#!/usr/bin/env bash

set -Eeuo pipefail

script_dir=$(CDPATH= cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd -P)
config_file="${script_dir}/cfsv2_manual.conf"
dry_run=0

# User-configurable directory containing the YYYYMMDD download directories.
# It can also be overridden from the environment when invoking this script.
#OUTPUT_DIR="${OUTPUT_DIR:-/home/tanpv/Progs/CLI/Check_data/data/CFS_Ope/00}"
OUTPUT_DIR="${OUTPUT_DIR:-/home/nampq/data/CFS_Ope/00}"

usage() {
    cat <<'EOF'
Usage:
  ./ope_download_manual.sh [options] YEAR MONTH DAY [HOUR]

Examples:
  ./ope_download_manual.sh 2026 08 21
  ./ope_download_manual.sh 2026 08 21 00
  ./ope_download_manual.sh --dry-run 2026 08 21

Options:
  --config FILE  Read an alternative configuration file.
  --dry-run      Validate and show URLs/paths without creating or downloading.
  -h, --help     Show this help.

The default HOUR is configured as 00. Output is stored below this script's
directory unless CFS_DATA_ROOT/CFS_LOG_ROOT are explicitly configured otherwise.
EOF
}

die() {
    printf 'ERROR: %s\n' "$*" >&2
    exit 1
}

while (($#)); do
    case "$1" in
        --config)
            (($# >= 2)) || die "--config requires a file"
            config_file=$2
            shift 2
            ;;
        --dry-run)
            dry_run=1
            shift
            ;;
        -h|--help)
            usage
            exit 0
            ;;
        --)
            shift
            break
            ;;
        -* )
            die "unknown option: $1"
            ;;
        *)
            break
            ;;
    esac
done

(($# == 3 || $# == 4)) || {
    usage >&2
    exit 2
}

year=$1
month=$2
day=$3
hour=${4:-}

[[ -r "$config_file" ]] || die "configuration is not readable: $config_file"
CFS_PACKAGE_DIR=$script_dir
# shellcheck source=cfsv2_manual.conf
source "$config_file"

hour=${hour:-$CFS_DEFAULT_HOUR}

[[ $year =~ ^[0-9]{4}$ ]] || die "YEAR must contain four digits"
[[ $month =~ ^[0-9]{1,2}$ ]] || die "MONTH must contain one or two digits"
[[ $day =~ ^[0-9]{1,2}$ ]] || die "DAY must contain one or two digits"
[[ $hour =~ ^[0-9]{1,2}$ ]] || die "HOUR must contain one or two digits"

printf -v month '%02d' "$((10#$month))"
printf -v day '%02d' "$((10#$day))"
printf -v hour '%02d' "$((10#$hour))"
ymd="${year}${month}${day}"
ymdh="${ymd}${hour}"

normalized=$(date -u -d "${year}-${month}-${day} ${hour}:00:00" +%Y%m%d%H 2>/dev/null) ||
    die "invalid UTC date: ${year}-${month}-${day} ${hour}:00"
[[ $normalized == "$ymdh" ]] || die "invalid UTC date: ${year}-${month}-${day} ${hour}:00"

case " $CFS_ALLOWED_HOURS " in
    *" $hour "*) ;;
    *) die "HOUR $hour is not in CFS_ALLOWED_HOURS: $CFS_ALLOWED_HOURS" ;;
esac

for command_name in curl date grep realpath sort wc wget; do
    command -v "$command_name" >/dev/null 2>&1 || die "required command not found: $command_name"
done

package_root=$(realpath -m -- "$script_dir")
log_root=$(realpath -m -- "$CFS_LOG_ROOT")

case "$log_root/" in
    "$package_root"/*) CFS_LOG_ROOT=$log_root ;;
    *) die "CFS_LOG_ROOT must stay below package directory: $package_root" ;;
esac

OUTPUT_DIR=$(realpath -m -- "$OUTPUT_DIR")
[[ $OUTPUT_DIR == /* ]] || die "OUTPUT_DIR must resolve to an absolute path"

remote_dir="${CFS_BASE_URL}/cfs.${ymd}/${hour}/${CFS_REMOTE_SUBDIR}"
cycle_output_dir="${OUTPUT_DIR}/${ymd}"
log_file="${CFS_LOG_ROOT}/cfs.${ymdh}.files"

printf '%s\n' \
    "CFSv2 manual download" \
    "  cycle       : $ymdh UTC" \
    "  remote      : $remote_dir/" \
    "  output      : $cycle_output_dir" \
    "  file list   : $log_file" \
    "  model mode  : $MODEL_MODE"

if ((dry_run)); then
    printf 'DRY RUN: validation completed; no directory or file was created.\n'
    exit 0
fi

[[ $MODEL_MODE == none ]] || die "review package supports MODEL_MODE=none only"

mkdir -p -- "$cycle_output_dir" "$CFS_LOG_ROOT"

curl_options=(
    --fail --silent --show-error --location
    --connect-timeout "$CFS_CONNECT_TIMEOUT"
    --retry "$CFS_RETRY_COUNT"
    --retry-delay "$CFS_RETRY_DELAY"
)
wget_options=(
    --continue
    --timeout="$CFS_CONNECT_TIMEOUT"
    --tries="$CFS_RETRY_COUNT"
)

if [[ $CFS_INSECURE_TLS == 1 ]]; then
    curl_options+=(--insecure)
    wget_options+=(--no-check-certificate)
fi

listing=$(curl "${curl_options[@]}" "$remote_dir/") ||
    die "cannot read remote directory: $remote_dir/"

printf '%s\n' "$listing" |
    grep -oE "$CFS_FILE_REGEX" |
    sort -u > "$log_file" || true

file_count=$(wc -l < "$log_file")
((file_count > 0)) || die "no flxf/pgbf GRIB2 files found at $remote_dir/"
printf 'Remote files found: %d\n' "$file_count"

# Treat a cycle as complete only when every file currently advertised by the
# remote directory already exists locally and is non-empty. A partial cycle
# still falls through to the normal loop so that only missing files are fetched.
all_files_downloaded=1
while IFS= read -r filename; do
    if [[ ! -s "${cycle_output_dir}/${filename}" ]]; then
        all_files_downloaded=0
        break
    fi
done < "$log_file"

if ((all_files_downloaded)); then
    printf 'Data of %s-%s-%s-%s have already been downloaded.\n' \
        "$year" "$month" "$day" "$hour"
    exit 0
fi

downloaded=0
skipped=0
failed=0

while IFS= read -r filename; do
    [[ $filename =~ ^(flxf|pgbf)[A-Za-z0-9._-]+\.grb2$ ]] || {
        printf '[REJECT] Unexpected filename: %s\n' "$filename" >&2
        ((failed += 1))
        continue
    }

    destination="${cycle_output_dir}/${filename}"
    if [[ -s "$destination" ]]; then
        printf '[SKIP] %s\n' "$filename"
        ((skipped += 1))
        continue
    fi

    printf '[GET ] %s\n' "$filename"
    if wget "${wget_options[@]}" "$remote_dir/$filename" -O "$destination"; then
        ((downloaded += 1))
    else
        printf '[FAIL] %s\n' "$filename" >&2
        ((failed += 1))
    fi
done < "$log_file"

printf '%s\n' \
    "Download summary" \
    "  downloaded : $downloaded" \
    "  skipped    : $skipped" \
    "  failed     : $failed" \
    "  output     : $cycle_output_dir"

((failed == 0)) || exit 1