rbtr-lang-sql 2026.9.0.dev1__py3-none-any.whl

rbtr_lang_sql/__init__.py

@@ -0,0 +1 @@
+"""SQL language plugin package."""

rbtr_lang_sql/plugin.py

@@ -0,0 +1,52 @@
+"""SQL language plugin.
+
+SQL has no functions/classes in the OO sense and no native import
+mechanism, so constructs map onto rbtr's capture conventions by shape:
+structural definitions are classes, routines and executable statements
+are functions, and standalone named objects are variables.  One chunk is
+produced per top-level statement, plus one per CTE.
+
+The tree-sitter query lives in `sql.scm`: DDL/ALTER/DROP verbs grouped
+into `[...]` alternations by chunk kind and name location, plus hand-
+written patterns for the DML statements those groups can't express.
+Verified against `tree-sitter-sql` 0.3.11.
+
+Extracted chunks::
+
+    CREATE TABLE users (...)            → class "users"
+    CREATE VIEW / MATERIALIZED VIEW     → class
+    CREATE TYPE mood AS ENUM (...)      → class "mood"
+
+    CREATE FUNCTION add(...) ...        → function "add"
+    SELECT ... FROM users               → function "users"
+    INSERT INTO logs ...                → function "logs"
+    UPDATE / DELETE ... t               → function "t"
+    WITH ranked AS (...)                → function "ranked"  (one per CTE)
+    ... UNION ...                       → function, unnamed
+
+    CREATE INDEX idx ON users(...)      → variable "idx"
+    CREATE SEQUENCE / SCHEMA / ROLE …   → variable
+    CREATE TRIGGER trg ...              → variable "trg"
+    ALTER / DROP <object>               → function "<object>"
+
+A statement with no nameable target (`SELECT 1`, a sub-query `FROM`, a
+`UNION`) carries no name.  `CREATE PROCEDURE` and `PRAGMA` are not
+extracted — the grammar has no node for them (they parse to `ERROR`).
+"""
+
+from __future__ import annotations
+
+from rbtr.languages.registration import LanguageRegistration, QueryExtraction, load_query
+
+# ── Plugin ───────────────────────────────────────────────────────────
+
+
+sql = LanguageRegistration(
+    id="sql",
+    extensions=frozenset({".sql"}),
+    grammar_module="tree_sitter_sql",
+    extraction=QueryExtraction(
+        query=load_query(__package__, "sql"),
+    ),
+    extraction_serial=3,
+)

rbtr_lang_sql/sql.scm

@@ -0,0 +1,70 @@
+; Comments (SQL: `--` line is `comment`, `/* */` block is `marginalia`).
+[(comment) (marginalia)] @comment
+
+; DDL/ALTER/DROP verbs, grouped by chunk kind and name location.
+;
+; object_reference names are left unanchored: each verb has exactly one
+; direct object_reference, and leaving it unanchored keeps IF NOT EXISTS /
+; OR REPLACE / TEMP working (anchoring to a keyword breaks when those
+; intervene). A bare (identifier) name (index/schema/role/…) is used when
+; the object_reference, if any, is a secondary reference (e.g. an index's
+; ON table). Verified against tree-sitter-sql 0.3.11.
+
+[
+  (create_table (object_reference name: (identifier) @_cls_name))
+  (create_view (object_reference name: (identifier) @_cls_name))
+  (create_materialized_view (object_reference name: (identifier) @_cls_name))
+  (create_type (object_reference name: (identifier) @_cls_name))
+] @class
+
+(create_sequence (object_reference name: (identifier) @_var_name)) @variable
+
+[
+  (create_function (object_reference name: (identifier) @_fn_name))
+  (alter_table (object_reference name: (identifier) @_fn_name))
+  (alter_view (object_reference name: (identifier) @_fn_name))
+  (alter_sequence (object_reference name: (identifier) @_fn_name))
+  (drop_table (object_reference name: (identifier) @_fn_name))
+  (drop_view (object_reference name: (identifier) @_fn_name))
+  (drop_sequence (object_reference name: (identifier) @_fn_name))
+  (drop_function (object_reference name: (identifier) @_fn_name))
+  (drop_type (object_reference name: (identifier) @_fn_name))
+] @function
+
+[
+  (create_index (identifier) @_var_name)
+  (create_schema (identifier) @_var_name)
+  (create_database (identifier) @_var_name)
+  (create_extension (identifier) @_var_name)
+  (create_role (identifier) @_var_name)
+] @variable
+
+[
+  (alter_index (identifier) @_fn_name)
+  (alter_schema (identifier) @_fn_name)
+  (alter_database (identifier) @_fn_name)
+  (alter_role (identifier) @_fn_name)
+  (alter_type (identifier) @_fn_name)
+  (drop_index (identifier) @_fn_name)
+  (drop_schema (identifier) @_fn_name)
+  (drop_database (identifier) @_fn_name)
+  (drop_role (identifier) @_fn_name)
+  (drop_extension (identifier) @_fn_name)
+] @function
+
+; DML statements, CTEs, triggers, and set operations — structures the DDL
+; groups can't express. DML patterns are anchored at (program (statement …))
+; so each top-level statement yields exactly one chunk: select and delete
+; keep their target in a sibling from, so the whole statement is captured.
+; A set_operation (UNION/INTERSECT/EXCEPT) has no single table → unnamed.
+
+(create_trigger (keyword_trigger) . (object_reference name: (identifier) @_var_name)) @variable
+
+(cte (identifier) @_fn_name) @function
+
+(program (statement (select)
+  (from (relation (object_reference name: (identifier) @_fn_name)))?) @function)
+(program (statement (set_operation) @function))
+(program (statement (insert (object_reference name: (identifier) @_fn_name))) @function)
+(program (statement (update (relation (object_reference name: (identifier) @_fn_name)))) @function)
+(program (statement (delete) (from (object_reference name: (identifier) @_fn_name))) @function)

