parse-md.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"question_number\": 7,\n",
      "  \"question_text\": \"The circle has centre $O$.\",\n",
      "  \"options\": [\n",
      "    {\n",
      "      \"number\": 1,\n",
      "      \"text\": \"$AE \\\\quad OC$\"\n",
      "    },\n",
      "    {\n",
      "      \"number\": 2,\n",
      "      \"text\": \"$AO \\\\quad BD$\"\n",
      "    },\n",
      "    {\n",
      "      \"number\": 3,\n",
      "      \"text\": \"$BD \\\\quad AE$\"\n",
      "    },\n",
      "    {\n",
      "      \"number\": 4,\n",
      "      \"text\": \"$OE \\\\quad FC$\"\n",
      "    }\n",
      "  ],\n",
      "  \"image_path\": \"https://cdn.mathpix.com/snip/images/EvYLPTzoMc2_WX4ixc6XAFmtTTafTJAkVe6bmcj6niI.original.fullsize.png\"\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "# use for single question testing\n",
    "# import json\n",
    "# import re\n",
    "# from typing import Dict, List, Any\n",
    "\n",
    "# def parse_mcq_question_text(text: str) -> Dict[str, Any]:\n",
    "#     # Extract question number and text\n",
    "#     match = re.match(r\"(\\d+)\\s+(.*?)\\n\", text)\n",
    "#     if not match:\n",
    "#         raise ValueError(\"Unable to extract question number and text\")\n",
    "#     question_number, question_text = match.groups()\n",
    "\n",
    "#     # Extract options\n",
    "#     options = re.findall(r\"\\((\\d+)\\)\\s+(.*?)(?=\\n\\(|\\n$|\\Z)\", text, re.DOTALL)\n",
    "    \n",
    "#     # Extract image path\n",
    "#     image_match = re.search(r\"!\\[\\]\\((.*?)\\)\", text)\n",
    "#     image_path = image_match.group(1) if image_match else None\n",
    "\n",
    "#     # Construct the JSON object\n",
    "#     question_data = {\n",
    "#         \"question_number\": int(question_number),\n",
    "#         \"question_text\": question_text.strip(),\n",
    "#         \"options\": [{\"number\": int(num), \"text\": text.strip()} for num, text in options],\n",
    "#         \"image_path\": image_path\n",
    "#     }\n",
    "\n",
    "#     return question_data\n",
    "\n",
    "# # Example usage\n",
    "# text = \"\"\"7 The circle has centre $O$.\n",
    "# AOE and COF are straight lines. Which pair of lines shows its radius and diameter?\n",
    "# ![](https://cdn.mathpix.com/snip/images/EvYLPTzoMc2_WX4ixc6XAFmtTTafTJAkVe6bmcj6niI.original.fullsize.png)\n",
    "# Radius $\\quad$ Diameter\n",
    "# (1) $AE \\quad OC$\n",
    "# (2) $AO \\quad BD$\n",
    "# (3) $BD \\quad AE$\n",
    "# (4) $OE \\quad FC$\"\"\"\n",
    "\n",
    "# parsed_json = parse_mcq_question_text(text)\n",
    "# print(json.dumps(parsed_json, indent=2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Error parsing question: Unable to extract question number and text for question 1\n",
      "JSONL file created: /Users/ob1/projects/sgllm/qna-proc/2021/questions.jsonl\n"
     ]
    }
   ],
   "source": [
    "import re\n",
    "import json\n",
    "from typing import List, Dict, Any\n",
    "\n",
    "def parse_mcq_questions_markdown_to_jsonl(file_path: str) -> str:\n",
    "    \"\"\"\n",
    "    Parse a markdown file containing multiple questions into a JSONL string.\n",
    "\n",
    "    Args:\n",
    "        file_path (str): Path to the markdown file.\n",
    "\n",
    "    Returns:\n",
    "        str: A JSONL string containing parsed question data.\n",
    "\n",
    "    Raises:\n",
    "        FileNotFoundError: If the specified file is not found.\n",
    "        ValueError: If there's an error parsing the markdown content.\n",
    "    \"\"\"\n",
    "    try:\n",
    "        with open(file_path, \"r\") as file:\n",
    "            content = file.read()\n",
    "    except FileNotFoundError:\n",
    "        raise FileNotFoundError(f\"The file at {file_path} was not found.\")\n",
