dirsql 0.3.124 → 0.3.126

package/README.md

@@ -89,7 +89,7 @@ Each event has `.action` (`'insert'` | `'update'` | `'delete'` | `'error'`), `.t
 
 ## CLI
 
-`npx dirsql` runs an HTTP server exposing the SDK over HTTP: `POST /query` for SQL and `GET /events` for a Server-Sent Events change stream. Requires **Node >= 20.11**. See the [CLI reference](https://thekevinscott.github.io/dirsql/reference/cli).
+`npx dirsql "<sql>"` runs one query and prints the rows as JSON — the default. `npx dirsql server` starts an HTTP server exposing the SDK over HTTP: `POST /query` for SQL and `GET /events` for a Server-Sent Events change stream. Requires **Node >= 20.11**. See the [CLI reference](https://thekevinscott.github.io/dirsql/reference/cli).
 
 ## License
 

package/docs/getting-started.md

@@ -1,11 +1,12 @@
 # Your first dirsql database
 
 In this tutorial you will turn a directory of three tiny markdown files into
-a SQL database you can query over HTTP — without writing any code. You will:
+a SQL database — with a single command, and without writing any code. You
+will:
 
 1. Create the directory and files.
-2. Start `dirsql` with zero configuration and query it with `curl`.
-3. Define your own table in a `.dirsql.toml` and query the new shape.
+2. Query them straight away with zero configuration.
+3. Declare your own table to name and reuse a shape, and query it.
 
 It takes about five minutes.
 
@@ -13,9 +14,9 @@ It takes about five minutes.
 them — so it is safe to point at a real directory of your own once you are
 done here. See [Read-only by design](./explanation#read-only-by-design).
 
-**You need:** a terminal with `curl` and [`jq`](https://jqlang.org/), and
-Node ≥ 20.11 (for `npx`). Every `npx dirsql` step below also has a `uvx`
-tab that behaves identically, if you prefer Python tooling
+**You need:** a terminal with [`jq`](https://jqlang.org/), and Node ≥ 20.11
+(for `npx`). Every `npx dirsql` step below also has a `uvx` tab that behaves
+identically, if you prefer Python tooling
 ([`uv`](https://docs.astral.sh/uv/)).
 
 ## 1. Create three files
@@ -59,61 +60,55 @@ notes/alice/welcome.md
 notes/bob/reading-list.md
 ```
 
-## 2. Start the server
+## 2. Query your files
 
-From inside `my-notes`, start `dirsql`:
+You wrote no configuration and no schema. Ask `dirsql` how many files are in
+this directory anyway — from inside `my-notes`, run one command:
 
 ::: code-group
 
 ```bash [npm]
-npx dirsql
+npx dirsql "SELECT COUNT(*) AS files FROM './'"
 ```
 
 ```bash [PyPI]
-uvx dirsql
+uvx dirsql "SELECT COUNT(*) AS files FROM './'"
 ```
 
 :::
 
 The first run downloads the package (`npx` asks for confirmation — answer
-`y`; `uvx` prints download progress), then the server starts:
+`y`; `uvx` prints download progress), then prints the result:
 
 ```
-Running at localhost:7117
+[{"files":3}]
 ```
 
-That one command scanned the directory, built an ephemeral SQLite database
-with one row per file, and started an HTTP server. Leave it running and
-open a **second terminal** for the next step.
+Three files, three rows. That one command scanned the directory, handed
+SQLite one row per file, ran your SQL, and printed the answer as JSON.
 
-## 3. Query your files
+There is no named table here — you never declared one. `'./'` is a
+[path-table](./reference/path-tables.md): a quoted path written where a table
+name goes. `'./'` means everything under the directory you ran the command
+in. The path *is* the query.
 
-You gave `dirsql` no configuration, so no named tables exist. Query the
-filesystem directly with a [path-table](./reference/path-tables.md) — a
-quoted path where a table name goes. `'./'` means everything under the
-index root. Ask it how many files there are:
+## 3. Select some columns
 
-```bash
-curl -s http://localhost:7117/query \
-  -H 'content-type: application/json' \
-  -d '{"sql":"SELECT COUNT(*) AS files FROM \'./\'"}'
-```
+The response is always a JSON array of row objects, so from here on we pipe
+it through `jq` to pretty-print. Ask for two columns instead of a count:
 
-```
-[{"files":3}]
-```
+::: code-group
 
-Three files, three rows. The response is always a JSON array of row
-objects ([HTTP API](./reference/http-api.md)) — from here on we pipe it
-through `jq` to pretty-print. Now select some columns:
+```bash [npm]
+npx dirsql query "SELECT path, size FROM './' ORDER BY path" | jq
+```
 
-```bash
-curl -s http://localhost:7117/query \
-  -H 'content-type: application/json' \
-  -d '{"sql":"SELECT path, size FROM \'./\' ORDER BY path"}' \
-  | jq
+```bash [PyPI]
+uvx dirsql query "SELECT path, size FROM './' ORDER BY path" | jq
 ```
 
