123 lines
3.8 KiB
Bash
Executable File
123 lines
3.8 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
# Source the script 'common-functions.sh'.
|
|
SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" &>/dev/null && pwd)
|
|
ROOT_DIR=$(grep --only-matching "^.*scripts[^/]*" <<<"$SCRIPT_DIR")
|
|
source "$ROOT_DIR/common-functions.sh"
|
|
|
|
_main() {
|
|
local input_files=""
|
|
local output_dir=""
|
|
|
|
# Execute initial checks.
|
|
_check_dependencies ""
|
|
input_files=$(_get_files "par_type=file; par_validate_conflict=true")
|
|
|
|
local dependencies=""
|
|
dependencies=$(_get_dependencies "$input_files")
|
|
_check_dependencies "$dependencies"
|
|
_display_wait_box "2"
|
|
output_dir=$(_get_output_dir "par_use_same_dir=true")
|
|
|
|
export -f _get_command _run_command
|
|
|
|
# Execute the function '_main_task' for each file in parallel.
|
|
_run_task_parallel "$input_files" "$output_dir"
|
|
_display_result_box "$output_dir"
|
|
}
|
|
|
|
_main_task() {
|
|
local input_file=$1
|
|
local output_dir=$2
|
|
local output_file=""
|
|
|
|
# Run the main process.
|
|
local command=""
|
|
command=$(_get_command "$input_file")
|
|
_run_command "$input_file" "$command" "$output_dir"
|
|
}
|
|
|
|
_get_dependencies() {
|
|
local input_files=$1
|
|
local dependencies=""
|
|
|
|
# Check dependencies for each file.
|
|
for input_file in $input_files; do
|
|
local command=""
|
|
command=$(_get_command "$input_file")
|
|
|
|
case $command in
|
|
"libreoffice_writer")
|
|
if ! _command_exists "libreoffice.writer" && ! _command_exists "libreoffice-writer"; then
|
|
dependencies+="package=libreoffice-writer"
|
|
fi
|
|
;;
|
|
"pandoc") dependencies+="command=pandoc" ;;
|
|
"pdftotext")
|
|
dependencies+="
|
|
command=pdftotext; pkg_manager=apt; package=poppler-utils |
|
|
command=pdftotext; pkg_manager=dnf; package=poppler-utils |
|
|
command=pdftotext; pkg_manager=pacman; package=poppler |
|
|
command=pdftotext; pkg_manager=zypper; package=poppler-tools"
|
|
;;
|
|
esac
|
|
dependencies+=$'\n'
|
|
done
|
|
|
|
printf "%s" "$dependencies"
|
|
}
|
|
|
|
_get_command() {
|
|
local input_file=$1
|
|
local command=""
|
|
|
|
case "${input_file,,}" in
|
|
*.doc | *.docx) command="libreoffice_writer" ;;
|
|
*.htm | *.html) command="libreoffice_writer" ;;
|
|
*.odf | *.fodf) command="libreoffice_writer" ;;
|
|
*.odt | *.fodt) command="libreoffice_writer" ;;
|
|
*.pdf) command="pdftotext" ;;
|
|
*.rtf) command="libreoffice_writer" ;;
|
|
*) command="pandoc" ;;
|
|
esac
|
|
|
|
printf "%s" "$command"
|
|
}
|
|
|
|
_run_command() {
|
|
local input_file=$1
|
|
local command=$2
|
|
local output_dir=$3
|
|
local output_format="txt"
|
|
|
|
output_file=$(_get_output_filename "$input_file" "$output_dir" "par_extension_opt=replace; par_extension=$output_format")
|
|
|
|
# Run the main process.
|
|
case $command in
|
|
"libreoffice_writer")
|
|
# NOTE: Workaround to fix the bug 37531 in LibreOffice.
|
|
# See the: https://bugs.documentfoundation.org/show_bug.cgi?id=37531
|
|
local temp_file=""
|
|
temp_file=$(_get_temp_file_dry)
|
|
|
|
# NOTE: LibreOffice does not support ' -- ' in the command line.
|
|
# NOTE: LibreOffice does not support defining the output file manually.
|
|
std_output=$(libreoffice --headless --convert-to "$output_format" "-env:UserInstallation=file://$temp_file" --outdir "$output_dir" "$input_file" 2>&1)
|
|
_check_output "$?" "$std_output" "$input_file" ""
|
|
|
|
# Remove the temporary file.
|
|
rm -f -- "$temp_file"
|
|
;;
|
|
"pandoc")
|
|
std_output=$(pandoc --standalone -o "$output_file" -- "$input_file" 2>&1)
|
|
_check_output "$?" "$std_output" "$input_file" "$output_file"
|
|
;;
|
|
"pdftotext")
|
|
std_output=$(pdftotext "$input_file" -- "$output_file" 2>&1)
|
|
_check_output "$?" "$std_output" "$input_file" "$output_file"
|
|
;;
|
|
esac
|
|
}
|
|
|
|
_main "$@"
|