personal-finance-database/notebooks/pdf_data_extraction_test.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "id": "b878d67d-c5a8-4817-99f3-4179edfd535a",
      "metadata": {
        "ExecuteTime": {
          "end_time": "2023-07-05T16:32:58.644004+00:00",
          "start_time": "2023-07-05T16:32:58.438814+00:00"
        },
        "noteable": {
          "cell_type": "code",
          "output_collection_id": "accfea1e-ca89-444f-9133-f39b76f10c8e"
        }
      },
      "outputs": [],
      "source": [
        "from pdfminer.high_level import extract_text\n",
        "import tabula\n",
        "\n",
        "class PdfParser:\n",
        "    def __init__(self, file_path):\n",
        "        self.file_path = file_path\n",
        "\n",
        "    def extract_text(self):\n",
        "        text = extract_text(self.file_path)\n",
        "        return text\n",
        "\n",
        "    def extract_table(self):\n",
        "        tables = tabula.read_pdf(self.file_path, pages='all')\n",
        "        return tables"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "id": "202c2c7a-a22f-468d-a67e-e10638fd96a2",
      "metadata": {
        "noteable": {
          "cell_type": "markdown"
        }
      },
      "source": [
        "Now, let's test the `PdfParser` class with a sample PDF file. Please replace `sample.pdf` with your actual PDF file path."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "id": "76db2644",
      "metadata": {},
      "outputs": [],
      "source": [
        "# pdf_file_path = r\"..\\data\\raw\\chris' statements\\savings\\sc\\sc_savings_eStatement_202306.pdf\"\n",
        "pdf_file_path = r\"..\\data\\raw\\chris' statements\\credit cards\\dbs\\dbs_card_9007_eStatement_202211.pdf\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e91a1abd-de76-4a12-adc5-fec5fd1e4301",
      "metadata": {
        "noteable": {
          "cell_type": "code"
        }
      },
      "outputs": [],
      "source": [
        "# replace 'sample.pdf' with your actual PDF file path\n",
        "pdf_parser = PdfParser(pdf_file_path)\n",
        "\n",
        "# extract text\n",
        "text = pdf_parser.extract_text()\n",
        "print(text)\n",
        "\n",
        "# extract tables\n",
        "tables = pdf_parser.extract_table()\n",
        "for table in tables:\n",
        "    display(table)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "60900e46",
      "metadata": {},
      "outputs": [],
      "source": [
        "len(tables)"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "finance",
      "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": "3.11.3"
    },
    "noteable": {
      "last_transaction_id": "73ebba71-e2fd-4f44-9b68-dbc9778c8fc8"
    },
    "noteable-chatgpt": {
      "create_notebook": {
        "openai_conversation_id": "179ca6e3-0377-5e6e-8f81-719779d73690",
        "openai_ephemeral_user_id": "ace69c27-3d03-5a21-855f-72ec4b037401",
        "openai_subdivision1_iso_code": "TW-TPE"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}