feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
package doc
|
|
|
|
|
|
|
|
|
|
import (
|
2026-05-21 10:42:27 +00:00
|
|
|
"cmp"
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
"context"
|
|
|
|
|
"fmt"
|
|
|
|
|
"io/fs"
|
|
|
|
|
"os"
|
|
|
|
|
"path/filepath"
|
|
|
|
|
|
|
|
|
|
"github.com/Tencent/WeKnora/cli/internal/cmdutil"
|
|
|
|
|
"github.com/Tencent/WeKnora/cli/internal/iostreams"
|
2026-05-21 10:42:27 +00:00
|
|
|
"github.com/Tencent/WeKnora/cli/internal/output"
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
)
|
|
|
|
|
|
2026-05-21 10:42:27 +00:00
|
|
|
// 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
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// runUploadRecursive walks dir, filters by Glob, and uploads each match
|
refactor(cli): drop aiclient package; align AGENTS.md with mainstream
Survey of 10 mainstream CLIs (gh, lark, stripe, vercel, supabase, aws,
azure, gcloud, openai/codex, github-copilot-cli) showed env-gated
per-command --help blurbs are a Stripe-only pattern; gh uses env detect
for telemetry only, and lark relies on installed agent Skills + MCP.
Our cmd/mcp/serve already covers the dominant 2025/26 path, so
internal/aiclient/ (136 LOC + 38 callsites) is net maintenance burden
without precedent.
- Drop internal/aiclient/ entirely (annotations + detect + tests)
- Remove 38 SetAgentHelp callsites + agentAwareHelpFunc / SetHelpFunc
wiring in cmd/root.go
- Migrate 4 command-level rules to standard Long help (visible to all,
not env-gated): doc upload mode mutex, kb edit at-least-one,
kb pin idempotent, search chunks channel mutex
- Rewrite AGENTS.md as a developer guide (gh-style 6 H2 / 167 lines):
audience preamble + Build / Architecture / Command Structure /
Testing / Code Style / Error Handling. Drops sections absent in
surveyed projects (Commit & PR Conventions, Who Uses This CLI)
- Clean 14 internal doc refs (ADR-N, spec §X, v0.X) in source comments
and docs that pointed at docs/superpowers/ — that directory is
local-only / uncommitted, so refs are dead for outside readers
- Drop forward-looking "once v0.2 ships" from README
2026-05-14 18:16:45 +00:00
|
|
|
// sequentially. Per-file errors do NOT abort the walk - they accumulate
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
// 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).
|
2026-05-17 17:38:30 +00:00
|
|
|
func runUploadRecursive(ctx context.Context, opts *UploadOptions, fopts *cmdutil.FormatOptions, svc UploadService, kbID, dir string) error {
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
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",
|
|
|
|
|
}
|
|
|
|
|
}
|
2026-05-15 18:38:48 +00:00
|
|
|
// 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
|
|
|
|
|
}
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
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
|
refactor(cli): drop aiclient package; align AGENTS.md with mainstream
Survey of 10 mainstream CLIs (gh, lark, stripe, vercel, supabase, aws,
azure, gcloud, openai/codex, github-copilot-cli) showed env-gated
per-command --help blurbs are a Stripe-only pattern; gh uses env detect
for telemetry only, and lark relies on installed agent Skills + MCP.
Our cmd/mcp/serve already covers the dominant 2025/26 path, so
internal/aiclient/ (136 LOC + 38 callsites) is net maintenance burden
without precedent.
- Drop internal/aiclient/ entirely (annotations + detect + tests)
- Remove 38 SetAgentHelp callsites + agentAwareHelpFunc / SetHelpFunc
wiring in cmd/root.go
- Migrate 4 command-level rules to standard Long help (visible to all,
not env-gated): doc upload mode mutex, kb edit at-least-one,
kb pin idempotent, search chunks channel mutex
- Rewrite AGENTS.md as a developer guide (gh-style 6 H2 / 167 lines):
audience preamble + Build / Architecture / Command Structure /
Testing / Code Style / Error Handling. Drops sections absent in
surveyed projects (Commit & PR Conventions, Who Uses This CLI)
- Clean 14 internal doc refs (ADR-N, spec §X, v0.X) in source comments
and docs that pointed at docs/superpowers/ — that directory is
local-only / uncommitted, so refs are dead for outside readers
- Drop forward-looking "once v0.2 ships" from README
2026-05-14 18:16:45 +00:00
|
|
|
// explicitly - no in-function default needed.
