MongoDB - Text Search

MongoDB - Full-Text Search

MongoDB provides powerful full-text search capabilities through text indexes. This feature allows you to efficiently search for words and phrases within string content. It includes case-insensitivity, support for multiple languages, and stemming (e.g., searching for ‘running’ also finds ‘run’).

Step 1: Create a Text Index

First, you need to define a text index on the field(s) you want to search. Let’s consider a collection of blog posts.

// Insert sample data
db.posts.insertMany([
   {
      "_id": 1,
      "content": "An introduction to the powerful features of MongoDB.",
      "tags": ["mongodb", "database", "nosql"]
   },
   {
      "_id": 2,
      "content": "MongoDB text search is a powerful tool for developers.",
      "tags": ["search", "full-text", "mongodb"]
   },
   {
      "_id": 3,
      "content": "We are exploring advanced database indexing techniques.",
      "tags": ["database", "performance"]
   }
]);

// Create a text index on the 'content' field
db.posts.createIndex({ content: "text" });

// You can also index multiple fields
// db.posts.createIndex({ content: "text", tags: "text" });

The index will be created in the background. MongoDB will parse the content, drop common stop words (like ‘a’, ‘the’, ‘is’), and stem words to their root form.

Step 2: Use the Text Index for Searching

To search using the index, you use the $text operator with $search.

Basic Word Search

This query finds all posts containing the word ‘mongodb’ or ‘powerful’.

// Search for posts containing 'mongodb' or 'powerful'
db.posts.find({ $text: { $search: "mongodb powerful" } });

Phrase Search

To search for an exact phrase, enclose it in double quotes.

// Search for the exact phrase "powerful tool"
db.posts.find({ $text: { $search: '"powerful tool"' } });

Excluding Words

You can exclude results containing a specific word by prefixing it with a hyphen -.

// Search for 'database' but not 'nosql'
db.posts.find({ $text: { $search: "database -nosql" } });

Ranking Search Results

A key feature of text search is the ability to rank results by relevance. You can access the relevance score using $meta: "textScore" and sort by it.

// Find posts about 'mongodb' and sort them by relevance
db.posts.find(
   { $text: { $search: "mongodb" } },
   { score: { $meta: "textScore" } } // Project the score into a 'score' field
).sort(
   { score: { $meta: "textScore" } } // Sort by the calculated text score
);

This query will return documents that match ‘mongodb’ best at the top of the result set.

Dropping a Text Index

If you no longer need the index, you can drop it by its name. First, find the index name using getIndexes().

> db.posts.getIndexes()
// The text index name is typically fieldName_text, e.g., 'content_text'

> db.posts.dropIndex("content_text")
{
    "nIndexesWas": 2,
    "ok": 1
}