PDF to JSON format using python

.ipynb_checkpoints/program-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

code/config.json

@@ -0,0 +1,84 @@
+{
+    "Identification": {
+        "Product Name": "Product Name[:]*\\s*(.*)",
+        "Cat No.": "Cat No.[:]*\\s*(.*)",
+        "Synonyms": "Synonyms[:]*\\s*(.*)",
+        "Recommended Use": "Recommended Use[:]*\\s*(.*)",
+        "Uses advised against": "Uses advised against[:]*\\s*(.*)",
+        "Company": {
+            "Name": "Company[:]*\\s*(.*)",
+            "Address": "Company[:]*\\s*.*\\s*(.*\\s.*\\s.*\\d{5})",
+            "Telephone": "Tel[:]*\\s*(\\(\\d{3}\\) \\d{3}-\\d{4})"
+        }
+
+    },
+    "Emergency Contact": {
+        "Emergency Telephone Number": "Emergency Telephone Number[:]*\\s*(.*)"
+    },
+    "Hazard Identification": {
+        "Classification": "Classification[:]*\\s*(.*)",
+        "Signal Word": "Signal Word[:]*\\s*(.*)",
+        "Hazard Statements": "Hazard Statements[:]*\\s*(.*)",
+        "Precautionary Statements": {
+            "Prevention": "Prevention[:]*\\s*((?:.*\\n)+?)(?=(Eyes|Disposal|Hazards not otherwise classified|\\Z))",
+            "Response": "Response[:]*\\s*(.*)(?=Eyes|\\nEyes)",
+            "Eyes": "Eyes[:]*\\s*(.*)",
+            "Disposal": "Disposal[:]*\\s*(.*)",
+            "Hazards not otherwise classified (HNOC)": "Hazards not otherwise classified[:]*\\s*(.*)"
+        }
+    },
+    "Composition/Information on Ingredients": {
+        "Component": "Component[:]*\\s*(.*)",
+        "CAS No": "CAS No[:]*\\s*(.*)",
+        "Weight %": "Weight %[:]*\\s*(.*)"
+    },
+    "First Aid Measures": {
+        "General Advice": "General Advice[:]*\\s*(.*)",
+        "Eye Contact": "Eye Contact[:]*\\s*(.*)",
+        "Skin Contact": "Skin Contact[:]*\\s*(.*)",
+        "Inhalation": "Inhalation[:]*\\s*(.*)",
+        "Ingestion": "Ingestion[:]*\\s*(.*)"
+    },
+    "Fire Fighting Measures": {
+        "Flash Point": "Flash Point[:]*\\s*(.*)",
+        "Autoignition Temperature": "Autoignition Temperature[:]*\\s*(.*)",
+        "Explosion Limits": "Explosion Limits[:]*\\s*(.*)"
+    },
+    "Handling and Storage": {
+        "Handling": "Handling[:]*\\s*(.*)",
+        "Storage": "Storage[:]*\\s*(.*)"
+    },
+    "Exposure Controls": {
+        "Engineering Measures": "Engineering Measures[:]*\\s*(.*)",
+        "Eye/face Protection": "Eye/face Protection[:]*\\s*(.*)",
+        "Skin Protection": "Skin Protection[:]*\\s*(.*)",
+        "Respiratory Protection": "Respiratory Protection[:]*\\s*(.*)"
+    },
+    "Physical and Chemical Properties": {
+        "Physical State": "Physical State[:]*\\s*(.*)",
+        "Appearance": "Appearance[:]*\\s*(.*)",
+        "Odor": "Odor[:]*\\s*(.*)",
+        "pH": "pH[:]*\\s*(.*)",
+        "Melting Point/Range": "Melting Point/Range[:]*\\s*(.*)",
+        "Boiling Point/Range": "Boiling Point/Range[:]*\\s*(.*)",
+        "Vapor Pressure": "Vapor Pressure[:]*\\s*(.*)",
+        "Solubility": "Solubility[:]*\\s*(.*)"
+    },
+    "Stability and Reactivity": {
+        "Stability": "Stability[:]*\\s*(.*)",
+        "Incompatible Materials": "Incompatible Materials[:]*\\s*(.*)"
+    },
+    "Toxicological Information": {
+        "Acute Toxicity": "Acute Toxicity[:]*\\s*(.*)",
+        "Chronic Effects": "Chronic Effects[:]*\\s*(.*)"
+    },
+    "Ecological Information": {
+        "Ecotoxicity": "Ecotoxicity[:]*\\s*(.*)"
+    },
+    "Disposal Considerations": {
+        "Waste Disposal Methods": "Waste Disposal Methods[:]*\\s*(.*)"
+    },
+    "Transport Information": {
+        "DOT": "DOT[:]*\\s*(.*)"
+    }
+}

