respect-validation/bin/update-iso-codes
Alexandre Gomes Gaigalas e2b6138bf6 Add PublicDomainSuffix Rule
- List will be auto-updated from https://publicsuffix.org/list/public_suffix_list.dat
 - Updated AbstractSearcher rules to be case insensitive
 - Updated PR creator bots
 - Docs and tests
2023-02-19 00:19:10 -03:00

103 lines
3.4 KiB
Bash
Executable file

#!/usr/bin/env bash
# Usage: {script} TLD_FILENAME
# Update list of TLD
set -euo pipefail
declare -r IFS=$'\n'
declare -r REPOSITORY_URL="https://salsa.debian.org/iso-codes-team/iso-codes.git"
declare -r REPOSITORY_DIRECTORY=".iso-codes-cache"
declare -r LIBRARY_DIRECTORY="library"
clone_repository()
{
echo "Cloning repository ${REPOSITORY_URL}"
if ! test -d "$REPOSITORY_DIRECTORY"
then
git clone --quiet "${REPOSITORY_URL}" "${REPOSITORY_DIRECTORY}"
fi
}
list_iso_3166()
{
local -r number_of_items=${1}
local -r filename=${2}
for index in $(seq 0 ${number_of_items}); do
local json=$(jq ".[][${index}]" < "${filename}")
local alpha_2=$(jq ".alpha_2" <<< "${json}" | tr '"' "'")
local alpha_3=$(jq ".alpha_3" <<< "${json}" | tr '"' "'")
local numeric=$(jq ".numeric" <<< "${json}" | tr '"' "'")
local name=$(jq -r ".name" <<< "${json}")
echo " [${alpha_2}, ${alpha_3}, ${numeric}], // ${name}"
done
}
update_country_codes()
{
local -r iso_3166_1_filename="${REPOSITORY_DIRECTORY}/data/iso_3166-1.json"
local -r iso_3166_1_count=$(grep "alpha_3" "${iso_3166_1_filename}" | wc --lines)
local -r iso_3166_3_filename="${REPOSITORY_DIRECTORY}/data/iso_3166-3.json"
local -r iso_3166_3_count=$(grep "alpha_3" "${iso_3166_3_filename}" | wc --lines)
local -r temporary_filename=$(mktemp)
local -r country_rule_filename="${LIBRARY_DIRECTORY}/Rules/CountryCode.php"
echo "Updating country codes using ISO 3166-1 and ISO 3166-3"
{
sed -n '/^</,/ \/\/ begin of auto-generated code/p' "${country_rule_filename}"
{
list_iso_3166 $[iso_3166_1_count - 1] ${iso_3166_1_filename}
list_iso_3166 $[iso_3166_3_count - 1] ${iso_3166_3_filename}
} | sort
sed -n '/^ \/\/ end of auto-generated code/,/^}/p' "${country_rule_filename}"
} > "${temporary_filename}"
mv "${temporary_filename}" "${country_rule_filename}"
}
update_subdivision_codes()
{
local -r iso_3166_1_filename="${REPOSITORY_DIRECTORY}/data/iso_3166-1.json"
local -r iso_3166_1_count=$(grep "\"alpha_3\"" "${iso_3166_1_filename}" | wc --lines)
local -r iso_3166_2_filename="${REPOSITORY_DIRECTORY}/data/iso_3166-2.json"
local -r iso_3166_2_count=$(grep --word-regexp "\"code\"" "${iso_3166_2_filename}" | wc --lines)
local -r temporary_filename=$(mktemp)
local last_country_code=""
local last_country_rule=""
rm -rf data/iso_3166-2/*
echo "Updating subdivision codes using ISO 3166-2"
for index in $(seq 0 $[iso_3166_1_count - 1]); do
local json=$(jq ".[][${index}]" < "${iso_3166_1_filename}")
local alpha_1="$(jq ".alpha_2" <<< "${json}" | tr -d '"')"
echo "$alpha_1" "$(jq ".name" <<< "${json}" | tr -d '"')"
done | while read -r country_line
do
local alpha_1="${country_line%% *}"
local cname="${country_line#$alpha_1 }"
local json="$(jq '."3166-2"[] | select(.code|test("^'"$alpha_1"'"))' < "${iso_3166_2_filename}")"
echo "<?php
return [
\"country\" => \"$cname\",
\"subdivisions\" => [" > "data/iso_3166-2/${alpha_1}.php"
jq -r "[.code, .name] | @tsv" <<< "${json}" | while read -r subdiv_line
do
local alpha_2="${subdiv_line%% *}"
local name="${subdiv_line#$alpha_2 }"
echo " \"${alpha_2#${alpha_1}-}\" => \"$name\","
done | sed '$ s/.$//' >> "data/iso_3166-2/${alpha_1}.php"
echo " ]
];" >> "data/iso_3166-2/${alpha_1}.php"
done
}
clone_repository
update_country_codes
update_subdivision_codes