Selectors

Selectors are the mechanism flextable uses to target specific cells when applying formatting, styling, or content operations. Nearly every formatting function — bold(), color(), bg(), padding(), align(), hline(), and more — accepts three selector arguments: i, j, and part. When all three default to NULL (rows) or their default string, the operation applies to the entire part.

The `part` argument

The part argument determines which section of the table is targeted.

part = "body"    # data rows (default for most functions)
part = "header"  # header rows
part = "footer"  # footer rows
part = "all"     # all three parts

ft <- flextable(head(mtcars)) |>
  bold(part = "header") |>        # bold every cell in the header
  bg(bg = "#FFFDE7", part = "all") # yellow tint across header, body, footer

When part = "all", the operation is applied to each part independently. Formula row selectors (i = ~ condition) cannot be used with part = "all", "header", or "footer" — these parts store only character representations of values, so numeric conditions cannot be evaluated on them.

The `i` argument — row selector

The i argument selects which rows within a part to target. When i = NULL (the default), all rows in the part are selected.

Formula (body only)
Integer index
Logical vector

Use a one-sided formula to filter rows by a condition evaluated against the body dataset. The expression uses column names from the original data.frame.

# Highlight rows where mpg is less than 22
ft <- flextable(head(mtcars, 10)) |>
  highlight(i = ~ mpg < 22, j = "mpg", color = "#ffe842")

# Color rows where a column equals a specific value
ft <- flextable(head(airquality, 20)) |>
  color(i = ~ Temp > 80, color = "red")

Formula selectors evaluate conditions using the actual data types from the dataset (numeric, logical, etc.). They only work with part = "body".

Pass an integer vector to select rows by position.

# Bold only the first and third body rows
ft <- flextable(head(mtcars)) |>
  bold(i = c(1, 3))

# Target the second header row
ft <- bold(ft, i = 2, part = "header")

Integer selection works with all parts, including header and footer.

Pass a logical vector of the same length as the number of rows in the part.

data <- head(mtcars, 6)
ft <- flextable(data) |>
  bg(i = c(TRUE, FALSE, TRUE, FALSE, TRUE, FALSE), bg = "#f0f0f0")

Targeting the last row with `nrow_part()`

nrow_part() returns the number of rows in a given part. It is the recommended way to programmatically select the last row:

ft <- flextable(iris[48:52, ])

# Bold the last body row
ft <- bold(ft, i = nrow_part(ft, part = "body"), part = "body")

# Add a footnote to the last row
ft <- footnote(
  ft,
  i = nrow_part(ft, part = "body"), j = 1:4,
  value = as_paragraph("Calculated mean")
)

The `j` argument — column selector

The j argument selects which columns to target. When j = NULL (the default), all columns are selected.

Character names
Formula
Integer index
Logical vector

The recommended approach — explicit column names are readable and resilient to column reordering:

# Color a single column
ft <- flextable(head(mtcars)) |>
  color(j = "mpg", color = "blue")

# Target multiple columns
ft <- flextable(head(mtcars)) |>
  bold(j = c("mpg", "cyl", "hp"))

A formula lists column names with +. This is useful for selecting or excluding multiple columns concisely:

# Select specific columns
ft <- flextable(head(mtcars)) |>
  bold(j = ~ vs + am + gear)

# Exclude a column using . - colname
ft <- flextable(head(mtcars)) |>
  bg(j = ~ . - mpg, bg = "#f5f5f5")

The formula is parsed as a model terms expression. Column names are extracted from the term labels.

Select columns by position:

# First and third columns
ft <- flextable(head(mtcars)) |>
  bold(j = c(1, 3))

Negative integers exclude columns:

# All columns except the first
ft <- flextable(head(mtcars)) |>
  bold(j = -1)

Pass a logical vector of the same length as ncol_keys(ft):

ncols <- ncol_keys(ft)
# Apply alternating background
ft <- bg(ft, j = rep(c(TRUE, FALSE), length.out = ncols), bg = "#f0f0f0")

Prefer character vectors (j = c("col1", "col2")) over integer positions. Character names are unambiguous when columns are reordered or when sharing code with others.

Combining `i`, `j`, and `part`

Selectors can be combined freely. The operation applies to the intersection of selected rows and selected columns in the targeted part:

ft <- flextable(head(mtcars, 10)) |>
  # Red text in body rows where mpg < 22, only in the mpg column
  color(i = ~ mpg < 22, j = "mpg", color = "red", part = "body") |>
  # Bold all header cells
  bold(part = "header") |>
  # Grey background for the last body row, all columns
  bg(
    i = nrow_part(ft, "body"),
    bg = "#eeeeee",
    part = "body"
  )

Selectors and merged cells

Selectors work with merged cells — they target the individual cells by row and column index, not by the visual span. When you format a merged cell, apply formatting to the top-left cell of the merged region (the anchor cell); other cells in the span are hidden and their formatting has no visual effect.

ft <- flextable(head(mtcars)) |>
  merge_h(i = 1, part = "body") |>   # merge all cells in row 1
  bold(i = 1, j = 1, part = "body")  # bold the anchor (top-left) cell

Best practices

Rows

Use i = ~ condition for conditional body formatting.
Use nrow_part(ft, "body") to target the last row without hard-coding a number.
Use integers or logical vectors to target header and footer rows.

Columns

Prefer character names (j = c("col1", "col2")) for maintainability.
Use formula syntax (j = ~ col1 + col2) when excluding columns is cleaner.
Avoid integer positions when column order may change.

Quick reference

Selector	Type	Example	Notes
`i`	Formula	`i = ~ mpg < 22`	Body only; uses original data types
`i`	Integer	`i = c(1, 3)`	Any part; positional
`i`	Logical	`i = c(TRUE, FALSE, ...)`	Length must match part row count
`i`	NULL	`i = NULL`	All rows (default)
`j`	Character	`j = c("mpg", "cyl")`	Recommended; column names
`j`	Formula	`j = ~ mpg + cyl`	Column list or exclusion
`j`	Integer	`j = 1:3`	Positional; negative excludes
`j`	Logical	`j = c(TRUE, FALSE, ...)`	Length must match `ncol_keys(ft)`
`j`	NULL	`j = NULL`	All columns (default)
`part`	String	`part = "body"`	`"body"`, `"header"`, `"footer"`, `"all"`

Get Started

Core Concepts

Formatting

Layout

Data Summaries

Export & Output

Advanced

The `part` argument

The `i` argument — row selector

Targeting the last row with `nrow_part()`

The `j` argument — column selector

Combining `i`, `j`, and `part`

Selectors and merged cells

Best practices

Rows

Columns

Quick reference

Build docs developers (and LLMs) love

Get Started

Core Concepts

Formatting

Layout

Data Summaries

Export & Output

Advanced

Documentation Index

​The part argument

​The i argument — row selector

​Targeting the last row with nrow_part()

​The j argument — column selector

​Combining i, j, and part

​Selectors and merged cells

​Best practices

Rows

Columns

​Quick reference

Build docs developers (and LLMs) love

The `part` argument

The `i` argument — row selector

Targeting the last row with `nrow_part()`

The `j` argument — column selector

Combining `i`, `j`, and `part`

Selectors and merged cells

Best practices

Quick reference