Diff: services/service-search.es
From d0b5b58 to d0b5b58
+0 / −0 lines
| Before | After |
|---|---|
| --- | --- |
| schema: foundry-doc-v1 | schema: foundry-doc-v1 |
| title: "service-search: Inverted Index" | title: "service-search: Inverted Index" |
| slug: service-search | slug: service-search |
| category: services | category: services |
| type: topic | type: topic |
| quality: complete | quality: complete |
| short_description: "service-search is the Ring 2 full-text search service built on the Tantivy Rust library, providing microsecond retrieval across millions of files through a static binary inverted index that requires no active database process." | short_description: "service-search is the Ring 2 full-text search service built on the Tantivy Rust library, providing microsecond retrieval across millions of files through a static binary inverted index that requires no active database process." |
| status: active | status: active |
| bcsc_class: public-disclosure-safe | bcsc_class: public-disclosure-safe |
| last_edited: 2026-05-08 | last_edited: 2026-05-08 |
| editor: pointsav-engineering | editor: pointsav-engineering |
| cites: [] | cites: [] |
| paired_with: service-search.es.md | paired_with: service-search.es.md |
| --- | --- |
| **service-search** answers full-text queries across millions of platform documents in microseconds, using a static binary inverted index built in Rust on the Tantivy library — and because the index is a file rather than a live database process, it can be copied to portable media and queried on any machine without additional dependencies. The service is a Ring 2 knowledge-and-processing component and conforms to the Data Archive and Retrieval Protocol (DARP) standard. It locates documents; it does not generate or classify them. | **service-search** answers full-text queries across millions of platform documents in microseconds, using a static binary inverted index built in Rust on the Tantivy library — and because the index is a file rather than a live database process, it can be copied to portable media and queried on any machine without additional dependencies. The service is a Ring 2 knowledge-and-processing component and conforms to the Data Archive and Retrieval Protocol (DARP) standard. It locates documents; it does not generate or classify them. |
| ## Architectural Baseline | ## Architectural Baseline |
| An inverted index works by building a compressed map from every word in the corpus to the list of documents that contain it — analogous to the index at the back of a reference book. At query time, the service looks up the query terms in this map and returns matching documents in microseconds, regardless of corpus size. Tantivy, the underlying Rust library, is designed for high-throughput indexing and low-latency retrieval on commodity hardware. | An inverted index works by building a compressed map from every word in the corpus to the list of documents that contain it — analogous to the index at the back of a reference book. At query time, the service looks up the query terms in this map and returns matching documents in microseconds, regardless of corpus size. Tantivy, the underlying Rust library, is designed for high-throughput indexing and low-latency retrieval on commodity hardware. |
| ## Ring and Role | ## Ring and Role |
| service-search occupies **Ring 2 — Knowledge and Processing** in the three-ring architecture. Ring 2 is multi-tenant via `moduleId` namespacing and operates deterministically without AI inference. service-search's role within Ring 2 is retrieval: it answers queries against the indexed corpus and returns ranked document references that Ring 2 or Ring 3 services use for downstream processing. The service does not generate content or classify documents — it locates them. | service-search occupies **Ring 2 — Knowledge and Processing** in the three-ring architecture. Ring 2 is multi-tenant via `moduleId` namespacing and operates deterministically without AI inference. service-search's role within Ring 2 is retrieval: it answers queries against the indexed corpus and returns ranked document references that Ring 2 or Ring 3 services use for downstream processing. The service does not generate content or classify documents — it locates them. |
| ## Structural Organization of Components | ## Structural Organization of Components |
| The index is built as a static binary file. Key architectural properties: | The index is built as a static binary file. Key architectural properties: |
| - **No active process required for queries.** The index is memory-mapped at query time; there is no database daemon to manage or restart. | - **No active process required for queries.** The index is memory-mapped at query time; there is no database daemon to manage or restart. |
| - **Portable.** The index file can be copied to USB storage or a different machine and queried immediately. | - **Portable.** The index file can be copied to USB storage or a different machine and queried immediately. |
| - **Compressed.** Tantivy's index format uses block-maximal encoding for term frequency data, keeping the index compact relative to corpus size. | - **Compressed.** Tantivy's index format uses block-maximal encoding for term frequency data, keeping the index compact relative to corpus size. |
| - **Updatable.** New documents are added to the index via a background indexing process that merges new segments. Queries can run against existing segments while new segments are being built. | - **Updatable.** New documents are added to the index via a background indexing process that merges new segments. Queries can run against existing segments while new segments are being built. |
| The service is integrated with `app-workplace-presentation` for interactive search in the workplace application and with the system contracts layer for programmatic retrieval. | The service is integrated with `app-workplace-presentation` for interactive search in the workplace application and with the system contracts layer for programmatic retrieval. |
| ## Configuration | ## Configuration |
| | Parameter | Purpose | | | Parameter | Purpose | |
| |---|---| | |---|---| |
| | Index path | Filesystem path for the Tantivy index directory | | | Index path | Filesystem path for the Tantivy index directory | |
| | Schema | Field definitions for indexed documents (title, body, date, category, moduleId) | | | Schema | Field definitions for indexed documents (title, body, date, category, moduleId) | |
| | Merge policy | Segment merge configuration controlling index compaction frequency | | | Merge policy | Segment merge configuration controlling index compaction frequency | |
| | Writer threads | Number of indexing threads for parallel document ingestion | | | Writer threads | Number of indexing threads for parallel document ingestion | |
| ## See Also | ## See Also |
| - [[service-extraction]] | - [[service-extraction]] |
| - [[service-slm]] | - [[service-slm]] |
| - [[service-people]] | - [[service-people]] |
| - [[trajectory-substrate]] | - [[trajectory-substrate]] |
| ## References | ## References |
| - §XI — Ring 2 knowledge-and-processing architecture | - §XI — Ring 2 knowledge-and-processing architecture |
| - `pointsav-monorepo/service-search/` — implementation crate | - `pointsav-monorepo/service-search/` — implementation crate |
| - Tantivy documentation — <https://docs.rs/tantivy/> | - Tantivy documentation — <https://docs.rs/tantivy/> |
| - DARP (Data Archive and Retrieval Protocol) — compliance requirement governing the index portability guarantee | - DARP (Data Archive and Retrieval Protocol) — compliance requirement governing the index portability guarantee |