Builder ergonomic constructor (NewPlugin().Observer().Wrap()
...MustBuild()) that enforces name/hookName grammar, hookName
uniqueness, and the Restrict ↔ FailClosed pairing regardless of
call order.
- Invocation is a read-only interface; the framework's concrete
invocation type lives in internal/hook so plugins cannot
fabricate denial / strict-mode / identity state. Args() returns
a defensive copy on every call so hook mutation cannot leak
into the original RunE.
- CommandDeniedError + AbortError carry structured fields for the
closed `command_denied` / `hook` envelope contract.
- ResetForTesting gated behind //go:build testing.
- README + godoc examples (Observer / Wrapper / Restrict) + two
runnable example forks (audit-observer, readonly-policy).
Host (internal/platform, internal/hook, internal/cmdpolicy):
- InstallAll: staged plugin registration with atomic commit, panic
isolation, FailOpen / FailClosed semantics, RequiredCLIVersion
semver check, single-Restrict invariant, duplicate-plugin-name
detection.
- hook.Install wraps every runnable cmd.RunE with:
Before observers (panic-safe) → denial guard → composed Wrap
chain → original RunE → After observers (always fire, even on
err). Denied commands physically bypass the Wrap chain so a
plugin Wrapper cannot suppress or rewrite a denial; observers
still see the attempt for audit.
- Recover shim around plugin Wrappers converts panics (including
the factory call) into a structured `hook` envelope with
reason_code=panic; namespacing shim attributes AbortError to
the namespaced hook name.
- cmdpolicy (renamed from internal/pruning) is the user-layer
command policy engine: walks the cobra tree, evaluates each
runnable command against a Rule's four-axis filter (Allow /
Deny / MaxRisk / Identities), produces parent-group aggregate
denials, and installs denyStubs. Rule.AllowUnannotated opts out
of the unannotated-deny gate for gradual adoption; risk_invalid
typos always deny with an edit-distance "did you mean"
suggestion.
- Strict-mode stub in cmd/prune.go composes the shared
detail.* / wrapped CommandDeniedError shape via cmdpolicy
helpers (BuildDenialError / CommandDeniedFromDenial /
DenialDetailMap), so command_denied envelopes from strict-mode
and user-layer policy carry the same closed-enum fields
(detail.layer / reason_code / policy_source). The historical
short Message + independent Hint are preserved unchanged.
- cmdpolicy/yaml: structural parsing of ~/.lark-cli/policy.yml
with KnownFields strict mode, including allow_unannotated.
- `config policy show` / `config policy validate` and the plugin
inventory diagnostic surface the resolved Rule (allow,
deny, max_risk, identities, allow_unannotated) and the hook
contributions per plugin.
Envelope contract (docs/extension/reason-codes.md):
- error.type is a closed set: command_denied, hook, plugin_install,
plugin_conflict, plugin_lifecycle.
- reason_code is a closed enum per error.type, dispatched on by
external agents and CI integrations.
- detail.layer = "policy" | "strict_mode" attributes the rejection.
Build / CI:
- Makefile unit-test / vet / coverage and ci.yml fast-gate +
unit-test + coverage now pass -tags testing so register_testing.go
is visible; ./extension/... is in the package list so the SDK's
own tests actually run.
- fmt-check and examples-build Makefile targets.
- bmatcuk/doublestar/v4 added as a direct dependency for `**` glob
matching in Rule.Allow / Rule.Deny.
Author-facing material:
- docs/extension/ (quickstart, plugin-author-guide, reason-codes)
is provided in the working tree but kept out of git tracking
per repo convention (.gitignore covers docs/).
Change-Id: I3b8ecc2923bd54c2dff19e5dce8a0855a6f9e703
* feat(extension/platform): plugin SDK with policy engine, hooks, and Builder
Introduces extension/platform — the in-process plugin SDK external
Go forks of lark-cli use to extend or restrict the command surface.
Plugins compile in via blank import; there is no dynamic loading
and no RPC isolation.
Public SDK (extension/platform):
- Plugin interface (Name / Version / Capabilities / Install).
- Registrar verbs: Observe, Wrap, On, Restrict.
- Hook types: Observer (side-effect, panic-safe, fires Before/After
RunE), Wrapper (middleware, may short-circuit via AbortError),
LifecycleHandler (Startup / Shutdown), Selector with nil-safe
And/Or/Not composition.
- Risk / Identity are defined string types with closed taxonomies;
ParseRisk / ParseIdentity convert raw strings with the
absent-vs-invalid distinction the engine relies on.
- Builder ergonomic constructor (NewPlugin().Observer().Wrap()
...MustBuild()) that enforces name/hookName grammar, hookName
uniqueness, and the Restrict ↔ FailClosed pairing regardless of
call order.
- Invocation is a read-only interface; the framework's concrete
invocation type lives in internal/hook so plugins cannot
fabricate denial / strict-mode / identity state. Args() returns
a defensive copy on every call so hook mutation cannot leak
into the original RunE.
