Merge pull request #293 from constructive-io/devin/1775777114-plpgsql-parse-package

feat: add plpgsql-parse package with comment preservation for PL/pgSQL function bodies

Author: pyramation

packages/plpgsql-parse/.gitignore

@@ -0,0 +1 @@
+dist/

packages/plpgsql-parse/README.md

@@ -0,0 +1,59 @@
+# plpgsql-parse
+
+<p align="center" width="100%">
+  <img height="250" src="https://raw.githubusercontent.com/constructive-io/constructive/refs/heads/main/assets/outline-logo.svg" />
+</p>
+
+Comment preserving PL/pgSQL parser. A wrapper around `plpgsql-parser` and `plpgsql-deparser` that preserves `--` line comments inside PL/pgSQL function bodies through parse-deparse round trips.
+
+## Installation
+
+```sh
+npm install plpgsql-parse
+```
+
+## Features
+
+* **Body Comment Preservation** -- Retains `--` line comments inside PL/pgSQL function bodies (`$$...$$`) through parse-deparse cycles
+* **Outer SQL Comment Preservation** -- Preserves comments and whitespace outside function definitions via `pgsql-parse`
+* **Idempotent Round-Trips** -- `parse -> deparse -> parse -> deparse` produces identical output
+* **Non-Invasive** -- Does not modify `plpgsql-parser`, `plpgsql-deparser`, or any other upstream packages
+
+## How It Works
+
+1. Uses `pgsql-parse` for outer SQL comment and whitespace preservation
+2. For each PL/pgSQL function, scans the `$$...$$` body to extract `--` comments with line numbers
+3. Associates each comment with the nearest following PL/pgSQL statement (anchor)
+4. On deparse, re-injects comments by matching statement keywords against the deparsed output
+
+## API
+
+### Parse
+
+```typescript
+import { parseSync, deparseSync, loadModule } from 'plpgsql-parse';
+
+await loadModule();
+
+const result = parseSync(`
+-- Create a counter function
+CREATE FUNCTION get_count() RETURNS int LANGUAGE plpgsql AS $$
+BEGIN
+  -- Count active users
+  RETURN (SELECT count(*) FROM users WHERE active);
+END;
+$$;
+`);
+
+// result.enhanced contains outer SQL comments/whitespace
+// result.functions contains body comments for each PL/pgSQL function
+const sql = deparseSync(result);
+// Output preserves both outer and body comments
+```
+
+## Credits
+
+Built on the excellent work of several contributors:
+
+* **[Dan Lynch](https://github.com/pyramation)** -- official maintainer since 2018 and architect of the current implementation
+* **[Lukas Fittl](https://github.com/lfittl)** for [libpg_query](https://github.com/pganalyze/libpg_query) -- the core PostgreSQL parser that powers this project