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
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 {
|
2026-05-17 17:38:30 +00:00
|
|
|
if fopts.WantsJSON() {
|
2026-05-21 10:42:27 +00:00
|
|
|
// Empty batch: emit an all-success envelope with zero items.
|
|
|
|
|
return output.WriteBatchEnvelope(iostreams.IO.Out, nil, fopts.TTY, cmdutil.GetProfile())
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
|
|
|
|
fmt.Fprintf(iostreams.IO.Out, "(no files matched %q under %s)\n", opts.Glob, dir)
|
|
|
|
|
return nil
|
|
|
|
|
}
|
|
|
|
|
|
2026-05-21 10:42:27 +00:00
|
|
|
// 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))
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
var firstFailCode cmdutil.ErrorCode
|
2026-05-21 10:42:27 +00:00
|
|
|
channel := cmp.Or(opts.Channel, uploadChannel)
|
|
|
|
|
|
|
|
|
|
outcomes, runErr := cmdutil.RunBatch(ctx, matches, func(ctx context.Context, p string) error {
|
2026-06-10 07:17:51 +00:00
|
|
|
k, err := svc.CreateKnowledgeFromFile(ctx, kbID, p, meta, opts.EnableMultimodel, "", channel, nil)
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
if err != nil {
|
|
|
|
|
code := cmdutil.ClassifyHTTPError(err)
|
|
|
|
|
if firstFailCode == "" {
|
|
|
|
|
firstFailCode = code
|
|
|
|
|
}
|
fix(cli): plug v0.3 final review findings (json + auth + path + bounds + kb)
Seven bugs surfaced via two audit rounds — parallel reviewer agents
plus a real-server end-to-end demo. Each fix arrives with a
regression test.
1. doc upload --recursive --json corrupted the envelope stream.
Per-file FAIL/OK plain lines printed unconditionally to stdout,
then a Success envelope, then on partial failure a typed error
that the root handler turned into a SECOND Failure envelope —
three outputs where one was expected. Fix: gate the plain lines
behind !opts.JSONOut, and add cmdutil.Error.Silent so the JSON-
path partial-failure preserves its typed exit code without
triggering PrintErrorEnvelope's default Failure-envelope write.
2. auth refresh / AuthRetryTransport misclassified HTTP failures as
network.error. RefreshAndPersist wrapped every refresher error
with CodeNetworkError, but the SDK emits "HTTP error 401: ..."
for a rejected refresh token — which should surface as
auth.token_expired. Switched to WrapHTTP for proper status-
derived classification. Affects both `auth refresh` and the
transport's refresh closure.
3. doc download accepted ".." as a server-suggested filename. The
rejection list covered "" / "." / filepath.Separator but not
bare ".." — filepath.Base("..") is "..", which slipped through
to os.Create and produced a confusing local.file_io wrap. Added
to the rejection set.
4. search chunks / docs / kb / sessions had no lower bound on
--limit. `-L 0` / `-L -1` was forwarded to the server with
undefined behavior. Added a 1..1000 bound at the RunE boundary
across all four (matching doc list / session list page-size
bounds). Internal callers in tests can still pass Limit==0 for
the "no client-side cap" runChunks path — the bound only applies
at the user-input layer.
5. cli/AGENTS.md ADR-3 verb-canon summary listed only v0.2 verbs as
"gh-canonical" and missed v0.3 additions (edit, pin, unpin,
download — all gh-canonical) plus locally-introduced ones
(empty, refresh, add, remove, link). Rewritten as an explicit
gh-canonical / locally-introduced split.
6. kb pin returned 404. Server registers /knowledge-bases/{id}/pin
as PUT (router.go:292); SDK was using POST. gin's router silently
404s on method-mismatch (treats it as path-not-found, not 405),
so the CLI classified the response as resource.not_found and
masked the real failure mode. Switched the SDK to http.MethodPut.
The asymmetry that hid this past round 1: kb unpin on a freshly-
created KB hits the no-op branch in cmd/kb/pin.go that skips the
SDK call entirely, so unpin "worked" without ever exercising the
broken path. Only the real-server demo, where kb pin actually
fires, surfaced it.
7. kb edit clobbered current Name when only --description was
passed. EditOptions used *string to distinguish "unset" from
"set to empty", but sdk.UpdateKnowledgeBaseRequest declares both
fields as plain string (no omitempty), so the JSON body always
carried `"name": ""`. Server requires Name → 400. Fix: runEdit
does fetch-then-update — GetKnowledgeBase first, build the PUT
body with current values, then overlay user-set fields. Same
TOCTOU window as kb pin / unpin.
