如何移除字符串中连续大写单词（含后续标点）并保留指定单词？

Got it, let's fix this problem properly. The key here is to target only sequences of 2+ pure uppercase words (and any trailing punctuation after the last word in the sequence), while leaving single uppercase words and mixed-case/number words untouched.

Step 1: The Right Regular Expression #

We need a regex pattern that specifically matches consecutive uppercase word clusters. Here's the pattern we'll use:

pattern <- "\\b[:upper:]+(\\s+[:upper:]+)+\\W*"

Let's break down what each part does:

• \\b[:upper:]+: Matches a single pure uppercase word (the word boundary \\b ensures we don't match partial words)
• (\\s+[:upper:]+)+: This captures one or more instances of "space + uppercase word" — this is what ensures we're targeting groups of 2 or more consecutive uppercase words
• \\W*: Matches any trailing non-word characters (like ?, ., !) that come right after the final uppercase word in the cluster, so we can remove those too

Step 2: Implement the Fix #

Let's apply this to your example string using stringr::str_remove_all():

library(stringr)

string <- "Lorem ipsum DOLOR SIT AMET? consectetuer adipiscing elit. Morbi gravida libero NEC velit. Morbi scelerisque luctus velit. ETIAM-123 dui sem, fermentum vitae, SAGITTIS ID? malesuada in, quam. Proin mattis lacinia justo. Vestibulum facilisis auctor urna. Aliquam IN LOREM SIT amet leo accumsan"

clean_string <- str_remove_all(string, pattern)

# Check the result
cat(clean_string)

Step 3: Verify the Result #

Running the code above will give you this cleaned string:

Lorem ipsum consectetuer adipiscing elit. Morbi gravida libero NEC velit. Morbi scelerisque luctus velit. ETIAM-123 dui sem, fermentum vitae, malesuada in, quam. Proin mattis lacinia justo. Vestibulum facilisis auctor urna. Aliquam amet leo accumsan

Notice that:

• The clusters DOLOR SIT AMET?, SAGITTIS ID?, and IN LOREM SIT are all removed (including the trailing ? in the first two)
• Single uppercase word NEC is preserved
• Mixed word ETIAM-123 is completely preserved (since it's not a pure uppercase word, our regex doesn't touch it)

Why Your Original Code Failed #

Your original pattern \\b[:upper:]+\\b matches all pure uppercase words, regardless of whether they're alone or in a group. That's why it removed NEC and chopped ETIAM-123 down to -123 — it only matched the ETIAM part. Our targeted regex avoids this by only going after clusters of 2+ pure uppercase words.

内容的提问来源于stack exchange，提问作者gebo-aa