如何在R中高效复用重复代码？寻求替代<<-的规范方案

我在R中合并两个数据集（爬取的科研作者机构隶属数据与Web of Science数据），由于ID变量格式混乱，需要多次执行「处理ID变量→合并→处理ID变量→合并」的流程。每次ID处理的逻辑各不相同，但合并环节的代码完全重复。我之前用<<-定义了全局赋值的函数来复用代码，但听说这是不良编码实践，想找不用<<-的替代方案，且因为任务特性无法用循环统一处理不同的ID操作。

原重复合并代码 #

b_suc <- merged_b[is.na(merged_b$cwts_organization_id)==F, ]
b_suc <- subset(b_suc, select = c(contribs, cwts_organization_id))
merged <- merge(merged, b_suc, by="contribs", all.x=T)
merged$cwts_organization_id.x <- 
  ifelse(is.na(merged$cwts_organization_id.x)==T & is.na(merged$cwts_organization_id.y)==F,
         merged$cwts_organization_id.y,
         merged$cwts_organization_id.x)
merged <- rename(merged, cwts_organization_id = cwts_organization_id.x)
merged <- subset(merged, select = -c(cwts_organization_id.y))

succes <- c(succes, sum(is.na(merged$cwts_organization_id)==F))
unsuc <- merged[is.na(merged$cwts_organization_id)==T, ]
unsuc <- subset(unsuc, select = -c(cwts_organization_id, short_name, city, id))

原全局赋值函数（使用<<-） #

append_end <- function(){
  b_suc <<- merged_b[is.na(merged_b$cwts_organization_id)==F, ]
  b_suc <<- subset(b_suc, select = c(contribs, cwts_organization_id))
  merged <<- merge(merged, b_suc, by="contribs", all.x=T)
  merged$cwts_organization_id.x <<- 
    ifelse(is.na(merged$cwts_organization_id.x)==T & is.na(merged$cwts_organization_id.y)==F,
           merged$cwts_organization_id.y,
           merged$cwts_organization_id.x)
  merged <<- rename(merged, cwts_organization_id = cwts_organization_id.x)
  merged <<- subset(merged, select = -c(cwts_organization_id.y))
  
  succes <<- c(succes, sum(is.na(merged$cwts_organization_id)==F))
  unsuc <<- merged[is.na(merged$cwts_organization_id)==T, ]
  unsuc <<- subset(unsuc, select = -c(cwts_organization_id, short_name, city))
}

替代方案：函数返回值+显式赋值 #

核心思路是让函数接收需要处理的输入数据，把所有处理后的结果打包成列表返回，然后在全局环境中显式更新变量，完全规避隐式全局赋值的问题。

改写后的函数 #

process_merge <- function(merged_current, merged_b_current, succes_current) {
  # 筛选出有有效机构ID的merged_b子集
  b_suc <- merged_b_current[!is.na(merged_b_current$cwts_organization_id), ]
  b_suc <- b_suc[, c("contribs", "cwts_organization_id"), drop = FALSE]
  
  # 执行左连接合并
  merged_new <- merge(merged_current, b_suc, by = "contribs", all.x = TRUE)
  
  # 用右侧的机构ID填充左侧的缺失值
  merged_new$cwts_organization_id.x <- ifelse(
    is.na(merged_new$cwts_organization_id.x) & !is.na(merged_new$cwts_organization_id.y),
    merged_new$cwts_organization_id.y,
    merged_new$cwts_organization_id.x
  )
  
  # 重命名列并移除多余的y列
  merged_new <- dplyr::rename(merged_new, cwts_organization_id = cwts_organization_id.x)
  merged_new <- merged_new[, !names(merged_new) %in% "cwts_organization_id.y", drop = FALSE]
  
  # 更新成功匹配的计数
  succes_new <- c(succes_current, sum(!is.na(merged_new$cwts_organization_id)))
  
  # 提取未匹配成功的记录子集
  unsuc_new <- merged_new[is.na(merged_new$cwts_organization_id), ]
  unsuc_new <- unsuc_new[, !names(unsuc_new) %in% c("cwts_organization_id", "short_name", "city", "id"), drop = FALSE]
  
  # 返回所有处理后的结果
  return(list(
    merged = merged_new,
    succes = succes_new,
    unsuc = unsuc_new
  ))
}

使用方式 #

每次处理完ID变量后，调用函数并显式更新全局变量：

# 调用函数，传入当前的merged、merged_b、succes
merge_result <- process_merge(merged, merged_b, succes)
# 依次更新全局变量
merged <- merge_result$merged
succes <- merge_result$succes
unsuc <- merge_result$unsuc

方案优势 #

1. 彻底抛弃<<-的隐式全局赋值，代码逻辑透明，调试时能清晰追踪变量变化
2. 函数的输出完全由输入决定，没有副作用，便于单独测试某一步的合并逻辑
3. 显式赋值的方式让全局变量的更新过程一目了然，避免意外修改其他变量

内容的提问来源于stack exchange，提问作者EmilA