Files
WeKnora/cli/cmd/doc/upload_recursive.go
T
nullkey d3c03b2f34 feat(cli)!: v0.10 reliability, agent-UX, and command-surface hardening
A hardening + finalization pass over the agent-first CLI: correctness fixes,
richer machine-readable signals, flag/naming consistency, and a symmetric
config surface. Pre-1.0, so it includes breaking renames.

Correctness:
- agent update: resolve/validate --model/--rerank-model (was storing a bogus
  name verbatim, corrupting config.model_id).
- doctor: honor WEKNORA_HOST / WEKNORA_API_KEY (headless path no longer reports
  "no host configured").
- session ask / MCP session_ask: text answer was empty on non-TTY — the agent
  stream sets Done=true on an intermediate agent_query frame before the answer,
  and AgentAccumulator treated the first Done as terminal. Terminate on the
  `complete` event (new sdk.AgentResponseTypeComplete), not a per-frame Done.
- batch exit codes: any per-item failure collapses to operation.failed (exit 1),
  including `doc upload --recursive` partial failures — a permanent per-file
  failure (e.g. a duplicate) no longer surfaces as a retryable exit 7 an agent
  would loop on; per-item typed errors stay in the envelope.

Agent-first signals & discovery:
- error.exit_code in the envelope (type + exit_code disambiguate the
  input.invalid_argument exit-2-vs-5 split in one JSON read).
- meta.hint on empty content search and on draft doc create; doc wait fails fast
  on a never-parsing draft instead of hanging to --timeout.
- retrieval-readiness is visible in the natural flow: kb status / kb check emit
  retrieval_ready, and kb create hints the fix when no embedding model is bound
  — an unconfigured KB no longer looks silently healthy.
- schema contract completeness: every leaf declares output + >=1 example
  (drift-guarded); output strings match the meta actually emitted; chunk list
  and search docs now emit meta.total_count (both previously dropped it).
- schema tolerates a quoted multi-word command label; zero-state auth — and
  `link` with no profile — point at profile setup / the headless WEKNORA_KB_ID
  path instead of looping on `auth login`.
- id-addressed reads tolerate a redundant --kb (doc view/wait, chunk list/view
  accept and ignore it, declared in schema) so a carried-over --kb doesn't
  exit 2; streaming commands warn that --jq does not apply to an NDJSON stream.
- keep JSON-always as the default; --jq hints spell out the .data path.

Consistency & gating:
- doc create: drop the deprecated --name alias (--title only; pre-1.0 break).
- chunk list --limit aligned to 1..10000; model list --limit/-L with
  has_more/total_count; api write-gates -X PUT/PATCH (exit 10); skills install
  expands a leading ~.
- docs corrected: search docs / doc list --keyword help is case-insensitive
  (server does LOWER LIKE); AGENTS.md risk-action list (no phantom kb.init; add
  model.update / kb.config.set) and batch example (failed item carries `error`);
  session resume --message id comes from `message list`, not the stream.
- auth/profile ergonomics: env credentials are now first-class — `auth token`
  prints the active WEKNORA_API_KEY / WEKNORA_TOKEN, and auth login/logout/refresh
  give an env-aware message instead of looping on "run auth login". `auth logout`
  clears credentials but keeps the profile registered (host preserved for
  re-login); deleting a profile is `profile remove`'s job (clean logout/remove
  separation, matching gh / lark).

Config surface (symmetric read/write, in-place model edits):
- kb config now returns a secret-free KBModelConfigView (was {}); `kb config`
  reads, new `kb config set` writes; `kb init` removed (misnomer).
- kb create --chat-model: retrieval-ready in one step.
- model update: edit a model in place (id preserved, references survive) —
  rotate --api-key-stdin, change base-url / display-name / etc.
- session continue-stream renamed to session resume.

Docs: AGENTS.md is the single wire-contract source; CHANGELOG slimmed; stale
kb-init / continue-stream references removed; skill wire-vocab guard extended.
AGENTS.md / weknora-shared SKILL document retrieval_ready (a KB needs an
embedding model to be searchable), that --jq does not apply to NDJSON
streams, and the env-credential-first auth path; the KB quickstart example
now creates a retrieval-ready KB.
2026-07-06 22:53:01 +08:00

