Filtering & Sorting

Once your data is loaded, you'll need to slice and dice it to find insights.

Filtering Rows

Molniya offers two main ways to filter data: the flexible JavaScript way, and the explicit structural way.

1. The Flexible Way (.filter)

Use a standard JavaScript arrow function. This is the most powerful method because you can write any logic you want.

// Keep users who are adults AND have an active subscription
const activeAdults = df.filter((row) => {
  return row.age >= 18 && row.subscription_status === 'active';
});

You can even use external helpers or complex regex:

const gmailUsers = df.filter(row => row.email.endsWith('@gmail.com'));

2. The Structural Way (.where)

If you have a simple condition, .where() is often more readable and shorter.

// All rows where 'status' is 'pending'
const pending = df.where('status', 'eq', 'pending');

// All rows where 'score' is greater than 80
const highScorers = df.where('score', 'gt', 80);

TIP

Performance Note: For extremely large datasets, chaining .where() conditions can sometimes be faster than a complex .filter() callback, but for most use cases, use whichever is more readable.

---

Sorting Data

Sorting is straightforward. By default, it sorts in ascending order (A-Z, 0-9).

Basic Sort

// Sort by price, lowest to highest
const cheapest = products.sort('price');

// Sort by price, highest to lowest (descending)
const mostExpensive = products.sort('price', 'desc');

Multi-Column Sort

Need to sort by Department first, then by Salary? Currently, Molniya optimizes for single-column sorts. To achieve multi-sort, chain them in reverse order of importance.

// Sort by Dept (primary), then Salary (secondary)
const sorted = employees
  .sort('salary', 'desc') // Secondary sort first
  .sort('department');    // Primary sort last (stable sort preserves order)

---

Selecting & Dropping Columns

Sometimes you have too much data (width is too high). Narrow it down to just what you need.

Select

Create a new DataFrame with only the specified columns.

// Create a "contact list" from a massive user table
const contacts = users.select('first_name', 'last_name', 'email');

Drop

Create a new DataFrame without specific columns. Great for removing sensitive data or temporary calculation columns.

// Remove internal IDs and password hashes
const publicView = users.drop('internal_id', 'password_hash');

---

Slicing (Head & Tail)

Quickly grab a chunk of rows from the start or end.

// Top 10 results
df.head(10).print();

// Final 5 entries
df.tail(5).print();

Chaining It All Together

The true power comes from combining these methods into a pipeline.

const report = sales
  .filter(r => r.region === 'EU')   // 1. Filter Region
  .sort('date', false)              // 2. Sort by Date (newest first)
  .drop('internal_code')            // 3. Clean up columns
  .head(5);                         // 4. Take top 5 recent sales

report.print();