Skip to contents

Convert PDFs to DOCX Using the Best Available Backend

Source: R/pdf_to_docx.R
convert_pdfs_to_docx.Rd

Backend preference is python (pdf2docx) -> LibreOffice -> Word COM (Windows only). When the python backend fails on an individual file and LibreOffice is present, that file is retried once with LibreOffice.

Usage

convert_pdfs_to_docx(
  conversions,
  backend = "auto",
  strict = TRUE,
  timeout = 600
)

Arguments

conversions

A data frame with character columns src and dest, and optionally temp_dest, a second location each converted file is copied to.

backend

One of "auto", "python", "libreoffice", "word_com".

strict

If TRUE (default), stop at the first bad row and, when every row is fine, error with an actionable message if no backend is available. Set FALSE to warn and skip instead; a skipped row comes back with converted = FALSE.

timeout

Seconds a converter may take on one file before it is stopped; it applies to every backend (python, LibreOffice and Word COM). A file that times out gets a warning that says so and comes back with converted = FALSE (the python backend then retries it once with LibreOffice, which gets its own timeout). Stopping a converter stops the process that was started; any process that one started in turn (LibreOffice's soffice.bin, or the Word instance behind Word COM) may keep running and has to be closed by hand.

Value

conversions with an added logical converted column.

Details

The rows of conversions are checked before any backend is looked for, so a bad row is reported the same way on every machine, whether or not a converter is installed. A row is bad when its src does not exist or is empty (0 bytes), when its dest is missing, or when its dest or temp_dest ends in .pdf (a DOCX is not a PDF, and the file that is there would be overwritten). A src that resolves to the same file as its dest or temp_dest, however the path is spelled (./a.pdf, another case), and a dest that is the same file as its temp_dest, are always errors, whatever strict is, because converting would overwrite the PDF or empty the copy. Rows that pass are then converted; if none pass, no backend is looked for at all.

A converter writes into a fresh file next to dest, and that file replaces dest only when the conversion succeeded and left a non-empty file. For pdf2docx and Word COM, succeeded means that the converter exited with status 0. LibreOffice's exit status is not looked at: it writes into a private, empty folder, so whatever it leaves there was written by this run, and a first start with a fresh profile may exit non-zero after a good conversion. A time-out is a failure for every backend. A DOCX that was already at dest (from an earlier run, say) is never mistaken for the result: when the conversion fails, dest is left exactly as it was (an old file is neither deleted nor reported as converted, and a half-written file from a crashed converter is discarded), and the row comes back with converted = FALSE.

See also

[convert_pdf_to_docx()], [check_docx_toolchain()], and [zip_reports()], whose docx_from_pdf argument accepts convert_pdf_to_docx.