+:::
+
 ```json
 [
   {
@@ -131,126 +126,123 @@ curl -s http://localhost:7117/query \
 ]
 ```
 
-`path` and `size` are two of the built-in file columns `dirsql` collects
-for every file — see [stat columns](./reference/columns.md#stat-columns)
-for the full list. (The `size` values are byte counts; they match the
-output above because you pasted the files exactly.)
+`path` and `size` are two of the built-in file columns `dirsql` collects for
+every file — see [stat columns](./reference/columns.md#stat-columns) for the
+full list. (The `size` values are byte counts; they match the output above
+because you pasted the files exactly.)
 
-You have a working SQL database over your files. Next, teach it the
-structure your folders already encode.
+You have a working SQL database over your files, and you never left the
+command line.
 
-## 4. Define a table
+## 4. Declare a table
 
-Look at the paths again: `notes/alice/ideas.md`, `notes/bob/reading-list.md`
-— the author's name is a directory segment. A config file can capture it as
-a real column.
+A declared table fixes a shape once: you give it a name, scope it to exactly
+the files you care about, and then query it by name instead of repeating a
+path in every question. It is also the on-ramp to everything a path-table
+can't do — a named table can be kept live by the watcher, persisted across
+restarts, and given a parser that reads inside your files.
 
-In your second terminal, still inside `my-notes`, create a `.dirsql.toml`:
+Still inside `my-notes`, create a `.dirsql.toml`:
 
 ```bash
 cat > .dirsql.toml <<'EOF'
 [[table]]
-ddl  = "CREATE TABLE notes (author TEXT, basename TEXT, size INTEGER)"
-glob = "notes/{author}/*.md"
+ddl  = "CREATE TABLE notes (dir TEXT, basename TEXT, size INTEGER)"
+glob = "notes/**/*.md"
 EOF
 ```
 
 Two keys define the table:
 
-- `glob` selects which files feed the table, and `{author}` is a
-  [glob capture](./reference/columns.md#glob-captures): whatever directory
-  name matches that segment becomes the row's `author` value.
+- `glob` selects which files feed the table — every `.md` at any depth under
+  `notes/`, relative to the directory the config sits in.
 - `ddl` is ordinary `CREATE TABLE` SQL naming the columns you want to keep.
+  Each is a [stat column](./reference/columns.md#stat-columns) `dirsql`
+  computes for every file.
 
-## 5. Restart and query the new shape
+## 5. Query the table
 
-Config is read at startup, so go back to the **first terminal**, stop the
-server with `Ctrl-C`, and start it again — this time pointing `dirsql` at
-your config with `-c` (`dirsql` does not auto-load a `.dirsql.toml` from the
-current directory; you always pass it explicitly):
+`dirsql` does not auto-load a `.dirsql.toml` from the current directory, so
+pass it explicitly with `-c`, **after** the SQL:
 
 ::: code-group
 
 ```bash [npm]
-npx dirsql -c .dirsql.toml
+npx dirsql "SELECT dir, basename, size FROM notes ORDER BY dir, basename" -c .dirsql.toml | jq
 ```
 
 ```bash [PyPI]
-uvx dirsql -c .dirsql.toml
+uvx dirsql "SELECT dir, basename, size FROM notes ORDER BY dir, basename" -c .dirsql.toml | jq
 ```
 
 :::
 
