Files
kebab/crates/kebab-chunk/src/tier2_shared.rs
altair823 b134ae9dd5 feat(chunk): integrate lindera korean morphological tokenizer
V009 의 tokenized_korean_text column 에 들어갈 morpheme sequence
를 lindera ko-dic 으로 분해. chunk builder pipeline 의 chunk 생성
직후 시점에서 호출 → chunk struct 의 field 에 pre-fill → store
의 put_chunks 가 단일 transaction 안에서 INSERT.

- crates/kebab-core/src/chunk.rs: Chunk struct 에
  tokenized_korean_text: Option<String> field 추가 (#[serde(default)]).
- crates/kebab-chunk/src/lib.rs: tokenize_korean_morphological()
  helper + OnceLock 캐싱 + fallback (None) 정책.
- crates/kebab-chunk/Cargo.toml: lindera features = ["embed-ko-dic"]
  추가 (DictionaryKind::KoDic 활성화에 필요).
- 모든 chunker (tier2_shared, md_heading_v1, pdf_page_v1, 9개
  code AST v1): Chunk 리터럴에 tokenized_korean_text pre-fill.
- crates/kebab-store-sqlite/src/documents.rs::put_chunks: INSERT
  SQL column list + placeholder + binding 갱신 (12번째 column).
- crates/kebab-chunk/tests/tokenize_korean.rs: 단위 테스트 2개.

lindera 3.0.7 API 정정: load_dictionary_from_kind →
load_embedded_dictionary, Token.text → Token.surface.

Spec: docs/superpowers/specs/2026-05-28-v0.20.x-korean-morphological-tokenizer-spec.md §6.2
Plan: docs/superpowers/plans/2026-05-28-v0.20.x-korean-morphological-tokenizer-plan.md (S3)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-28 10:22:15 +00:00