- CommandDeniedError + AbortError carry structured fields for the
closed `command_denied` / `hook` envelope contract.
- ResetForTesting gated behind //go:build testing.
- README + godoc examples (Observer / Wrapper / Restrict) + two
runnable example forks (audit-observer, readonly-policy).
Host (internal/platform, internal/hook, internal/cmdpolicy):
- InstallAll: staged plugin registration with atomic commit, panic
isolation, FailOpen / FailClosed semantics, RequiredCLIVersion
semver check, single-Restrict invariant, duplicate-plugin-name
detection.
- hook.Install wraps every runnable cmd.RunE with:
Before observers (panic-safe) → denial guard → composed Wrap
chain → original RunE → After observers (always fire, even on
err). Denied commands physically bypass the Wrap chain so a
plugin Wrapper cannot suppress or rewrite a denial; observers
still see the attempt for audit.
- Recover shim around plugin Wrappers converts panics (including
the factory call) into a structured `hook` envelope with
reason_code=panic; namespacing shim attributes AbortError to
the namespaced hook name.
- cmdpolicy (renamed from internal/pruning) is the user-layer
command policy engine: walks the cobra tree, evaluates each
runnable command against a Rule's four-axis filter (Allow /
Deny / MaxRisk / Identities), produces parent-group aggregate
denials, and installs denyStubs. Rule.AllowUnannotated opts out
of the unannotated-deny gate for gradual adoption; risk_invalid
typos always deny with an edit-distance "did you mean"
suggestion.
- Strict-mode stub in cmd/prune.go composes the shared
detail.* / wrapped CommandDeniedError shape via cmdpolicy
helpers (BuildDenialError / CommandDeniedFromDenial /
DenialDetailMap), so command_denied envelopes from strict-mode
and user-layer policy carry the same closed-enum fields
(detail.layer / reason_code / policy_source). The historical
short Message + independent Hint are preserved unchanged.
- cmdpolicy/yaml: structural parsing of ~/.lark-cli/policy.yml
with KnownFields strict mode, including allow_unannotated.
- `config policy show` / `config policy validate` and the plugin
inventory diagnostic surface the resolved Rule (allow,
deny, max_risk, identities, allow_unannotated) and the hook
contributions per plugin.
Envelope contract (docs/extension/reason-codes.md):
- error.type is a closed set: command_denied, hook, plugin_install,
plugin_conflict, plugin_lifecycle.
- reason_code is a closed enum per error.type, dispatched on by
external agents and CI integrations.
- detail.layer = "policy" | "strict_mode" attributes the rejection.
Build / CI:
- Makefile unit-test / vet / coverage and ci.yml fast-gate +
unit-test + coverage now pass -tags testing so register_testing.go
is visible; ./extension/... is in the package list so the SDK's
own tests actually run.
- fmt-check and examples-build Makefile targets.
- bmatcuk/doublestar/v4 added as a direct dependency for `**` glob
matching in Rule.Allow / Rule.Deny.
Author-facing material:
- docs/extension/ (quickstart, plugin-author-guide, reason-codes)
is provided in the working tree but kept out of git tracking
per repo convention (.gitignore covers docs/).
Change-Id: I3b8ecc2923bd54c2dff19e5dce8a0855a6f9e703
* refactor(policy): remove validate command and update diagnostics
* fix(extension/platform): address PR review must-fix items
- cmdpolicy: skip AnnotationPureGroup commands in EvaluateAll,
aggregateParents, and hasRunnableDescendant so user-layer policy
no longer blocks `<group> --help` after the unknown-subcommand
guard attaches RunE to every parent
- cmd/root: tag guarded parent groups with AnnotationPureGroup
- extension/platform: drop `//go:build testing` from register_testing.go
so `go test ./...` works without an extra build tag
- extension/platform/README: inline reason_code reference, fix plugin
lifecycle diagram order (init/Register precede RegisteredPlugins)
- cmd/platform_bootstrap: route userPolicyPath through
core.GetBaseConfigDir so LARKSUITE_CLI_CONFIG_DIR is honoured
- cmdpolicy: add RedactHomeDir helper, fold base config dir and
$HOME prefixes for config policy show + resolver errors
- internal/platform: reject unrecognised FailurePolicy values with
invalid_capability instead of silently fail-open
- cmd/config: surface diagnostic policy/plugins commands in
`config --help` Long text
- CHANGELOG: document command_denied error.type rename and
unknown_subcommand exit-2 behavior change
* fix(extension/platform): address CodeRabbit review comments + CI gofmt
- hook/install: propagate wrapper-injected ctx to invokeOriginal so
RunE/Run see context values added by upstream Wrappers
