Assembling reports into one document

library(rtfreporter)

A clinical deliverable is many tables, listings and figures, each written to its own .rtf by its own program. assemble_rtf() joins them into one document: a cover page, a table of contents whose entries are links, each output starting on its own page with its own running header and footer, and a bookmark per output, which Word and LibreOffice turn into the PDF’s bookmarks.

The Gallery’s demographics, adverse events, vital signs and listing, assembled with a cover page and a table of contents, then converted to PDF with LibreOffice:

Each entry of the table of contents is a link. Click the adverse events entry and the reader opens that table’s first page:

Show code

files <- c("t_dm.rtf", "t_ae.rtf", "t_vs.rtf", "l_sae.rtf")   # the Gallery's outputs
assemble_rtf(
  files, "XYZ-001_tables.rtf", overwrite = TRUE,
  cover = list(title = "Study XYZ-001", subtitle = "Tables and Listings",
               version = "Draft 1"),
  toc = data.frame(
    file    = files,
    heading = c("DEMOGRAPHICS", "SAFETY", "SAFETY", "LISTINGS"),
    label   = c("Table 14.1.1  Demographic Characteristics",
                "Table 14.3.1.1  Treatment-Emergent Adverse Events by SOC and PT",
                "Table 14.3.5.1  Vital Signs by Visit",
                "Listing 16.2.7.2  Severe Treatment-Emergent Adverse Events")),
  toc_page_numbering = "roman")
system2("soffice", c("--headless", "--convert-to", "pdf", "XYZ-001_tables.rtf"))

The outputs first

Each output is written as it always is – its own program, its own generate_rtfreport(). Here, two small ones:

Show code

make <- function(title, path) {
  d <- data.frame(A = c("1", "2"), B = c("x", "y"), stringsAsFactors = FALSE)
  doc <- rtf_document() |> rtf_tables(as_rtftables(d), titles = list(c(title)))
  generate_rtfreport(doc, path, overwrite = TRUE); path
}
f1 <- make("Table 14.1.1 Demographics",   tempfile(fileext = ".rtf"))
f2 <- make("Table 14.2.1 Adverse Events",  tempfile(fileext = ".rtf"))

A table of contents from the titles: toc = "auto"

toc = "auto" builds one TOC entry per file, taking each label from the file’s first title:

pkg <- tempfile(fileext = ".rtf")
assemble_rtf(c(f1, f2), pkg, toc = "auto", overwrite = TRUE)
file.exists(pkg)
#> [1] TRUE

A cover page and a table of contents of your own: cover =, toc =

For sections and explicit labels, pass the table of contents as a table: one row per file, its label, and a heading printed above the entry whenever it changes (level sets the indent). Each entry creates an RTF bookmark, so the TOC is navigable in the word processor and the entries jump to the right table:

Show code

pkg2 <- tempfile(fileext = ".rtf")
assemble_rtf(
  c(f1, f2), pkg2,
  cover = list(title = "Study XYZ-001", subtitle = "Final Tables", version = "v1.0"),
  toc = data.frame(
    file    = c(f1, f2),
    heading = c("EFFICACY ANALYSES", "SAFETY ANALYSES"),
    label   = c("Table 14.1.1 Demographics", "Table 14.2.1 Adverse Events")
  ),
  toc_page_numbering = "roman",
  overwrite = TRUE
)
file.exists(pkg2)

#> [1] TRUE

assemble_rtf() also takes a cover page and toc_leader / toc_page_numbering options; see ?assemble_rtf.

Page numbers

Each output keeps its own running footer, so a table says Page 2 of 5 of itself, whichever page of the deliverable it is on: write it with the static {PAGE} / {TOTAL_PAGES} tokens, filled in when that output is written. toc_page_numbering numbers the table of contents’ own pages ("roman", "decimal"), and the page each entry points to is the page in the assembled document.

Assembling a folder: assemble_folder()

To combine every RTF in a directory, assemble_folder(dir, output_file) reads each file’s table number and title into a table of contents and assembles them. Without output_file it returns that table instead: edit the order, headings and labels (or save it with spec_file = and edit it in Excel), then pass it – or its path – to assemble_rtf(toc = ). See ?assemble_folder.

Bookmarks in the PDF, and Word and LibreOffice

The assembled file is RTF. Each table of contents entry is a link to a bookmark at the start of its file, and each file also starts with a heading at an outline level, printed invisibly (white, one point high). Word and LibreOffice show those headings as the PDF’s bookmarks when they export the file to PDF, so the PDF can be navigated table by table, as an eCTD submission expects. With LibreOffice from R:

system2("soffice", c("--headless", "--convert-to", "pdf", pkg2))
pdftools::pdf_toc(sub("[.]rtf$", ".pdf", pkg2))   # one bookmark a file

The page numbers in the table of contents are fields that already hold the page rtfreporter computed for each file; Word and LibreOffice update them when they refresh fields. The links and the bookmarks above were checked in the PDF that LibreOffice writes. In Word itself a link follows on Ctrl+click, its default setting.

See ?assemble_rtf and ?assemble_folder for every argument.

Where next

• Rendering and post-processing — writing one output, and editing the written file
• Headers, footers and tokens — the running header and footer each output keeps
• The pharmaverse catalog — tables of three packages, assembled