code/program.ipynb

@@ -0,0 +1,114 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "106932cf",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Processed edta--ph--lt.pdf and saved JSON data to edta--ph--lt.json\n"
+     ]
+    }
+   ],
+   "source": [
+    "import pdfplumber\n",
+    "import json\n",
+    "import re\n",
+    "import os\n",
+    "\n",
+    "def extract_text_from_pdf_plumber(pdf_path):\n",
+    "    with pdfplumber.open(pdf_path) as pdf:\n",
+    "        text = \"\"\n",
+    "        for page in pdf.pages:\n",
+    "            page_text = page.extract_text()\n",
+    "            text += page_text + \"\\n\"\n",
+    "    return text\n",
+    "\n",
+    "def load_patterns(config_path):\n",
+    "    with open(config_path, 'r') as config_file:\n",
+    "        return json.load(config_file)\n",
+    "\n",
+    "def parse_sds_text(text, patterns):\n",
+    "    sds_dict = {}\n",
+    "\n",
+    "    for section, field_patterns in patterns.items():\n",
+    "        sds_dict[section] = {}\n",
+    "\n",
+    "        for field, pattern in field_patterns.items():\n",
+    "            if isinstance(pattern, dict):\n",
+    "                # Handle nested dictionary (e.g., Company information)\n",
+    "                sds_dict[section][field] = {}\n",
+    "                for sub_field, sub_pattern in pattern.items():\n",
+    "                    match = re.search(sub_pattern, text)\n",
+    "                    if match:\n",
+    "                        sds_dict[section][field][sub_field] = match.group(1).strip()\n",
+    "            else:\n",
+    "                # Handle regular pattern matching\n",
+    "                match = re.search(pattern, text)\n",
+    "                if match:\n",
+    "                    sds_dict[section][field] = match.group(1).strip()\n",
+    "\n",
+    "    return sds_dict\n",
+    "\n",
+    "\n",
+    "def process_multiple_pdfs(pdf_folder, config_path, output_folder):\n",
+    "    # Load regex patterns from config\n",
+    "    section_patterns = load_patterns(config_path)\n",
+    "    \n",
+    "    # Ensure output folder exists\n",
+    "    if not os.path.exists(output_folder):\n",
+    "        os.makedirs(output_folder)\n",
+    "    \n",
+    "    # Loop through each PDF file in the folder\n",
+    "    for pdf_filename in os.listdir(pdf_folder):\n",
+    "        if pdf_filename.endswith(\".pdf\"):\n",
+    "            pdf_path = os.path.join(pdf_folder, pdf_filename)\n",
+    "            \n",
+    "            # Extract text from the PDF\n",
+    "            extracted_text = extract_text_from_pdf_plumber(pdf_path)\n",
+    "            \n",
+    "            # Parse the extracted text using the patterns\n",
+    "            sds_data = parse_sds_text(extracted_text, section_patterns)\n",
+    "            \n",
+    "            # Save the parsed data as JSON\n",
+    "            json_filename = os.path.splitext(pdf_filename)[0] + \".json\"\n",
+    "            json_path = os.path.join(output_folder, json_filename)\n",
+    "            with open(json_path, 'w') as json_file:\n",
+    "                json.dump(sds_data, json_file, indent=4)\n",
+    "            \n",
+    "            print(f\"Processed {pdf_filename} and saved JSON data to {json_filename}\")\n",
+    "\n",
+    "pdf_folder = \"/home/sabrina/Documents/Project/data/\"\n",
+    "config_path = \"/home/sabrina/Documents/Project/config.json\"\n",
+    "output_folder = \"/home/sabrina/Documents/Project/json_output/\"\n",
+    "\n",
+    "process_multiple_pdfs(pdf_folder, config_path, output_folder)\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "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": "3.6.15"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

