dirsql 0.4.26 → 0.4.28

package/docs/howto/define-tables.md

@@ -107,7 +107,9 @@ maintain nothing. `name` still has to be the row table (`posts`); the virtual
 table lives beside it under its own name. The batch runs once, when the table
 is created, and editing any part of it rebuilds a
 [persistent cache](./persist.md) from scratch. Full rules:
-[Batch `ddl`](../reference/config.md#batch-ddl).
+[Batch `ddl`](../reference/config.md#batch-ddl); pasteable FTS5 and vector
+templates, with the trigger mistakes that fail silently, are in
+[Add a search index to a table](./search-indexes.md).
 
 ## Going further
 

package/docs/howto/query-without-config.md

@@ -55,7 +55,9 @@ question over a few hundred files, wasteful for a large tree you query
 repeatedly. When that day comes, [declare a table](./define-tables.md): it is
 indexed once, kept fresh by the watcher, and can persist across restarts. The
 zero-config query is the floor; a named table is the escalation path — nothing
-you write here has to be thrown away to get there.
+you write here has to be thrown away to get there. The two sides in full, and
+why indexes belong to only one of them, are in
+[how `dirsql` thinks](../explanation.md#two-table-kinds-opposite-trade-offs).
 
 ## Notes
 

package/docs/howto/search-indexes.md

@@ -0,0 +1,220 @@
+# Add a search index to a table
+
+Make a declared table answer search queries instead of scanning: a B-tree for
+lookups, an FTS5 index for keywords, a `vec0` index for meaning. All three are
+statements in the table's [`ddl` batch](../reference/config.md#batch-ddl), so
+SQLite does the work and `dirsql` maintains nothing.
+
+The templates below are meant to be pasted and edited. Both are shaped around
+one worked example — notes in `notes/*.md`, parsed into a `notes` table by an
+`extract.py` that prints `{"slug": …, "title": …, "body": …}` per file, exactly
+as in [Define tables for your files](./define-tables.md).
+
+## Why triggers, and why only two
+
+`dirsql` writes file rows with plain `INSERT` and `DELETE`; an update is a
+delete and an insert in one transaction, and there is no `UPDATE` path on user
+rows. So an insert trigger and a delete trigger cover every way a row can
+change — during the initial build, on each [watcher](./react-to-changes.md)
+event, and when a [persistent cache](./persist.md) reconciles a tree that
+moved on. There is no third trigger to write and no maintenance command to run.
+
+The triggers are also the part that goes wrong, which is why the templates
+spell them out rather than hiding them behind an abstraction.
+
+## Keyword search (FTS5)
+
+FTS5 ships inside SQLite — no extension, no install. Declare an
+external-content index beside the row table and let the two triggers feed it:
+
+```toml
+[[table]]
+name    = "notes"
+glob    = "notes/**/*.md"
+on-file = "python3 extract.py {path}"
+ddl     = '''
+CREATE TABLE notes (slug TEXT, title TEXT, body TEXT);
+
+CREATE VIRTUAL TABLE notes_fts USING fts5(
+  body,
+  content='notes',
+  content_rowid='rowid',
+  tokenize='porter unicode61'
+);
+CREATE TRIGGER notes_ai AFTER INSERT ON notes BEGIN
+  INSERT INTO notes_fts(rowid, body) VALUES (new.rowid, new.body);
+END;
+CREATE TRIGGER notes_ad AFTER DELETE ON notes BEGIN
+  INSERT INTO notes_fts(notes_fts, rowid, body)