185 lines
5.9 KiB
Go

package doc
import (
"cmp"
"context"
"fmt"
"io/fs"
"os"
"path/filepath"
"github.com/Tencent/WeKnora/cli/internal/cmdutil"
"github.com/Tencent/WeKnora/cli/internal/iostreams"
"github.com/Tencent/WeKnora/cli/internal/output"
)
// uploadedFile holds the server-side result for a successful per-file upload.
// Keyed by file path in the uploadResults map populated by the RunBatch closure.
type uploadedFile struct {
ID string
Name string
}
// runUploadRecursive walks dir, filters by Glob, and uploads each match
// sequentially. Per-file errors do NOT abort the walk - they accumulate
// and the final return aggregates them so the user sees the full picture
// in one run. Exit semantics: nil error on full success, a typed *cmdutil.Error
// when ≥1 file failed (the typed code mirrors the first failure's
// classification so callers can still branch).
func runUploadRecursive(ctx context.Context, opts *UploadOptions, fopts *cmdutil.FormatOptions, svc UploadService, kbID, dir string) error {
if opts.Name != "" {
return &cmdutil.Error{
Code: cmdutil.CodeInputInvalidArgument,
Message: "--name cannot be combined with --recursive (one name can't apply to N files)",
Hint: "drop --name or upload files one at a time",
}
}
// Parse --metadata up front so a malformed value aborts before the
// first SDK call - otherwise a typo in `key=value` would only surface
// per-file as repeated identical errors.
meta, err := parseMetadataKV(opts.Metadata)
if err != nil {
return err
}
info, err := os.Stat(dir)
if err != nil {
if os.IsNotExist(err) {
return cmdutil.Wrapf(cmdutil.CodeUploadFileNotFound, err, "directory not found: %s", dir)
}
return cmdutil.Wrapf(cmdutil.CodeLocalFileIO, err, "stat %s", dir)
}
if !info.IsDir() {
return &cmdutil.Error{
Code: cmdutil.CodeInputInvalidArgument,
Message: fmt.Sprintf("not a directory: %s (drop --recursive to upload a single file)", dir),
}
}
// Sanity-check the pattern up front so a typo doesn't show up as "no
// files matched" per-file. Cobra populates --glob; tests pass it
// explicitly - no in-function default needed.
if _, err := filepath.Match(opts.Glob, ""); err != nil {
return &cmdutil.Error{
Code: cmdutil.CodeInputInvalidArgument,
Message: fmt.Sprintf("invalid --glob %q: %v", opts.Glob, err),
}
}
matches, err := walkMatches(dir, opts.Glob)
if err != nil {
return cmdutil.Wrapf(cmdutil.CodeLocalFileIO, err, "walk %s", dir)
}
if len(matches) == 0 {
if fopts.WantsJSON() {
// Empty batch: emit an all-success envelope with zero items.
return output.WriteBatchEnvelope(iostreams.IO.Out, nil, fopts.TTY, cmdutil.GetProfile())
}
fmt.Fprintf(iostreams.IO.Out, "(no files matched %q under %s)\n", opts.Glob, dir)
return nil
}
// uploaded captures per-path server results for successful uploads.
// Populated by the RunBatch closure; read by the resultFn below.
uploaded := make(map[string]uploadedFile, len(matches))
channel := cmp.Or(opts.Channel, uploadChannel)
outcomes, runErr := cmdutil.RunBatch(ctx, matches, func(ctx context.Context, p string) error {
k, err := svc.CreateKnowledgeFromFile(ctx, kbID, p, meta, opts.EnableMultimodel, "", channel, nil)
if err != nil {
// Per-file progress lines are human progress signal; suppress
// under --format json so they don't precede the JSON object on stdout.
if !fopts.WantsJSON() {
fmt.Fprintf(iostreams.IO.Out, "FAIL %s: %v\n", filepath.Base(p), err)
}
return err
}
id, name := "", ""
if k != nil {
id = k.ID
name = k.FileName
}
uploaded[p] = uploadedFile{ID: id, Name: name}
if !fopts.WantsJSON() {
fmt.Fprintf(iostreams.IO.Out, "OK %s (id: %s)\n", filepath.Base(p), id)
}
return nil
})
failures := 0
for _, oc := range outcomes {
if oc.Err != nil {
failures++
}
}
if fopts.WantsJSON() {
if err := cmdutil.EmitBatch(outcomes, fopts, iostreams.IO.Out, func(p string) any {
f := uploaded[p]
return map[string]any{"id": f.ID, "name": f.Name}
}); err != nil {
return err
}
} else {
fmt.Fprintf(iostreams.IO.Out, "Uploaded %d, Failed %d\n", len(outcomes)-failures, failures)
}
if runErr != nil {
// Silent on the --format json path: the success object above already
// carries per-file detail; without Silent the root error handler would
// print to stderr in addition. ExitCode still walks Code so the typed
// exit-code-by-class contract holds.
// Any per-file failure collapses to operation.failed (exit 1) — the
// per-file batch envelope carries each file's typed error, which is the
// authoritative signal (matches doc/session/chunk batch delete). A
// cancelled / timed-out batch keeps its context class (124 / 130) so a
// permanent partial failure (e.g. a duplicate) is never misreported as a
// retryable 5xx that an agent would loop on.
code := cmdutil.CodeOperationFailed
if ctxErr := ctx.Err(); ctxErr != nil {
code = cmdutil.ClassifyContextErr(ctxErr)
}
return &cmdutil.Error{
Code: code,
Message: fmt.Sprintf("%d of %d uploads failed", failures, len(matches)),
Silent: fopts.WantsJSON(),
}
}
return nil
}
// walkMatches returns every regular file under root whose base name matches
// pattern. Order is filepath.WalkDir's lexical order (stdlib guarantee on
// every supported FS), which is deterministic for test assertions.
func walkMatches(root, pattern string) ([]string, error) {
var out []string
err := filepath.WalkDir(root, func(p string, d fs.DirEntry, err error) error {
if err != nil {
return err
}
if d.IsDir() {
return nil
}
// Skip non-regular files (sockets, devices); the SDK can't upload
// them and they'd show as opaque server errors.
info, ierr := d.Info()
if ierr != nil {
return ierr
}
if !info.Mode().IsRegular() {
return nil
}
ok, merr := filepath.Match(pattern, d.Name())
if merr != nil {
return merr
}
if ok {
out = append(out, p)
}
return nil
})
if err != nil {
return nil, err
}
return out, nil
}