Remove lines matching regex

Remove Lines Matching Regex filters multiline text by testing each individual line against a regular expression and discarding every match. It is useful when the lines you need to remove share a pattern rather than an exact value: a log level, generated comment, timestamp, identifier prefix, or structured record type. The result preserves every retained line exactly, including its original line ending, and reports how many lines were removed and kept so automated workflows can verify what changed.

Write a pattern for the complete line you want to remove

Provide the source in text and a JavaScript regular expression in pattern. Write the expression itself without slash delimiters: use <code>^DEBUG\b</code>, for example, rather than <code>/^DEBUG\b/</code>. Each line is tested separately, and a line is removed when any part of it matches.

Add <code>^</code> when the relevant text must occur at the beginning, <code>$</code> when it must occur at the end, and both anchors when the entire line must fit the rule. A pattern such as <code>^\s*#</code> removes lines whose first non-whitespace character is a hash, while <code>temporary$</code> removes lines ending with that word. Matching is case-sensitive and uses standard JavaScript regular-expression syntax.

Start with a narrow expression, inspect the counts and filtered text, then broaden it only when the result confirms that the intended lines were selected.

Understand exactly what happens to line endings and empty lines

The filter recognizes Unix line feeds, Windows carriage-return plus line-feed pairs, and standalone carriage returns. It retains the original ending attached to every line that remains, which means mixed or platform-specific input is not silently normalized. The regular expression is tested against the content of a line only; newline characters are not included in the match.

Empty lines are real lines and can be removed with a pattern such as <code>^$</code>. If the source ends with a newline, that delimiter remains attached to the preceding retained line unless that line itself is removed. Consequently, removing the last unterminated line can leave the earlier trailing delimiter intact, faithfully reflecting the bytes associated with the retained line.

An empty source is treated as one empty line, so <code>^$</code> removes it and returns zero kept lines; a nonmatching expression keeps it. No trimming, case conversion, or whitespace cleanup occurs.

Use deterministic filtering safely in logs, code, and pipelines

This capability is well suited to preprocessing data before comparison, indexing, review, or import. In logs, remove known health checks or a noisy severity prefix before sending the meaningful events to an analyst. In generated code, discard stable banner comments before comparing builds.

In structured exports, remove records whose line prefix identifies a header, trailer, or unwanted category. Because the same deterministic algorithm runs without network access, repeated calls with identical input produce identical output. The expression is compiled once and then applied to each line, but regular expressions can still become computationally expensive when they contain ambiguous nested repetition.

Prefer anchored, specific patterns for large inputs, and avoid expressions known to cause excessive backtracking. Review the returned text when designing a rule, and use the counts as a guard in automation—for example, stop a deployment when a cleanup intended to remove one generated line reports hundreds.