+    VALUES ('delete', old.rowid, old.body);
+END;
+'''
+```
+
+`content='notes'` makes the index store only its own search structures and
+read column values back from `notes`, so the text is not duplicated.
+`tokenize='porter unicode61'` stems English on top of the default
+case-folding, so *deploying* matches *deploy*; drop the `porter` half to match
+whole words only.
+
+Query the index and join back to the row table on `rowid`:
+
+```bash
+dirsql query "
+SELECT n.slug,
+       bm25(notes_fts) AS score,
+       snippet(notes_fts, 0, '[', ']', '…', 8) AS excerpt
+FROM notes_fts JOIN notes AS n ON n.rowid = notes_fts.rowid
+WHERE notes_fts MATCH 'deploy'
+ORDER BY score" -c ./.dirsql.toml
+```
+
+```json
+[{"excerpt":"Rebase before you [deploy]. Keep pull requests small.","score":-1.0476190476190478e-6,"slug":"branches"},
+ {"excerpt":"…the spaghetti goes in. [Deploy] the garlic late.","score":-8.461538461538463e-7,"slug":"pasta"}]
+```
+
+**`bm25()` returns negative scores, and better matches are more negative** —
+so `ORDER BY score` ascending is best-first, with no `DESC`. `snippet()`'s
+arguments are the table, the column index, the open and close markers, the
+ellipsis, and the token budget.
+
+### Check the delete trigger
+
+A wrong or missing delete trigger fails **silently**: the index keeps rows for
+files that are gone, and they keep matching. Nothing errors.
+
+The symptom is a hit that has no row behind it, which a `LEFT JOIN` exposes:
+
+```bash
+dirsql query "
+SELECT n.slug
+FROM notes_fts LEFT JOIN notes AS n ON n.rowid = notes_fts.rowid
+WHERE notes_fts MATCH 'frost'" -c ./.dirsql.toml --persist
+```
+
+```json
+[{"slug":null}]
+```
+
+A `null` slug is a stale index entry — the base row is gone and the index did
+not hear about it. With the `notes_ad` trigger above, the deleted file's entry
+is gone too and the query returns nothing.
+
+Note the `--persist` flag: without it every run rebuilds the index from
+scratch, so a broken delete trigger cannot show itself. Deletes are visible
+against a [warm cache](./persist.md) and through the watcher, which is exactly
+where the staleness would have bitten.
+
+## Vector search (`vec0`)
+
+Two pieces beyond SQLite: the
+[`sqlite-vec`](https://github.com/asg017/sqlite-vec) extension supplies the
+`vec0` virtual table, and
+[`dirsql-plugin-embeddings`](../plugins.md#dirsql-plugin-embeddings) supplies
+`embed()`. Installing the plugin brings both — its config fragment declares
+the extension too, and the launcher
+[discovers it](../reference/cli.md#plugins):
+
+```bash
+uvx --with dirsql-plugin-embeddings dirsql query "…" -c ./.dirsql.toml
+```
+
+Loading `sqlite-vec` by hand instead (another runtime, a pinned build) is
+[Load a SQLite extension](./load-extension.md).
+
+### Find your model's dimension
+
+A `vec0` column declares a fixed width, so the template needs the number of
+components your model emits. `embed()` returns the vector as JSON text, so ask
+it:
+
+```bash
+uvx --with dirsql-plugin-embeddings dirsql query \
+  "SELECT json_array_length(embed('probe', 'minishlab/potion-retrieval-32M')) AS dims"
+```
+
+Run it once per model id you intend to use, and substitute the answer for the
+`512` in the template below. (The first call for a model downloads it — see
