#!/usr/bin/env bash
set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
BINARY="${SCRIPT_DIR}/../src/target/release/obikmer"
IDX_DIR="${SCRIPT_DIR}/specimen_index_presence"
OUTPUT="${SCRIPT_DIR}/global_index_presence"
STATS_DIR="${SCRIPT_DIR}/stats/merge_presence"

mkdir -p "${STATS_DIR}"

run_n=$(printf '%03d' "$(find "${STATS_DIR}" -maxdepth 1 -name 'run_*.csv' | wc -l | tr -d ' ')")
CSV="${STATS_DIR}/run_${run_n}.csv"

printf 'run,n_sources,bootstrap_wall_s,bootstrap_rss_b,spectrums_wall_s,spectrums_rss_b,merge_partitions_wall_s,merge_partitions_rss_b,pack_wall_s,pack_rss_b,total_wall_s,total_rss_b\n' >"${CSV}"

parse_reporter() {
    local run="$1" n_sources="$2" logfile="$3"
    python3 - "$run" "$n_sources" "$logfile" <<'PYEOF'
import sys, re

run, n_sources, logfile = sys.argv[1], sys.argv[2], sys.argv[3]

def strip_ansi(s):
    return re.sub(r'\x1b\[[\x30-\x3f]*[\x20-\x2f]*[\x40-\x7e]', '', s)

def parse_wall(s):
    s = s.strip()
    if s.endswith('ms'): return float(s[:-2]) / 1000.0
    if s.endswith('s'):  return float(s[:-1])
    return 0.0

def parse_rss(s):
    m = re.match(r'([\d.]+)\s*(GB|MB|KB|B)', s.strip())
    if not m: return 0
    return int(float(m.group(1)) * {'GB': 1<<30, 'MB': 1<<20, 'KB': 1024, 'B': 1}[m.group(2)])

def is_sep(s):
    return bool(s) and not re.search(r'[A-Za-z0-9]', s)

stats = {}
state = 'scan'

with open(logfile, errors='replace') as fh:
    for raw in fh:
        line = strip_ansi(raw.rstrip('\n'))
        s    = line.strip()

        if state == 'scan':
            if re.search(r'\bstage\b.*\bwall\b', line):
                state = 'in_header'
        elif state == 'in_header':
            if is_sep(s):
                state = 'rows'
        elif state == 'rows':
            if is_sep(s):
                state = 'total'
            elif s:
                parts = re.split(r'  +', s)
                if len(parts) >= 4:
                    stats[parts[0]] = (parse_wall(parts[1]), parse_rss(parts[3]))
        elif state == 'total':
            if s:
                parts = re.split(r'  +', s)
                if len(parts) >= 3:
                    stats[parts[0]] = (parse_wall(parts[1]),
                                       parse_rss(parts[3]) if len(parts) > 3 else 0)
            break

STAGE_ORDER = ['bootstrap', 'spectrums', 'merge_partitions', 'pack']
row = [run, n_sources]
for stage in STAGE_ORDER:
    w, r = stats.get(stage, ('', ''))
    row += [f'{w:.3f}' if isinstance(w, float) else '', str(r)]
tw, tr = stats.get('TOTAL', ('', ''))
row += [f'{tw:.3f}' if isinstance(tw, float) else '', str(tr)]
print(','.join(row))
PYEOF
}

mapfile -t sources < <(find "${IDX_DIR}" -mindepth 1 -maxdepth 1 -type d | sort)

if [[ ${#sources[@]} -eq 0 ]]; then
    echo "ERROR: no indexes found in ${IDX_DIR}" >&2
    exit 1
fi

echo "Merging ${#sources[@]} presence indexes → ${OUTPUT}"
printf '  %s\n' "${sources[@]}"

STDERR_LOG=$(mktemp)
trap 'rm -f "${STDERR_LOG}"' EXIT

"${BINARY}" merge \
    --output          "${OUTPUT}" \
    --force \
    --force-presence \
    "${sources[@]}" \
    2>"${STDERR_LOG}"

cat "${STDERR_LOG}" >&2
parse_reporter "${run_n}" "${#sources[@]}" "${STDERR_LOG}" >>"${CSV}"

echo "Done. Run ${run_n} → ${CSV}"