packages/plpgsql-parse/__tests__/__snapshots__/roundtrip.test.ts.snap

@@ -0,0 +1,161 @@
+// Jest Snapshot v1, https://jestjs.io/docs/snapshot-testing
+
+exports[`fixture round-trip tests exception-handler.sql deparsed output matches snapshot 1`] = `
+"-- Function with comments in exception handler
+CREATE FUNCTION safe_divide(
+  a numeric,
+  b numeric
+) RETURNS numeric LANGUAGE plpgsql AS $$
+DECLARE
+  v_result numeric;
+BEGIN
+  -- Attempt the division
+  v_result := a / b;
+  RETURN v_result;
+EXCEPTION
+  WHEN division_by_zero THEN
+    -- Log the error and return null
+    RAISE NOTICE 'Division by zero: % / %', a, b;
+    RETURN NULL;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests loop-with-comments.sql deparsed output matches snapshot 1`] = `
+"-- Function with comments inside loops
+CREATE FUNCTION process_batch(
+  p_batch_size int
+) RETURNS int LANGUAGE plpgsql AS $$
+DECLARE
+  v_processed integer := 0;
+  r RECORD;
+BEGIN
+  -- Process items in batches
+  FOR r IN SELECT id, data FROM pending_items LIMIT p_batch_size LOOP
+    -- Process each item
+    PERFORM process_item(r.id, r.data);
+    v_processed := v_processed + 1;
+  END LOOP;
+
+  -- Return the count of processed items
+  RETURN v_processed;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests multi-function.sql deparsed output matches snapshot 1`] = `
+"-- Multiple functions in one file
+-- with outer SQL comments preserved
+
+CREATE FUNCTION add_numbers(
+  a int,
+  b int
+) RETURNS int LANGUAGE plpgsql AS $$
+BEGIN
+  -- Simple addition
+  RETURN a + b;
+END;
+$$;
+
+-- Second function with its own body comments
+CREATE FUNCTION multiply_numbers(
+  a int,
+  b int
+) RETURNS int LANGUAGE plpgsql AS $$
+BEGIN
+  -- Multiply the inputs
+  RETURN a * b;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests multiple-comments.sql deparsed output matches snapshot 1`] = `
+"-- Function with multiple comment groups
+CREATE FUNCTION process_order(
+  p_order_id int
+) RETURNS void LANGUAGE plpgsql AS $$
+DECLARE
+  v_total numeric;
+  v_status text;
+BEGIN
+  -- First, calculate the order total
+  SELECT sum(amount) INTO v_total FROM order_items WHERE order_id = p_order_id;
+
+  -- Then update the order status
+  -- based on the total amount
+  IF v_total > 1000 THEN
+    v_status := 'premium';
+  ELSE
+    v_status := 'standard';
+  END IF;
+
+  -- Finally, record the result
+  UPDATE orders SET status = v_status, total = v_total WHERE id = p_order_id;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests nested-blocks.sql deparsed output matches snapshot 1`] = `
+"-- Function with nested blocks and comments
+CREATE FUNCTION complex_logic(
+  p_id int
+) RETURNS text LANGUAGE plpgsql AS $$
+DECLARE
+  v_result text;
+BEGIN
+  -- Initialize result
+  v_result := 'unknown';
+
+  -- Try the main logic
+  BEGIN
+    -- Fetch and process
+    SELECT status INTO v_result FROM items WHERE id = p_id;
+  EXCEPTION
+    WHEN no_data_found THEN
+      -- Handle missing item
+      v_result := 'not_found';
+  END;
+
+  RETURN v_result;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests no-comments.sql deparsed output matches snapshot 1`] = `
+"CREATE FUNCTION get_one() RETURNS int LANGUAGE plpgsql AS $$
+BEGIN
+  RETURN 1;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests simple-function.sql deparsed output matches snapshot 1`] = `
+"-- Simple function with body comments
+CREATE FUNCTION get_user_count() RETURNS int LANGUAGE plpgsql AS $$
+DECLARE
+  v_count integer;
+BEGIN
+  -- Count all active users
+  SELECT count(*) INTO v_count FROM users WHERE is_active = true;
+  RETURN v_count;
+END;
+$$;"
+`;
+
+exports[`fixture round-trip tests trigger-function.sql deparsed output matches snapshot 1`] = `
+"-- Trigger function with comments in body
+CREATE FUNCTION audit_trigger() RETURNS trigger LANGUAGE plpgsql AS $$
+BEGIN
+  -- Set the updated_at timestamp
+  NEW.updated_at := now();
+
+  -- Record the change in audit log
+  IF TG_OP = 'UPDATE' THEN
+    INSERT INTO audit_log (table_name, operation, old_data, new_data)
+    VALUES (TG_TABLE_NAME, TG_OP, row_to_json(OLD), row_to_json(NEW));
+  END IF;
+
+  RETURN NEW;
+END;
+$$;"
+`;

packages/plpgsql-parse/__tests__/fixtures/exception-handler.sql

@@ -0,0 +1,17 @@
+-- Function with comments in exception handler
+CREATE FUNCTION safe_divide(a numeric, b numeric) RETURNS numeric
+LANGUAGE plpgsql
+AS $$
+DECLARE
+  v_result numeric;
+BEGIN
+  -- Attempt the division
+  v_result := a / b;
+  RETURN v_result;
+EXCEPTION
+  WHEN division_by_zero THEN
+    -- Log the error and return null
+    RAISE NOTICE 'Division by zero: % / %', a, b;
+    RETURN NULL;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/loop-with-comments.sql

