菜单
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.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
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.
Use when storing large, growable, or randomly-writable byte objects (files, uploads, streamed content) in SQLite with gosqlite — the blobstore package, plus when to reach for the fixed-size Conn.OpenBlob / ext/blobio instead.
| name | extensions |
| description | 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. |
ext/)Each extension is an independent sub-package. Two registration shapes:
// Pool-wide: blank-import the /auto variant (every conn picks it up).
import _ "gosqlite.org/ext/regexp/auto"
// Per-conn: register explicitly.
import "gosqlite.org/ext/regexp"
regexp.Register(conn) // conn is *sqlite.Conn
After registration, just use the SQL:
db.Where("email REGEXP ?", `^.*@example\.com$`)
db.Select(`hex(sha256(body)) AS digest`)
Scalars / aggregates / collations (blank-import /auto, use in any SQL): regexp (RE2 + regexp_*), uuid (v1–v7 + uuid_str/uuid_blob), hash (md5/sha*/blake2b/blake3/xxh64), ipaddr (ipcontains/ipfamily), zorder (Morton), stats (variance/percentile/regr_*/median/mode), unicode (upper/lower/normalize/unaccent + NOCASE_UNICODE), encode (base64/32/16/ascii85/url), text (text_reverse/lpad/split/…), fuzzy (levenshtein/jaro_winkler/soundex/caverphone), decimal (exact base-10 + decimal_sum), money (fixed-2dp + money_format), time (time_unix/add/diff/part), eval (dynamic SQL — trusted input only).
Virtual tables (CREATE VIRTUAL TABLE … USING module(…)): array (Go slice via sqlite.Pointer), csv, lines, statement, closure (graph), pivot, rtree (spatial), series (generate_series).
Stores / I/O: bloom (persistent Bloom filter), spellfix1 (fuzzy vocabulary), blobio (incremental BLOB I/O), fileio (sandboxed file functions).
csv.Table, lines.Table, closure.Graph, bloom.Filter, spellfix1.Vocab, rtree.Table) that hides the DDL — prefer it for new code.csv / lines / fileio need an fs.FS sandbox at registration, which /auto can't carry — register via RegisterFS(conn, fsys) or a Driver.RegisterConnectionHook. (blobio is not filesystem-bound: register it with plain blobio.Register(conn) — it does incremental BLOB I/O over columns.)Conn.RegisterFunc / RegisterAggregator / RegisterWindowFunction (no ext package needed).Conn.CreateModule / CreateEponymousModule (implement VTab / VTabCursor, call DeclareVTab in the constructor). For an efficient module, the planner hooks are VTabDistinct(info) — from BestIndex, how far the query relaxes ordering/duplication so you can skip work — and VTabFunctionFinder (implement FindFunction on your VTab) to override a SQL function applied to your columns — declare the name with Conn.OverloadFunction(name, nArg) so it prepares (e.g. a MATCH operator). Worked examples: vtab-planning (VTabDistinct), vtab-overload (FindFunction + OverloadFunction).Full catalog + status: docs/extensions/index.md, dev/coverage/ext.md.