- hook/testing: SetStderrForTesting returns a restore func; tests now
defer it via t.Cleanup to avoid cross-test sink leakage
- cmdpolicy/active: deep-copy ActivePolicy.Rule on SetActive/GetActive
so callers can't mutate the stored global through shared slices
- platform/inventory: deep-copy Inventory + nested Plugins / HookEntry
/ RuleView slices on SetActiveInventory / GetActiveInventory
- platform/staging: Restrict clones the plugin-supplied Rule before
retaining it so the plugin can't mutate it after Install returns
- platform/version: reject RequiredCLIVersion with more than three
numeric components instead of silently truncating 1.2.3.4 to 1.2.3
- cmd/platform_bootstrap: clear cmdpolicy.SetActive on yaml resolver
error so config policy show doesn't surface a stale rule
- cmd/platform_bootstrap_test: tmpHome pins LARKSUITE_CLI_CONFIG_DIR
so host env can't bleed into the policy test fixtures
- cmdpolicy/apply: installDenyStub returns bool; Apply count no longer
over-reports when strict-mode short-circuits the install
- cmdpolicy/engine: aggregateParents now returns the runnable hybrid's
own denial status when all children are placeholder branches
- cmdpolicy/resolver_test: use t.TempDir()-rooted missing path instead
of hardcoded /nonexistent for hermetic missing-file assertion
- cmd/config/plugins: empty-inventory branch emits total: 0 so the
JSON schema stays stable across populated/empty cases
- cmd/platform_guards_test: select leaf by RunE != nil (not Runnable)
so the test doesn't nil-deref on Run-only commands
- gofmt run on previously committed cmdpolicy/path*.go (CI fast-gate)
* fix(cmdpolicy): replace filepath.Abs with filepath.Clean for lint policy
The depguard / forbidigo rule blocks filepath.Abs in internal/ on the
grounds that it accesses the filesystem (Getwd) directly. Switch
RedactHomeDir + foldPrefix to operate on filepath.Clean strings; real
callers pass already-absolute paths (resolver builds yamlPath via
filepath.Join on the absolute config root), so the redaction outcome
is unchanged for production inputs. Relative inputs fall through to
the unchanged branch — filepath.Rel rejects the mixed-absoluteness
case with an error, which the foldPrefix helper already treats as
"not a hit".
* refactor(cmdpolicy): pure Resolve + drop path redaction & verbose comments
- Resolve becomes a pure function; I/O moves to LoadYAMLPolicy so
precedence selection can be unit-tested without vfs mocks
- ActivePolicy drops YAMLPath; config policy show JSON loses yaml_path
and yaml_shadowed (and the TOCTOU stat that surfaced them)
- RedactHomeDir and path_test.go removed: the home-dir folding was only
earning its keep through the now-deleted yaml_path field
- cmd/build.go bootstrap block trimmed from 71 to 39 lines by cutting
PR-rationale comments; one note kept for the fail-CLOSED-vs-fail-OPEN
business rule
- cmd/config/config.go: parent Long no longer hard-codes hidden command
hints, matching their Hidden:true intent
Change-Id: Icfbb818ce3ef523c63286bfbed34c49be08ed6a2
* refactor(platform): drop StrictMode/Identity from Invocation interface
These two accessors were documented in the public SDK as "After observers
always see ok=true" but the framework never plumbed values to them, so they
always returned ("", false). Zero internal/example/test callers; a plugin
author trusting the doc would silently get wrong behaviour.
Identity is also fundamentally unsuited for Before observers (per-command
identity resolves inside RunE via f.AuthFor, after Before fires). StrictMode
is a global value better placed on a Framework/Environment interface than
per-Invocation. Removing is non-breaking now (no callers); adding later is
non-breaking too.
Change-Id: Ice200543e9bca3bda759ad98a6e34a56df69e915
* fix(prune): preserve original metadata on strict-mode denial stubs
strictModeStubFrom built a fresh *cobra.Command from scratch, dropping
the original command's annotations (risk_level, lark:supportedIdentities,
cmdmeta.domain) and help text. cobraCommandView is a live proxy walking
parent annotations, so after the Remove+Add replacement, audit observers
firing on a strict-mode-denied command saw Cmd().Risk()=("",false) and
Cmd().Identities()=nil -- breaking the first-class use case for
audit/compliance plugins.
Copy child.Annotations into the stub (stamping the denial annotations on
top) and propagate Short/Long for help-text parity with
cmdpolicy/apply.go::installDenyStub, which preserves these by virtue of
mutating in place.
Regression test asserts risk_level / supportedIdentities / Short / Long
all survive replacement, alongside the denial annotations.