@@ -0,0 +1,19 @@
+-- Function with comments inside loops
+CREATE FUNCTION process_batch(p_batch_size integer) RETURNS integer
+LANGUAGE plpgsql
+AS $$
+DECLARE
+  v_processed integer := 0;
+  r RECORD;
+BEGIN
+  -- Process items in batches
+  FOR r IN SELECT id, data FROM pending_items LIMIT p_batch_size LOOP
+    -- Process each item
+    PERFORM process_item(r.id, r.data);
+    v_processed := v_processed + 1;
+  END LOOP;
+
+  -- Return the count of processed items
+  RETURN v_processed;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/multi-function.sql

@@ -0,0 +1,21 @@
+-- Multiple functions in one file
+-- with outer SQL comments preserved
+
+CREATE FUNCTION add_numbers(a integer, b integer) RETURNS integer
+LANGUAGE plpgsql
+AS $$
+BEGIN
+  -- Simple addition
+  RETURN a + b;
+END;
+$$;
+
+-- Second function with its own body comments
+CREATE FUNCTION multiply_numbers(a integer, b integer) RETURNS integer
+LANGUAGE plpgsql
+AS $$
+BEGIN
+  -- Multiply the inputs
+  RETURN a * b;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/multiple-comments.sql

@@ -0,0 +1,23 @@
+-- Function with multiple comment groups
+CREATE FUNCTION process_order(p_order_id integer) RETURNS void
+LANGUAGE plpgsql
+AS $$
+DECLARE
+  v_total numeric;
+  v_status text;
+BEGIN
+  -- First, calculate the order total
+  SELECT sum(amount) INTO v_total FROM order_items WHERE order_id = p_order_id;
+
+  -- Then update the order status
+  -- based on the total amount
+  IF v_total > 1000 THEN
+    v_status := 'premium';
+  ELSE
+    v_status := 'standard';
+  END IF;
+
+  -- Finally, record the result
+  UPDATE orders SET status = v_status, total = v_total WHERE id = p_order_id;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/nested-blocks.sql

@@ -0,0 +1,23 @@
+-- Function with nested blocks and comments
+CREATE FUNCTION complex_logic(p_id integer) RETURNS text
+LANGUAGE plpgsql
+AS $$
+DECLARE
+  v_result text;
+BEGIN
+  -- Initialize result
+  v_result := 'unknown';
+
+  -- Try the main logic
+  BEGIN
+    -- Fetch and process
+    SELECT status INTO v_result FROM items WHERE id = p_id;
+  EXCEPTION
+    WHEN no_data_found THEN
+      -- Handle missing item
+      v_result := 'not_found';
+  END;
+
+  RETURN v_result;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/no-comments.sql

@@ -0,0 +1,7 @@
+CREATE FUNCTION get_one() RETURNS integer
+LANGUAGE plpgsql
+AS $$
+BEGIN
+  RETURN 1;
+END;
+$$;

packages/plpgsql-parse/__tests__/fixtures/simple-function.sql

@@ -0,0 +1,12 @@
+-- Simple function with body comments
+CREATE FUNCTION get_user_count() RETURNS integer
+LANGUAGE plpgsql
+AS $$
+DECLARE
+  v_count integer;
+BEGIN
+  -- Count all active users
+  SELECT count(*) INTO v_count FROM users WHERE is_active = true;
+  RETURN v_count;
+END;
+$$;