    "\n",
    "    # Split the content into questions\n",
    "    questions = re.split(r\"\\n\\n(?=\\d+\\s)\", content.strip())\n",
    "\n",
    "    jsonl_data: List[str] = []\n",
    "\n",
    "    for i, question in enumerate(questions):\n",
    "        try:\n",
    "            question_data = parse_single_question(i, question)\n",
    "            jsonl_data.append(json.dumps(question_data))\n",
    "        except ValueError as e:\n",
    "            print(f\"Error parsing question: {e}\")\n",
    "\n",
    "    return \"\\n\".join(jsonl_data)\n",
    "\n",
    "def parse_single_question(i: int, question: str) -> Dict[str, Any]:\n",
    "    \"\"\"\n",
    "    Parse a single question from markdown format into a dictionary.\n",
    "\n",
    "    Args:\n",
    "        question (str): The markdown text of a single question.\n",
    "\n",
    "    Returns:\n",
    "        Dict[str, Any]: A dictionary containing parsed question data.\n",
    "\n",
    "    Raises:\n",
    "        ValueError: If unable to extract required information from the question.\n",
    "    \"\"\"\n",
    "    # Extract question number and text\n",
    "    match = re.match(r\"(\\d+)\\s+(.*?)(?:\\n\\(1\\)|\\n\\$\\$|\\Z)\", question, re.DOTALL)\n",
    "    if not match:\n",
    "        raise ValueError(f\"Unable to extract question number and text for question {i+1}\")\n",
    "    question_number, question_text = match.groups()\n",
    "\n",
    "    # Extract options\n",
    "    options = re.findall(r\"\\((\\d+)\\)\\s+(.*?)(?=\\n\\(\\d+\\)|\\n\\$\\$|\\Z)\", question, re.DOTALL)\n",
    "\n",
    "    # Extract LaTeX if present\n",
    "    latex_match = re.search(r\"\\$\\$(.*?)\\$\\$\", question, re.DOTALL)\n",
    "    latex = latex_match.group(1).strip() if latex_match else None\n",
    "\n",
    "    # Extract image path if present\n",
    "    image_match = re.search(r\"!\\[\\]\\((.*?)\\)\", question)\n",
    "    image_path = image_match.group(1) if image_match else None\n",
    "\n",
    "    # Construct the question data dictionary\n",
    "    question_data: Dict[str, Any] = {\n",
    "        \"question_number\": int(question_number),\n",
    "        \"question_text\": question_text.strip(),\n",
    "        \"options\": [{\"option_number\": int(num), \"option_text\": text.strip()} for num, text in options],\n",
    "        \"latex\": latex,\n",
    "        \"image_path\": image_path\n",
    "    }\n",
    "\n",
    "    return question_data\n",
    "\n",
    "# Usage TODO:\n",
    "file_path = \"/Users/ob1/projects/sgllm/qna-proc/2021/raw_mmd.md\"\n",
    "jsonl_output = parse_mcq_questions_markdown_to_jsonl(file_path)\n",
    "\n",
    "# Write to a new file\n",
    "output_file = \"/Users/ob1/projects/sgllm/qna-proc/2021/questions.jsonl\"\n",
    "with open(output_file, \"w\") as f:\n",
    "    f.write(jsonl_output)\n",
    "\n",
    "print(f\"JSONL file created: {output_file}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Error parsing question: Unable to extract question number and text for question 1\n",
      "JSONL file created: /Users/ob1/projects/sgllm/qna-proc/2021/p1-bookB/questions.jsonl\n"
     ]
    }
   ],
   "source": [
    "import re\n",
    "import json\n",
    "from typing import List, Dict, Any\n",
    "\n",
    "def parse_openended_questions_markdown_to_jsonl(file_path: str) -> str:\n",
    "    \"\"\"\n",
    "    Parse a markdown file containing multiple questions into a JSONL string.\n",
    "\n",
    "    Args:\n",
    "        file_path (str): Path to the markdown file.\n",
    "\n",
    "    Returns:\n",
    "        str: A JSONL string containing parsed question data.\n",
    "\n",
    "    Raises:\n",
    "        FileNotFoundError: If the specified file is not found.\n",
    "        ValueError: If there's an error parsing the markdown content.\n",
