#!/usr/bin/env bash
# okf-normalize-types.sh — normaliza type taxonomy para vocabulário canónico OKF
# Uso: okf-normalize-types.sh [--dry-run]
# §5 do OKF-Compliance-Plan-Global.md

set -euo pipefail

HUB_ROOT="/media/ealmeida/Dados/Hub"
DRY_RUN=false
[[ "${1:-}" == "--dry-run" ]] && DRY_RUN=true

# Canonical type mappings (lowercase → Canonical)
declare -A TYPE_MAP=(
    ["procedimento"]="Playbook"
    ["procedure"]="Playbook"
    ["playbook"]="Playbook"
    ["reference"]="Reference"
    ["api"]="Reference"
    ["spec"]="Specification"
    ["index"]="Index"
    ["changelog"]="Changelog"
    ["status"]="Status"
    ["document"]="Document"
    ["template"]="Template"
    ["client profile"]="Client Profile"
    ["note"]="note"
)

is_excluded() {
    local rel="$1"
    [[ "$rel" == *99-Arquivo/* ]] && return 0
    [[ "$rel" == *.stversions/* ]] && return 0
    [[ "$rel" == *.obsidian/* ]] && return 0
    [[ "$rel" == *.ijfw/* ]] && return 0
    [[ "$rel" == *_templates/* ]] && return 0
    [[ "$rel" == *90-Templates/* ]] && return 0
    [[ "$rel" == *.git/* ]] && return 0
    [[ "$rel" == *node_modules/* ]] && return 0
    [[ "$rel" == *.github/* ]] && return 0
    [[ "$rel" == *.wayland/* ]] && return 0
    [[ "$rel" == *.hermes/* ]] && return 0
    local base
    base="$(basename "$rel")"
    [[ "$base" == "MEMORY.md" ]] && return 0
    [[ "$base" == "CLAUDE.md" ]] && return 0
    [[ "$base" == "GEMINI.md" ]] && return 0
    [[ "$base" == "AGENTS.md" ]] && return 0
    return 1
}

CHANGED=0
TOTAL=0

echo "Normalizing type taxonomy in $HUB_ROOT..."
[[ "$DRY_RUN" == "true" ]] && echo "(DRY-RUN mode — no files modified)"
echo ""

while IFS= read -r -d '' f; do
    rel="${f#$HUB_ROOT/}"
    is_excluded "$rel" && continue

    # Get current type
    current_type="$(awk 'BEGIN{n=0} /^---[[:space:]]*$/{n++; if(n==2) exit} n==1 && /^type:/{print; exit}' "$f" 2>/dev/null || true)"
    [[ -z "$current_type" ]] && continue

    # Extract value after "type:"
    type_val="$(echo "$current_type" | sed 's/^type: *//' | tr -d '"' | tr -d "'" | xargs)"

    # Check if needs normalization
    type_lower="$(echo "$type_val" | tr '[:upper:]' '[:lower:]')"
    canonical="${TYPE_MAP[$type_lower]:-}"

    if [[ -n "$canonical" ]] && [[ "$canonical" != "$type_val" ]]; then
        ((TOTAL++)) || true
        echo "  $rel: '$type_val' → '$canonical'"
        if [[ "$DRY_RUN" != "true" ]]; then
            sed -i "s/^type:.*$/type: $canonical/" "$f"
            ((CHANGED++)) || true
        fi
    fi
done < <(find "$HUB_ROOT" -name '*.md' -not -path '*/.git/*' -not -path '*/node_modules/*' -print0)

echo ""
echo "────────────────────────────────────"
echo "To normalize: $TOTAL | Changed: $CHANGED"
[[ "$DRY_RUN" == "true" ]] && echo "(Re-run without --dry-run to apply)"