#!/usr/bin/env bash
# vendor_enricher_python.sh — copia las funciones Python del registry
# que un enricher declara en `uses_functions` a su directorio
# `_vendored/`. El run.py importa de `_vendored/` en lugar de
# `<registry_root>/python/functions/`, lo que hace al binario
# distribuible sin acceso al fn_registry.
#
# Issue 0033b.
#
# Uso:
#   tools/vendor_enricher_python.sh <enricher_dir> [<registry_root>]
#
# Lee `uses_functions` del manifest YAML, filtra IDs `*_py_*`,
# resuelve `file_path` desde registry.db, copia los .py y todas las
# importaciones siblings dentro del mismo dominio (transitivo).
#
# Genera `.vendor.lock` con `<id>  <sha256>  <src_path>` para
# auditoria. Idempotente — si los hashes coinciden, no copia.
#
# Salida:
#   <enricher_dir>/_vendored/__init__.py
#   <enricher_dir>/_vendored/<domain>/__init__.py
#   <enricher_dir>/_vendored/<domain>/<filename>.py
#   <enricher_dir>/.vendor.lock

set -euo pipefail

ENR_DIR="${1:?enricher_dir requerido}"
REGISTRY_ROOT="${2:-${REGISTRY_ROOT:-$(pwd)}}"

if [[ ! -f "$ENR_DIR/manifest.yaml" ]]; then
    echo "ERROR: $ENR_DIR/manifest.yaml no existe" >&2
    exit 1
fi
if [[ ! -f "$REGISTRY_ROOT/registry.db" ]]; then
    echo "ERROR: $REGISTRY_ROOT/registry.db no existe (REGISTRY_ROOT incorrecto)" >&2
    exit 2
fi

VENDOR="$ENR_DIR/_vendored"
LOCK="$ENR_DIR/.vendor.lock"

# ----------------------------------------------------------------------------
# Leer uses_functions del manifest (subset YAML soportado por el indexer C++).
# Acepta forma inline `[a, b]` o lista en lineas indentadas con `- `.
# ----------------------------------------------------------------------------
ids=$(awk '
    /^uses_functions:[[:space:]]*\[/ {
        line = $0
        sub(/^uses_functions:[[:space:]]*\[/, "", line)
        sub(/\].*$/, "", line)
        gsub(/[",]/, " ", line)
        print line
        exit
    }
    /^uses_functions:[[:space:]]*$/ { collecting = 1; next }
    collecting && /^[[:space:]]*-[[:space:]]+/ {
        sub(/^[[:space:]]*-[[:space:]]+/, "")
        sub(/[[:space:]]*#.*$/, "")
        gsub(/[\047"]/, "")
        print
        next
    }
    collecting && /^[^[:space:]-]/ { collecting = 0 }
' "$ENR_DIR/manifest.yaml" | tr ' ' '\n' | awk 'NF' | grep '_py_' || true)

if [[ -z "$ids" ]]; then
    # No hay nada que vendorizar — limpiamos vendor/ y lock por si
    # quedaron de un manifest anterior.
    rm -rf "$VENDOR" "$LOCK"
    echo "vendor: $ENR_DIR — sin uses_functions Python"
    exit 0
fi

# ----------------------------------------------------------------------------
# Para cada ID, obtener file_path de registry.db. Construir lista de
# (id, abs_src_path, domain, basename) — la unidad atomica de copia.
# ----------------------------------------------------------------------------
declare -A SEEN     # paths absolutos ya procesados (dedup)
declare -A LOCK_NEW # nueva tabla de lock: id -> sha256 src_path

queue=()
for id in $ids; do
    fp=$(sqlite3 "$REGISTRY_ROOT/registry.db" \
        "SELECT file_path FROM functions WHERE id='$id';")
    if [[ -z "$fp" ]]; then
        echo "WARN: $id no esta en registry.db (skip)" >&2
        continue
    fi
    abs="$REGISTRY_ROOT/$fp"
    if [[ ! -f "$abs" ]]; then
        echo "WARN: file_path '$fp' no existe (skip $id)" >&2
        continue
    fi
    queue+=("$id|$abs")
done

if [[ ${#queue[@]} -eq 0 ]]; then
    rm -rf "$VENDOR" "$LOCK"
    echo "vendor: $ENR_DIR — ninguna funcion Python resoluble"
    exit 0
fi