+[model](../plugins.md#model).) Getting it wrong is at least loud — the build fails naming both
+numbers:
+
+```
+dirsql query: failed to load config: SQLite error: Dimension mismatch for
+inserted vector for the "embedding" column. Expected 8 dimensions but received 4.
+```
+
+### The template
+
+```toml
+[[table]]
+name    = "notes"
+glob    = "notes/**/*.md"
+on-file = "python3 extract.py {path}"
+ddl     = '''
+CREATE TABLE notes (slug TEXT, title TEXT, body TEXT);
+
+-- Width must equal what the probe printed for the model id named below.
+CREATE VIRTUAL TABLE notes_vec USING vec0(embedding float[512]);
+CREATE TRIGGER notes_vi AFTER INSERT ON notes BEGIN
+  INSERT INTO notes_vec(rowid, embedding)
+    VALUES (new.rowid, embed(new.body, 'minishlab/potion-retrieval-32M'));
+END;
+CREATE TRIGGER notes_vd AFTER DELETE ON notes BEGIN
+  DELETE FROM notes_vec WHERE rowid = old.rowid;
+END;
+'''
+```
+
+The delete side is an ordinary `DELETE`, not FTS5's `'delete'` command row —
+`vec0` is a normal-looking table that way.
+
+Query it with `sqlite-vec`'s KNN form, joined back on `rowid`:
+
+```bash
+uvx --with dirsql-plugin-embeddings dirsql query "
+SELECT n.slug, v.distance
+FROM notes_vec AS v JOIN notes AS n ON n.rowid = v.rowid
+WHERE v.embedding MATCH embed('how do I cook pasta?', 'minishlab/potion-retrieval-32M')
+  AND k = 3
+ORDER BY v.distance" -c ./.dirsql.toml
+```
+
+`MATCH` plus `k = 3` is what makes this a top-k scan rather than a full one —
+`vec0` uses `k`, not `LIMIT`. `distance` is supplied by the virtual table,
+closest first.
+
+**Name the same model id in the trigger and the query.** Nothing checks that
+the two agree, and vectors from different models are not comparable. If the
+models happen to share a width the mismatch is entirely silent — rankings just
+get worse. Spelling the id out in both places, rather than leaning on the
+default in either, is the cheap defence; it also puts the model in the
+[config hash](../reference/config.md#batch-ddl), so changing it rebuilds the
+index instead of mixing old vectors with new.
+
+Rows are embedded once, at ingest, and cached on disk by content and model
+([vector cache](../plugins.md#vector-cache)). A search then costs one
+`embed()` call for the query text plus an in-process scan.
+
+::: warning Editing `ddl` wedges a persisted `vec0` cache
+Under `--persist`, editing any part of a `ddl` batch whose cache holds a
+`vec0` table currently fails with `SQLite error: no such module: vec0`, and
+keeps failing until the cache is deleted (`rm -rf <root>/.dirsql`). Tracked in
+[#1008](https://github.com/thekevinscott/dirsql/issues/1008); FTS5 is
+unaffected.
+:::
+
+## What a rebuild does
+
+`ddl` runs once, when the table is created. Editing any character of it
+changes the config hash, which drops a persisted cache and re-ingests every
+file — so a new index, a different tokenizer or a changed model id all rebuild
+from scratch rather than leaving a half-migrated index behind. The full rules
+are in [Batch `ddl`](../reference/config.md#batch-ddl).
+
+## Going further
+
+- Ranked semantic search over files with no config at all —
+  [Search documents by meaning](./search-by-meaning.md).
+- Why indexes belong to declared tables and not to path-tables —
+  [how `dirsql` thinks](../explanation.md).