Audit-flagged items intentionally NOT changed:
- kb pin / unpin check-then-toggle TOCTOU: documented; the clean
fix would be a server-side setter and belongs in a separate API
change.
- AuthRetryTransport singleflight test gap for one concurrency
scenario; v0.4 polish.
- cli/README.md:50 "once v0.2 ships" and CHANGELOG.md:8
"10 top-level commands": v0.2-PR artifacts, not v0.3-introduced.
- kb edit / kb pin are v0.3-new commands, so neither bug needs a
cli/CHANGELOG.md Fixed entry — the v0.3 release ships them
working as the Added bullets advertise.
2026-05-13 07:39:09 +00:00
|
|
|
// Per-file progress lines are human progress signal; suppress
|
2026-05-17 17:38:30 +00:00
|
|
|
// under --format json so they don't precede the JSON object on stdout.
|
|
|
|
|
if !fopts.WantsJSON() {
|
fix(cli): plug v0.3 final review findings (json + auth + path + bounds + kb)
Seven bugs surfaced via two audit rounds — parallel reviewer agents
plus a real-server end-to-end demo. Each fix arrives with a
regression test.
1. doc upload --recursive --json corrupted the envelope stream.
Per-file FAIL/OK plain lines printed unconditionally to stdout,
then a Success envelope, then on partial failure a typed error
that the root handler turned into a SECOND Failure envelope —
three outputs where one was expected. Fix: gate the plain lines
behind !opts.JSONOut, and add cmdutil.Error.Silent so the JSON-
path partial-failure preserves its typed exit code without
triggering PrintErrorEnvelope's default Failure-envelope write.
2. auth refresh / AuthRetryTransport misclassified HTTP failures as
network.error. RefreshAndPersist wrapped every refresher error
with CodeNetworkError, but the SDK emits "HTTP error 401: ..."
for a rejected refresh token — which should surface as
auth.token_expired. Switched to WrapHTTP for proper status-
derived classification. Affects both `auth refresh` and the
transport's refresh closure.
3. doc download accepted ".." as a server-suggested filename. The
rejection list covered "" / "." / filepath.Separator but not
bare ".." — filepath.Base("..") is "..", which slipped through
to os.Create and produced a confusing local.file_io wrap. Added
to the rejection set.
4. search chunks / docs / kb / sessions had no lower bound on
--limit. `-L 0` / `-L -1` was forwarded to the server with
undefined behavior. Added a 1..1000 bound at the RunE boundary
across all four (matching doc list / session list page-size
bounds). Internal callers in tests can still pass Limit==0 for
the "no client-side cap" runChunks path — the bound only applies
at the user-input layer.
5. cli/AGENTS.md ADR-3 verb-canon summary listed only v0.2 verbs as
"gh-canonical" and missed v0.3 additions (edit, pin, unpin,
download — all gh-canonical) plus locally-introduced ones
(empty, refresh, add, remove, link). Rewritten as an explicit
gh-canonical / locally-introduced split.
6. kb pin returned 404. Server registers /knowledge-bases/{id}/pin
as PUT (router.go:292); SDK was using POST. gin's router silently
404s on method-mismatch (treats it as path-not-found, not 405),
so the CLI classified the response as resource.not_found and
masked the real failure mode. Switched the SDK to http.MethodPut.
The asymmetry that hid this past round 1: kb unpin on a freshly-
created KB hits the no-op branch in cmd/kb/pin.go that skips the
SDK call entirely, so unpin "worked" without ever exercising the
broken path. Only the real-server demo, where kb pin actually
fires, surfaced it.
7. kb edit clobbered current Name when only --description was
passed. EditOptions used *string to distinguish "unset" from
"set to empty", but sdk.UpdateKnowledgeBaseRequest declares both
fields as plain string (no omitempty), so the JSON body always
carried `"name": ""`. Server requires Name → 400. Fix: runEdit
does fetch-then-update — GetKnowledgeBase first, build the PUT
body with current values, then overlay user-set fields. Same
TOCTOU window as kb pin / unpin.
Audit-flagged items intentionally NOT changed:
- kb pin / unpin check-then-toggle TOCTOU: documented; the clean
fix would be a server-side setter and belongs in a separate API
change.
- AuthRetryTransport singleflight test gap for one concurrency
scenario; v0.4 polish.
- cli/README.md:50 "once v0.2 ships" and CHANGELOG.md:8
"10 top-level commands": v0.2-PR artifacts, not v0.3-introduced.