output/acetone-acs-l (1).json

@@ -0,0 +1,80 @@
+{
+    "Identification": {
+        "Product Name": "Acetone",
+        "Cat No.": ": A9-4; A9-20; A9-200; A11-1; A11-4; A11-20; A11-200; A11S-4; A13-20;",
+        "Synonyms": "2-Propanone; Dimethyl ketone; (Certified ACS, HPLC, OPTIMA, Histological,",
+        "Recommended Use": "Laboratory chemicals.",
+        "Uses advised against": "Food, drug, pesticide or biocidal product use.",
+        "Company": {
+            "Name": "Fisher Scientific Company",
+            "Address": "One Reagent Lane\nFair Lawn, NJ 07410",
+            "Telephone": "(201) 796-7100"
+        }
+    },
+    "Emergency Contact": {
+        "Emergency Telephone Number": "CHEMTREC\u00d2, Inside the USA: 800-424-9300"
+    },
+    "Hazard Identification": {
+        "Classification": "______________________________________________________________________________________________",
+        "Signal Word": "Danger",
+        "Hazard Statements": "Highly flammable liquid and vapor",
+        "Precautionary Statements": {
+            "Prevention": "Wash face, hands and any exposed skin thoroughly after handling\nDo not breathe dust/fume/gas/mist/vapors/spray\nUse only outdoors or in a well-ventilated area\nKeep away from heat/sparks/open flames/hot surfaces. - No smoking\nKeep container tightly closed\nGround/bond container and receiving equipment\nUse explosion-proof electrical/ventilating/lighting equipment\nUse only non-sparking tools\nTake precautionary measures against static discharge\nWear protective gloves/protective clothing/eye protection/face protection\nKeep cool\nResponse\nGet medical attention/advice if you feel unwell\nInhalation\nIF INHALED: Remove victim to fresh air and keep at rest in a position comfortable for breathing\nCall a POISON CENTER or doctor/physician if you feel unwell\nSkin\nIF ON SKIN (or hair): Take off immediately all contaminated clothing. Rinse skin with water/shower",
+            "Eyes": "IF IN EYES: Rinse cautiously with water for several minutes. Remove contact lenses, if present and easy to do. Continue rinsing",
+            "Disposal": "______________________________________________________________________________________________",
+            "Hazards not otherwise classified (HNOC)": "(HNOC)"
+        }
+    },
+    "Composition/Information on Ingredients": {
+        "Component": "CAS No Weight %",
+        "CAS No": "67-64-1",
+        "Weight %": "Acetone 67-64-1 >95"
+    },
+    "First Aid Measures": {
+        "General Advice": "If symptoms persist, call a physician.",
+        "Eye Contact": "Rinse immediately with plenty of water, also under the eyelids, for at least 15 minutes. Get",
+        "Skin Contact": "Wash off immediately with plenty of water for at least 15 minutes. If skin irritation persists,",
+        "Inhalation": "IF INHALED: Remove victim to fresh air and keep at rest in a position comfortable for breathing",
+        "Ingestion": "Clean mouth with water and drink afterwards plenty of water."
+    },
+    "Fire Fighting Measures": {
+        "Flash Point": "-20  \u00b0C  /  -4  \u00b0F",
+        "Autoignition Temperature": "465  \u00b0C  /  869  \u00b0F",
+        "Explosion Limits": "Upper 12.8 vol %"
+    },
+    "Handling and Storage": {
+        "Handling": "and storage",
+        "Storage": "Store in a well-ventilated place. Keep container tightly closed"
+    },
+    "Exposure Controls": {
+        "Engineering Measures": "Ensure adequate ventilation, especially in confined areas. Ensure that eyewash stations",
+        "Eye/face Protection": "Wear appropriate protective eyeglasses or chemical safety goggles as described by",
+        "Respiratory Protection": "Follow the OSHA respirator regulations found in 29 CFR 1910.134 or European Standard"
+    },
+    "Physical and Chemical Properties": {
+        "Physical State": "Liquid",
+        "Appearance": "Colorless",
+        "Odor": "sweet",
+        "pH": "7",
+        "Melting Point/Range": "-95  \u00b0C  /  -139  \u00b0F",
+        "Boiling Point/Range": "56  \u00b0C  /  132.8  \u00b0F",
+        "Vapor Pressure": "247 mbar  @  20 \u00b0C",
+        "Solubility": "Soluble in water"
+    },
+    "Stability and Reactivity": {
+        "Stability": "and reactivity",
+        "Incompatible Materials": ".  Strong oxidizing agents."
+    },
+    "Toxicological Information": {
+        "Acute Toxicity": "Product Information"
+    },
+    "Ecological Information": {
+        "Ecotoxicity": "."
+    },
+    "Disposal Considerations": {
+        "Waste Disposal Methods": "Chemical waste generators must determine whether a discarded chemical is classified as a"
+    },
+    "Transport Information": {
+        "DOT": "UN-No UN1090"
+    }
+}

