TextService

Full-text search operations backed by tantivy (BM25 scoring, 30+ languages).

Proto source

query/text.proto

Transport

All methods are available via:

• gRPC on port 7080 — native high-performance API
• REST/JSON on port 7081 — HTTP/JSON transcoding via embedded structured-proxy

Methods

TextSearch

Search nodes matching a full-text query within a label. Searches all text-indexed properties for the label and returns BM25-scored results with optional HTML-highlighted snippets. If no text index is registered for the label, returns an empty result set (no error). Query syntax: boolean (AND, OR, NOT), phrase ("exact phrase"), prefix wildcard (term*), per-term boosting (term^N).

HTTP: POST /v1/query/text/search

Request: TextSearchRequest

Field	Type	Description
label	string	Label to search within (e.g., "Article", "Document"). Must have at least one text index registered; returns empty results otherwise.
query	string	Full-text query string. Supports boolean operators (AND, OR, NOT), phrase search ("exact phrase"), prefix wildcards (term*), and per-term boosting (term^N). Examples: "rust AND graph" "machine learning" "data*" "name^3 OR description^1"
limit	uint32	Maximum number of results to return. 0 = default (10). Capped at 1000.
fuzzy	bool	If true, enable Levenshtein-1 fuzzy matching for individual query terms. Increases recall at the cost of precision. Not applied to quoted phrases or terms that already contain operators.
language	string	Language for query tokenization and stemming (e.g., "english", "russian", "chinese_jieba", "japanese_lindera", "none"). Empty string = use the index's default language (typically "english" unless overridden at index creation time).

Response: TextSearchResponse

Field	Type	Description
results	TextResult[]	Results ordered by BM25 score descending (most relevant first).

Types

TextResult

A single text search result with BM25 score and optional highlighted snippet.

Field	Type	Description
node_id	uint64	Node ID of the matching node.
score	float	BM25 relevance score. Higher = more relevant to the query. Not normalized; values depend on corpus size and per-field weights.
snippet	string	HTML-tagged snippet of matching text with <b> tags around matched terms. Derived from the indexed field text using tantivy's SnippetGenerator. Empty when language-specific search is used (highlight requires default language tokenization) or when the indexed field has no stored content. Strip <b> and </b> tags for plain-text display.

TextSearchRequest

Request to search indexed text properties across a node label.

Field	Type	Description
label	string	Label to search within (e.g., "Article", "Document"). Must have at least one text index registered; returns empty results otherwise.
query	string	Full-text query string. Supports boolean operators (AND, OR, NOT), phrase search ("exact phrase"), prefix wildcards (term*), and per-term boosting (term^N). Examples: "rust AND graph" "machine learning" "data*" "name^3 OR description^1"
limit	uint32	Maximum number of results to return. 0 = default (10). Capped at 1000.
fuzzy	bool	If true, enable Levenshtein-1 fuzzy matching for individual query terms. Increases recall at the cost of precision. Not applied to quoted phrases or terms that already contain operators.
language	string	Language for query tokenization and stemming (e.g., "english", "russian", "chinese_jieba", "japanese_lindera", "none"). Empty string = use the index's default language (typically "english" unless overridden at index creation time).

TextSearchResponse

Response containing BM25-scored text search results.

Field	Type	Description
results	TextResult[]	Results ordered by BM25 score descending (most relevant first).