{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "30a7b117",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import os\n",
    "\n",
    "# Define the file paths\n",
    "file_paths = [\n",
    "    '/home/mshahidul/readctrl/data/reasoning/refined_evaluated_support_0_100_qwen3-32B.json',\n",
    "    '/home/mshahidul/readctrl/data/reasoning/refined_evaluated_support_100_200_qwen3-32B.json',\n",
    "    '/home/mshahidul/readctrl/data/reasoning/refined_evaluated_support_200_300_qwen3-32B.json'\n",
    "]\n",
    "\n",
    "merged_data = []\n",
    "\n",
    "# Loop through and append data\n",
    "for file_path in file_paths:\n",
    "    if os.path.exists(file_path):\n",
    "        with open(file_path, 'r', encoding='utf-8') as f:\n",
    "            data = json.load(f)\n",
    "            # Assuming each file contains a list of objects\n",
    "            if isinstance(data, list):\n",
    "                merged_data.extend(data)\n",
    "            else:\n",
    "                merged_data.append(data)\n",
    "        print(f\"Successfully loaded: {file_path}\")\n",
    "    else:\n",
    "        print(f\"Warning: File not found: {file_path}\")\n",
    "\n",
    "# Save the merged result\n",
    "output_path = '/home/mshahidul/readctrl/data/reasoning/refined_evaluated_support_merged_0_300_qwen3-32B.json'\n",
    "with open(output_path, 'w', encoding='utf-8') as f:\n",
    "    json.dump(merged_data, f, indent=4)\n",
    "\n",
    "print(f\"\\nTotal records merged: {len(merged_data)}\")\n",
    "print(f\"Merged file saved to: {output_path}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "27ab3270",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "# Define file paths\n",
    "readability_path = '/home/mshahidul/readctrl/data/classified_readability/classified_multiclinsum_test_en.json'\n",
    "reasoning_path = '/home/mshahidul/readctrl/data/reasoning/refined_evaluated_support_merged_0_300_qwen3-32B.json'\n",
    "output_path = '/home/mshahidul/readctrl/data/reasoning/merged_readability_reasoning_en_final.json'\n",
    "\n",
    "# 1. Load the readability data and create a lookup map\n",
    "with open(readability_path, 'r') as f:\n",
    "    readability_data = json.load(f)\n",
    "\n",
    "# Create a dictionary for O(1) lookup: {id: score}\n",
    "readability_lookup = {item['id']: item['readability_score'] for item in readability_data}\n",
    "\n",
    "# 2. Load the reasoning data\n",
    "with open(reasoning_path, 'r') as f:\n",
    "    reasoning_data = json.load(f)\n",
    "\n",
    "# 3. Merge the scores into the reasoning data\n",
    "merged_count = 0\n",
    "for entry in reasoning_data:\n",
    "    entry_id = entry.get('id')\n",
    "    if entry_id in readability_lookup:\n",
    "        # Add the score to the existing dictionary\n",
    "        entry['readability_score'] = readability_lookup[entry_id]\n",
    "        merged_count += 1\n",
    "    else:\n",
    "        # Optional: Handle cases where an ID is missing in the readability file\n",
    "        entry['readability_score'] = None\n",
    "\n",
    "# 4. Save the merged result\n",
    "with open(output_path, 'w') as f:\n",
    "    json.dump(reasoning_data, f, indent=4)\n",
    "\n",
    "print(f\"Successfully merged {merged_count} records. Saved to {output_path}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2ef2e0b6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Threshold set to: 90.0%\n",
      "Successfully saved 192 records to: /home/mshahidul/readctrl/data/final_result/processed_threshold_results.json\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import os\n",
    "\n",
    "# Configuration\n",
    "input_file = '/home/mshahidul/readctrl/data/reasoning/merged_readability_reasoning_en_final.json'\n",
    "output_dir = '/home/mshahidul/readctrl/data/final_result'\n",
    "output_filename = 'processed_threshold_results.json'\n",
    "\n",
    "# Set your threshold here (e.g., 0.90 for 90%, 0.85 for 85%)\n",
    "SUPPORT_THRESHOLD = 0.90 \n",
    "\n",
    "def process_with_threshold(threshold):\n",
    "    # Ensure the output folder exists\n",
    "    if not os.path.exists(output_dir):\n",
    "        os.makedirs(output_dir)\n",