Change-Id: I19810a34575996344b63e839066888c154d69335
* chore(platform): align docs with implementation; fold home in yaml warnings
Followup cleanup to the previous three refactor commits, addressing review
fallout where public docs / examples / contract notes still pointed at
deleted symbols or unimplemented designs:
- cmd/build.go: Build() docstring now mentions the plugin install + Startup
emit side effects; Shutdown only fires on Execute path
- extension/platform/doc.go, lifecycle.go, invocation.go: drop references
to the deleted StrictMode/Identity methods, restore minimal Godoc on
Cmd/Args/Started
- extension/platform/view.go, cmd/platform_bootstrap.go,
internal/hook/install.go: rewrite "snapshot before pruning" promise to
match the actual contract (live view + strict-mode stub metadata
preservation)
- cmd/platform_guards_test.go: stubInvocation drops the two old methods
- cmd/platform_bootstrap.go: redactHome() last-mile folds $HOME -> ~ in
warnPolicyError so an os.PathError carrying the absolute policy path
does not leak the user's home dir to stderr / agent / CI logs
- examples/readonly-policy/README.md: drop yaml_path from the sample
`config policy show` envelope (the field was removed in
|
||
|
JackZhao10086
|
a6de8360f0 |
feat(auth): add scope hint for missing authorization errors (#776)
* feat(auth): add scope hint for missing authorization errors * fix(auth): handle existing hints in missing scope error * refactor(auth): centralize user authorization error detection * fix(auth): handle nil error case in IsNeedUserAuthorizationError |
||
|
liuxinyanglxy
|
4d4508dfd7 |
feat(event): add event subscription & consume system (#654)
* feat(event): add event subscription & consume system with orphan bus detection
Introduces end-to-end Feishu event consumption via a new `lark-cli event`
command family. Users can subscribe to and consume real-time events
(IM messages, chat/member lifecycle, reactions, ...) in a forked bus
daemon architecture with orphan detection, reflected + overrideable JSON
schemas, and AI-friendly `--json` / `--jq` output.
Commands
--------
- `event list [--json]` list subscribable EventKeys
- `event schema <key>` Parameters + Output Schema + auth info
- `event consume <key>` foreground blocking consume; SIGINT/SIGTERM
/stdin-EOF shutdown; `--max-events` /
`--timeout` bounded; `--jq` projection;
`--output-dir` spool; `--param` KV inputs
- `event status [--fail-on-orphan] [--json]` bus daemon health
- `event stop [--all] [--force] [--json]` stop bus daemon(s)
- `event _bus` (hidden) forked daemon entrypoint
Architecture
------------
- Bus daemon (internal/event/bus): per-AppID forked process that holds
the Feishu long-poll connection and fans events out to 1..N local
consumers over an IPC socket. Drop-oldest backpressure, TOCTOU-safe
cleanup via AcquireCleanupLock, idle-timeout self-shutdown, graceful
SIGTERM.
- Consume client (internal/event/consume): fork+dial the daemon,
handshake, remote preflight (HTTP /open-apis/event/v1/connection),
JQ projection, sequence-gap detection, health probe. Bounded
execution (`--max-events` / `--timeout`) for AI/script usage.
- Wire protocol (internal/event/protocol): newline-delimited JSON
frames with 1 MB size cap and 5 s write deadlines. Hello / HelloAck /
PreShutdownCheck / Shutdown / StatusQuery control messages.
- Orphan detection (internal/event/busdiscover): OS process-table scan
(ps on Unix, PowerShell on Windows) with two-gate cmdline filter
(lark-cli + event _bus) that naturally rejects pid-reused unrelated
processes.
- Transport (internal/event/transport): Unix socket on darwin/linux,
Windows named pipe on windows.
- Schema system (internal/event, internal/event/schemas): SchemaDef with
mutually-exclusive Native (framework wraps V2 envelope) or Custom
(zero-touch) specs. Reflection reads `desc` / `enum` / `kind` struct
tags, with array elements diving into `items`. FieldOverrides overlay
engine addresses paths via JSON Pointer (including `/*` array
wildcard) and runs post-reflect, post-envelope. Lint guards orphan
override paths.
- IM events (events/im): 11 keys — receive / read / recalled, chat and
member lifecycle, reactions — all with per-field open_id / union_id /
user_id / chat_id / message_id / timestamp_ms format annotations.
Robustness
----------
- Bus idle-timer race fix: re-check live conn count under lock before
honoring the tick; Stop+drain before Reset per timer contract.
- Protocol frame cap: replace `br.ReadBytes('\n')` with `ReadFrame` that
rejects frames > MaxFrameBytes (1 MB). Closes a DoS path where any
local peer could grow the reader's buffer unbounded.
- Control-message writes gated by WriteTimeout (5 s) so a wedged peer
kernel buffer can't stall writers indefinitely.
- Consume signal goroutine: `signal.Stop` + `ctx.Done` select, no leak
across repeated invocations in the same process.
- JQ pre-flight compile so bad expressions fail before the bus fork and
any server-side PreConsume side effects.
- `f.NewAPIClient`'s `*core.ConfigError` now passes through unwrapped
so the actionable "run lark-cli config init" hint reaches the user.
Subprocess / AI contract
------------------------
- `event consume` emits `[event] ready event_key=<key>` on stderr once
the bus handshake completes and events will flow. Parent processes
block-read stderr until this line before reading stdout — no `sleep`
fallback needed.
