Consolidates PDFs, man pages, and web pages into a single PDF file.

pdfpack.sh

#!/bin/sh
set -e

# Default values
output=""
pdfdocs=""
manpages=""
webpages=""
sources=""

# Function to display usage information
usage() {
  echo "Consolidates PDFs, man pages, and web pages into a single PDF file."
  echo "Usage: $0 [-o OUTPUT_FILE] [-m MANPAGE1,MANPAGE2,...] [-w WEBPAGE1,WEBPAGE2,...]"
  echo "  -o OUTPUT_FILE : Specify the output PDF file name"
  echo "  -p PDFPAGES    : Comma-separated list of pdf pages to include"
  echo "  -m MANPAGES    : Comma-separated list of man pages to convert to PDF"
  echo "  -w WEBPAGES    : Comma-separated list of web pages to convert to PDF"
  echo "  -s SOURCES     : Comma-separated list of source files to convert to put in PDF code blocks"
  echo "  -h             : Display this help message"
}

# Parse command line options
while getopts ":o:p:m:w:s:h" opt; do
  case $opt in
    o)
      output="$OPTARG"
      ;;
    p)
      pdfdocs="$OPTARG"
      ;;
    m)
      manpages="$OPTARG"
      ;;
    w)
      webpages="$OPTARG"
      ;;
    s)
      sources="$OPTARG"
      ;;
    h)
      usage
      exit 0
      ;;
    \?)
      echo "Invalid option: -$OPTARG" >&2
      usage >&2
      exit 1
      ;;
    :)
      echo "Option -$OPTARG requires an argument." >&2
      usage
      exit 1
      ;;
  esac
done

# Check if required commands are installed
required_programs='mandoc man pandoc pdfjam'
for cmd in $required_programs; do
  if ! command -v "$cmd" > /dev/null; then
    echo "Error: $cmd is not installed." >&2
    echo "Required programs: $required_programs" >&2
    exit 1
  fi
done

# Check if output file is specified
if [ -z "$output" ]; then
  output=/dev/stdout
fi

# Convert comma-separated strings to space-separated for iteration
manpages=$(echo "$manpages" | tr ',' ' ')
webpages=$(echo "$webpages" | tr ',' ' ')
sources=$(echo "$sources" | tr ',' ' ')

docs=$(echo "$pdfdocs" | tr ',' ' ')

# Process manpages
for m in $manpages; do 
  path=/tmp/$m.pdf
  mandoc -T pdf "$(man -w "$m")" > "$path"
  docs="$docs $path"
done

# Process webpages
for w in $webpages; do
  path=/tmp/$(basename "$w" .html).pdf
  pandoc -o "$path" "$w"
  docs="$docs $path"
done

extension() {
  path=$1
	extension="${path##*.}"
	if [ "$extension" != "$path" ]
  then echo "$extension"
	fi
}

# Process sources
code_block() {
  s=$1
  echo file: "$s"
  printf '```%s\n' "$(extension "$s")"
  cat "$s"
  echo '```'
}
for s in $sources; do
  path=/tmp/$(basename "$s" ".$(extension "$s")").pdf
  echo path: "$path"
  code_block "$s" | pandoc -f markdown -o "$path"
  docs="$docs $path"
done

if [ -z "$docs" ]; then
  echo "Error: no documents specified." >&2
  usage >&2
  exit 1
fi

# Combine all documents into one PDF
# shellcheck disable=SC2086
pdfjam -o "$output" -- $docs