-```
-Running at localhost:7117
-```
-
-This time `dirsql` loaded your `.dirsql.toml` and served the `notes` table
-you defined. Query it from the second terminal:
-
-```bash
-curl -s http://localhost:7117/query \
-  -H 'content-type: application/json' \
-  -d '{"sql":"SELECT author, basename, size FROM notes ORDER BY author, basename"}' \
-  | jq
-```
-
 ```json
 [
   {
     "basename": "ideas.md",
-    "size": 52,
-    "author": "alice"
+    "dir": "notes/alice",
+    "size": 52
   },
   {
     "basename": "welcome.md",
-    "size": 66,
-    "author": "alice"
+    "dir": "notes/alice",
+    "size": 66
   },
   {
     "basename": "reading-list.md",
-    "size": 41,
-    "author": "bob"
+    "dir": "notes/bob",
+    "size": 41
   }
 ]
 ```
 
-Every row now carries an `author` column extracted from its path — no
-extraction code, just a glob. And it is a real SQL column, so you can
-aggregate on it:
+You queried `FROM notes` by name — no path, no glob to repeat. And because
+`dir` is a real SQL column, you can aggregate on it. Count each author's
+notes by their folder:
 
-```bash
-curl -s http://localhost:7117/query \
-  -H 'content-type: application/json' \
-  -d '{"sql":"SELECT author, COUNT(*) AS notes FROM notes GROUP BY author"}' \
-  | jq
+::: code-group
+
+```bash [npm]
+npx dirsql query "SELECT dir, COUNT(*) AS notes FROM notes GROUP BY dir ORDER BY dir" -c .dirsql.toml | jq
+```
+
+```bash [PyPI]
+uvx dirsql query "SELECT dir, COUNT(*) AS notes FROM notes GROUP BY dir ORDER BY dir" -c .dirsql.toml | jq
 ```
 
+:::
+
 ```json
 [
   {
-    "author": "alice",
+    "dir": "notes/alice",
     "notes": 2
   },
   {
-    "author": "bob",
+    "dir": "notes/bob",
     "notes": 1
   }
 ]
 ```
 
-That's the whole loop: files in a directory, a declarative table on top,
-SQL over HTTP.
+That's the whole loop: files in a directory, an instant query with no
+configuration, and a declared table when you want a named shape to reuse.
 
 ## Where to go next
 
-- [Configuration file](./reference/config.md) — the complete `.dirsql.toml`
-  reference: more tables, ignore patterns, persistence, hooks.
-- [CLI](./reference/cli.md) — flags like `--port` and `--config`, plus
-  `dirsql init`.
-- [HTTP API](./reference/http-api.md) — `POST /query` in full, plus
-  `GET /events`, a live stream of row changes as files change.
+- [Query files without a config](./howto/query-without-config.md) — more
+  path-table questions you can ask with no setup at all.
+- [Define tables for your files](./howto/define-tables.md) — the full
+  `[[table]]` recipe: multiple tables, ignore patterns.
+- [Extract rows from file contents](./howto/extract-from-contents.md) —
+  pull columns out of *inside* your files with an `on-file` parser.
+- [CLI](./reference/cli.md) — every flag, plus running `dirsql` as a
+  long-lived HTTP server instead of one-shot queries.
+- [HTTP API](./reference/http-api.md) — `POST /query`, plus `GET /events`,
+  a live stream of row changes as files change.
 - [SDK](./reference/sdk.md) — embed `dirsql` in a Python, Rust, or
-  TypeScript program instead of running the server.
-- Why is the database rebuilt from your files on every startup? See
+  TypeScript program instead of running the CLI.
+- Why is the database rebuilt from your files on every query? See
   [how `dirsql` thinks](./explanation.md).
+```

package/docs/howto/load-extension.md

@@ -53,7 +53,7 @@ entrypoint = "sqlite3_vec_init"
 ```
 
 ```bash
-uvx --with sqlite-vec dirsql -c ./.dirsql.toml
+uvx --with sqlite-vec dirsql server -c ./.dirsql.toml
 ```
 