201 lines
6.4 KiB
Rust

//! p10-2: Tier 2 chunker shared helpers (oversize fallback + Chunk build).
//!
//! Mirrors `code_rust_ast_v1`'s Chunk-construction pattern exactly so that
//! id / hashes / token-count / ChunkPolicy semantics stay identical across
//! Tier 1 (AST) and Tier 2 (resource-aware) chunkers.
use anyhow::Result;
use kebab_core::{
BlockId, CanonicalDocument, Chunk, ChunkPolicy, ChunkerVersion, DocumentId, SourceSpan,
id_for_chunk,
};
pub(crate) const AST_CHUNK_MAX_LINES: u32 = 200;
const BYTES_PER_TOKEN: usize = 3;
const POLICY_HASH_HEX_LEN: usize = 16;
/// Compute the policy hash the same way `code_rust_ast_v1` does.
pub(crate) fn policy_hash(policy: &ChunkPolicy) -> String {
let bytes = serde_json_canonicalizer::to_vec(policy)
.expect("canonical JSON serialization of ChunkPolicy must not fail");
let hex = blake3::hash(&bytes).to_hex().to_string();
hex[..POLICY_HASH_HEX_LEN].to_string()
}
/// Emit one chunk for `(text, line_start..=line_end, symbol, lang)`, splitting
/// into line-windows of at most `AST_CHUNK_MAX_LINES` if the slice is oversize.
/// Mirrors the oversize path in `code_rust_ast_v1`'s `chunk` impl.
///
/// `base_split_key` is used as the `split_key` for the non-oversize single-chunk
/// case. Callers that emit multiple chunks from the same document (e.g.
/// `K8sManifestResourceV1Chunker` — one call per k8s resource) MUST pass
/// `Some(line_start)` so that each call produces a distinct `chunk_id`.
/// Single-chunk callers (dockerfile-file-v1, manifest-file-v1) pass `None` to
/// keep chunk_ids stable (no sibling can collide when there's only one chunk).
#[allow(clippy::too_many_arguments)]
pub(crate) fn push_chunks_with_oversize(
out: &mut Vec<Chunk>,
doc: &CanonicalDocument,
policy: &ChunkPolicy,
text: &str,
line_start: u32,
line_end: u32,
symbol: &str,
lang: &str,
chunker_version: &str,
base_split_key: Option<u32>,
) -> Result<()> {
let n_lines = (line_end - line_start + 1).max(1);
let cv = ChunkerVersion(chunker_version.to_string());
let base_policy_hash = policy_hash(policy);
if n_lines <= AST_CHUNK_MAX_LINES {
out.push(build_chunk(
doc,
&cv,
&base_policy_hash,
text,
line_start,
line_end,
symbol,
lang,
base_split_key,
));
return Ok(());
}
let lines: Vec<&str> = text.lines().collect();
let total = lines.len();
let mut window_start = line_start;
let mut i = 0usize;
while i < total {
let take = (AST_CHUNK_MAX_LINES as usize).min(total - i);
let window_text = lines[i..i + take].join("\n");
let window_end = window_start + take as u32 - 1;
out.push(build_chunk(
doc,
&cv,
&base_policy_hash,
&window_text,
window_start,
window_end,
symbol,
lang,
Some(window_start),
));
i += take;
window_start = window_end + 1;
}
Ok(())
}
/// Build a single `Chunk`, mirroring `make_chunk` in `code_rust_ast_v1.rs`
/// exactly (same id recipe, same token estimate, same field set).
///
/// `split_key` is `Some(line_start_of_window)` for oversize splits, `None`
/// for normal single-chunk emission. Mirrors the `Some(part_ls)` / `None`
/// split_key pattern in 1A-2.
#[allow(clippy::too_many_arguments)]
pub(crate) fn build_chunk(
doc: &CanonicalDocument,
chunker_version: &ChunkerVersion,
base_policy_hash: &str,
text: &str,
line_start: u32,
line_end: u32,
symbol: &str,
lang: &str,
split_key: Option<u32>,
) -> Chunk {
let span = SourceSpan::Code {
line_start,
line_end,
symbol: Some(symbol.to_string()),
lang: Some(lang.to_string()),
};
build_chunk_from_span(
doc,
chunker_version,
base_policy_hash,
text,
span,
split_key,
)
}
/// Like `build_chunk` but emits `symbol: None`. Used by Tier 3 (per spec §9.3).
///
/// Accepts `policy: &ChunkPolicy` and `chunker_version: &str` (string slice)
/// so callers don't need to pre-compute the hash and version wrapper.
/// `split_key` is `Some(window_start)` for oversize line-window splits.
#[allow(clippy::too_many_arguments)]
pub(crate) fn build_chunk_no_symbol(
doc: &CanonicalDocument,
policy: &ChunkPolicy,
text: &str,
line_start: u32,
line_end: u32,
lang: &str,
chunker_version: &str,
split_key: Option<u32>,
) -> Chunk {
let cv = ChunkerVersion(chunker_version.to_string());
let base_policy_hash = policy_hash(policy);
let span = SourceSpan::Code {
line_start,
line_end,
symbol: None,
lang: Some(lang.to_string()),
};
build_chunk_from_span(doc, &cv, &base_policy_hash, text, span, split_key)
}
/// Core chunk-building logic shared by `build_chunk` and `build_chunk_no_symbol`.
///
/// Takes a pre-built `SourceSpan` so the only difference between the two
/// public helpers is whether `symbol` is `Some` or `None`. All id/hash/
/// token mechanics are identical.
fn build_chunk_from_span(
doc: &CanonicalDocument,
chunker_version: &ChunkerVersion,
base_policy_hash: &str,
text: &str,
span: SourceSpan,
split_key: Option<u32>,
) -> Chunk {
// id_hash mirrors code_rust_ast_v1's make_chunk logic:
// split_key Some(k) => "{base_policy_hash}#L{k}"
// split_key None => base_policy_hash
let id_hash = match split_key {
Some(k) => format!("{base_policy_hash}#L{k}"),
None => base_policy_hash.to_string(),
};
// block_ids: Tier 2/3 chunkers have no per-block structure (the whole file
// is one Block::Code), so we pass an empty slice — same as using the doc-
// level slice without explicit block granularity.
let block_ids: Vec<BlockId> = vec![];
let chunk_id = id_for_chunk(
&DocumentId(doc.doc_id.0.clone()),
chunker_version,
&block_ids,
&id_hash,
);
let token_estimate = text.len().div_ceil(BYTES_PER_TOKEN);
Chunk {
chunk_id,
doc_id: DocumentId(doc.doc_id.0.clone()),
block_ids,
tokenized_korean_text: crate::tokenize_korean_morphological(text),
text: text.to_string(),
heading_path: Vec::new(),
source_spans: vec![span],
token_estimate,
chunker_version: chunker_version.clone(),
policy_hash: base_policy_hash.to_string(),
}
}