- kb edit / kb pin are v0.3-new commands, so neither bug needs a
cli/CHANGELOG.md Fixed entry — the v0.3 release ships them
working as the Added bullets advertise.
2026-05-13 07:39:09 +00:00
|
|
|
fmt.Fprintf(iostreams.IO.Out, "FAIL %s: %v\n", filepath.Base(p), err)
|
|
|
|
|
}
|
2026-05-21 10:42:27 +00:00
|
|
|
return err
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
2026-05-21 10:42:27 +00:00
|
|
|
id, name := "", ""
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
if k != nil {
|
|
|
|
|
id = k.ID
|
2026-05-21 10:42:27 +00:00
|
|
|
name = k.FileName
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
2026-05-21 10:42:27 +00:00
|
|
|
uploaded[p] = uploadedFile{ID: id, Name: name}
|
2026-05-17 17:38:30 +00:00
|
|
|
if !fopts.WantsJSON() {
|
fix(cli): plug v0.3 final review findings (json + auth + path + bounds + kb)
Seven bugs surfaced via two audit rounds — parallel reviewer agents
plus a real-server end-to-end demo. Each fix arrives with a
regression test.
1. doc upload --recursive --json corrupted the envelope stream.
Per-file FAIL/OK plain lines printed unconditionally to stdout,
then a Success envelope, then on partial failure a typed error
that the root handler turned into a SECOND Failure envelope —
three outputs where one was expected. Fix: gate the plain lines
behind !opts.JSONOut, and add cmdutil.Error.Silent so the JSON-
path partial-failure preserves its typed exit code without
triggering PrintErrorEnvelope's default Failure-envelope write.
2. auth refresh / AuthRetryTransport misclassified HTTP failures as
network.error. RefreshAndPersist wrapped every refresher error
with CodeNetworkError, but the SDK emits "HTTP error 401: ..."
for a rejected refresh token — which should surface as
auth.token_expired. Switched to WrapHTTP for proper status-
derived classification. Affects both `auth refresh` and the
transport's refresh closure.
3. doc download accepted ".." as a server-suggested filename. The
rejection list covered "" / "." / filepath.Separator but not
bare ".." — filepath.Base("..") is "..", which slipped through
to os.Create and produced a confusing local.file_io wrap. Added
to the rejection set.
4. search chunks / docs / kb / sessions had no lower bound on
--limit. `-L 0` / `-L -1` was forwarded to the server with
undefined behavior. Added a 1..1000 bound at the RunE boundary
across all four (matching doc list / session list page-size
bounds). Internal callers in tests can still pass Limit==0 for
the "no client-side cap" runChunks path — the bound only applies
at the user-input layer.
5. cli/AGENTS.md ADR-3 verb-canon summary listed only v0.2 verbs as
"gh-canonical" and missed v0.3 additions (edit, pin, unpin,
download — all gh-canonical) plus locally-introduced ones
(empty, refresh, add, remove, link). Rewritten as an explicit
gh-canonical / locally-introduced split.
6. kb pin returned 404. Server registers /knowledge-bases/{id}/pin
as PUT (router.go:292); SDK was using POST. gin's router silently
404s on method-mismatch (treats it as path-not-found, not 405),
so the CLI classified the response as resource.not_found and
masked the real failure mode. Switched the SDK to http.MethodPut.
The asymmetry that hid this past round 1: kb unpin on a freshly-
created KB hits the no-op branch in cmd/kb/pin.go that skips the
SDK call entirely, so unpin "worked" without ever exercising the
broken path. Only the real-server demo, where kb pin actually
fires, surfaced it.
7. kb edit clobbered current Name when only --description was
passed. EditOptions used *string to distinguish "unset" from
"set to empty", but sdk.UpdateKnowledgeBaseRequest declares both
fields as plain string (no omitempty), so the JSON body always
carried `"name": ""`. Server requires Name → 400. Fix: runEdit
does fetch-then-update — GetKnowledgeBase first, build the PUT
body with current values, then overlay user-set fields. Same
TOCTOU window as kb pin / unpin.
Audit-flagged items intentionally NOT changed:
- kb pin / unpin check-then-toggle TOCTOU: documented; the clean
fix would be a server-side setter and belongs in a separate API
change.
- AuthRetryTransport singleflight test gap for one concurrency
scenario; v0.4 polish.
- cli/README.md:50 "once v0.2 ships" and CHANGELOG.md:8
"10 top-level commands": v0.2-PR artifacts, not v0.3-introduced.