rbtr_lang_sql/tests/__snapshots__/test_samples/test_edges_match_snapshot.json

@@ -0,0 +1 @@
+[]

rbtr_lang_sql/tests/__snapshots__/test_samples/test_extraction_matches_snapshot.json

@@ -0,0 +1,268 @@
+[
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- Greeter schema — tables, views, and statements for greetings.\n--\n-- The SQL plugin extracts one chunk per top-level statement: CREATE\n-- TABLE/VIEW (as classes), CREATE FUNCTION and DML/CTEs (as functions),\n-- and CREATE SCHEMA/INDEX/SEQUENCE (as variables). There is no import\n-- concept. CREATE PROCEDURE and PRAGMA do not parse and live in the\n-- xfail registry instead.",
+    "line_start": 1,
+    "line_end": 7,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "fcb24b55c55b71af"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "variable",
+    "name": "greet",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE SCHEMA greet",
+    "line_start": 9,
+    "line_end": 9,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "2bf9543268256917"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- trailing comment: its own chunk",
+    "line_start": 9,
+    "line_end": 9,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "98756ee52a748b1e"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- Standalone note, separated by blank lines from any statement.\n-- Second line of the same block.",
+    "line_start": 11,
+    "line_end": 12,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "0c438a2a19d6d59e"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "class",
+    "name": "greeters",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE TABLE greeters (\n    id INTEGER PRIMARY KEY,\n    prefix TEXT NOT NULL\n)",
+    "line_start": 14,
+    "line_end": 17,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "119e8882c4026de1"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "variable",
+    "name": "idx_greeters_prefix",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE INDEX idx_greeters_prefix ON greeters (prefix)",
+    "line_start": 19,
+    "line_end": 19,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "ab3497f2cf28a336"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "variable",
+    "name": "greeting_id_seq",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE SEQUENCE greeting_id_seq",
+    "line_start": 21,
+    "line_end": 21,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "1fdfd3531b021310"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "class",
+    "name": "active_greeters",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE VIEW active_greeters AS\nSELECT id, prefix FROM greeters WHERE prefix IS NOT NULL",
+    "line_start": 23,
+    "line_end": 24,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "42074e751143db09"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "shout",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE FUNCTION shout(message TEXT) RETURNS TEXT AS 'SELECT upper(message)'",
+    "line_start": 26,
+    "line_end": 26,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "e809c9a3ee5d5686"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "recent",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "WITH recent AS (\n    SELECT id FROM greeters\n)\n\nSELECT id FROM recent",
+    "line_start": 28,
+    "line_end": 32,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "330f6d8ac830cc9f"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "recent",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "recent AS (\n    SELECT id FROM greeters\n)",
+    "line_start": 28,
+    "line_end": 30,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "a2b86ca2dd9035b0"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "greeters",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "INSERT INTO greeters (id, prefix) VALUES (1, 'Hello')",
+    "line_start": 34,
+    "line_end": 34,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "c68ef3a9070b8b98"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "greeters",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "UPDATE greeters SET prefix = 'Hi' WHERE id = 1",
+    "line_start": 36,
+    "line_end": 36,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "f9b1fae7ce538957"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql.sql",
+    "kind": "function",
+    "name": "greeters",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "DELETE FROM greeters WHERE id = 1",
+    "line_start": 38,
+    "line_end": 38,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "6af17e06c316e1a8"
+  }
+]

rbtr_lang_sql/tests/__snapshots__/test_samples/test_sql_dialect_extraction_matches_snapshot[sql_clickhouse].json