output/acetone-acs-l.json

@@ -0,0 +1,80 @@
+{
+    "Identification": {
+        "Product Name": "Acetone",
+        "Cat No.": ": A9-4; A9-20; A9-200; A11-1; A11-4; A11-20; A11-200; A11S-4; A13-20;",
+        "Synonyms": "2-Propanone; Dimethyl ketone; (Certified ACS, HPLC, OPTIMA, Histological,",
+        "Recommended Use": "Laboratory chemicals.",
+        "Uses advised against": "Food, drug, pesticide or biocidal product use.",
+        "Company": {
+            "Name": "Fisher Scientific Company",
+            "Address": "One Reagent Lane\nFair Lawn, NJ 07410",
+            "Telephone": "(201) 796-7100"
+        }
+    },
+    "Emergency Contact": {
+        "Emergency Telephone Number": "CHEMTREC\u00d2, Inside the USA: 800-424-9300"
+    },
+    "Hazard Identification": {
+        "Classification": "______________________________________________________________________________________________",
+        "Signal Word": "Danger",
+        "Hazard Statements": "Highly flammable liquid and vapor",
+        "Precautionary Statements": {
+            "Prevention": "Wash face, hands and any exposed skin thoroughly after handling\nDo not breathe dust/fume/gas/mist/vapors/spray\nUse only outdoors or in a well-ventilated area\nKeep away from heat/sparks/open flames/hot surfaces. - No smoking\nKeep container tightly closed\nGround/bond container and receiving equipment\nUse explosion-proof electrical/ventilating/lighting equipment\nUse only non-sparking tools\nTake precautionary measures against static discharge\nWear protective gloves/protective clothing/eye protection/face protection\nKeep cool\nResponse\nGet medical attention/advice if you feel unwell\nInhalation\nIF INHALED: Remove victim to fresh air and keep at rest in a position comfortable for breathing\nCall a POISON CENTER or doctor/physician if you feel unwell\nSkin\nIF ON SKIN (or hair): Take off immediately all contaminated clothing. Rinse skin with water/shower",
+            "Eyes": "IF IN EYES: Rinse cautiously with water for several minutes. Remove contact lenses, if present and easy to do. Continue rinsing",
+            "Disposal": "______________________________________________________________________________________________",
+            "Hazards not otherwise classified (HNOC)": "(HNOC)"
+        }
+    },
+    "Composition/Information on Ingredients": {
+        "Component": "CAS No Weight %",
+        "CAS No": "67-64-1",
+        "Weight %": "Acetone 67-64-1 >95"
+    },
+    "First Aid Measures": {
+        "General Advice": "If symptoms persist, call a physician.",
+        "Eye Contact": "Rinse immediately with plenty of water, also under the eyelids, for at least 15 minutes. Get",
+        "Skin Contact": "Wash off immediately with plenty of water for at least 15 minutes. If skin irritation persists,",
+        "Inhalation": "IF INHALED: Remove victim to fresh air and keep at rest in a position comfortable for breathing",
+        "Ingestion": "Clean mouth with water and drink afterwards plenty of water."
+    },
+    "Fire Fighting Measures": {
+        "Flash Point": "-20  \u00b0C  /  -4  \u00b0F",
+        "Autoignition Temperature": "465  \u00b0C  /  869  \u00b0F",
+        "Explosion Limits": "Upper 12.8 vol %"
+    },
+    "Handling and Storage": {
+        "Handling": "and storage",
+        "Storage": "Store in a well-ventilated place. Keep container tightly closed"
+    },
+    "Exposure Controls": {
+        "Engineering Measures": "Ensure adequate ventilation, especially in confined areas. Ensure that eyewash stations",
+        "Eye/face Protection": "Wear appropriate protective eyeglasses or chemical safety goggles as described by",
+        "Respiratory Protection": "Follow the OSHA respirator regulations found in 29 CFR 1910.134 or European Standard"
+    },
+    "Physical and Chemical Properties": {
+        "Physical State": "Liquid",
+        "Appearance": "Colorless",
+        "Odor": "sweet",
+        "pH": "7",
+        "Melting Point/Range": "-95  \u00b0C  /  -139  \u00b0F",
+        "Boiling Point/Range": "56  \u00b0C  /  132.8  \u00b0F",
+        "Vapor Pressure": "247 mbar  @  20 \u00b0C",
+        "Solubility": "Soluble in water"
+    },
+    "Stability and Reactivity": {
+        "Stability": "and reactivity",
+        "Incompatible Materials": ".  Strong oxidizing agents."
+    },
+    "Toxicological Information": {
+        "Acute Toxicity": "Product Information"
+    },
+    "Ecological Information": {
+        "Ecotoxicity": "."
+    },
+    "Disposal Considerations": {
+        "Waste Disposal Methods": "Chemical waste generators must determine whether a discarded chemical is classified as a"
+    },
+    "Transport Information": {
+        "DOT": "UN-No UN1090"
+    }
+}