R中使用dplyr按v/w任意顺序匹配，基于另一dataframe更新e列

实现思路 #

由于匹配规则不区分v、w两列的顺序，我们可以先为两张表的每行生成统一排序后的匹配键：将每行的v和w值按字典序排序后拼接成字符串，作为两张表的关联依据。关联后再根据箭头类型、行列顺序调整最终的e值即可。

完整实现代码 #

library(dplyr)

# 构造示例数据（真实使用时可替换为你自己的数据源）
t <- tibble(
  v = c("dave", "floria", "floria", "floria", "quinn"),
  w = c("adam", "adam", "dave", "quinn", "adam"),
  e = c("->", "->", "->", "->", "->")
)
data <- tibble(
  v = c("dave", "quinn"),
  w = c("floria", "adam"),
  e = c("->", "<->")
)

# 1. 预处理参考表data，生成匹配键，重名字段避免关联时冲突
data_processed <- data %>%
  rowwise() %>%
  mutate(match_key = paste(sort(c(v, w)), collapse = "|")) %>%
  ungroup() %>%
  rename(data_v = v, data_w = w, data_e = e)

# 2. 处理主表t，关联后更新e列
result <- t %>%
  rowwise() %>%
  mutate(match_key = paste(sort(c(v, w)), collapse = "|")) %>%
  ungroup() %>%
  left_join(data_processed, by = "match_key") %>%
  mutate(
    e = case_when(
      # 未匹配到的行保留原e值
      is.na(data_e) ~ e,
      # 双向箭头不区分顺序直接复用
      data_e == "<->" ~ data_e,
      # 单向箭头根据列顺序调整方向
      data_e == "->" ~ case_when(
        v == data_v & w == data_w ~ "->",
        v == data_w & w == data_v ~ "<-"
      )
    )
  ) %>%
  # 清理多余字段，保留原表结构
  select(v, w, e)

输出结果 #

执行代码后得到的result和预期输出完全一致：

# A tibble: 5 × 3
  v      w     e    
  <chr>  <chr> <chr>
1 dave   adam  ->   
2 floria adam  ->   
3 floria dave  <-   
4 floria quinn ->   
5 quinn  adam  <-> 

内容的提问来源于stack exchange，提问作者bananaboy