- All list-like commands have `--json` for structured consumption.
- Skill docs in `skills/lark-event/` (SKILL.md + references/) brief AI
agents on the command surface, JQ against Output Schema, bounded
execution, and subprocess lifecycle.
Testing
-------
Unit tests across bus/hub, consume loop, protocol codec, dedup,
registry, transport (Unix + Windows), schema reflection, field
overrides, pointer resolver. Integration tests cover fork startup,
shutdown, orphan detection, probe, stdin EOF, preflight, bounded
execution, and Windows busdiscover PowerShell compatibility.
Change-Id: Ib69d6d8409b33b99790081e273d4b5b01b7dbf80
* fix(event): address CodeRabbit findings + lift patch coverage above 60%
CodeRabbit comments (PR #654)
-----------------------------
1. bus/dedup: IsDuplicate dropped legitimate (post-TTL) events after
cleanupExpired fired. The run-every-1000-inserts cleanup removed
TTL-expired IDs from the `seen` map but left them in the ring;
IsDuplicate's ring-scan fallback then rediscovered them and falsely
reported "duplicate", and bus.Publish silently dropped the event.
Removed the ring-scan branch — `seen` is the sole authority, the ring
only bounds map size via overflow eviction. New regression test
TestDedupFilter_TTLExpiryAfterCleanupRunRespected exercises the 10-
insert + cleanup path and guards the fix.
2. consume/remote_preflight: the decoder only read `data.online_instance_
cnt`. A non-zero business code with no data payload decoded to 0 and
callers treated it as "verified zero", forking a local bus that would
duplicate events. Added Code / Msg fields and promoted code != 0 into
an error so the caller distinguishes verified-zero from check-failed.
3. cmd/event/stop: swapped os.ReadDir / os.Stat to vfs.ReadDir / vfs.Stat
in discoverAppIDs per project guideline (enables test mocking). New
TestDiscoverAppIDs_* lifts discoverAppIDs from 0% to 100%.
4. cmd/event/appmeta_err: narrowed authURLPattern from
feishu.cn|feishu.net|larksuite.com|larkoffice.com to the two hosts
consoleScopeGrantURL actually produces. Kept the allowlist pinned to
ResolveEndpoints' output with a comment flagging the synchrony.
5. cmd/event/list: moved "No EventKeys registered." and "Use 'event
schema <key>' for details." hints to stderr so `event list | jq`
style pipelines don't ingest them as data.
6. cmd/event/schema: runSchema is a RunE entry point; swapped the bare
fmt.Errorf on resolveSchemaJSON failure to output.Errorf so AI
agents parse a structured error envelope.
Coverage bumps (patch ~50% -> ~60%)
-----------------------------------
internal/event/consume/loop_test.go: loop.go was 0% at patch time.
New tests cover consumeLoop end-to-end via net.Pipe (events -> sink,
max-events -> ctx.Done -> PreShutdownCheck/Ack), seq-gap warning,
jq filtering + early compile failure, isTerminalSinkError classifier.
Takes consumeLoop from 0% to ~74%.
internal/event/protocol/messages_test.go: all NewXxx constructors,
Encode/Decode roundtrip per message type, EncodeWithDeadline deadline
enforcement, ReadFrame MaxFrameBytes rejection + EOF propagation.
Takes protocol from 28% to ~86%.
Also bundles small UX polish:
- cmd/event/consume: --output-dir flag doc flags path-traversal behavior;
jq-validation failures now re-wrap with an event-specific hint
pointing at `event schema` for payload shape.
- internal/event/consume.validateParams: error now names the EventKey
and lists valid param names inline so AI callers recover without a
second `event schema` round-trip.
- skills/lark-event: description expanded to mention
listener/subscribe/consume synonyms + the IM scope set explicitly;
lark-event-im reference polished; obsolete lark-event-subscribe
reference removed.
Verified with go test -race -timeout 120s across ./cmd/event/...,
./events/..., ./internal/event/...; gofmt clean; go vet clean.
Change-Id: I3837b8645ea1d7529c9a8fd4c2bbfa965ae1b519
* test(event): cover format helpers + cobra factories
Adds cmd/event/format_helpers_test.go covering the pure output helpers
and factory wire-ups that RunE-level tests would need a live bus to
exercise:
- writeStopJSON: shape assertions + nil → [] (scripts expecting
.results | length must not see null).
- writeStopText: stdout vs stderr routing — stopped / no-bus lines to
stdout, refused / errored lines to stderr.
- busState.String: all three discriminator values.
- humanizeDuration: each bucket boundary (seconds / minutes / hours / days).
- writeStatusText: covers stateNotRunning / stateRunning (with consumer
table) / stateOrphan (with kill hint).
- writeStatusJSON: orphan entry carries suggested_action + issue;
running entry must NOT carry those fields (hint-leak guard for
scripts that key on issue != "").