    "    \"\"\"\n",
    "    try:\n",
    "        with open(file_path, \"r\") as file:\n",
    "            content = file.read()\n",
    "    except FileNotFoundError:\n",
    "        raise FileNotFoundError(f\"The file at {file_path} was not found.\")\n",
    "\n",
    "    # Split the content into questions\n",
    "    questions = re.split(r\"\\n\\n(?=\\d+\\s)\", content.strip())\n",
    "\n",
    "    jsonl_data: List[str] = []\n",
    "\n",
    "    for i, question in enumerate(questions):\n",
    "        try:\n",
    "            question_data = parse_single_question(i, question)\n",
    "            jsonl_data.append(json.dumps(question_data))\n",
    "        except ValueError as e:\n",
    "            print(f\"Error parsing question: {e}\")\n",
    "\n",
    "    return \"\\n\".join(jsonl_data)\n",
    "\n",
    "def parse_single_question(i: int, question: str) -> Dict[str, Any]:\n",
    "    \"\"\"\n",
    "    Parse a single question from markdown format into a dictionary.\n",
    "\n",
    "    Args:\n",
    "        question (str): The markdown text of a single question.\n",
    "\n",
    "    Returns:\n",
    "        Dict[str, Any]: A dictionary containing parsed question data.\n",
    "\n",
    "    Raises:\n",
    "        ValueError: If unable to extract required information from the question.\n",
    "    \"\"\"\n",
    "    # Extract question number and text\n",
    "    match = re.match(r\"(\\d+)\\s+(.*?)(?:\\n\\(1\\)|\\n\\$\\$|\\Z)\", question, re.DOTALL)\n",
    "    if not match:\n",
    "        raise ValueError(f\"Unable to extract question number and text for question {i+1}\")\n",
    "    question_number, question_text = match.groups()\n",
    "\n",
    "    # Extract options\n",
    "    # options = re.findall(r\"\\((\\d+)\\)\\s+(.*?)(?=\\n\\(\\d+\\)|\\n\\$\\$|\\Z)\", question, re.DOTALL)\n",
    "\n",
    "    # Extract LaTeX if present\n",
    "    # latex_match = re.search(r\"\\$\\$(.*?)\\$\\$\", question, re.DOTALL)\n",
    "    # latex = latex_match.group(1).strip() if latex_match else None\n",
    "\n",
    "    # Extract answer if present (e.g. \\nAns: $\\qquad$) and remove it from question text\n",
    "    answer_match = re.search(r\"Ans:\\s+(.*)\", question)\n",
    "    answer_units = answer_match.group(1).strip() if answer_match else None\n",
    "    question_text = question_text.replace(f\"Ans: {answer_units}\", \"\").strip()\n",
    "\n",
    "    # Extract image path if present\n",
    "    image_match = re.search(r\"!\\[\\]\\((.*?)\\)\", question)\n",
    "    image_path = image_match.group(1) if image_match else None\n",
    "    # remove image_path from question text\n",
    "    question_text = question_text.replace(f\"![]({image_path})\", \"\").strip()\n",
    "\n",
    "    # Construct the question data dictionary\n",
    "    question_data: Dict[str, Any] = {\n",
    "        \"question_number\": int(question_number),\n",
    "        \"question_text\": question_text.strip(),\n",
    "        # \"options\": [{\"option_number\": int(num), \"option_text\": text.strip()} for num, text in options],\n",
    "        # \"latex\": latex,\n",
    "        \"image_path\": image_path,\n",
    "        # \"answer_units\": answer_units\n",
    "    }\n",
    "\n",
    "    return question_data\n",
    "\n",
    "# Usage TODO:\n",
    "file_path = \"/Users/ob1/projects/sgllm/qna-proc/2021/p1-bookB/rawmd.md\"\n",
    "jsonl_output = parse_openended_questions_markdown_to_jsonl(file_path)\n",
    "\n",
    "# Write to a new file\n",
    "output_file = \"/Users/ob1/projects/sgllm/qna-proc/2021/p1-bookB/questions.jsonl\"\n",
    "with open(output_file, \"w\") as f:\n",
    "    f.write(jsonl_output)\n",
    "\n",
    "print(f\"JSONL file created: {output_file}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}