القائمة
Use when combining vector (semantic) and full-text (lexical) search results into one ranking in a Go app with gosqlite — Reciprocal Rank Fusion via the fusion package.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
Use when the task needs a SQL function or virtual table from the gosqlite ext/ catalog — regexp, uuid, hash, ipaddr, zorder, stats, unicode, encode, text, fuzzy, decimal, money, time, eval, csv, lines, statement, closure, pivot, rtree, series, array, bloom, spellfix1, fileio, blobio.
Use when debugging surprising behaviour with gosqlite, or as a pre-ship checklist — the gotchas around per-conn hooks, the connection pool, vec quirks, bind limits, and what is not supported.
Use when the task needs encryption at rest, storing a database compressed at rest, page-checksum corruption detection, an in-memory database with isolation, or serving a database from an embed.FS / byte buffer in a Go app with gosqlite.
Use when using gorm.io/gorm with gosqlite — the gosqlite.org/gorm dialector, a CGo-free drop-in for glebarez/sqlite and gorm.io/driver/sqlite, shipped as its own module.
Use when switching an existing Go project to gosqlite from mattn/go-sqlite3, modernc.org/sqlite, glebarez/sqlite, gorm.io/driver/sqlite, or ncruces/go-sqlite3.
Use when opening a SQLite database in Go with gosqlite.org, choosing the "sqlite" vs "sqlite3" driver name, or deciding which sub-package to reach for. The starting point for any task using this driver.
استنادا إلى تصنيف SOC المهني
| name | hybrid-search |
| description | Use when combining vector (semantic) and full-text (lexical) search results into one ranking in a Go app with gosqlite — Reciprocal Rank Fusion via the fusion package. |
The fusion package merges two ranked result sets via Reciprocal Rank Fusion (Cormack 2009). Pure Go, no SQLite dependency.
import "gosqlite.org/fusion"
vecHits, _ := tbl.KNNSlice(ctx, queryVec, 50) // see vector-search
ftsHits, _ := idx.SearchSlice(ctx, fts.Term("brown fox"), fts.WithLimit(50)) // see full-text-search
vecKeys := make([]int64, len(vecHits))
for i, h := range vecHits { vecKeys[i] = h.Rowid }
ftsKeys := make([]int64, len(ftsHits))
for i, h := range ftsHits { ftsKeys[i] = h.Key }
top, _ := fusion.RRF([][]int64{vecKeys, ftsKeys}, fusion.WithLimit(20))
for _, r := range top { fmt.Println(r.Key, r.Score) }
fusion.RRF([][]K{...}, opts...) — fuse any number of ranked key lists.fusion.RRF2(a, b) — convenience for exactly two lists.int64 rowid or string).Run the two searches with a generous limit (50–100) each, fuse, then take the top N. Prerequisites: the vector-search and full-text-search skills. Full reference: docs/guides/hybrid-search.md.