- exitForOrphan: flag-off never errors; flag-on errors iff any orphan
is present, with ExitValidation code.
- NewCmdConsume / NewCmdStatus / NewCmdStop / NewCmdList / NewCmdBus:
flag registration + RunE presence, so review catches flag-name drift.
NewCmdBus check also pins Hidden=true.
Lifts cmd/event coverage 51.7% → 61.1%; aggregate event-package
coverage crosses the 60% codecov patch threshold (62% locally).
Change-Id: I9ecf3d905a8f9607b9441ee8a61e746496e2be63
* fix(event): address lint + deadcode CI failures
4 golangci-lint findings + 1 deadcode finding flagged on PR #654.
lint
----
1. cmd/event/stop.go:86 (ineffassign): `targets := []string{}` is
overwritten by both branches of the `if o.all` below, so the empty-
slice initializer is dead. Switched to `var targets []string`.
2. cmd/event/consume.go nilerr: the user-identity scope preflight
swallows a non-nil ResolveToken error and returns nil. This is
intentional — a missing/expired user token must not block consume;
the bus handshake will surface the real auth error with actionable
hints. Added `//nolint:nilerr` with a 4-line comment pinning the
reasoning.
3. events/im/message_receive.go:62 nilerr: malformed JSON payload
returns the original bytes + nil so consumers still see the event
(the WARN breadcrumb lives in the outer loop). Added
`//nolint:nilerr` with a one-line comment.
4. internal/event/schemas/fromtype_test.go:26 unused: `unexportedStr`
is a reflection-test fixture — its presence (not value) exercises
the FromType skip-unexported path verified at the "unexported
field should not be in schema" assertion. Added `//nolint:unused`
and a 4-line comment pointing at the guarded assertion.
deadcode
--------
5. internal/event/testutil/testutil.go: NewTCPFake has no callers in
the repo. Removed the constructor plus the `inner == nil` TCP-mode
branches from Listen / Dial / Cleanup. FakeTransport now only
supports the wrapped-overlay mode (NewWrappedFake), which is the
one every existing test uses. Doc comment simplified accordingly.
Verified locally: go test -race -timeout 120s across ./cmd/event/...,
./events/..., ./internal/event/... all green; gofmt clean; go vet
clean.
Change-Id: Ie8a2270827a0bde6b8159ab70aaf5c1e9ca7d5b9
* fix(event): drop stale enum + simplify protocol test type helper
- events/im/message_receive.go: dropped the `enum` tag on
ImMessageReceiveOutput.MessageType. convertlib registers many more
message types than the old 11-item list (video / location /
calendar / todo / vote / hongbao / merge_forward / folder / ...),
so a partial enum would tell AI consumers that valid values like
"video" are invalid and produce false-negative JQ filters.
- internal/event/protocol/messages_test.go: collapsed the
typeOf → reflectTypeName → stringType chain in
TestEncode_DecodeRoundtripAllTypes to a single fmt.Sprintf("%T", v).
The hand-maintained type switch silently returned "<unknown>" for
any new message type, which would have let future Decode bugs slip
past the roundtrip assertion. Also removed a dead `cases` table at
the top of TestConstructors_PinTypeField left over from an earlier
refactor.
Change-Id: I831e96f8417e80637596030d652a559de0d33122
* docs(event): polish skill docs + rename root_path_hint to jq_root_path
- skills/lark-event/SKILL.md, lark-event-im.md: translated to English,
reorganized around a top-level "Core commands" table, scenario
recipes tightened.
- cmd/event/schema.go: renamed the writeSchemaJSON hint field
RootPathHint / "root_path_hint" -> JQRootPath / "jq_root_path" to
make its purpose (a jq path prefix) obvious at the call site; no
external consumer depends on the old name yet.
Change-Id: I00c14061ca33caedc0975bfeadc4b26d3dcd314d
* chore(event): strip excessive comments
Change-Id: I8f44f36f5dbdba3ef95dfc67069dc796232f91ec
* fix(event): dedup self-eviction race + protocol oversized-frame test
dedup: in IsDuplicate, the ring-slot eviction step deleted seen[id] even
when ring[pos] equalled the freshly-recorded id (post-TTL reinsertion
landing on its own historical slot). Net result: ring still held id but
seen did not, so the next IsDuplicate(id) returned false and the
duplicate was delivered. Skip the delete when old == eventID. New
TestDedupFilter_SelfEvictionPreservesFreshEntry pins the invariant by
pre-loading the ring slot and asserting the second call still reports
duplicate.
protocol: TestReadFrame_RejectsOversized used strings.Contains feeding
t.Logf, so any non-nil error passed — including a future regression
that returned io.ErrUnexpectedEOF while silently keeping the buffer
unbounded. Promoted MaxFrameBytes overflow to a sentinel
ErrFrameTooLarge and the test now asserts via errors.Is.