package/docs/reference/cli.md

@@ -9,9 +9,9 @@ The `dirsql` binary has these modes:
 | `dirsql query "<sql>"` | Explicit synonym for the default one-shot query. |
 | `dirsql server` | Start a long-lived HTTP server exposing a SQL view of a directory. See [HTTP API](./http-api.md). |
 | `dirsql init` | Generate a `.dirsql.toml`. |
+| `dirsql` (bare) | Open a [REPL](#the-repl) over the current directory, reading statements until EOF. |
 
-Bare `dirsql` with no SQL is a usage error pointing at `dirsql server` — it
-does **not** start the server.
+Bare `dirsql` does **not** start the server — that is `dirsql server`.
 
 ## Installation
 
@@ -47,6 +47,186 @@ dirsql "SELECT basename, size FROM './' ORDER BY size DESC LIMIT 5"
 pipeline, same flags, same output. See that section for config discovery,
 `--persist`, `--on-file`, hooks, and exit codes.
 
+## The REPL
+
+`dirsql` with no subcommand and no SQL reads statements until EOF:
+
+```bash
+dirsql
+# dirsql 0.2.7 — this directory is a database.
+#
+#   SELECT basename, size FROM './' ORDER BY size DESC LIMIT 5;
+#   SELECT path FROM './**/*.md' WHERE content LIKE '%TODO%';
+#
+# `exit`, `quit`, or Ctrl-D to leave.
+#
+# dirsql> SELECT count(*) AS files FROM './';
+# files
+# -----
+# 128
+#
+# 1 row
+# dirsql>
+```
+
+Statements go through the same pipeline as [`dirsql query`](#dirsql-query) and
+`POST /query`, so a statement typed at the prompt and one passed on the command
+line return identical rows. Every config flag the default mode takes — `-c`,
+`--persist`, `--no-ignore`, `--on-file` — applies unchanged:
+
+```bash
+dirsql -c .dirsql.toml --persist
+```
+
+The index is built **once**, before the first prompt: statements share one scan
+rather than re-walking the directory each time, and the live watcher keeps it
+fresh between them. Files the scan had to skip are named on stderr once, up
+front.
+
+### Output format
+
+Rows go where they are useful: a **table** when stdout is a terminal, the
+**JSON array** when it is piped or redirected. `SELECT * FROM './'` in a
+5000-file tree should not put a 5000-element JSON array in front of a person,
+and `dirsql "…" | jq` should not have to parse a table.
+
+`--format` overrides that, in both directions, and is valid in the REPL and in
+[`dirsql query`](#dirsql-query) alike:
+
+| Value | Renders |
+|---|---|
+| `auto` (default) | Table if stdout is a terminal, JSON otherwise. |
+| `table` | Always a table — including into a pipe or a file. |
+| `json` | Always the JSON array — including at a terminal. |
+
+```bash
+dirsql "SELECT basename, size FROM './' ORDER BY basename" --format table
+# basename  size
+# --------  ----
+# a.md      6
+# bb.md     10
+#
+# 2 rows
+```
+
+There is no `.mode`: dirsql has no dot-commands to extend (see
+[Leaving](#leaving)), and a flag serves the one-shot query too. `dirsql server`
+does not take `--format` — its transport is JSON over HTTP.
+
+**`auto` keys on stdout, not stdin.** `dirsql > rows.json` typed at a terminal
+is still headed for a file, and the file gets JSON.
+
+Table rendering is deliberately plain: aligned columns, a rule under the
+header, a row count, and `NULL` spelled out so it cannot be confused with an
+empty string. Two things happen to a value on its way into a cell, both
+because a `content` column holds a whole file: **newlines, tabs and other
+control characters are escaped** (`\n`, `\t`, `\u{…}`) so one row cannot span
+several lines, and **anything longer than 60 characters is truncated with
+`…`** so one column cannot set the width of every row. `--format json`
+returns the values unaltered.
+
+Laying the table out to the terminal's width, and paging a long result, are
+both out of scope; pipe to `less` for the latter.
+
+### Where a statement ends
+
+At its semicolon — the same rule `sqlite3` uses, and **SQLite's own tokenizer**
+decides where that semicolon is. So a statement can be laid out over as many
+lines as it needs, and a `;` inside a string literal, a comment, or a
+`BEGIN … END` body is not mistaken for the end of one:
+
+```
+dirsql> SELECT basename, size
+   ...> FROM './'
+   ...> ORDER BY size DESC
+   ...> LIMIT 5;
+```
+
+The `...>` prompt says the statement is still open. `exit`, `quit`, and a blank
+line are not SQL, so they are taken as typed rather than waiting for a
+terminator.
+
+### Editing and history
+
+The prompt is a full line editor ([reedline](https://github.com/nushell/reedline)),
+with the emacs bindings a shell prompt has:
+
+| Key | Does |
+|---|---|
+| ↑ / ↓ | Walk back and forth through history. |
+| Ctrl-R | Reverse-search history; type to narrow, Enter to accept. |
+| Ctrl-A / Ctrl-E | Jump to the start / end of the line. |