 **Node (`npx dirsql`, TypeScript SDK).** Use the *npm package name* — but

package/docs/howto/parse-files-into-columns.md

@@ -0,0 +1,122 @@
+# Parse your files into columns
+
+A [path-table](../reference/path-tables.md) gives you one row per file, with the
+stat columns (`path`, `size`, `mtime`, …). When the columns you actually want
+live *inside* each file, attach a parser with
+[`--on-file`](../reference/cli.md#on-file-command): prototype it inline on the
+path-table, then paste the **same** command into a config file the day you want
+a watcher, persistence, or more than one table. The parser command never
+changes across that move — that is the whole point.
+
+## 1. Start from a bare path-table
+
+You have a directory of Markdown posts with YAML frontmatter:
+
+```
+posts/hello-world.md
+posts/on-recursion.md
+```
+
+A path-table already answers questions about the files themselves:
+
+```bash
+dirsql query "SELECT basename, size FROM './posts/*.md' ORDER BY basename"
+```
+
+```json
+[{"basename":"hello-world.md","size":65},{"basename":"on-recursion.md","size":102}]
+```
+
+But `title` and `author` are inside the frontmatter, not in the stat columns.
+To get them, you need a parser.
+
+## 2. Attach a parser with `--on-file`
+
+Any program that reads one file and prints a **JSON array of row objects** on
+stdout is a parser. Here is a small one, `extract.py`, that reads a post's
+frontmatter:
+
+```python
+#!/usr/bin/env python3
+import json, re, sys
+
+text = open(sys.argv[1], encoding="utf-8").read()
+m = re.match(r"^---\n(.*?)\n---", text, re.DOTALL)
+fields = dict(
+    (k.strip(), v.strip())
+    for k, _, v in (line.partition(":") for line in (m.group(1).splitlines() if m else []))
+)
+print(json.dumps([{"title": fields.get("title"), "author": fields.get("author")}]))
+```
+
+Point the path-table at it with `--on-file`:
+
+```bash
+dirsql query "SELECT title, author FROM './posts/*.md' ORDER BY title" \
+  --on-file 'python3 extract.py {path}'
+```
+
+```json
+[{"author":"Ada Lovelace","title":"Hello World"},{"author":"Alan Turing","title":"On Recursion"}]
+```
+
+Now the parser's output *is* the table. `--on-file` runs the command once per
+matched file; `{path}` is the file's absolute path, one of the placeholders in
+the shared [`on-file` hook contract](../reference/hooks.md#on-file) (argv
+splitting, timeout, and per-file failure isolation all come from there). The
+stat columns are no longer reachable — a parser that wants the path emits it,
+since it already has `{path}`. See
+[Parsing rows with `--on-file`](../reference/path-tables.md#parsing-rows-with-on-file)
+for the full behavior.
+
+`--on-file` applies to every path-table in the query and may be given at most
+once. It is a `query`-only flag: there is no config file involved yet, so it is
+the fastest way to see whether a parser produces the rows you expect.
+
+## 3. Graduate to a config file
+
+The inline flag re-scans and re-parses every file on every query, defines one
+parser for the whole query, and forgets everything when the process exits. When
+you want a **watcher** that keeps the rows fresh, **persistence** across
+restarts, or **different parsers for different file sets**, move the parser into
+a `.dirsql.toml` [`[[table]]`](../reference/config.md#table) — and paste the
+command in verbatim:
+
+```toml
+[[table]]
+ddl     = "CREATE TABLE posts (title TEXT, author TEXT)"
+glob    = "posts/*.md"
+on-file = "python3 extract.py {path}"
+```
+
+The `on-file` value is byte-for-byte the string you passed to `--on-file`. The
+[`on-file` hook contract](../reference/hooks.md#on-file) is the same contract
+the flag used — the flag and the config key are two spellings of the same
+attachment. What you gain by moving to config is everything around the parser:
+
+```bash
+dirsql query "SELECT title, author FROM posts ORDER BY title" -c ./.dirsql.toml
+```
+
+```json
+[{"author":"Ada Lovelace","title":"Hello World"},{"author":"Alan Turing","title":"On Recursion"}]
+```
+
+Same command, same rows. The difference is what a declared table brings: it is
+indexed on build, kept fresh by the watcher, survives restarts with
+[`--persist`](./persist.md), and a config can declare
+[many tables](./define-tables.md) — each with its own `on-file` — where the flag
+gives every path-table one parser. A declared table also merges the filesystem
+facts back onto each row, so a `path TEXT` column in the DDL is populated by
+`dirsql` even though the parser did not emit it
+([precedence](../reference/columns.md#precedence)).
+
+## Going further
+
+- The full stat-vs-parsed behavior, failure isolation, and skip rules:
+  [Parsing rows with `--on-file`](../reference/path-tables.md#parsing-rows-with-on-file).
+- Starting from a declared table instead of a path-table?
+  [Extract rows from file contents](./extract-from-contents.md) covers the
+  config-first path.
+- One row per record *within* a file (JSONL, multiple frontmatter blocks): the
+  parser prints one object per row — the array length is the row count.