Change-Id: I50281dad392152b0ca083fd30c38eb0695e63bd3
* docs(event): clarify .content shape per message_type + add sender filter recipe
Change-Id: I619fd15c1a362e42e6602fd3e3316bbc75eddc5e
* fix(event): replace cmdline-regex bus discovery with PID file + close concurrent fork race
Bus discovery previously walked the OS process table and parsed `--profile cli_*` from
cmdline; the regex rejected any non-cli_ profile name (D-03a). Replace with per-AppID
bus.pid + bus.alive.lock under events/<AppID>/, probed via try-lock. AppID round-trips
through the directory name, so the profile-vs-AppID confusion is gone by construction.
Also fix B-07 (two consumers each fork an independent bus, halving event delivery):
- forkBus holds bus.fork.lock until child is dial-able, not just until cmd.Start
- bus daemon takes alive.lock before binding the socket; cleanup-TOCTOU race can no
longer leave two listeners on different inodes
status.go renders an orphan with PID=0 distinctly (live bus but pid file unreadable)
so we never print "Action: kill 0".
Change-Id: I3bf0a6cf1d91fb274ac5a6df83d66896aafb291f
* style(event): gofmt bus.go
Trailing blank line introduced when appending acquireAliveLock helper.
Change-Id: I4ae1b4a4363dc6c89dcbd6a170f4563117490ba3
* fix(event): swap os.Remove/Rename for vfs.* and silence forbidigo on internal diagnostics
golangci-lint forbidigo blocks os.* in internal/. Switch the pid-file write to vfs.Remove/vfs.Rename and add a nolint marker on the two stderr diagnostics in busdiscover, matching the existing pattern in consume/*.
Change-Id: Ia6768be62aefeb8ca40f991d3130a78ef2ec0ea5
* fix(event): cross-platform --all + clean SIGPIPE shutdown for consume
- stop --all: replace bus.sock-file probe with busdiscover lock-based
scan; previously skipped Windows entirely (named-pipe transport, no
socket on disk) and misidentified Unix stale sockets as live. Same
win for `event status` (shares discoverAppIDs).
- consume: ignore SIGPIPE so a closed stdout pipe (e.g. `... | head -n 1`)
surfaces as EPIPE error and reaches the existing isTerminalSinkError
cleanup path (log "output pipe closed", lastForKey query, hub
unregister), instead of being killed by Go's default fd 1/2 SIGPIPE
handler with exit 141 and zero deferred cleanup.
Build-tagged: real on unix, no-op on windows (no SIGPIPE there).
Change-Id: I453b19f05c489fd9d5c1a9ba3bdc35e127c15b83
* docs(event): translate IM EventKey descriptions and field tags to English
Aligns with the rest of the codebase (titles, struct names, README) which
are already in English. Surfaces in `event list` / `event schema` and is
also consumed by AI agents.
- events/im/message_receive.go: 11 desc tags on ImMessageReceiveOutput
- events/im/native.go: 10 description fields on Native EventKeys
- events/im/register.go: im.message.receive_v1 Description
Change-Id: I6f46950b4793f137e0129c1f06019a3419195443
* docs(event): drop misleading AuthTypes[0] auto-default claim
The KeyDefinition comment and SKILL.md flag table both stated that
`--as auto` resolves to `AuthTypes[0]`. It does not — ResolveAs goes
through global rules (config default_as / credential hint / `bot`
fallback) without consulting the EventKey. AuthTypes is only used by
CheckIdentity as a post-resolve whitelist.
Reword the field comment to plain whitelist semantics and have SKILL.md
defer `--as` documentation to lark-shared.
Change-Id: Ia5d3d3790aed05813a0fa72d6b43518224e2055b
* revert(comments): restore original comments on 3rd-party files
e61482a stripped comments across 105 files. Restore the four files
authored by others (cmd/build.go, shortcuts/common/{types,runner}.go,
shortcuts/event/subscribe.go) to their pre-strip state so unrelated
documentation isn't churned in this PR.
Change-Id: Ie2527b06bfaf5b3861b0b9dff1e19bbfe7dde456
|
||
|
tuxedomm
|
fbed6beac3 |
refactor: split Execute into Build + Execute with explicit IO and keychain injection (#371)
* refactor(cmd): split Execute into Build with IO/Keychain injection
Introduce a public cmd.Build entry point so external consumers (cli-server,
MCP server, other embedders) can assemble the full CLI command tree without
going through os.Args or the platform keychain. Build takes an
InvocationContext plus functional BuildOptions:
* WithIO(in, out, errOut) — inject custom streams; terminal detection
is derived from the input's underlying *os.File when present.
* WithKeychain(kc) — swap the credential store.
* HideProfile(bool) — registered later in cmd.HideProfile.
The existing Execute() keeps using the internal buildInternal (which
still returns the Factory so error handling can attribute exit codes),
and SetDefaultFS replaces the global VFS implementation at startup.