- kb edit / kb pin are v0.3-new commands, so neither bug needs a
cli/CHANGELOG.md Fixed entry — the v0.3 release ships them
working as the Added bullets advertise.
2026-05-13 07:39:09 +00:00
|
|
|
fmt.Fprintf(iostreams.IO.Out, "OK %s (id: %s)\n", filepath.Base(p), id)
|
|
|
|
|
}
|
2026-05-21 10:42:27 +00:00
|
|
|
return nil
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
failures := 0
|
|
|
|
|
for _, oc := range outcomes {
|
|
|
|
|
if oc.Err != nil {
|
|
|
|
|
failures++
|
|
|
|
|
}
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
|
|
|
|
|
2026-05-17 17:38:30 +00:00
|
|
|
if fopts.WantsJSON() {
|
2026-05-21 10:42:27 +00:00
|
|
|
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 {
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
return err
|
|
|
|
|
}
|
|
|
|
|
} else {
|
2026-05-21 10:42:27 +00:00
|
|
|
fmt.Fprintf(iostreams.IO.Out, "Uploaded %d, Failed %d\n", len(outcomes)-failures, failures)
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
|
|
|
|
|
2026-05-21 10:42:27 +00:00
|
|
|
if runErr != nil {
|
2026-05-17 17:38:30 +00:00
|
|
|
// Silent on the --format json path: the success object above already
|
2026-05-21 10:42:27 +00:00
|
|
|
// 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.
|
|
|
|
|
code := firstFailCode
|
|
|
|
|
if code == "" {
|
|
|
|
|
code = cmdutil.ClassifyContextErr(ctx.Err())
|
|
|
|
|
}
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
return &cmdutil.Error{
|
2026-05-21 10:42:27 +00:00
|
|
|
Code: code,
|
|
|
|
|
Message: fmt.Sprintf("%d of %d uploads failed", failures, len(matches)),
|
2026-05-17 17:38:30 +00:00
|
|
|
Silent: fopts.WantsJSON(),
|
feat(cli): doc download + upload --recursive
Roadmap items 3-9 (download) and 3-10 (recursive upload).
SDK addition (additive, non-breaking):
- OpenKnowledgeFile(ctx, id) (filename, body io.ReadCloser, err) —
the new primitive that returns the body as a stream plus the
server-suggested Content-Disposition filename. The existing path-
form DownloadKnowledgeFile is now a thin wrapper (also gained
partial-file-on-error cleanup, a pre-existing bug exposed by the
reshape).
doc download <id>:
Borrows shape from `gh release download` (positional id, output flag,
`-` sentinel for stdout). Flag names match gh canon verified against
the gh manual: `-O, --output <file>` for destination; `--clobber` for
overwrite control.
- Default: writes to cwd under the server-suggested filename. If the
server didn't send one, errors with input.missing_flag.
- --output FILE / -O FILE: writes to FILE. Refuses overwrite without
--clobber.
- --output -: stream to stdout (binary-safe).
- Partial writes on error are cleaned up.
doc upload --recursive <dir> --glob '*.pdf':
NOTE on upstream parity: `gh release upload` does NOT support
--recursive (verified — it takes individual file args only). `aws s3
cp --recursive` does, but uses `--include`/`--exclude` glob pattern
pairs rather than a single `--glob`. weknora's single positive `--glob`
is a deliberate simplification, not a direct mirror of either tool.
- Walks the tree, filters by base-name glob, uploads each match
sequentially. Per-file line output: OK / FAIL with the underlying
error. Exit 0 only on full success; on partial failure returns the
first failure's typed code so callers can branch. Rejects --name
with --recursive.
- --dry-run lists matches without uploading.
- --json emits {kb_id, uploaded[], failed[]} envelope at completion.
Bugs caught in the post-commit reviewer round:
- SECURITY: server-supplied filename was used in os.Rename without
sanitization. A malicious / buggy server returning
"../../etc/shadow" could escape cwd. Now filepath.Base'd; "." / "/"
/ "" rejected. Regression test added.
- Wasted-bytes path eliminated via the SDK reshape: the CLI now
inspects filename and applies refuseIfExists BEFORE streaming.
Two-phase temp+rename gone.
- refuseIfExists(path, clobber) helper extracted.
- --json honored in --recursive (uploadOutcome was JSON-tagged but
the envelope was never emitted).
7 + 7 unit tests for download (+ path-traversal regression) and
recursive upload (+ JSON envelope regression).
Roadmap: 3-9, 3-10.
2026-05-12 15:36:34 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
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
|
|
|
|
|
}
|