# ----------------------------------------------------------------------------
# Idempotencia: comparar hashes vs lock existente. Si todos coinciden,
# salir sin tocar nada.
# ----------------------------------------------------------------------------
state_hash=""
for entry in "${queue[@]}"; do
    id="${entry%%|*}"
    abs="${entry#*|}"
    sha=$(sha256sum "$abs" | cut -d' ' -f1)
    state_hash+="$id|$sha"$'\n'
done
state_sha=$(echo -n "$state_hash" | sha256sum | cut -d' ' -f1)

if [[ -f "$LOCK" ]]; then
    cur_state=""
    while IFS= read -r line; do
        # formato: "<id>  <sha256>  <path>"
        eid=$(echo "$line" | awk '{print $1}')
        esha=$(echo "$line" | awk '{print $2}')
        cur_state+="$eid|$esha"$'\n'
    done < "$LOCK"
    cur_sha=$(echo -n "$cur_state" | sha256sum | cut -d' ' -f1)
    if [[ "$cur_sha" == "$state_sha" ]]; then
        echo "vendor: $ENR_DIR — sin cambios (.vendor.lock OK)"
        exit 0
    fi
fi

# ----------------------------------------------------------------------------
# Copia con expansion transitiva de imports siblings dentro del mismo
# dominio. Si un .py vendorizado tiene `from X import Y` o
# `import X` donde X es un modulo del mismo directorio, X.py tambien
# se copia (solo si existe junto al fuente original).
# ----------------------------------------------------------------------------
rm -rf "$VENDOR"
mkdir -p "$VENDOR"
touch "$VENDOR/__init__.py"

copy_with_siblings() {
    local abs="$1"
    if [[ -n "${SEEN[$abs]:-}" ]]; then return 0; fi
    SEEN[$abs]=1

    # Inferir dominio del path: <root>/python/functions/<domain>/<file>.py
    local rel
    rel=$(realpath --relative-to="$REGISTRY_ROOT" "$abs")
    local domain
    domain=$(echo "$rel" | awk -F/ '{print $(NF-1)}')
    local fname
    fname=$(basename "$abs")
    local dst_dir="$VENDOR/$domain"
    mkdir -p "$dst_dir"
    touch "$dst_dir/__init__.py"
    cp "$abs" "$dst_dir/$fname"

    # Escanear imports siblings: lineas `from <name> import` o
    # `import <name>` donde <name>.py existe en el mismo dir que abs.
    local src_dir
    src_dir=$(dirname "$abs")
    local sibling_names
    sibling_names=$(grep -E '^[[:space:]]*(from [a-zA-Z_][a-zA-Z0-9_]+ import|import [a-zA-Z_][a-zA-Z0-9_]+)' "$abs" \
        | sed -E 's/^[[:space:]]*from ([a-zA-Z_][a-zA-Z0-9_]+).*/\1/; s/^[[:space:]]*import ([a-zA-Z_][a-zA-Z0-9_]+).*/\1/' \
        | sort -u)
    for name in $sibling_names; do
        local sib="$src_dir/$name.py"
        if [[ -f "$sib" && "$sib" != "$abs" ]]; then
            copy_with_siblings "$sib"
        fi
    done
}

> "$LOCK.tmp"
for entry in "${queue[@]}"; do
    id="${entry%%|*}"
    abs="${entry#*|}"
    copy_with_siblings "$abs"
    sha=$(sha256sum "$abs" | cut -d' ' -f1)
    rel=$(realpath --relative-to="$REGISTRY_ROOT" "$abs")
    echo "$id  $sha  $rel" >> "$LOCK.tmp"
done

mv "$LOCK.tmp" "$LOCK"
n=$(wc -l < "$LOCK")
echo "vendor: $ENR_DIR — $n funcs declaradas, $(find "$VENDOR" -name '*.py' | wc -l) archivos copiados"