#!/usr/bin/env bash
# =============================================================================
# orphan-wiki.sh — find source pages whose raw source no longer exists.
# Reads source_path from each wiki/sources/*.md frontmatter. If the raw is gone,
# the page is orphaned. Emits JSON envelope: {status, genome, count, files[], detail[]}.
# Read-only: no lock needed (same policy as pending-raw).
# =============================================================================
set -euo pipefail

genome="${1:?usage: orphan-wiki.sh <genome>}"
base_dir="${GENOMES_ROOT:-${HOME}/genomes}"
cd "${base_dir}/${genome}" 2>/dev/null || { echo '{"status":"error","reason":"unknown genome"}'; exit 1; }

# Clean start on the configured base (single source of truth in lib/clean-start.sh).
: "${KG_LIB_DIR:=${HOME}/knowledge-genome-orchestrator/lib}"
source "${KG_LIB_DIR}/clean-start.sh" 2>/dev/null \
  || { echo '{"status":"error","reason":"clean-start.sh not found"}'; exit 1; }
clean_start || { echo '{"status":"error","reason":"clean-start failed"}'; exit 1; }

declare -a ORPH=()
for page in wiki/sources/*.md; do
  [[ -e "$page" ]] || continue
  sp="$(sed -n 's/^source_path:[[:space:]]*//p' "$page" | tr -d '\r' | head -n1)"
  # Pages without source_path are pre-Step-2 legacy: ignore, don't false-positive.
  [[ -n "$sp" ]] || continue
  [[ -f "$sp" ]] || ORPH+=("$page")
done

if [[ ${#ORPH[@]} -eq 0 ]]; then
  echo '{"status":"ok","genome":"'"$genome"'","count":0,"files":[],"detail":[]}'
else
  for x in "${ORPH[@]}"; do printf '%s\torphan\n' "$x"; done \
    | jq -R 'split("\t") | {path: .[0], reason: .[1]}' \
    | jq -s --arg g "$genome" '{status:"ok", genome:$g, count:length, files:[.[].path], detail:.}'
fi