Merge pull request #84 from opencultureconsulting:felixlohmeier/import-tsv-26

import tsv command
This commit is contained in:
Felix Lohmeier 2022-11-16 23:23:19 +01:00 committed by GitHub
commit e507573121
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 287 additions and 6 deletions

224
orcli
View File

@ -133,7 +133,8 @@ orcli_import_usage() {
echo
# :command.usage_commands
printf "Commands:\n"
echo " csv import comma-separated values (CSV)"
echo " csv import character-separated values (CSV)"
echo " tsv import tab-separated values (TSV)"
echo
# :command.long_usage
@ -151,11 +152,11 @@ orcli_import_usage() {
# :command.usage
orcli_import_csv_usage() {
if [[ -n $long_usage ]]; then
printf "orcli import csv - import comma-separated values (CSV)\n"
printf "orcli import csv - import character-separated values (CSV)\n"
echo
else
printf "orcli import csv - import comma-separated values (CSV)\n"
printf "orcli import csv - import character-separated values (CSV)\n"
echo
fi
@ -222,6 +223,74 @@ orcli_import_csv_usage() {
fi
}
# :command.usage
orcli_import_tsv_usage() {
if [[ -n $long_usage ]]; then
printf "orcli import tsv - import tab-separated values (TSV)\n"
echo
else
printf "orcli import tsv - import tab-separated values (TSV)\n"
echo
fi
printf "Usage:\n"
printf " orcli import tsv [FILE...] [OPTIONS]\n"
printf " orcli import tsv --help | -h\n"
echo
# :command.long_usage
if [[ -n $long_usage ]]; then
printf "Options:\n"
# :command.usage_fixed_flags
echo " --help, -h"
printf " Show this help\n"
echo
# :command.usage_flags
# :flag.usage
echo " --encoding ENCODING"
printf " set character encoding\n"
echo
# :flag.usage
echo " --trimStrings"
printf " trim leading & trailing whitespace from strings\n"
echo
# :flag.usage
echo " --projectName PROJECTNAME"
printf " set a name for the OpenRefine project\n"
echo
# :flag.usage
echo " --quiet, -q"
printf " suppress log output, print errors only\n"
echo
# :command.usage_args
printf "Arguments:\n"
# :argument.usage
echo " FILE..."
printf " Path to one or more files or URLs. When FILE is -, read standard input.\n"
printf " Default: -\n"
echo
# :command.usage_examples
printf "Examples:\n"
printf " orcli import tsv \"file\"\n"
printf " orcli import tsv \"file1\" \"file2\"\n"
printf " cat \"file\" | orcli import tsv\n"
printf " orcli import tsv \"https://git.io/fj5hF\"\n"
printf " orcli import tsv \"file\" \\\\\n --separator \";\" \\\\\n --encoding \"ISO-8859-1\" \\\\\n --trimStrings \\\\\n --projectName \"duplicates\"\n"
echo
fi
}
# :command.usage
orcli_list_usage() {
if [[ -n $long_usage ]]; then
@ -783,6 +852,10 @@ send_completions() {
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --projectName --quiet --separator --trimStrings -h -q")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'import tsv\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --projectName --quiet --trimStrings -h -q")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'export tsv\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --output --quiet -h -q")" -- "$cur" )'
echo $' ;;'
@ -792,7 +865,7 @@ send_completions() {
echo $' ;;'
echo $''
echo $' \'import\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--help -h csv")" -- "$cur" )'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--help -h csv tsv")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'export\'*)'
@ -869,6 +942,42 @@ orcli_import_csv_command() {
}
# :command.function
orcli_import_tsv_command() {
# src/import_tsv_command.sh
# shellcheck shell=bash
# call init_import function to eval args and to set basic post data
init_import
# check if stdin is present if selected
if [[ ${args[file]} == '-' ]] || [[ ${args[file]} == '"-"' ]]; then
if ! read -u 0 -t 0; then
orcli_import_tsv_usage
exit 1
fi
fi
# assemble specific post data (some options require json format)
data+=("format=text/line-based/*sv")
options='{ '
options+="\"separator\": \"\\t\""
if [[ ${args[--encoding]} ]]; then
options+=', '
options+="\"encoding\": \"${args[--encoding]}\""
fi
if [[ ${args[--trimStrings]} ]]; then
options+=', '
options+="\"trimStrings\": true"
fi
options+=' }'
data+=("options=${options}")
# call post_import function to post data and validate results
post_import "${data[@]}"
}
# :command.function
orcli_list_command() {
# src/list_command.sh
@ -1328,6 +1437,13 @@ orcli_import_parse_requirements() {
shift $#
;;
tsv )
action="tsv"
shift
orcli_import_tsv_parse_requirements "$@"
shift $#
;;
# :command.command_fallback
"" )
orcli_import_usage >&2
@ -1472,6 +1588,98 @@ orcli_import_csv_parse_requirements() {
}
# :command.parse_requirements
orcli_import_tsv_parse_requirements() {
# :command.fixed_flags_filter
case "${1:-}" in
--help | -h )
long_usage=yes
orcli_import_tsv_usage
exit
;;
esac
# :command.command_filter
action="import tsv"
# :command.parse_requirements_while
while [[ $# -gt 0 ]]; do
key="$1"
case "$key" in
# :flag.case
--encoding )
# :flag.case_arg
if [[ -n ${2+x} ]]; then
args[--encoding]="$2"
shift
shift
else
printf "%s\n" "--encoding requires an argument: --encoding ENCODING" >&2
exit 1
fi
;;
# :flag.case
--trimStrings )
# :flag.case_no_arg
args[--trimStrings]=1
shift
;;
# :flag.case
--projectName )
# :flag.case_arg
if [[ -n ${2+x} ]]; then
args[--projectName]="$2"
shift
shift
else
printf "%s\n" "--projectName requires an argument: --projectName PROJECTNAME" >&2
exit 1
fi
;;
# :flag.case
--quiet | -q )
# :flag.case_no_arg
args[--quiet]=1
shift
;;
-?* )
printf "invalid option: %s\n" "$key" >&2
exit 1
;;
* )
# :command.parse_requirements_case
# :command.parse_requirements_case_repeatable
if [[ -z ${args[file]+x} ]]; then
args[file]="\"$1\""
shift
else
args[file]="${args[file]} \"$1\""
shift
fi
;;
esac
done
# :command.default_assignments
[[ -n ${args[file]:-} ]] || args[file]="-"
}
# :command.parse_requirements
orcli_list_parse_requirements() {
# :command.fixed_flags_filter
@ -1915,6 +2123,14 @@ run() {
orcli_import_csv_command
fi
elif [[ $action == "import tsv" ]]; then
if [[ ${args[--help]:-} ]]; then
long_usage=yes
orcli_import_tsv_usage
else
orcli_import_tsv_command
fi
elif [[ $action == "list" ]]; then
if [[ ${args[--help]:-} ]]; then
long_usage=yes

