寻求更简洁优雅的R代码实现论文嵌套式前置元数据展示

我正在用Obsidian和R清理数十篇论文的YAML属性，已经用上了dplyr、tidyr、purrr、ymlthis这些包，但处理嵌套列表还是有点棘手。下面是一份论文元数据的嵌套列表（转自YAML）：

li <- list(
    aliases = "Al-Kadem2016",    
    type = "paper",
    title = "Real-Time Estimation of Flow Rate in Dry Gas Wells - A New Field",
    citationKey = "Al-Kadem2016", 
    id = "SPE-182754-MS",
    DOI = "https://doi.org/10.2118/182754-MS",
    ISBN = "",
    volume = "",
    url = "",
    authors = list(
        "Mohammad S. Al-Kadem",
        "Mohammad S. Al Dabbous",
        "Ali S. Al Mashhad",
        "Hassan A. Alsadah",
        "Dhafer Al Shehri"
    ),
    institutions = list(
        '[[Saudi Aramco]]',
        '[[KFUPM]]'
    ),
    published = '2016-04-25',
    conferenceName = "Saudi Arabia Annual Technical Symposium and Exhibition",
    location = "Dammam, Saudi Arabia",
    publisher = "SPE",
    created = '2023-01-26',
    cdow = "Thursday",
    downloaded =  '2023-01-26',
    summary =  "An empirical correlation was developed to calculate the real-time flow rate in dry gas wells at the surface utilizing the most appropriate parameters: upstream flowing wellhead pressure, downstream flowing wellhead pressure, upstream flowing wellhead temperature, and choke size.",
    critique = "Shared with VFM team. Contains **Panhandle choke correlation original and
  modified**. Covers some empirical choke correlations, applicability and limitations.",
    rating = 4,
    pages = 10, 
    references = 5,
    review = "",
    tags = "",
    has_glossary = FALSE,
    cited_by = "",
    up = '[[Choke Correlation]]', 
    down = "",
    related = list(
    '[[Gas Wells]]',
    '[[Atlas/Concepts/Empirical Correlations]]',
    '[[Atlas/Concepts/Flow Rate Estimation]]',
    '[[Atlas/Objects/Venturi Flowmeter]]',
    '[[Atlas/Objects/Plant Information]]'        
            ),
    research = "CHK",
    notes = "what-makes(iiii) + doi + authors + tags(iiii) + snips(iiii) + view + defs(iiiii)",
    cover = '[[poster-20231026193936.png]]',
    poster = '[[poster-20231026193936.png]]',
    pdf_annotated = FALSE,
    zotero_up = FALSE,
    zotero_notes = FALSE,
    zotero_highlights = FALSE,
    uuid = '20230126065407',
    added = '2023-01-01',
    pdf_attached = "spe-000000.pdf",
    pdf_highlights = FALSE
)

我自己写了一段双层循环的代码，能输出符合需求的键值对结果，但代码不够简洁高效，想找更优雅的实现方案，核心是学习嵌套列表的遍历技术，不用ymlthis::as_yml进行格式化。原代码如下：

# get values from frontmatter
fm_pluck <- li

for(i in 1:length(fm_pluck)) {
    element  <- fm_pluck[i]
    key      <- names(element)
    values   <- unlist(element)
    n_values <- length(values)
    cat(sprintf("%2d %2d %-15s", i, n_values, key))
    if (n_values > 1) {
        cat("\n")
        for (v in 1:n_values) {
            cat(sprintf("%20d %-12s \n", v, values[v]))    
        }
        
    } else {
        cat(sprintf("%-20s \n", values))
    }
}

优化方案：用purrr实现函数式遍历 #

利用你已经在使用的purrr包，可以用函数式编程的方式替代显式循环，代码更简洁易读。

方法1：iwalk+匿名函数

iwalk是purrr中专门用于遍历列表并执行副作用（比如打印输出）的函数，会自动传入元素值、键和索引：

library(purrr)

iwalk(li, function(val, key, idx) {
  vals <- unlist(val)
  n_vals <- length(vals)
  
  # 打印键的头部信息
  cat(sprintf("%2d %2d %-15s", idx, n_vals, key))
  
  if (n_vals > 1) {
    cat("\n")
    # 遍历多值项，用walk2同时处理索引和值
    walk2(seq_len(n_vals), vals, ~cat(sprintf("%20d %-12s \n", .x, .y)))
  } else {
    cat(sprintf("%-20s \n", vals))
  }
})

方法2：紧凑版匿名函数

如果不需要复用处理逻辑，可以把代码写得更紧凑：

library(purrr)

iwalk(li, ~{
  vals <- unlist(.x)
  n <- length(vals)
  cat(sprintf("%2d %2d %-15s", .y, n, names(.x)))
  if (n > 1) {
    cat("\n")
    walk2(seq_len(n), vals, ~cat(sprintf("%20d %-12s \n", ..1, ..2)))
  } else {
    cat(sprintf("%-20s \n", vals))
  }
})

方法3：拆分独立函数（便于复用）

如果需要在多个地方复用处理逻辑，把打印逻辑拆成独立函数更清晰：

library(purrr)

print_key_value <- function(val, key, idx) {
  vals <- unlist(val)
  n_vals <- length(vals)
  
  cat(sprintf("%2d %2d %-15s", idx, n_vals, key))
  
  if (n_vals > 1) {
    cat("\n")
    walk2(seq_len(n_vals), vals, ~cat(sprintf("%20d %-12s \n", .x, .y)))
  } else {
    cat(sprintf("%-20s \n", vals))
  }
}

# 调用函数遍历列表
iwalk(li, print_key_value)

这些方案都避免了显式的双层循环，利用purrr的函数式工具实现遍历，代码更符合tidyverse生态的风格，也更易维护和扩展。

内容的提问来源于stack exchange，提问作者f0nzie