Files
kebab/crates/kebab-app/Cargo.toml
altair823 8f7b6ee538 feat(embed): candle 임베딩 provider (NUMA-안전, opt-in) + v0.22.0
duo-socket NUMA 서버에서 fastembed(onnxruntime)가 intra-op 스레드를 48개로
하드코딩해 NUMA 힙 손상 → double-free 로 ingest 가 죽는 문제를 회피하기 위해,
같은 multilingual-e5-large 모델을 순수 Rust(candle)로 돌리는 opt-in 임베딩
provider 를 추가한다.

- 신규 crate kebab-embed-candle: CandleEmbedder (kebab_core::Embedder).
  hf-hub safetensors → XLMRobertaModel forward → mask mean-pool → L2 → e5
  prefix. candle 의존성 트리를 이 crate 에 격리 (core/config 외 kebab-* 의존 0).
- 스레드 캡: [models.embedding].num_threads + env KEBAB_EMBED_THREADS →
  글로벌 rayon 풀 1회 캡 (NUMA-안전 레버).
- kebab-app::embedder() 가 provider 분기 (fastembed/onnx/"" → 기존 경로 불변,
  candle → CandleEmbedder, 미지값 → 에러).
- Phase 0 스파이크 crate 제거 (production 흡수).
- 버전 0.21.1 → 0.22.0 (신규 config surface, pre-1.0 minor bump).

패리티: cosine_min=1.000000, max abs diff=2.01e-7 (< 1e-5) → embedding_version
유지, 재색인 0. fastembed default 동작/벡터 불변. wire schema 변경 없음.

검증(파일+exit code): clippy -D warnings EXIT=0(warning 0), test EXIT=0
(candle unit 5 + thread_cap rayon=4 + config 68), parity #[ignore] EXIT=0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 14:52:25 +00:00

106 lines
5.2 KiB
TOML

[package]
name = "kebab-app"
version = { workspace = true }
edition = { workspace = true }
rust-version = { workspace = true }
license = { workspace = true }
repository = { workspace = true }
description = "Facade — orchestrates components for kb-cli/tui/desktop"
[dependencies]
kebab-core = { path = "../kebab-core" }
kebab-config = { path = "../kebab-config" }
kebab-source-fs = { path = "../kebab-source-fs" }
kebab-parse-md = { path = "../kebab-parse-md" }
kebab-chunk = { path = "../kebab-chunk" }
kebab-store-sqlite = { path = "../kebab-store-sqlite" }
kebab-store-vector = { path = "../kebab-store-vector" }
kebab-search = { path = "../kebab-search" }
kebab-embed = { path = "../kebab-embed" }
kebab-embed-local = { path = "../kebab-embed-local" }
kebab-embed-candle = { path = "../kebab-embed-candle" }
kebab-llm = { path = "../kebab-llm" }
kebab-llm-local = { path = "../kebab-llm-local" }
kebab-rag = { path = "../kebab-rag" }
# p9-fb-41 PR-9c-2: facade construction of OnnxNliVerifier when
# `[rag] nli_threshold > 0`. Trait-only consumption via kebab-rag's
# `with_verifier`; no kebab-nli internals leak into kebab-app code
# beyond the construction site in `open_with_config`.
kebab-nli = { path = "../kebab-nli" }
# P6-4: image extractor + OCR + caption adapters live here. App
# threads them into the per-asset dispatch (see `ingest_one_asset`
# image branch). Trait-only consumption — no `kebab-parse-image`
# internals leak into kb-app code.
kebab-parse-image = { path = "../kebab-parse-image" }
# P7-3: PDF text extractor lives here. App threads it into the
# per-asset dispatch (see `ingest_one_asset` PDF branch) and runs the
# resulting `CanonicalDocument` through `kebab-chunk::PdfPageV1Chunker`.
kebab-parse-pdf = { path = "../kebab-parse-pdf" }
lopdf = { workspace = true }
# Enhancement 1 (v0.20.x r2): JPEG dimension decode in pdf_ocr_apply.rs.
# jpeg feature added explicitly (F3 closure-r1) rather than relying on
# feature unification via kebab-parse-image.
image = { version = "0.25", default-features = false, features = ["png", "jpeg"] }
# p10-1A-2: Rust AST extractor lives here. App threads it into the
# per-asset dispatch (see `ingest_one_asset` Code branch) and runs the
# resulting `CanonicalDocument` through `kebab-chunk::CodeRustAstV1Chunker`.
kebab-parse-code = { path = "../kebab-parse-code" }
anyhow = { workspace = true }
blake3 = { workspace = true }
serde = { workspace = true }
serde_json = { workspace = true }
time = { workspace = true }
uuid = { workspace = true }
tracing = { workspace = true }
tracing-subscriber = { version = "0.3", features = ["env-filter", "fmt", "json"] }
tracing-appender = "0.2"
toml = "0.8"
dirs = "5"
# p9-fb-19: in-process LRU cache for `App::search`. Capacity from
# `config.search.cache_capacity` (default 256, ~1.3 MB cap).
lru = { workspace = true }
# p9-fb-19: NFKC-normalize cache-key queries so `"Foo"` / `"FOO"` /
# `" foo "` collapse to one entry. Same crate kebab-normalize +
# kebab-core already use, no version drift.
unicode-normalization = "0.1"
# p9-fb-31: GitignoreBuilder for .kebabignore matching in ingest_file_with_config.
# Same version as kebab-source-fs (0.4) to avoid duplicate dep versions.
ignore = "0.4"
# p9-fb-34: opaque pagination cursor encodes payload as base64.
base64 = { workspace = true }
# Enhancement 3 (v0.20.x r2): direct SQL queries for inspect_ocr_stats/failures.
rusqlite = { workspace = true }
[dev-dependencies]
kebab-config = { path = "../kebab-config" }
# doc-side expansion (Phase 2) Task 4: ExpansionGenerator unit tests build
# MockLanguageModel (gated behind kebab-llm's `mock` feature, default OFF in
# [dependencies]). Enabling it here turns it on for the test build only.
kebab-llm = { path = "../kebab-llm", features = ["mock"] }
rusqlite = { workspace = true }
filetime = "0.2"
tempfile = { workspace = true }
# Image-pipeline integration tests use wiremock to stub Ollama for OCR
# / caption HTTP calls. Async runtime to host the mock server only;
# the kb-app code under test stays sync.
wiremock = { workspace = true }
tokio = { workspace = true, features = ["rt-multi-thread"] }
image = { version = "0.25", default-features = false, features = ["png", "jpeg"] }
# P7-3 PDF integration tests build in-memory PDF fixtures via the same
# lopdf builder pattern `kebab-parse-pdf::tests::common` uses; pinned
# to the same major (0.32) so byte output is identical between the two
# fixture surfaces.
lopdf = { workspace = true }
# error_wire::tests::llm_unreachable_classifies_to_model_unreachable needs a real
# reqwest::Error (private constructor) — built from a connect-refused call.
reqwest = { version = "0.12", default-features = false, features = ["blocking", "rustls-tls"] }
[features]
# Marker feature — spec §6.3 Option A (단순): lindera 는 kebab-chunk 가 default dep 으로 소유.
# disable path 없음; 이 feature 는 spec §6.3 명시를 honor 하는 role 만.
default = ["fts_korean_morphological"]
fts_korean_morphological = []
[lints]
workspace = true