Files
Enterprise-Knowledge-Base/cli/cmd/search/docs.go
T
nullkey 5adcedf170 refactor(cli): v0.3 cross-cutting cleanup
Cross-cutting findings surfaced by the branch-completion review.

Perf bug:
- Factory.Client closure was not memoized. Factory.ResolveKB internally
  calls f.Client() to resolve --kb name → id, then the command's RunE
  calls f.Client() again. Two SDK clients, two keyring reads, two
  AuthRetryTransport allocations per name-resolved invocation, with
  *independent* token state (a refresh in one was invisible to the
  other). Switched to sync.Once like Secrets already does.

Silent bug bait:
- cmdutil.NormalizeHost docstring claimed CodeInputMissingFlag for the
  empty case; code returned CodeInputInvalidArgument. Aligned doc to
  code (present-but-empty is a bad value, not a missing flag).

Agent contract gaps:
- Five user-facing subcommands lacked SetAgentHelp: auth login /
  logout / list / status and chat. Added concise strings with error-
  code call-outs so agents can branch without parsing human strings.

Helper extraction (≥3 callers):
- text.KnowledgeDisplayName(fileName, title, id) — byte-identical
  formatter that was in both cmd/doc/list.go and cmd/search/docs.go.
  Takes fields directly so internal/text stays SDK-free.
- cmdutil.WrapHTTP(cause, fmt, args...) *Error — replaces the
  `Wrapf(ClassifyHTTPError(err), err, ...)` pattern across 24 SDK
  call sites. Sed-driven migration; off-pattern shapes in chat.go
  (used streamErr) and cmdutil/kb.go (in-package) hand-edited.
  Contract test gains a comment update: post-migration the dominant
  pattern is WrapHTTP which the AST scanner skips entirely (only
  NewError/Wrapf selectors inspected); ClassifyHTTPErrorOutputs()
  bridge still covers the dynamic codes those paths can yield.

UX consistency:
- cmd/doc/list.go --page-size help now reads "Items per page
  (1..1000)" matching cmd/session/list.go. The bounds validation
  already enforced 1..1000; the help text was the last drift.

Comment-discipline sweep:
- Deleted the WHAT-only "*Options captures `weknora ...` flag state"
  docstring across 23 files (context, kb, auth, doc, session, search,
  chat, doctor, link). Where the line carried a real WHY clause
  (kb/delete, doc/delete, session/delete, kb/edit), kept the WHY and
  dropped only the leading WHAT phrase.
- Stripped third-party project-name attribution from inline comments
  and one user-visible flag-help string across ~40 files in cli/cmd
  and cli/internal (plus 4 test-file comments). Removed phrases like
  "Mirrors `gh X`", "borrowed from lark-cli", "kubectl-style",
  "gcloud `--project`", "Stripe pattern", and the embedded GitHub
  URLs pointing at those projects. Behavioral descriptions and the
  WHY behind each comment are preserved; only the upstream-name
  attribution is gone. Inspiration / north-star references belong in
  cli/AGENTS.md (the design doc) and commit messages, not scattered
  through every file.

  Triggered by an audit round that surfaced several false / fragile
  parity claims (e.g. "Mirrors `gh repo edit`" — gh repo edit has no
  --name flag; "matches gcloud `--project` id-or-name" — gcloud's
  --project accepts ID only). Rather than fix them one by one, the
  whole category of in-comment external-project references was
  stripped uniformly.
2026-05-14 10:57:17 +08:00