package/docs/howto/persist.md

@@ -1,7 +1,7 @@
 # Keep the index across restarts
 
 By default the database is ephemeral: rebuilt from your files on every
-startup and discarded on exit. The [`--persist [PATH]`](../reference/cli.md#server-mode)
+startup and discarded on exit. The [`--persist [PATH]`](../reference/cli.md#dirsql-server)
 flag keeps the SQLite index on disk instead, so a restart only re-parses
 files that actually changed — the difference between seconds and
 milliseconds on large trees, and between re-running and skipping expensive

package/docs/howto/react-to-changes.md

@@ -18,8 +18,8 @@ glob = "**/*"
 
 ## 1. Open the stream
 
-With the server running (`npx dirsql -c ./.dirsql.toml` /
-`uvx dirsql -c ./.dirsql.toml`), subscribe from another terminal:
+With the server running (`npx dirsql server -c ./.dirsql.toml` /
+`uvx dirsql server -c ./.dirsql.toml`), subscribe from another terminal:
 
 ```bash
 curl -N http://localhost:7117/events

package/docs/howto/search-by-meaning.md

@@ -10,7 +10,7 @@ embed each question at query time.
 ::: tip Just want it working?
 [`dirsql-plugin-embeddings`](https://pypi.org/project/dirsql-plugin-embeddings/)
 packages exactly what this guide builds, ready to install:
-`uvx --with dirsql-plugin-embeddings dirsql`. Keep reading to see how it's
+`uvx --with dirsql-plugin-embeddings dirsql server`. Keep reading to see how it's
 built — the same three pieces, from scratch.
 :::
 

package/docs/index.md

@@ -34,7 +34,7 @@ db = DirSQL(
     "./my-project",
     tables=[
         Table(
-            ddl="CREATE TABLE files (name TEXT, size INTEGER, type TEXT)",
+            ddl="CREATE TABLE records (name TEXT, size INTEGER, type TEXT)",
             glob="data/*.json",
             on_file=lambda path: [json.loads(open(path, encoding="utf-8").read())],
         ),
@@ -42,7 +42,7 @@ db = DirSQL(
 )
 
 # SQL queries over your filesystem
-large = db.query("SELECT * FROM files WHERE size > 1000")
+large = db.query("SELECT * FROM records WHERE size > 1000")
 ```
 
 ```rust [Rust]
@@ -52,14 +52,14 @@ let db = DirSQL::new(
     "./my-project",
     vec![
         Table::new(
-            "CREATE TABLE files (name TEXT, size INTEGER, type TEXT)",
+            "CREATE TABLE records (name TEXT, size INTEGER, type TEXT)",
             "data/*.json",
             |path| vec![serde_json::from_str(&std::fs::read_to_string(path).unwrap()).unwrap()],
         ),
     ],
 )?;
 
-let large = db.query("SELECT * FROM files WHERE size > 1000")?;
+let large = db.query("SELECT * FROM records WHERE size > 1000")?;
 ```
 
 ```typescript [TypeScript]
@@ -70,14 +70,14 @@ const db = new DirSQL({
   root: './my-project',
   tables: [
     new Table({
-      ddl: 'CREATE TABLE files (name TEXT, size INTEGER, type TEXT)',
+      ddl: 'CREATE TABLE records (name TEXT, size INTEGER, type TEXT)',
       glob: 'data/*.json',
       onFile: (path) => [JSON.parse(readFileSync(path, 'utf8'))],
     }),
   ],
 });
 
-const large = await db.query('SELECT * FROM files WHERE size > 1000');
+const large = await db.query('SELECT * FROM records WHERE size > 1000');
 ```
 
