如何精确匹配含特殊字符的Git提交信息并获取提交哈希

To write a Bash function get_commits_by_message() that returns commit hashes where the entire commit message exactly matches your input (even when it contains special characters, spaces, or newlines), here's a robust implementation:

get_commits_by_message() {
    local target_message="$1"
    
    # Use git log to output each commit's hash and full message, separated by a unique delimiter
    git log --all --format="%H%n%B%n===COMMIT_BOUNDARY===" | awk -v target="$target_message" '
        BEGIN {
            current_hash = ""
            current_msg = ""
            in_msg = 0
        }
        $0 == "===COMMIT_BOUNDARY===" {
            # Compare accumulated message to target (exact match)
            if (current_msg == target) {
                print current_hash
            }
            # Reset for next commit
            current_hash = ""
            current_msg = ""
            in_msg = 0
            next
        }
        !in_msg {
            current_hash = $0
            in_msg = 1
            next
        }
        in_msg {
            # Append lines to message, preserving newlines
            current_msg = current_msg (current_msg != "" ? "\n" : "") $0
        }
    '
}

How It Works #

1. Git Log Formatting: We use git log --all --format="%H%n%B%n===COMMIT_BOUNDARY===" to output:

   • %H: The full commit hash
   • %B: The complete commit message (including all lines, not just the subject)
   • A unique delimiter (===COMMIT_BOUNDARY===) to separate each commit's data. Pick something that won't appear in any of your commit messages.

2. Awk Processing: The awk script parses the git log output:

   • Tracks the current commit hash and accumulates every line of the message.
   • When it hits the delimiter, it checks if the accumulated message exactly matches your target input.
   • If there's a match, it prints the commit hash.

Alternative: Simpler (But Slower for Large Repos) #

If you prefer a more straightforward approach (easier to debug, but less efficient for repos with thousands of commits), you can loop through each commit and compare messages directly:

get_commits_by_message() {
    local target_message="$1"
    
    git rev-list --all | while read -r commit_hash; do
        # Extract the full commit message (skip the commit header)
        local commit_msg=$(git cat-file commit "$commit_hash" | sed '1,/^$/d')
        
        # Exact string comparison (handles newlines and special characters)
        if [ "$commit_msg" = "$target_message" ]; then
            echo "$commit_hash"
        fi
    done
}

Key Notes #

• Multi-Line Messages: Both implementations correctly handle multi-line commit messages, as they preserve and compare all lines of the message.
• Special Characters: By using exact string comparison (instead of regex), we avoid issues with special characters like *, ?, [], or $ that would break regex-based searches.
• Trailing Newlines: Git typically appends a trailing newline to commit messages. If your target message doesn't include this newline, the comparison may fail. To fix this, you can add a trailing newline to target_message (e.g., target_message="$1"$'\n').
• Scope: Both scripts use --all to search all branches. If you want to limit to a specific branch, replace --all with the branch name (e.g., main).

内容的提问来源于stack exchange，提问作者Marvin Irwin