Got it, let's tackle this problem. The core requirement is to replace every standalone single digit (0-9) with its English word equivalent, while leaving multi-digit numbers untouched. We also need to handle digits at the start or end of a line correctly.

The Sed Command #

Here's the complete command that does exactly what you need:

sed -E -e 's/(^|[^0-9])0([^0-9]|$)/\1zero\2/g' \
       -e 's/(^|[^0-9])1([^0-9]|$)/\1one\2/g' \
       -e 's/(^|[^0-9])2([^0-9]|$)/\1two\2/g' \
       -e 's/(^|[^0-9])3([^0-9]|$)/\1three\2/g' \
       -e 's/(^|[^0-9])4([^0-9]|$)/\1four\2/g' \
       -e 's/(^|[^0-9])5([^0-9]|$)/\1five\2/g' \
       -e 's/(^|[^0-9])6([^0-9]|$)/\1six\2/g' \
       -e 's/(^|[^0-9])7([^0-9]|$)/\1seven\2/g' \
       -e 's/(^|[^0-9])8([^0-9]|$)/\1eight\2/g' \
       -e 's/(^|[^0-9])9([^0-9]|$)/\1nine\2/g' input.txt

How It Works #

Let's break down the logic behind each replacement rule:

• -E: Enables extended regular expressions, so we don't have to escape parentheses (()) which makes the pattern cleaner.
• The regex pattern (^|[^0-9]) matches either the start of a line (^) or a non-digit character ([^0-9]). This ensures we don't match the first digit of a multi-number (like the "2" in "23").
• The middle part ([0-9]) targets the single digit we want to replace.
• The final part ([^0-9]|$) matches either a non-digit character or the end of a line ($), ensuring we don't match the last digit of a multi-number (like the "3" in "23").
• \1 and \3 preserve the characters before and after the digit (if any), so we don't mess up the surrounding text. We just swap the digit in \2 with its English equivalent.

Testing with Your Example #

If you run the command on your input string:

5 word 23 word 1 word

You'll get the expected output:

five word 23 word one word

Edge Cases Handled #

This command works for all edge scenarios you might encounter:

• Line-start digits: 3 hello → three hello
• Line-end digits: hello 7 → hello seven
• Isolated single-digit lines: 9 → nine
• Digits adjacent to non-alphanumeric characters: !4? → !four?

Note: If you're using BSD sed (common on macOS), replace -E with -r—the rest of the command stays the same.

内容的提问来源于stack exchange，提问作者Hilbert Erwin