 :::

package/docs/reference/cli.md

@@ -1,39 +1,56 @@
 # CLI
 
-The `dirsql` binary has three modes:
+Query is the default: `dirsql "<sql>"` runs one query and prints JSON rows.
+The `dirsql` binary has these modes:
 
 | Invocation | Behavior |
 |---|---|
-| `dirsql` (no subcommand) | Start a long-lived HTTP server exposing a SQL view of a directory. See [HTTP API](./http-api.md). |
-| `dirsql query "<sql>"` | Query the file system as JSON |
-| `dirsql init` | Generate a `.dirsql.toml` |
+| `dirsql "<sql>"` | Run one query over the directory and print the rows as JSON. The default; identical to `dirsql query "<sql>"`. |
+| `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`. |
+
+Bare `dirsql` with no SQL is a usage error pointing at `dirsql server` — it
+does **not** start the server.
 
 ## Installation
 
 ::: code-group
 
 ```bash [npm]
-npx dirsql
+npx dirsql "SELECT * FROM './'"
 ```
 
 ```bash [PyPI]
-uvx dirsql
+uvx dirsql "SELECT * FROM './'"
 ```
 
 ```bash [Cargo]
 # The `cli` feature is opt-in; this installs the binary only.
 cargo install dirsql --features cli
-dirsql
+dirsql "SELECT * FROM './'"
 ```
 
 :::
 
 The npm launcher requires **Node ≥ 20.11**.
 
-## Server mode
+## Default query mode
 
 ```bash
-dirsql
+# No -c: query the filesystem with a path-table.
+dirsql "SELECT basename, size FROM './' ORDER BY size DESC LIMIT 5"
+# [{"basename":"model.bin","size":104857600}, …]
+```
+
+`dirsql "<sql>"` is exactly [`dirsql query "<sql>"`](#dirsql-query) — same
+pipeline, same flags, same output. See that section for config discovery,
+`--persist`, `--on-file`, hooks, and exit codes.
+
+## `dirsql server`
+
+```bash
+dirsql server
 # Running at localhost:7117
 ```
 
@@ -43,12 +60,15 @@ requests, closes open `/events` streams, and exits.
 