    "\n",
    "    # Load the source data\n",
    "    try:\n",
    "        with open(input_file, 'r') as f:\n",
    "            data = json.load(f)\n",
    "    except FileNotFoundError:\n",
    "        print(f\"Error: Source file not found at {input_file}\")\n",
    "        return\n",
    "\n",
    "    final_output = []\n",
    "\n",
    "    for item in data:\n",
    "        evals = item.get('subclaim_evaluations', [])\n",
    "        \n",
    "        if not evals:\n",
    "            continue  # Skip items with no subclaims to evaluate\n",
    "            \n",
    "        # Calculate the percentage of supported subclaims\n",
    "        supported_count = sum(1 for sub in evals if sub.get('support_label') == 'supported')\n",
    "        support_ratio = supported_count / len(evals)\n",
    "        \n",
    "        # Keep if it meets the threshold (e.g., 0.90)\n",
    "        if support_ratio >= threshold:\n",
    "            clean_item = item.copy()\n",
    "            \n",
    "            # Map readability_score to difficulty\n",
    "            score = clean_item.get('readability_score', 0)\n",
    "            if score >= 4:\n",
    "                clean_item['difficulty'] = 'easy'\n",
    "            elif score == 3:\n",
    "                clean_item['difficulty'] = 'medium'\n",
    "            else:\n",
    "                clean_item['difficulty'] = 'hard'\n",
    "            \n",
    "            # Add metadata about the support ratio for transparency\n",
    "            clean_item['support_percentage'] = round(support_ratio * 100, 2)\n",
    "            \n",
    "            # Remove the subclaim_evaluations field\n",
    "            if 'subclaim_evaluations' in clean_item:\n",
    "                del clean_item['subclaim_evaluations']\n",
    "            \n",
    "            final_output.append(clean_item)\n",
    "\n",
    "    # Save to a single JSON file\n",
    "    target_path = os.path.join(output_dir, output_filename)\n",
    "    with open(target_path, 'w', encoding='utf-8') as out_f:\n",
    "        json.dump(final_output, out_f, indent=4, ensure_ascii=False)\n",
    "    \n",
    "    print(f\"Threshold set to: {threshold * 100}%\")\n",
    "    print(f\"Successfully saved {len(final_output)} records to: {target_path}\")\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    process_with_threshold(SUPPORT_THRESHOLD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "295a4a2a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Success! Merged data saved to: /home/mshahidul/readctrl/data/factual_testing/merged_evaluated_support_0_300.json\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import os\n",
    "\n",
    "# List of file paths to merge\n",
    "file_paths = [\n",
    "    '/home/mshahidul/readctrl/data/factual_testing/evaluated_support_0_100_qwen3-32B.json',\n",
    "    '/home/mshahidul/readctrl/data/factual_testing/evaluated_support_100_200_qwen3-32B.json',\n",
    "    '/home/mshahidul/readctrl/data/factual_testing/evaluated_support_200_300_qwen3-32B.json'\n",
    "]\n",
    "\n",
    "merged_data = []\n",
    "\n",
    "# Iterate through each file and append its contents to the list\n",
    "for file_path in file_paths:\n",
    "    if os.path.exists(file_path):\n",
    "        with open(file_path, 'r', encoding='utf-8') as f:\n",
    "            data = json.load(f)\n",
    "            # If the JSON is a list, extend the merged list\n",
    "            if isinstance(data, list):\n",
    "                merged_data.extend(data)\n",
    "            # If the JSON is a single dictionary, append it\n",
    "            else:\n",
    "                merged_data.append(data)\n",
    "    else:\n",
    "        print(f\"Warning: File not found - {file_path}\")\n",
    "\n",
    "# Save the combined data to a new file\n",
    "output_file = '/home/mshahidul/readctrl/data/factual_testing/merged_evaluated_support_0_300.json'\n",
    "\n",
    "with open(output_file, 'w', encoding='utf-8') as f:\n",
    "    json.dump(merged_data, f, indent=4)\n",
    "\n",
    "print(f\"Success! Merged data saved to: {output_file}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e7ba1534",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Updating scores for 100 documents...\n",