Hardening applied up front:
* cmdutil.NewIOStreams(in, out, errOut) centralizes terminal detection
so SystemIO() and WithIO share one path.
* cmdutil.NewDefault normalizes partial IOStreams — callers may pass
&IOStreams{Out: buf} without tripping nil-writer panics in the
RoundTripper warnings, Cobra, or the credential provider.
* Build guards against nil functional options.
* An API contract test (cmd/build_api_test.go) exercises Build +
WithIO + WithKeychain + HideProfile + SetDefaultFS so the public
surface is reachable by deadcode analysis.
Change-Id: I7c895e6019817401accbde2db3ef800da40ad319
* feat(schema): filter methods by strict mode in schema output
When strict mode is active, schema output now excludes methods that
are incompatible with the forced identity. This applies to both
pretty and JSON output formats at the resource and method levels.
Change-Id: I39647d5578466c3e23dc545bfb917ae075203ad7
* refactor: centralize strict-mode as flag registration
Change-Id: Iec11151c5002c2f58a8aa067d08747db2e4d2d8c
* fix(cmd): align strict-mode completion and build context; drop dead register shims
Thread a context.Context through RegisterShortcuts, RegisterServiceCommands,
and service.registerService/Resource/Method by introducing explicit
*WithContext variants. Pass that context into NewCmdServiceMethodWithContext
so shortcut and service command construction can honor cancellation and
strict-mode pruning consistently.
Also drop the context-less registerMethod and registerResource shims —
they became unreachable once the WithContext variants took over, and
were the source of new deadcode warnings. registerService is retained
because service_test.go still calls it directly.
Change-Id: I3fe5673aed663c7383bbbc5b0ae94d1f3491f22d
* refactor(cmd): hide --profile in single-app mode via build option
- GlobalOptions gains HideProfile; RegisterGlobalFlags stays pure and reads
the policy off the struct. No boolean-trap parameter, one call per site.
- buildConfig holds GlobalOptions inline so HideProfile(bool) BuildOption
mutates it directly. buildInternal stays a pure assembly function and
requires callers to supply WithIO — no implicit os.Std* fallback.
- Add WithIO BuildOption (wrapping raw io.Reader/Writer with automatic
*os.File TTY detection); Execute injects streams explicitly and decides
profile visibility via HideProfile(isSingleAppMode()).
- installTipsHelpFunc force-shows hidden root flags while rendering the
root command's own help, so single-app users still discover --profile
via lark-cli --help without it polluting subcommand helps.
Change-Id: I7755387e993992ca969e0a4a6f54441cc1993eef
* feat(transport): extension abort hook and shared base transport
Two transport-layer changes bundled because both reshape the base
round-tripper contract used by the HTTP client, the Lark SDK client,
and the in-process updater.
1. Extension abort hook (PreRoundTripE).
Extensions implementing exttransport.AbortableInterceptor can now
return an error from PreRoundTripE to skip the built-in chain. The
post hook still fires with (nil, reason) so extensions can unwind
resources. extensionMiddleware captures the provider name so the
returned *AbortError carries attribution.
2. Shared base transport to stop RPC leak.
util.NewBaseTransport cloned http.DefaultTransport on every call, so
each cmdutil.Factory produced a fresh *http.Transport whose
persistConn readLoop/writeLoop goroutines lingered until
IdleConnTimeout (~90s). Invisible in a single-process CLI, but the
fork is consumed by cli-server where each RPC request constructs a
new Factory, causing linear memory + goroutine growth under load.
Replace NewBaseTransport with SharedTransport — returns
http.DefaultTransport (the stdlib-wide singleton) by default, and
a cached proxy-disabled clone only when LARK_CLI_NO_PROXY is set.
Return type is http.RoundTripper to discourage in-place mutation of
the shared instance. FallbackTransport is kept as a thin
*http.Transport wrapper so existing callers in internal/auth and
internal/cmdutil transport decorators (which were already on the
singleton path) do not have to migrate.
Leak-site migrations: factory_default.go (HTTP + SDK base) and
update.go now call SharedTransport directly.
Change-Id: Ia82462134c5c5ee838be878b887860f41446a235
* fix: unblock Build() zero-opts path and sidecar demo build
Two regressions surfaced on refactor/build-execute-split:
1. cmd.Build(ctx, inv) without WithIO panicked at rootCmd.SetIn/Out/Err
because cfg.streams stayed nil — NewDefault normalized internally
but cmd/build.go never saw the normalized value. Default cfg.streams
to cmdutil.SystemIO() before the root command wires them, and add a
TestBuild_NoOptions regression guard.
2. sidecar/server-demo/main.go still called cmdutil.NewDefault(inv),
so `go build -tags authsidecar_demo ./sidecar/server-demo` failed
with "not enough arguments". Pass nil for the new streams parameter
to preserve the prior behavior (NewDefault substitutes SystemIO).
Change-Id: I20227b2355cde7d19e22eba3eb841c6d8611e8a7
|