+| Alt-B / Alt-F | Move back / forward a word. |
+| Ctrl-W, Ctrl-K, Ctrl-Y | Kill the previous word, kill to end of line, yank it back. |
+| Ctrl-C | Abandon the line and return to a fresh prompt. **Does not exit.** |
+| Ctrl-D | Leave. |
+
+History is kept in one file for every directory — a query worked out in one
+project is worth recalling in the next, the same way `sqlite3` keeps a single
+`~/.sqlite_history`. It holds the last 1000 statements, at
+`$XDG_DATA_HOME/dirsql/history`, falling back to
+`~/.local/share/dirsql/history` (`%APPDATA%\dirsql\history` on Windows). If
+none of those resolve, history is kept in memory for the session only.
+
+### Terminal vs. pipe
+
+The prompt, banner, editor, and history exist only when **stdin is a terminal**.
+From a pipe or a redirect there is none of that, and the terminator rule does
+not apply either: a redirected script is not being typed, so there is no
+continuation prompt to hang it off. **One statement per line, no `;` needed:**
+
+```bash
+printf "SELECT 1 AS n\nSELECT 2 AS n\n" | dirsql
+# [{"n":1}]
+# [{"n":2}]
+
+dirsql < queries.sql > rows.jsonl
+```
+
+Blank lines do nothing in either mode.
+
+### Leaving
+
+`exit`, `quit` (either case), or Ctrl-D. There are no dot-commands: the `.`
+prefix exists in `sqlite3` to namespace meta-commands against SQL, and with no
+meta-commands there is nothing to namespace. Schema questions are ordinary SQL:
+
+```sql
+SELECT name FROM sqlite_master WHERE type = 'table';
+```
+
+### Errors
+
+A statement that fails prints its diagnostic — the same string the HTTP
+`{"error": …}` body carries — to stderr, and **the session continues**. This is
+the one behavioral difference from `dirsql query`, which exits `1` on the first
+failure:
+
+```
+dirsql> SELECT nope FROM missing;
+dirsql: SQLite error: no such table: missing
+dirsql> SELECT 1 AS n;
+n
+-
+1
+
+1 row
+```
+
+A config that cannot be loaded is different in kind: it fails identically for
+every statement, so it is reported once and exits `1` before the first prompt.
+
+### Exit codes
+
+| Code | Meaning |
+|---|---|
+| `0` | Clean EOF (Ctrl-D, `exit`, `quit`, or the end of a piped script) — **including when statements failed**. Matches interactive `sqlite3`; use [`dirsql query`](#dirsql-query) when a script needs a statement's exit status. |
+| `1` | The index could not be built (a bad `-c`, an unresolvable `--on-file`), or stdin could not be read. Nothing was executed. |
+
+`23` (partial scan) is not produced here: skipped files are reported before the
+first prompt, and a session's exit code describes the session rather than one
+scan.
+
 ## `dirsql server`
 
 ```bash
@@ -204,6 +384,12 @@ Errors print the same diagnostic the HTTP `{"error": …}` body carries —
 config failures, SQL errors, rejected reads, hook failures, timeouts — to
 stderr, with exit code `1`.
 
+#### `--format {auto,table,json}`
+
+How to render the result rows — the same flag [the REPL](#output-format)
+takes, with the same `auto` default. A one-shot query is usually piped, so
+`auto` usually means JSON; `--format table` is there for the times it is not.
+
 ### Exit codes
 
 | Code | Meaning |

package/docs/reference/path-tables.md

@@ -190,7 +190,13 @@ That is the right trade for a hundreds-of-files, run-it-once question. When the
 same tree is queried repeatedly, or is large, declare a
 [table](/reference/config) for it instead — a declared table is indexed on
 build, kept fresh by the watcher, and (with `--persist`) survives restarts, so
-its rows are read from SQLite rather than re-walked each time.
+its rows are read from SQLite rather than re-walked each time. It is also the
+only place a keyword or vector index can live
+([templates](/howto/search-indexes)).
+
+Neither kind is the better one: each buys what the other gives up, and the
+choice is stated as one trade in
+[how `dirsql` thinks](/explanation#two-table-kinds-opposite-trade-offs).
 
 ## Skip rules
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dirsql",
-  "version": "0.4.26",
+  "version": "0.4.28",
   "description": "Ephemeral SQL index over a local directory",
   "license": "MIT",
   "repository": "https://github.com/thekevinscott/dirsql",
@@ -213,10 +213,10 @@
     ]
   },
   "optionalDependencies": {
-    "@dirsql/lib-linux-x64-gnu": "0.4.26",
-    "@dirsql/lib-linux-arm64-gnu": "0.4.26",
-    "@dirsql/lib-darwin-x64": "0.4.26",
-    "@dirsql/lib-darwin-arm64": "0.4.26",
-    "@dirsql/lib-win32-x64-msvc": "0.4.26"
+    "@dirsql/lib-linux-x64-gnu": "0.4.28",
+    "@dirsql/lib-linux-arm64-gnu": "0.4.28",
+    "@dirsql/lib-darwin-x64": "0.4.28",
+    "@dirsql/lib-darwin-arm64": "0.4.28",
+    "@dirsql/lib-win32-x64-msvc": "0.4.28"
   }
 }