{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "bed45d12-7681-4ba4-9c89-48a3515704e2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "File ‘hinglishNorm.json’ already there; not retrieving.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget -nc https://raw.githubusercontent.com/piyushmakhija5/hinglishNorm/master/dataset/hinglishNorm.json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "965589a9-c62e-4659-a6bc-6f0a2bad5d19",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['id', 'inputText', 'tags', 'normalizedText'], dtype='object')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_json(\"./hinglishNorm.json\")\n",
    "df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b5c7c7c7-b9a6-4ea2-a5ef-edaf982ae0ad",
   "metadata": {},
   "source": [
    "### Required columns\n",
    "- target_hinglish\n",
    "- source_hindi\n",
    "- parallel_english\n",
    "- annotations\n",
    "- raw_input\n",
    "- alternates\n",
    "\n",
    "> For **hinglishNorm**, only `target_hinglish`, `raw_input` and `annotations` are valid\n",
    "\n",
    "### Mappings\n",
    "- `normalizedText` _=>_ `target_hinglish`\n",
    "- `inputText` _=>_ `raw_input`\n",
    "- `tags` _=>_ `annotations` (after json.loads)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b54fdd52-1ab0-4c84-89e5-0bcb8fcbfbeb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add empty columns\n",
    "df[\"source_hindi\"] = \\\n",
    "  df[\"parallel_english\"] = \\\n",
    "  df[\"alternates\"] = None\n",
    "\n",
    "# Remove unnecessary columns\n",
    "df = df.drop(\"id\", axis=1)\n",
    "\n",
    "# Rename columns\n",
    "df = df.rename(columns={\n",
    "    \"normalizedText\": \"target_hinglish\", \n",
    "    \"inputText\": \"raw_input\", \n",
    "    \"tags\": \"annotations\", })\n",
    "\n",
    "# Parse annotations json\n",
    "import json\n",
    "df[\"annotations\"] = df[\"annotations\"].map(lambda x: json.loads(x.replace(\"'\", '\"')))\n",
    "\n",
    "# Split dataset\n",
    "from sklearn.model_selection import train_test_split\n",
    "_train_eval_df, test_df = train_test_split(df, test_size=0.1)\n",
    "train_df, eval_df = train_test_split(_train_eval_df, test_size=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "6e804366-34cd-45c7-b3c6-46b7b8c1b420",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting tables\n",
      "  Downloading tables-3.7.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (5.9 MB)\n",
      "     |████████████████████████████████| 5.9 MB 4.9 MB/s            \n",
      "\u001b[?25hRequirement already satisfied: numpy>=1.19.0 in /opt/conda/lib/python3.7/site-packages (from tables) (1.19.5)\n",
      "Requirement already satisfied: packaging in /opt/conda/lib/python3.7/site-packages (from tables) (21.3)\n",
      "Collecting numexpr>=2.6.2\n",
      "  Downloading numexpr-2.8.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (379 kB)\n",
      "     |████████████████████████████████| 379 kB 73.4 MB/s            \n",
      "\u001b[?25hRequirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /opt/conda/lib/python3.7/site-packages (from packaging->tables) (3.0.6)\n",
      "Installing collected packages: numexpr, tables\n",
      "Successfully installed numexpr-2.8.1 tables-3.7.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/pandas/core/generic.py:2718: PerformanceWarning: \n",
      "your performance may suffer as PyTables will pickle object types that it cannot\n",
      "map directly to c-types [inferred_type->mixed,key->block0_values] [items->Index(['raw_input', 'annotations', 'target_hinglish', 'source_hindi',\n",
      "       'parallel_english', 'alternates'],\n",
      "      dtype='object')]\n",
      "\n",
      "  encoding=encoding,\n"
     ]
    }
   ],
   "source": [
    "!pip install tables\n",
    "\n",
    "# Save to hdfs files\n",
    "train_df.to_hdf(\"./data.h5\", \"train\", complevel=9)\n",
    "test_df.to_hdf(\"./data.h5\", \"test\", complevel=9)\n",
    "eval_df.to_hdf(\"./data.h5\", \"eval\", complevel=9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c8908455-c76f-4ee2-9608-b215f6fafa7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Confirm that everything worked as expected\n",
    "\n",
    "# Load from hdfs files\n",
    "_train_df = pd.read_hdf(\"./data.h5\", \"train\")\n",
    "_test_df = pd.read_hdf(\"./data.h5\", \"test\")\n",
    "_eval_df = pd.read_hdf(\"./data.h5\", \"eval\")\n",
    "\n",
    "assert (len(_train_df) == len(train_df)) == \\\n",
    "  (len(_eval_df) == len(eval_df)) == \\\n",
    "  (len(_test_df) == len(test_df))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "60461121-bed5-4ba0-ba7d-dd46256c62e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "!rm hinglishNorm.json"
   ]
  }
 ],
 "metadata": {
  "environment": {
   "kernel": "python3",
   "name": "managed-notebooks.m87",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/base-cu110:latest"
  },
  "kernelspec": {
   "display_name": "Python (Local)",
   "language": "python",
   "name": "local-base"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}