131 lines
4.5 KiB
Go

package search
import (
"context"
"fmt"
"sort"
"strings"
"text/tabwriter"
"github.com/spf13/cobra"
"github.com/Tencent/WeKnora/cli/internal/agent"
"github.com/Tencent/WeKnora/cli/internal/cmdutil"
"github.com/Tencent/WeKnora/cli/internal/format"
"github.com/Tencent/WeKnora/cli/internal/iostreams"
"github.com/Tencent/WeKnora/cli/internal/text"
sdk "github.com/Tencent/WeKnora/client"
)
// docsPageSize is how many entries we pull per ListKnowledge round-trip
// when paging through a KB to filter client-side. Server caps page_size at
// 1000 (per the doc/list bound this branch already added).
const docsPageSize = 200
type DocsSearchOptions struct {
Query string
KB string // raw --kb (UUID or name)
KBID string // resolved id; populated before listing
Limit int
JSONOut bool
}
// DocsSearchService is the narrow SDK surface this command depends on.
// Server has no fuzzy-document-name endpoint, so the CLI pages through
// ListKnowledge and filters by Title / FileName client-side.
type DocsSearchService interface {
ListKnowledge(ctx context.Context, kbID string, page, pageSize int, tagID string) ([]sdk.Knowledge, int64, error)
}
// NewCmdDocs builds `weknora search docs "<query>" --kb <id-or-name>`.
// Pages through the KB's documents and surfaces every entry whose title
// or filename contains the query (case-insensitive). Useful for finding
// a specific upload to download or delete.
func NewCmdDocs(f *cmdutil.Factory) *cobra.Command {
opts := &DocsSearchOptions{}
cmd := &cobra.Command{
Use: `docs "<query>"`,
Short: "Find documents in a knowledge base by name (client-side substring match)",
Example: ` weknora search docs "Q3 forecast" --kb finance
weknora search docs "spec" --kb engineering --limit 5`,
Args: cobra.ExactArgs(1),
RunE: func(c *cobra.Command, args []string) error {
opts.Query = strings.TrimSpace(args[0])
if opts.Query == "" {
return cmdutil.NewError(cmdutil.CodeInputInvalidArgument, "query argument cannot be empty")
}
cli, err := f.Client()
if err != nil {
return err
}
kbID, err := cmdutil.ResolveKBFlag(c.Context(), cli, opts.KB)
if err != nil {
return err
}
opts.KBID = kbID
return runDocsSearch(c.Context(), opts, cli)
},
}
cmd.Flags().StringVar(&opts.KB, "kb", "", "Knowledge base UUID or name (required)")
cmd.Flags().IntVarP(&opts.Limit, "limit", "L", 20, "Maximum results to return")
cmd.Flags().BoolVar(&opts.JSONOut, "json", false, "Output JSON envelope")
_ = cmd.MarkFlagRequired("kb")
agent.SetAgentHelp(cmd, "Lists documents in --kb whose title or file_name contains the query. Pages through the KB sequentially; stops once limit hits found. Returns the full Knowledge object so agents can derive id / file_size / processed_at without a second call.")
return cmd
}
func runDocsSearch(ctx context.Context, opts *DocsSearchOptions, svc DocsSearchService) error {
needle := strings.ToLower(opts.Query)
var matches []sdk.Knowledge
// Page through the KB until limit matches found or pagination exhausted.
// The server returns total; stop when (page-1)*pageSize >= total.
for page := 1; ; page++ {
items, total, err := svc.ListKnowledge(ctx, opts.KBID, page, docsPageSize, "")
if err != nil {
return cmdutil.WrapHTTP(err, "list documents")
}
for _, k := range items {
if matchKnowledge(k, needle) {
matches = append(matches, k)
if opts.Limit > 0 && len(matches) >= opts.Limit {
goto done
}
}
}
if int64(page*docsPageSize) >= total || len(items) == 0 {
break
}
}
done:
sortKnowledgeByRecency(matches)
if opts.JSONOut {
return format.WriteEnvelope(iostreams.IO.Out, format.Success(matches, &format.Meta{KBID: opts.KBID}))
}
if len(matches) == 0 {
fmt.Fprintln(iostreams.IO.Out, "(no matches)")
return nil
}
tw := tabwriter.NewWriter(iostreams.IO.Out, 0, 0, 2, ' ', 0)
fmt.Fprintln(tw, "ID\tFILE\tTYPE\tUPDATED")
for _, k := range matches {
name := text.Truncate(50, text.KnowledgeDisplayName(k.FileName, k.Title, k.ID))
fmt.Fprintf(tw, "%s\t%s\t%s\t%s\n", k.ID, name, k.FileType, k.UpdatedAt.Format("2006-01-02"))
}
return tw.Flush()
}
// matchKnowledge reports whether title or filename contains needle (already
// lowercased by caller).
func matchKnowledge(k sdk.Knowledge, needle string) bool {
return text.ContainsFold(needle, k.Title, k.FileName)
}
// sortKnowledgeByRecency sorts in place by UpdatedAt desc.
func sortKnowledgeByRecency(items []sdk.Knowledge) {
sort.Slice(items, func(i, j int) bool {
return items[i].UpdatedAt.After(items[j].UpdatedAt)
})
}