 ### Flags
 
+Config flags are subcommand-local: pass them after `server`
+(`dirsql server -c <cfg>`).
+
 | Flag | Default | Description |
 |---|---|---|
-| `-c, --config <path>` | none | Path to a [config file](./config.md). **Repeatable** (`-c a -c b`): the configs load and merge in argv order — see [Composing multiple configs](./config.md#composing-multiple-configs). The index is always rooted at the **invocation directory** (the current working directory), regardless of where a config lives — so `--config /elsewhere/.dirsql.toml` still indexes the directory you ran `dirsql` from. With none given, **no named tables are defined** — query the filesystem with a [path-table](./path-tables.md) (`FROM './'`). A `./.dirsql.toml` on disk is **not** auto-loaded; pass it explicitly. A `-c` naming a file that does not exist is an [error](#degraded-mode). |
+| `-c, --config <path>` | none | Path to a [config file](./config.md). **Repeatable** (`-c a -c b`): the configs load and merge in argv order — see [Composing multiple configs](./config.md#composing-multiple-configs). The index is always rooted at the **invocation directory** (the current working directory), regardless of where a config lives — so `--config /elsewhere/.dirsql.toml` still indexes the directory you ran `dirsql server` from. With none given, **no named tables are defined** — query the filesystem with a [path-table](./path-tables.md) (`FROM './'`). A `./.dirsql.toml` on disk is **not** auto-loaded; pass it explicitly. A `-c` naming a file that does not exist is an [error](#degraded-mode). |
 | `--host <addr>` | `localhost` | Bind address. |
 | `--port <n>` | `7117` | TCP port to bind. |
-| `--persist [<path>]` | off | Keep the SQLite index on disk between runs so a restart only re-parses files that actually changed. Bare `--persist` caches at `<root>/.dirsql/cache.db`; `--persist <path>` caches at `<path>`. Off by default (the index is ephemeral). Also available on [`dirsql query`](#dirsql-query), passed after the subcommand. See [Keep the index across restarts](../howto/persist.md). |
+| `--persist [<path>]` | off | Keep the SQLite index on disk between runs so a restart only re-parses files that actually changed. Bare `--persist` caches at `<root>/.dirsql/cache.db`; `--persist <path>` caches at `<path>`. Off by default (the index is ephemeral). Also available on [`dirsql query`](#dirsql-query). See [Keep the index across restarts](../howto/persist.md). |
 | `--extension <path>` | none | Load a SQLite extension by literal path, overriding the config's `[[dirsql.extension]]` entries. Repeatable. Format: `<path>` or `<path>::<entrypoint>`. Internal plumbing for the pip/npm launchers, which resolve package-name extensions and pass the resolved paths here — not intended for direct use. When any `--extension` is present, the config file's own extension entries are not loaded. |
 | `--version` | | Print the version and exit. |
 | `--help` | | Print usage and exit. |
@@ -124,8 +144,9 @@ dirsql query "SELECT COUNT(*) AS n FROM posts" -c ./.dirsql.toml | jq '.[0].n'
 Pass `-c`/`--config`, `--persist`, and `--extension` **after** `query`
 (`dirsql query "<sql>" -c <cfg>`). A config flag placed *before* the subcommand
 is a hard error — `error: the subcommand 'query' cannot be used with
-'--config <CONFIG>'` — never silently dropped. (In server mode, with no
-subcommand, the same flags are passed directly: `dirsql -c <cfg>`.)
+'--config <CONFIG>'` — never silently dropped. (The default mode without the
+`query` keyword takes the same flags after the SQL: `dirsql "<sql>" -c <cfg>`;
+for the server they follow the subcommand: `dirsql server -c <cfg>`.)
 :::
 
 The subcommand builds the index, runs the SQL, prints the result rows as a
@@ -168,7 +189,9 @@ timeout. The parser's output is the whole schema; the stat columns are not
 reachable on a parsed path-table. `--on-file` may be given **at most once** (a
 repeat is an error pointing at config files) and never touches config-declared
 tables. It is a `query`-only flag — server mode rejects it as an unknown
-argument.
+argument. The command string is copy-paste identical to a `[[table]]`
+`on-file` key, so an inline parser graduates to a config file unchanged — see
+[Parse your files into columns](../howto/parse-files-into-columns.md).
 
 Errors print the same diagnostic the HTTP `{"error": …}` body carries —
 config failures, SQL errors, rejected reads, hook failures, timeouts — to
@@ -197,7 +220,7 @@ The output does **not** auto-load. Once you've tweaked it, pass it explicitly
 to run against it:
 
 ```bash
-dirsql -c ./.dirsql.toml
+dirsql "SELECT * FROM files" -c ./.dirsql.toml
 ```
 
