{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c1dc1edc",
   "metadata": {},
   "source": [
    "# Load a knowledge base into an empty DAS"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96908c71",
   "metadata": {},
   "source": [
    "This notebook shows how to start an empty DAS and load a knowledge base into it.\n",
    "\n",
    "The first cell just imports the relevant class and instantiates a DAS object."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "755c65e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from das.distributed_atom_space import DistributedAtomSpace\$n",
    "import warnings\$n",
    "# avoids an annoying warning message from the Couchbase lib\$n",
    "warnings.filterwarnings('ignore')\n",
    "das = DistributedAtomSpace()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70362877",
   "metadata": {},
   "source": [
    "Point `KNOWLEDGE_BASE` to the file or folder where the knowledge base is. No tarballs or zip files here, only plain `.metta` or `.scm` files or folders with multiple files."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "50b6daa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "KNOWLEDGE_BASE = \"/tmp/samples\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5efb59cb",
   "metadata": {},
   "source": [
    "Select between the two load methods according to the knowledge base format. `load_canonical_knowledge_base()` is a lot faster but can be used only for `.metta` files which follows some extra assumptions:\n",
    "\n",
    "- The DBs are empty.\n",
    "- All MeTTa files have exactly one toplevel expression per line.\n",
    "- There are no empty lines.\n",
    "- Every \"named\" expressions (e.g. nodes) mentioned in a given\$n",
    "  expression is already mentioned in a typedef (i.e. something\$n",
    "  like (: \"my_node_name\" my_type)' previously IN THE SAME FILE).\n",
    "- Every type mentioned in a typedef is already defined IN THE SAME FILE.\n",
    "- All expressions are normalized (regarding separators, parenthesis etc)\n",
    "  like (: \"my_node_name\" my_type)' or\$n",
    "  (Evaluation \"name\" (Evaluation \"name\" (List \"name\" \"name\")))'. No tabs,\n",
    "  no double spaces, no spaces after (', etc.\n",
    "- All typedefs appear before any regular expressions\$n",
    "- Among typedefs, any terminal types (e.g. (: \"my_node_name\" my_type)') appear\$n",
    "  after all actual type definitions (e.g. (: Concept Type)')\n",
    "- No \"(\" or \")\" in atom names\$n",
    "- Flat type hierarchy (i.e. all types inherit from Type)\n",
    "\n",
    "Usually, \"canonical\" files are generated automatically by some conversion tool (e.g. `flybase2metta`)\n",
    "\n",
    "If the knowledge base is `.scm` or regular `.metta` file(s) then you should select `das.load_knowledge_base()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9480f15f",
   "metadata": {},
   "outputs": [],
   "source": [
    "das.clear_database()\n",
    "das.load_knowledge_base(KNOWLEDGE_BASE)\n",
    "#das.load_canonical_knowledge_base(KNOWLEDGE_BASE)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14317d6c",
   "metadata": {},
   "source": [
    "If the knowledge base is large, the load can take a time to finish. You can follow the progress by looking at `/tmp/das.log`. Once it's done, you should execute a count just to make sure it worked OK."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8443903f",
   "metadata": {},
   "outputs": [],
   "source": [
    "das.count_atoms()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}