clawdie-ai/scripts/memory/memory-pg.sh

#!/usr/bin/env bash
# memory-pg.sh — PostgreSQL-backed memory operations
#
# Usage:
#   ./memory-pg.sh store "summary text" [topics] [importance]
#   ./memory-pg.sh search "query text" [limit]
#   ./memory-pg.sh recent [limit]
#   ./memory-pg.sh important [limit]
#   ./memory-pg.sh count

. "$(dirname "$0")/common.sh"

DISPLAY_DATETIME_SQL=$(pg_display_datetime_expr "created_at")

# Sanitize an integer argument (for LIMIT clauses which don't support psql variables)
safe_int() {
    local val="$1"
    if ! [[ "$val" =~ ^[0-9]+$ ]]; then
        echo "Error: expected integer, got '$val'" >&2
        exit 1
    fi
    echo "$val"
}

# ── Store a new memory with automatic chunking + embedding ──
cmd_store() {
    local summary="$1"
    local topics="${2:-}"
    local importance="${3:-2}"

    # Parse topics into PostgreSQL array
    local pg_topics="{}"
    if [ -n "$topics" ]; then
        # Convert comma-separated to PostgreSQL array: "a,b,c" → "{a,b,c}"
        pg_topics="{$topics}"
    fi

    # Escape summary for SQL
    local escaped_summary
    escaped_summary=$(python3 -c "import sys; print(sys.argv[1].replace(\"'\", \"''\"))" "$summary")
    local safe_importance
    safe_importance=$(safe_int "$importance")

    # Insert memory
    local memory_id
    memory_id=$(pg -c "
        INSERT INTO memories (summary, topics, importance)
        VALUES ('$escaped_summary', '$pg_topics'::text[], $safe_importance)
        RETURNING id;
    ")

    if [ -z "$memory_id" ]; then
        echo "Error: failed to insert memory" >&2
        exit 1
    fi

    echo "Stored memory: $memory_id"

    # Chunk the summary
    local chunk_order=0
    while IFS= read -r chunk; do
        [ -z "$chunk" ] && continue

        local content_hash
        content_hash=$(echo -n "$chunk" | md5)

        # Escape chunk text for SQL
        local escaped_chunk
        escaped_chunk=$(python3 -c "import sys; print(sys.argv[1].replace(\"'\", \"''\"))" "$chunk")

        # Insert chunk
        local chunk_id
        chunk_id=$(pg -c "
            INSERT INTO memory_chunks (memory_id, chunk_order, chunk_text, content_hash)
            VALUES ('$memory_id', $chunk_order, '$escaped_chunk', '$content_hash')
            RETURNING id;
        ")

        # Generate and store embedding
        local embedding
        embedding=$("$SCRIPT_DIR/embed.sh" "$chunk")

        pg -c "
            INSERT INTO memory_embeddings (chunk_id, embedding, embedding_provider, embedding_model)
            VALUES ('$chunk_id', '$embedding'::vector, '$EMBED_PROVIDER', '$EMBED_MODEL');
        " >/dev/null

        echo "  chunk $chunk_order embedded (${#chunk} chars)"
        chunk_order=$((chunk_order + 1))
    done < <("$SCRIPT_DIR/chunk.sh" "$summary")

    echo "Done: $chunk_order chunks embedded"
}

# ── Hybrid search ──
cmd_search() {
    local query="$1"
    local safe_limit
    safe_limit=$(safe_int "${2:-5}")

    # Generate query embedding
    local query_embedding
    query_embedding=$("$SCRIPT_DIR/embed.sh" "$query")

    # Escape query for SQL (single quotes → '')
    local escaped_query
    escaped_query=$(python3 -c "import sys; print(sys.argv[1].replace(\"'\", \"''\"))" "$query")

    pg_pretty -c "
        SELECT
            left(chunk_text, 80) AS match,
            importance,
            topics,
            round(combined_score::numeric, 6) AS score,
            ${DISPLAY_DATETIME_SQL} AS created
        FROM search_memories('$escaped_query', '$query_embedding'::vector, $safe_limit);
    "
}

# ── Recent memories ──
cmd_recent() {
    local safe_limit
    safe_limit=$(safe_int "${1:-5}")

    pg_pretty -c "
        SELECT
            ${DISPLAY_DATETIME_SQL} AS created,
            importance AS imp,
            left(summary, 80) AS summary,
            topics
        FROM memories
        ORDER BY created_at DESC
        LIMIT $safe_limit;
    "
}

# ── High-importance memories ──
cmd_important() {
    local safe_limit
    safe_limit=$(safe_int "${1:-10}")

    pg_pretty -c "
        SELECT
            ${DISPLAY_DATETIME_SQL} AS created,
            importance AS imp,
            left(summary, 80) AS summary,
            topics
        FROM memories
        WHERE importance >= 4
        ORDER BY importance DESC, created_at DESC
        LIMIT $safe_limit;
    "
}

# ── Count ──
cmd_count() {
    echo "memories:  $(pg -c "SELECT count(*) FROM memories;")"
    echo "chunks:    $(pg -c "SELECT count(*) FROM memory_chunks;")"
    echo "embeddings:$(pg -c "SELECT count(*) FROM memory_embeddings;")"
}

# ── Main dispatcher ──
case "${1:-help}" in
    store)     cmd_store "${2:?Summary required}" "${3:-}" "${4:-2}" ;;
    search)    cmd_search "${2:?Query required}" "${3:-5}" ;;
    recent)    cmd_recent "${2:-5}" ;;
    important) cmd_important "${2:-10}" ;;
    count)     cmd_count ;;
    *)
        echo "Usage: memory-pg.sh <command> [args]"
        echo ""
        echo "Commands:"
        echo "  store <summary> [topics] [importance]  Store a memory with auto-chunking + embedding"
        echo "  search <query> [limit]                 Hybrid search (full-text + vector)"
        echo "  recent [limit]                         Show recent memories"
        echo "  important [limit]                      Show high-importance memories"
        echo "  count                                  Show table counts"
        ;;
esac