Ruby正则表达式需求：匹配特定GitLab提交消息（排除ed/ing结尾词）

Got it, let's tackle this GitLab commit message regex requirement for Ruby. Here's a solution that fits your needs perfectly:

First, here's the Ruby regex that will match valid commit messages and reject the ones with past-tense (ed ending) or present-participle (ing ending) first words after the colon:

/\A(?:\b(MAINT|STRY-\d+): (\w+(?<!ed|ing))\b(?:\s+|\z))+\z/

If you want to restrict the first word to only letters (no numbers or underscores) for stricter validation, use this version instead:

/\A(?:\b(MAINT|STRY-\d+): ([a-zA-Z]+(?<!ed|ing))\b(?:\s+|\z))+\z/

How It Works #

Let's break down the regex piece by piece to understand why it meets your requirements:

• \A and \z: These anchor the regex to the absolute start and end of the string, ensuring we validate the entire commit message (no partial matches slip through).
• (?: ... )+: A non-capturing group that repeats for each "prefix: content" block in your message (like MAINT: ... or STRY-1: ...).
• \b(MAINT|STRY-\d+)\b:: Matches your valid prefixes with word boundaries (\b) to avoid accidental matches (e.g., MAINTX or STRY-123abc won't be recognized as valid prefixes). The trailing colon matches the required separator.
• (\w+(?<!ed|ing))\b: This is the core validation logic:
  • \w+ matches the first word after the colon (swap with [a-zA-Z]+ if you want only letters).
  • (?<!ed|ing) is a negative lookbehind assertion—it checks that the end of the word does NOT end with ed or ing, blocking the past-tense and present-participle words you want to exclude.
  • \b ensures we're checking a complete word, not just a substring.
• (?:\s+|\z): Matches either whitespace between commit blocks or the end of the string, so multiple blocks are properly separated.

Testing It Out #

Here's how you can implement this regex in Ruby to validate your commit messages:

# Valid message (will return true)
valid_msg = "MAINT: Refactor something STRY-1: Add something STRY-2: Update something"

# Invalid messages (will return false)
invalid_ed_msg = "MAINT: Refactored something STRY-1: Added something STRY-2: Updated something"
invalid_ing_msg = "MAINT: Refactoring something STRY-3: Adding something STRY-4: Updating something"

regex = /\A(?:\b(MAINT|STRY-\d+): (\w+(?<!ed|ing))\b(?:\s+|\z))+\z/

puts regex.match?(valid_msg)       # => true
puts regex.match?(invalid_ed_msg)  # => false
puts regex.match?(invalid_ing_msg) # => false

Customization Tips #

• If you use other prefixes (like FIX, FEAT, or DOC), just add them to the prefix group: \b(MAINT|STRY-\d+|FIX|FEAT|DOC)\b:
• If your first words allow hyphens (e.g., add-user), adjust the word matching part to [\w-]+(?<!ed|ing)—the lookbehind will still correctly check the ending of the word.

内容的提问来源于stack exchange，提问作者Moon