为何stringr::str_extract对某字符向量始终返回NA？

尝试用str_extract从世界贸易组织（WTO）网站抓取的数据中提取日期时，目标字符向量始终返回NA，但手动输入相同字符串却能正常提取。具体代码及运行结果如下：

> country_comparison$status[1:10]
 [1] "Settled or terminated (withdrawn, mutually agreed solution) on 29 March 1995" "Implementation notified by respondent on 25 September 1997"                  
 [3] "In consultations on 4 April 1995"                                             "Implementation notified by respondent on 25 September 1997"                  
 [5] "Settled or terminated (withdrawn, mutually agreed solution) on 20 July 1995"  "Settled or terminated (withdrawn, mutually agreed solution) on 19 July 1995" 
 [7] "Settled or terminated (withdrawn, mutually agreed solution) on 5 July 1996"   "Mutually acceptable solution on implementation notified on 9 January 1998"   
 [9] "Panel established, but not yet composed on 11 October 1995"                   "Mutually acceptable solution on implementation notified on 9 January 1998"   

> country_comparison$status[1:10] %>% str_extract(pattern = "[0-9]{1,2} [A-Za-z]+ [0-9]{4}")
 [1] NA NA NA NA NA NA NA NA NA NA

> c("Settled or terminated (withdrawn, mutually agreed solution) on 29 March 1995", "Implementation notified by respondent on 25 September 1997") %>% str_extract(pattern = "[0-9]{1,2} [A-Za-z]+ [0-9]{4}")
[1] "29 March 1995"     "25 September 1997"

核心原因分析 #

• 非标准空格字符：网页抓取的文本中，日期前后的空格大概率不是普通半角空格（ ，ASCII码32），而是网页常用的不间断空格（ ，ASCII码160）或全角空格。你的正则表达式用的是普通空格，无法匹配这些特殊空格，导致返回NA；而手动输入的字符串用的是普通空格，所以匹配成功。

• 隐藏不可见字符：网页文本可能夹杂零宽空格、控制字符等不可见内容，这些字符夹在数字、月份、年份之间，破坏了正则的匹配逻辑，但控制台打印时无法察觉，看起来和手动输入的字符串一致。

验证与解决方法 #

1. 检查特殊字符：用charToRaw()对比抓取字符串和手动输入字符串的原始字节，就能定位差异：

   # 查看抓取字符串的原始字节
   charToRaw(country_comparison$status[1])
   # 查看手动输入字符串的原始字节
   charToRaw("Settled or terminated (withdrawn, mutually agreed solution) on 29 March 1995")
   

2. 修改正则兼容空白字符：把正则里的普通空格换成\\s（匹配所有空白字符，包括特殊空格、制表符等）：

   country_comparison$status[1:10] %>% str_extract(pattern = "[0-9]{1,2}\\s[A-Za-z]+\\s[0-9]{4}")
   

   或者直接指定匹配普通空格和不间断空格：

   country_comparison$status[1:10] %>% str_extract(pattern = "[0-9]{1,2}[ \\xA0][A-Za-z]+[ \\xA0][0-9]{4}")
   

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