Full-text search provides a robust mechanism for querying unstructured text data stored within a database. This technique operates by constructing an inverted index where each significant word is mapped to its occurrences and positions across various documents. When a user initiates a search, the system leverages this optimized index to swiftly pinpoint and retrieve highly relevant documents, significantly enhancing search performance over sequential scanning.

MongoDB introduced native full-text search support starting from version 2.4, and it has since become a standard, integral feature. The platform supports text indexing for a diverse set of languages, ensuring better linguistic accuracy and relevance across different content types. Supported languages include:

• Danish
• Dutch
• English
• Finnish
• French
• German
• Hungarian
• Italian
• Norwegian
• Portuguese
• Romanian
• Russian
• Spanish
• Swedish
• Turkish

Enabling Full-Text Search

For MongoDB installations running version 2.6 or later, full-text search functionality is activated by default, requiring no explicit configuration steps. However, if you are operating an older MongoDB version (specifically 2.4.x), you might need to enable it manually. This can be achieved either by including a parameter during the mongod process startup or by executing an administrative command:

Via mongod Startup

mongod --setParameter textSearchEnabled=true

Via Database Command

db.adminCommand({ setParameter: "textSearchEnabled", value: true })

Creating a Full-Text Index

To enable efficient text queries on your collection, you must first define a text index on the specific field(s) that contain the textual content you wish to search. Let's consider a collection named documents, where each document has a structure similar to this:

{
   "docTitle": "Exploring MongoDB Indexing Techniques",
   "docBody": "This comprehensive document details various indexing strategies for MongoDB, including full-text search.",
   "tags": [ "mongodb", "indexing", "database" ]
}

To enable full-text searching on the docBody field, you would create the index using the createIndex() method:

db.documents.createIndex({ docBody: "text" })

MongoDB also allows creating a wildcard text index, which indexes all string fields within a document:

db.documents.createIndex({ "$**": "text" })

Inspecting Text Indexes

To confirm that your text index has been successfully created or to ascertain its asigned name, you can retrieve a list of all index associated with a collection. Use the getIndexes() method on your collection:

db.documents.getIndexes()

This command will output an array detailling all indexes on the documents collection. The text index will typically appear with a generated name such as docBody_text.

Executing Full-Text Queries

With a text index in place, you can perform full-text searches using the $text query operator in conjunction with the $search operator. For example, to locate documents containing the term "MongoDB" within the indexed docBody field:

db.documents.find({ $text: { $search: "MongoDB" } })

Executing this query would yield results similar to our example document:

{
   "_id" : ObjectId("60c4a1e7b2d3c4e5f6g71890"),
   "docTitle" : "Exploring MongoDB Indexing Techniques",
   "docBody" : "This comprehensive document details various indexing strategies for MongoDB, including full-text search.",
   "tags" : [ "mongodb", "indexing", "database" ]
}

It's worth noting that for very old MongoDB versions (prior to 3.2), full-text searches were performed using the direct text command. This method is now deprecated in favor of the more integrated $text query operator:

db.documents.runCommand("text", { search: "MongoDB" })

Removing a Text Index

Should a text index no longer be required, you can remove it to optimize database resources. First, identify the exact name of the index by using getIndexes():

db.documents.getIndexes()

Assuming the text index on docBody is named docBody_text, you can remove it using the dropIndex() method:

db.documents.dropIndex("docBody_text")

While it's possible to drop all text indexes on a collection with a specific index specification, dropping by the exact name is generally recommended for clarity and safety, especially when managing multiple indexes.