 ### Flags

package/docs/reference/config.md

@@ -40,7 +40,7 @@ hook-timeout = 300
 ```
 
 Persistence is not a config key. Keep the SQLite index on disk between runs
-with the [`--persist [PATH]` CLI flag](./cli.md#server-mode) — a machine-local
+with the [`--persist [PATH]` CLI flag](./cli.md#dirsql-server) — a machine-local
 operational choice that belongs to the runner, not to shareable config.
 
 ## `[[dirsql.extension]]`

package/docs/reference/hooks.md

@@ -99,6 +99,17 @@ every watched change. The command reads the file itself and prints a JSON
 array of row objects; see [`[[table]]`](./config.md#table) for the
 row-mapping rules.
 
+The same command is attachable two ways, over the same contract: the
+`on-file` config key on a `[[table]]`, and the
+[`--on-file <command>`](./cli.md#on-file-command) flag on `dirsql query`,
+which attaches it to every [path-table](./path-tables.md#parsing-rows-with-on-file)
+in the query. The command string is identical between the two spellings — the
+flag is the inline form, the config key the declared form (see
+[Parse your files into columns](../howto/parse-files-into-columns.md)). The one
+behavioral difference is not in this contract but in the surrounding table: a
+declared `[[table]]` merges the filesystem facts onto each parsed row, while a
+parsed path-table exposes only the parser's output.
+
 | Placeholder | Value |
 |---|---|
 | `{path}` | The matched file's **absolute** path. `on-file = "extract.py {path}"` — self-sufficient from any working directory, so the command resolves it even when the config lives outside the index. |

package/docs/reference/http-api.md

@@ -1,6 +1,6 @@
 # HTTP API
 
-The [`dirsql` server](./cli.md#server-mode) (default `localhost:7117`)
+The [`dirsql` server](./cli.md#dirsql-server) (default `localhost:7117`)
 exposes two endpoints: `POST /query` and `GET /events`.
 
 ## `POST /query`
@@ -27,7 +27,7 @@ The `200` response is a JSON array of row objects keyed by column name:
 ```bash
 curl -s http://localhost:7117/query \
   -H 'content-type: application/json' \
-  -d '{"sql":"SELECT COUNT(*) AS n FROM files"}' \
+  -d '{"sql":"SELECT COUNT(*) AS n FROM \'./\'"}' \
   | jq
 ```
 

package/docs/reference/path-tables.md

@@ -198,9 +198,9 @@ pattern if you would rather not see them.
 Path-tables are ordinary SQLite tables once resolved, so they join freely:
 
 ```sql
-SELECT p.basename, f.size
+SELECT p.basename, d.size
 FROM './docs/*.md' AS p
-JOIN files AS f ON f.path = p.path;
+JOIN pages AS d ON d.path = p.path;
 ```
 
 A zero-match path-table is not an error — it is an empty table, and the query

package/docs/reference/sdk.md

@@ -130,7 +130,7 @@ shortcut was removed in #603 — use
   ephemeral, rebuilt every startup). The cache lives at
   `<root>/.dirsql/cache.db` by default; on restart, only files whose stat
   changed are re-parsed. (The CLI exposes the same switch as the
-  [`--persist [PATH]`](./cli.md#server-mode) flag; it is not a config key.)
+  [`--persist [PATH]`](./cli.md#dirsql-server) flag; it is not a config key.)
 - `persist_path` / `persistPath` — Override the cache location. Ignored when
   persistence is off. Constructor values are used as given. In Rust these two
   parameters collapse into a single builder method: `.persist(None)` enables

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dirsql",
-  "version": "0.3.124",
+  "version": "0.3.126",
   "description": "Ephemeral SQL index over a local directory",
   "license": "MIT",
   "repository": "https://github.com/thekevinscott/dirsql",
@@ -212,15 +212,15 @@
     ]
   },
   "optionalDependencies": {
-    "@dirsql/lib-linux-x64-gnu": "0.3.124",
-    "@dirsql/lib-linux-arm64-gnu": "0.3.124",
-    "@dirsql/lib-darwin-x64": "0.3.124",
-    "@dirsql/lib-darwin-arm64": "0.3.124",
-    "@dirsql/lib-win32-x64-msvc": "0.3.124",
-    "@dirsql/cli-linux-x64-gnu": "0.3.124",
-    "@dirsql/cli-linux-arm64-gnu": "0.3.124",
-    "@dirsql/cli-darwin-x64": "0.3.124",
-    "@dirsql/cli-darwin-arm64": "0.3.124",
-    "@dirsql/cli-win32-x64-msvc": "0.3.124"
+    "@dirsql/lib-linux-x64-gnu": "0.3.126",
+    "@dirsql/lib-linux-arm64-gnu": "0.3.126",
+    "@dirsql/lib-darwin-x64": "0.3.126",
+    "@dirsql/lib-darwin-arm64": "0.3.126",
+    "@dirsql/lib-win32-x64-msvc": "0.3.126",
+    "@dirsql/cli-linux-x64-gnu": "0.3.126",
+    "@dirsql/cli-linux-arm64-gnu": "0.3.126",
+    "@dirsql/cli-darwin-x64": "0.3.126",
+    "@dirsql/cli-darwin-arm64": "0.3.126",
+    "@dirsql/cli-win32-x64-msvc": "0.3.126"
   }
 }