      "Successfully updated scores for 100 documents.\n",
      "File saved to: /home/mshahidul/readctrl/data/reasoning/updated_scores/refined_v2_full_evaluation_200_300_qwen3-32B.json\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import argparse\n",
    "import os\n",
    "\n",
    "def calculate_scores(data):\n",
    "    \"\"\"\n",
    "    Recalculates factual_attribution and completeness scores based on \n",
    "    the updated labels in attribution_details and completeness_details.\n",
    "    \"\"\"\n",
    "    updated_count = 0\n",
    "\n",
    "    for doc in data:\n",
    "        # 1. Recalculate Factual Attribution Score\n",
    "        attribution_list = doc.get('attribution_details', [])\n",
    "        if attribution_list:\n",
    "            supported_attr = sum(1 for item in attribution_list if item.get('label') == 'supported')\n",
    "            doc['scores']['factual_attribution'] = supported_attr / len(attribution_list)\n",
    "        else:\n",
    "            doc['scores']['factual_attribution'] = 0.0\n",
    "\n",
    "        # 2. Recalculate Completeness Score\n",
    "        completeness_list = doc.get('completeness_details', [])\n",
    "        if completeness_list:\n",
    "            supported_comp = sum(1 for item in completeness_list if item.get('present_in_summary') == 'supported')\n",
    "            doc['scores']['completeness'] = supported_comp / len(completeness_list)\n",
    "        else:\n",
    "            doc['scores']['completeness'] = 0.0\n",
    "            \n",
    "        updated_count += 1\n",
    "\n",
    "    return data, updated_count\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    # parser = argparse.ArgumentParser(description=\"Update scores in refined clinical evaluation JSON.\")\n",
    "    # parser.add_argument(\"--input_file\", type=str, required=True, help=\"Path to the refined JSON file.\")\n",
    "    # parser.add_argument(\"--output_file\", type=str, help=\"Path to save the updated JSON. If omitted, overwrites input.\")\n",
    "    # args = parser.parse_args()\n",
    "    input_file = '/home/mshahidul/readctrl/data/reasoning/refined_v2_full_evaluation_200_300_qwen3-32B.json'\n",
    "    output_path = \"/home/mshahidul/readctrl/data/reasoning/updated_scores\"\n",
    "    output_file = os.path.join(output_path, os.path.basename(input_file))\n",
    "    # Load data\n",
    "    with open(input_file, 'r') as f:\n",
    "        data = json.load(f)\n",
    "\n",
    "    print(f\"Updating scores for {len(data)} documents...\")\n",
    "    \n",
    "    # Process\n",
    "    updated_data, count = calculate_scores(data)\n",
    "\n",
    "   \n",
    "    \n",
    "    # Save results\n",
    "    with open(output_file, 'w') as f:\n",
    "        json.dump(updated_data, f, indent=2, ensure_ascii=False)\n",
    "\n",
    "    print(f\"Successfully updated scores for {count} documents.\")\n",
    "    print(f\"File saved to: {output_file}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "612109dc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['index', 'id', 'fulltext', 'fulltext_subclaims', 'summary', 'summary_subclaims', 'diff_label_texts', 'diff_label_subclaims', 'readability_score'])\n",
      "dict_keys(['low_health_literacy', 'intermediate_health_literacy', 'proficient_health_literacy'])\n",
      "dict_keys(['low_health_literacy', 'intermediate_health_literacy', 'proficient_health_literacy'])\n"
     ]
    }
   ],
   "source": [
    "# /home/mshahidul/readctrl/data/extracting_subclaim/extracted_subclaims_syn_data_with_gs_summary_en.json\n",
    "import json\n",
    "with open('/home/mshahidul/readctrl/data/extracting_subclaim/extracted_subclaims_syn_data_with_gs_summary_en.json', 'r') as f:\n",
    "    anno_data = json.load(f)\n",
    "print(anno_data[0].keys())\n",
    "print(anno_data[0]['diff_label_texts'].keys())\n",
    "print(anno_data[0]['diff_label_subclaims'].keys())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "un",
   "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": "3.11.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}