Commit graph

5 commits

Author SHA1 Message Date
Andrew Altshuler
45500a690a
refactor(cli): collapse export + graphs-list onto GraphClient (RFC-009 Phase 3c) (#213)
The last two embedded-vs-remote forks move onto the enum, so every such
`if` in the CLI now lives in client.rs — the point of the refactor.

- `export<W: Write>`: the streaming verb 3b deferred (writes to a writer,
  chunks the HTTP response body, rather than returning a DTO). Embedded
  calls db.export_jsonl_to_writer; Remote streams the chunked body through.
  Opens WITHOUT policy (like reads), so it routes via resolve().
- `list_graphs`: remote-only by design (no local enumeration endpoint), so
  the Embedded arm keeps the loud "requires a remote multi-graph server"
  bail verbatim. Routing it through the enum still buys the shared
  resolve() addressing/token preamble the arm hand-rolled.

Retire the now-orphaned execute_export_to_writer /
execute_export_remote_to_writer pair, and sweep two pre-existing dead fns
while in the files: inferred_config_path (helpers.rs) and yaml_string
(output.rs, shadowed by test-local copies).

parity_matrix gains one row, parity_export — the single intended matrix
change in this phase. Export is a JSONL stream, not a single --json doc,
so it compares the two arms' output line-wise (sorted; twin graphs are
byte-copies so rows need no scrubbing). graphs-list gets no row: its
remote-only behavior is a documented exclusion, not an equality case.

Full workspace tests pass; all 12 parity rows green.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-13 21:03:45 +03:00
aaltshuler
adbb2a181c refactor(cli): consume omnigraph-api-types directly; unify the load mapping
The CLI's wire-DTO imports repoint from omnigraph_server::api to
omnigraph-api-types (the server's other exports — queries registry,
config types — still come from omnigraph-server). The local Load arm's
inline LoadOutput hand-construction in main.rs is extracted into
load_output_from_result next to load_output_from_tables in output.rs, so
both '-> LoadOutput' mappings (engine LoadResult for local, wire
IngestOutput for remote) live in one place.

Deviation from the plan, with reason: LoadOutput stays CLI-side rather
than moving into the wire-DTO crate — it is a rendered CLI output type,
not an HTTP wire DTO, and its mapping consumes a CLI clap type
(CliLoadMode). The shared crate stays strictly wire DTOs. Shapes
unchanged: the parity matrix passes textually unchanged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-13 17:05:32 +03:00
aaltshuler
cd1f175396 feat(cli): omnigraph config migrate — the RFC-008 split (stage 2)
Reads a legacy omnigraph.yaml and produces the three-section split: team
half as a ready-to-review cluster.yaml proposal (graphs with TODO schema
pointers — the legacy file never knew schemas — per-graph queries
directories, policies with applies_to bindings), personal half as an
operator-config merge (actor, output/table defaults — OperatorDefaults
gains the two table keys with their cascade hops — remote graphs with
bearer_token_env become servers entries plus a printed login step, and
legacy aliases split per the RFC: content to the catalog as a manual
step, binding to an operator alias), plus a dropped-keys section with
reasons. Touches nothing without --write; with it, the operator merge is
key-level (existing entries always win; prior file backed up), and
cluster.yaml is emitted only when absent (else cluster.yaml.proposed).
--json emits the report structurally.

The completeness contract is a unit test: every top-level key of the
legacy schema must classify somewhere, or the RFC-008 map has a bug.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 23:32:05 +03:00
aaltshuler
a819ab500e feat(cli): keyed credentials — servers:, the token chain, login/logout (RFC-007 PR 2)
The operator config gains servers: (name -> url; never a token). A remote
command whose URL prefix-matches an operator server resolves its bearer
token through the keyed chain first — OMNIGRAPH_TOKEN_<NAME> env, then the
[<name>] section of ~/.omnigraph/credentials (created 0600 via temp+rename,
#139 finding 7; group/world-readable files refused loudly) — falling
through to the legacy chain unchanged. URL keying makes §D5 rule 3
structural: a token is only ever sent to the server it is keyed to.
Longest-prefix matching with a path-boundary check (http://h:8080 never
matches http://h:8080-evil). Inserting the keyed hop above the legacy chain
is safe by construction — no existing setup can have servers: defined.

omnigraph login <name> stores/rotates one section (token from --token or
one stdin line — the pipe flow keeps secrets out of shell history);
omnigraph logout removes it, idempotently; logging in before declaring the
server warns instead of failing (the gh model).

Coverage: URL-match/no-substring-trap, credentials round-trip preserving
sibling sections, 0600 write + over-permissive refusal, env-name mapping;
the legacy resolve test is now hermetic against a real ~/.omnigraph and
asserts byte-identical legacy behavior with no servers defined; one
spawned-binary e2e walks the whole lifecycle against an authed server:
refusal -> wrong-token login (stdin) -> rotate (--token) -> authorized read
-> env-beats-file -> non-matching-URL negative -> logout revokes.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 21:24:51 +03:00
aaltshuler
916015c416 refactor(cli): split main.rs into cli/helpers/output modules
Verbatim moves: the clap surface (every command/subcommand/arg struct) to
cli.rs, resolution helpers (config/actor/graph/branch/query, remote HTTP,
env/token, scaffolding) to helpers.rs, human/JSON formatting to output.rs,
the in-source test mod to main_tests.rs via #[path]. main.rs (1,184 lines)
keeps main() and the dispatch match. Visibility bumps only; 22 binary
tests green.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 15:14:27 +03:00