Build multi-tenant RAG with Neon's database-per-user model — no nosy neighbors, max isolation, minimal costs
Docs/Neon Postgres guides/Functions/String functions/regexp_replace

Postgres regexp_replace() function

Replace substrings matching a regular expression pattern

The Postgres regexp_replace() function replaces substrings that match a regular expression pattern with the specified replacement string.

This function is particularly useful for complex string manipulations, and data cleaning/formatting tasks. Consider scenarios where you'd want to remove or replace specific patterns in text or transform data to meet certain requirements. For instance, you might use it to format phone numbers consistently, remove HTML tags from text, or anonymize sensitive information in logs.

Try it on Neon!

Neon is Serverless Postgres built for the cloud. Explore Postgres features and functions in our user-friendly SQL editor. Sign up for a free account to get started.

Sign Up

Function signature

The regexp_replace() function has the following syntax:

regexp_replace(source text, pattern text, replacement text [, flags text]) -> text
  • source: The input string to perform replacements on.
  • pattern: The regular expression pattern to match.
  • replacement: The string to replace matched substrings with.
  • flags (optional): A string of one or more single-letter flags that modify how the regex is interpreted.

It returns the input string with occurrence(s) of the pattern replaced by the replacement string.

More recent versions of Postgres (starting with Postgres 16) also support additional parameters to further control the replacement operation:

regexp_replace(source text, pattern text, replacement text [, start int, [, N int]] [, flags text]) -> text
  • start: The position in the source string to start searching for matches (default is 1).
  • N: If specified, only the Nth occurrence of the pattern is replaced. If N is 0, or the g flag is used, all occurrences are replaced.

Example usage

Consider a customer_data table with a phone_number column containing phone numbers in different formats. We can use regexp_replace() to standardize these numbers to a consistent format.

WITH customer_data AS (
  SELECT '(555) 123-4567' AS phone_number
  UNION ALL
  SELECT '555.987.6543' AS phone_number
  UNION ALL
  SELECT '555-321-7890' AS phone_number
)
SELECT
  phone_number AS original_number,
  regexp_replace(phone_number, '[^\d]', '', 'g') AS cleaned_number
FROM customer_data;

This query removes all non-digit characters from the phone numbers, standardizing them to a simple string of digits.

original_number | cleaned_number
-----------------+----------------
 (555) 123-4567  | 5551234567
 555.987.6543    | 5559876543
 555-321-7890    | 5553217890
(3 rows)

Advanced examples

Use regexp_replace() with backreferences

You can use backreferences in the replacement string to include parts of the matched pattern in the replacement.

WITH log_data AS (
  SELECT '2023-05-15 10:30:00 - User john.doe@example.com logged in' AS log_entry
  UNION ALL
  SELECT '2023-05-15 11:45:30 - User jane.smith@example.org logged out' AS log_entry
)
SELECT
  log_entry AS original_log,
  regexp_replace(log_entry, '(.*) - User (.+@.+) (.+)$', '\1 - User [REDACTED] \3') AS anonymized_log
FROM log_data;

This query anonymizes email addresses in log entries by replacing them with [REDACTED] while preserving the rest of the log structure.

original_log                         |              anonymized_log
--------------------------------------------------------------+-------------------------------------------
 2023-05-15 10:30:00 - User john.doe@example.com logged in    | 2023-05-15 10:30:00 - User [REDACTED] in
 2023-05-15 11:45:30 - User jane.smith@example.org logged out | 2023-05-15 11:45:30 - User [REDACTED] out
(2 rows)

Modify the behavior of regexp_replace() using flags

The flags parameter allows you to modify how the function operates. Common flags include:

  • g: Global replacement (replace all occurrences)
  • i: Case-insensitive matching
  • n: Newline-sensitive matching
WITH product_descriptions AS (
  SELECT 'Red Apple: sweet and crisp' AS description
  UNION ALL
  SELECT 'Green Apple: tart and juicy apple' AS description
  UNION ALL
  SELECT 'Yellow Apple: mild and sweet' AS description
)
SELECT
  description AS original_description,
  regexp_replace(description, 'apple', 'pear', 'gi') AS modified_description
FROM product_descriptions;

This query replaces all occurrences of "apple" (case-insensitive) with "pear" in the product descriptions.

original_description        |      modified_description
-----------------------------------+---------------------------------
 Red Apple: sweet and crisp        | Red pear: sweet and crisp
 Green Apple: tart and juicy apple | Green pear: tart and juicy pear
 Yellow Apple: mild and sweet      | Yellow pear: mild and sweet
(3 rows)

Use regexp_replace() for complex pattern matching and replacement

regexp_replace() can handle complex patterns for sophisticated text processing tasks. For example, the query below removes all HTML tags from the given markup, producing plain text.

WITH html_content AS (
  SELECT '<p>This is <b>bold</b> and <i>italic</i> text.</p>' AS content
  UNION ALL
  SELECT '<div>Another <span style="color: red;">example</span> here.</div>' AS content
)
SELECT
  content AS original_html,
  regexp_replace(content, '<[^>]+>', '', 'g') AS plain_text
FROM html_content;

This query produces the following output:

original_html                           |          plain_text
-------------------------------------------------------------------+-------------------------------
 <p>This is <b>bold</b> and <i>italic</i> text.</p>                | This is bold and italic text.
 <div>Another <span style="color: red;">example</span> here.</div> | Another example here.
(2 rows)

Additional considerations

Performance implications

While regexp_replace() is powerful, complex regular expressions or operations on large text fields can be computationally expensive. For frequently used operations, consider preprocessing the data or using simpler string functions if possible.

Alternative functions

  • replace(): A simpler function for straightforward string replacements without regular expressions.
  • translate(): Useful for character-by-character replacements.
  • regexp_matches(): Returns an array of all substrings matching a regular expression pattern, which can be useful in conjunction with other functions for complex transformations.

Resources

Last updated on

Was this page helpful?