如何在PDF的kableExtra表格中显示竖线（pipe）字符？

问题背景 #

我复现了一篇数据验证教程的步骤，使用flextable展示验证错误时，在R Notebook或HTML预览中显示正常，但编织为PDF时效果极差。于是转向kable/kableExtra生成紧凑表格，却发现验证规则中的竖线（|）字符（例如表达式(DBP == 0 | SBP ==0) & Outcome == 'Survived'），在经过kableExtra样式处理后，PDF中会显示为|。

使用环境版本 #

• R 4.3.2
• dplyr 1.1.4
• Flextable 0.9.4
• kableExtra 1.4.0
• validate 1.1.5

最小可复现示例（MWE） #

---
title: "kableExtra vs flextable"
output:
   pdf_document:
     toc: true
     number_sections: true
   html_notebook:
     toc: true
     number_sections: true
---

# Load Libraries
library(dplyr)
library(validate)
library(flextable)
library(kableExtra)

# Create dataset
myData <- data.frame(PatientID = c('P001', 'P002', 'P003' , NA, 'P005', 'P006', 'P007', 'P008', 
                                    'P009', 'P010', 'P008', 'P11'), 
 Age = c(23, 12, 5, 8, 245, NA, 23, 45, 87, 121, 56, 130), 
 Outcome = c('Died',  'Died', NA, 'Survived', 'Survived', 'Survived', 'Survived',
                               'Survived', 'Survived', '?', 'Survived', 'Unknown'), 
 SBP = c(0, 0, 120, 80, 45, 67, 100, 130, 350, 120, 46, 120), 
 DBP = c(0, 0, 80, 70, 30, 40, 80, 210, NA, 80, 0, 80)
                  )

# Add an id field in the data. 
myData <- myData %>%
 dplyr::mutate(id  = row_number())

# Create validation rules
myrules <- validator( "Patient Id Unique"   = is_unique(PatientID),
                       "Patient ID"          = is.na(PatientID),
                       "Age in range"        = Age >= 0 & Age <= 120,
                       "Outcome validity "   = Outcome %in% c('Survived', 'Died'),
                       "SBP in range"        = SBP >= 0 & SBP <= 300,
                       "DBP in range"        = in_range (DBP, min = 0 , max =  200),
                       "Survived with No BP" = (DBP == 0 | SBP ==0)  & Outcome == 'Survived' )

# Create validation results with the confront function
output <- confront(myData, myrules, key  ="id")

# Show errors in table
# Convert our output into a dataframe first
dout <- as.data.frame(output)

# Errors for each record id
dErrors <- dout %>%
   dplyr::filter(! value == TRUE)%>%
   dplyr::select(id, name, expression)%>%
   dplyr::arrange(id)

# Display validation results in flextable
# This part works well in a R Notebook or HTML, but not in PDF.
ft <- flextable(dErrors)%>%
   theme_booktabs()%>%
   merge_v(j = ~id)%>%
   set_header_labels(name       = "Error",
                     id         =  "Record ID",
                     expression = "Validation expression")
ft

# Try to reproduce the flextable with kable/kableExtra
# This produces a more compact table but unfortunately without the pipe character
dout %>% 
   filter(!value==TRUE) %>% 
   select(id, name, expression) %>% 
   arrange(id) %>% 
   kable(booktabs = TRUE, escape = TRUE) %>% 
   kable_classic(full_width = FALSE) %>% 
   column_spec(3, width = "10cm") %>% 
   collapse_rows(columns = 1, valign = "top")

异常现象排查 #

• 仅使用kable不经过kableExtra处理时，竖线符号能正常显示在PDF表格中：

  dout %>% 
    filter(!value==TRUE) %>% 
    select(id, name, expression) %>% 
    arrange(id) %>% 
    kable(booktabs = TRUE, escape = TRUE)
  

• 但只要添加任何kableExtra样式（如collapse_rows()或kable_styling()），竖线符号就会变为|
• 尝试过向kable传入format = 'latex'或escape = TRUE，均无法解决问题；更新R及相关包也无效

其他尝试（未解决） #

尝试修复flextable的PDF显示问题：

• 使用ft <- fit_to_width(ft, max_width = 15, unit = "cm")调整宽度，对PDF中的表格无影响，文本仍拥挤
• 在YAML头部指定latex_engine: xelatex，也无效果

需求 #

我更倾向于使用kableExtra，请问如何在使用kableExtra时让竖线（|）字符在PDF中正常显示？

内容的提问来源于stack exchange，提问作者Ouburg