Rhombic NPM | npm.io

Utilities for parsing, analysing and manipulating SQL.

Getting started

npm install rhombic

Description

Rhombic can parse SQL with 2 different parsers with different sets of operations applicable in each case:

Chevrotain based parser built from ground up to support simple statements for ANSI SQL dialect. Parsed SQL tree can then be manipulated (adding projection items, ordering, filters) and serialized back to SQL text. For details and available operations see src/index.ts
Antlr based parser (generated with antlr4ts) from SQL grammar derived from Apache Spark SQL grammar with the goal to support most SQL dialects with broad functionality. Currenly this mode can be used to extract SQL column level lineage. For details and available operations see src/antlr/index.ts

Antlr parser - lineage

To build SQL column level lineage for an SQL query using Antlr-based parser:

import { antlr, TablePrimary } from "rhombic";

try {
  const parsingOptions = {
    // if double quotes should quote identifiers:
    doubleQuotedIdentifier: true
  };
  const q = antlr.parse("SELECT * FROM abc;", parsingOptions);

  console.log(q.getUsedTables()); // [{ tableName: "abc" }];
  const getTable = (table: TablePrimary) => {
    /* Logic to retrieve table & columns metadata */
  };

  // Whether to use "mergedLeaves" or "tree" lineage type
  const mergedLeaves = true;
  const lineageOptions = {
    positionalRefsEnabled: false
  };
  const lineage = q.getLineage(getTable, mergedLeaves, lineageOptions);
} catch (e) {
  // Parsing errors
}

You can then use something like react-flow to draw a nice visualization of your lineage -

Chevrotain parser - SQL manipulation

import rhombic from "rhombic";

try {
  const query = rhombic
    .parse("SELECT * FROM abc;")
    .addProjectionItem("city")
    .toString();

  console.log(query); // SELECT city FROM abc;
} catch (e) {
  // Parsing errors
}