View File

@ -38,7 +38,7 @@ commands:
commands:
- name: csv
help: import comma-separated values (CSV)
help: import character-separated values (CSV)
args:
- name: file
help: Path to one or more files or URLs. When FILE is -, read standard input.
@ -72,6 +72,37 @@ commands:
--trimStrings \\\\
--projectName "duplicates"
- name: tsv
help: import tab-separated values (TSV)
args:
- name: file
help: Path to one or more files or URLs. When FILE is -, read standard input.
default: "-"
repeatable: true
flags:
- long: --encoding
help: set character encoding
arg: encoding
- long: --trimStrings
help: trim leading & trailing whitespace from strings
- long: --projectName
arg: projectName
help: set a name for the OpenRefine project
- long: --quiet
short: -q
help: suppress log output, print errors only
examples:
- orcli import tsv "file"
- orcli import tsv "file1" "file2"
- cat "file" | orcli import tsv
- orcli import tsv "https://git.io/fj5hF"
- |-
orcli import tsv "file" \\\\
--separator ";" \\\\
--encoding "ISO-8859-1" \\\\
--trimStrings \\\\
--projectName "duplicates"
- name: list
help: list projects on OpenRefine server

30
src/import_tsv_command.sh Normal file
View File

@ -0,0 +1,30 @@
# shellcheck shell=bash
# call init_import function to eval args and to set basic post data
init_import
# check if stdin is present if selected
if [[ ${args[file]} == '-' ]] || [[ ${args[file]} == '"-"' ]]; then
if ! read -u 0 -t 0; then
orcli_import_tsv_usage
exit 1
fi
fi
# assemble specific post data (some options require json format)
data+=("format=text/line-based/*sv")
options='{ '
options+="\"separator\": \"\\t\""
if [[ ${args[--encoding]} ]]; then
options+=', '
options+="\"encoding\": \"${args[--encoding]}\""
fi
if [[ ${args[--trimStrings]} ]]; then
options+=', '
options+="\"trimStrings\": true"
fi
options+=' }'
data+=("options=${options}")
# call post_import function to post data and validate results
post_import "${data[@]}"

View File

@ -38,6 +38,10 @@ send_completions() {
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --projectName --quiet --separator --trimStrings -h -q")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'import tsv\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --projectName --quiet --trimStrings -h -q")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'export tsv\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--encoding --help --output --quiet -h -q")" -- "$cur" )'
echo $' ;;'
@ -47,7 +51,7 @@ send_completions() {
echo $' ;;'
echo $''
echo $' \'import\'*)'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--help -h csv")" -- "$cur" )'
echo $' while read -r; do COMPREPLY+=( "$REPLY" ); done < <( compgen -W "$(_orcli_completions_filter "--help -h csv tsv")" -- "$cur" )'
echo $' ;;'
echo $''
echo $' \'export\'*)'