@@ -0,0 +1,59 @@
+[
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_clickhouse.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- ClickHouse dialect sample — MergeTree engine, ORDER BY, special types.",
+    "line_start": 1,
+    "line_end": 1,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "0202263c4558210b"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_clickhouse.sql",
+    "kind": "class",
+    "name": "events",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE TABLE events (\n    event_date Date,\n    user_id UInt64,\n    name LowCardinality(String),\n    props Array(String),\n    value Nullable(Float64)\n)\nENGINE = MergeTree\nPARTITION BY toYYYYMM(event_date)",
+    "line_start": 3,
+    "line_end": 11,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "b7a5c7af04875e96"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_clickhouse.sql",
+    "kind": "class",
+    "name": "events_mv",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE MATERIALIZED VIEW events_mv\nENGINE = SummingMergeTree\nORDER BY user_id\nAS SELECT user_id, count() AS c FROM events GROUP BY user_id",
+    "line_start": 14,
+    "line_end": 17,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "b6ef13e06c2afd39"
+  }
+]

rbtr_lang_sql/tests/__snapshots__/test_samples/test_sql_dialect_extraction_matches_snapshot[sql_duckdb].json

@@ -0,0 +1,78 @@
+[
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_duckdb.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- DuckDB dialect sample — LIST/STRUCT types, CTAS, CREATE MACRO.",
+    "line_start": 1,
+    "line_end": 1,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "0202263c4558210b"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_duckdb.sql",
+    "kind": "class",
+    "name": "users",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE TABLE users (\n    id BIGINT PRIMARY KEY,\n    email VARCHAR NOT NULL,\n    tags VARCHAR[],\n    profile STRUCT(name VARCHAR, age INTEGER)",
+    "line_start": 3,
+    "line_end": 7,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "bbc0c1d3c758c83c"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_duckdb.sql",
+    "kind": "class",
+    "name": "recent",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE TABLE recent AS SELECT * FROM users LIMIT 10",
+    "line_start": 10,
+    "line_end": 10,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "9a7eceaf6e76be85"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_duckdb.sql",
+    "kind": "class",
+    "name": "emails",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE VIEW emails AS SELECT email FROM users",
+    "line_start": 14,
+    "line_end": 14,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "c401891794cc2bc1"
+  }
+]

rbtr_lang_sql/tests/__snapshots__/test_samples/test_sql_dialect_extraction_matches_snapshot[sql_mysql].json

@@ -0,0 +1,59 @@
+[
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_mysql.sql",
+    "kind": "comment",
+    "name": "",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "-- MySQL dialect sample — backtick identifiers, UNSIGNED AUTO_INCREMENT, ENGINE.",
+    "line_start": 1,
+    "line_end": 1,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "0202263c4558210b"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_mysql.sql",
+    "kind": "class",
+    "name": "`users`",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE TABLE `users` (\n    `id` INT UNSIGNED NOT NULL AUTO_INCREMENT,\n    `email` VARCHAR(255) NOT NULL,\n    `created_at` TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,\n    PRIMARY KEY (`id`),\n    UNIQUE KEY `uq_email` (`email`)\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4",
+    "line_start": 3,
+    "line_end": 9,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "4cab96a45cc21ec3"
+  },
+  {
+    "blob_sha": "sha1",
+    "file_path": "sql_mysql.sql",
+    "kind": "variable",
+    "name": "`idx_email`",
+    "scope": "",
+    "language": "sql",
+    "file_language": "sql",
+    "content": "CREATE INDEX `idx_email` ON `users` (`email`)",
+    "line_start": 11,
+    "line_end": 11,
+    "metadata": {
+      "module": "",
+      "names": "",
+      "dots": "",
+      "language_hint": ""
+    },
+    "id": "1b1d6760ac055d12"
+  }
+]