Medical Form Implemented

app.py

@@ -1,40 +1,40 @@
 import gradio as gr
-from pipeline import extract_info, extract_child_fee_info
-from pdf2image import convert_from_path
+from pipeline import extract_info_batch, extract_child_fee_info,extract_medical_info
+from PIL import Image
+
+
 
 
 
 with gr.Blocks() as demo:
     with gr.Tabs():
-        with gr.Tab("Single Upload"):
+        with gr.Tab("Receipts Upload"):
             with gr.Row():
                 with gr.Column(scale=2):
-                    img_input = gr.Image(type="pil",label="Image Upload",elem_id="upload-img",show_label=False,height=512,width=512)
+                    batch_img_input = gr.File(
+                        file_types=["image"],
+                        label="Batch Image Upload",
+                        elem_id="batch-upload-img",
+                        show_label=True,
+                        file_count="multiple"
+                    )
 
                 with gr.Column(scale=2):
-                    output_box = gr.Markdown(
-                        value="""```json
-                                    {
-                                    "merchant": "",
-                                    "date": "",
-                                    "total_amount": null,
-                                    "items": [
-                                        {
-                                            "description": "",
-                                            "amount": null
-                                        }
-                                    ]
-                                    }
-                                    ```""",
-
-                        label="Extracted Info", elem_id="output-box", show_label=False
+                    batch_output_box = gr.Markdown(
+                        value="Upload Images to extract information",
+                        label="Batch Extracted Info",
+                        elem_id="batch-output-box",
+                        show_label=True
+                    )
+                    batch_img_input.change(
+                        fn=extract_info_batch,
+                        inputs=batch_img_input,
+                        outputs=batch_output_box
                     )
-                    img_input.upload(fn=extract_info, inputs=img_input, outputs=output_box)
         
 
 
-
-        with gr.Tab("Reimbursement Form"):
+        with gr.Tab("Child Fee Reimbursement Form"):
             with gr.Row():
                 with gr.Column(scale=2):
                     img_input = gr.Image(
@@ -48,17 +48,12 @@ with gr.Blocks() as demo:
 
                 with gr.Column(scale=2):
                     # Dropdown for form names
-                    form_dropdown = gr.Dropdown(
-                        choices=["Child Fee Reimbursement", "Medical Reimbursement", "Other Form"],
-                        label="Select Form",
-                        value="Child Fee Reimbursement",
-                        multiselect=False,
-                        interactive=True,
-                    )
+                    gr.Markdown("## Child Fee Reimbursement")
                     
                     # 2x2 grid for info fields
                     with gr.Row():
                         emp_name = gr.Textbox(label="Employee Name")
+                
                         emp_code = gr.Textbox(label="Employee Code")
                     with gr.Row():
                         department = gr.Textbox(label="Department")
@@ -72,11 +67,45 @@ with gr.Blocks() as demo:
                         outputs=preview_output
                     )
 
+       
+        with gr.Tab("Medical Reimbursement Form"):
+            with gr.Row():
+                with gr.Column(scale=2):
+                    medical_img_input = gr.Image(
+                        type="pil",
+                        label="Image Upload",
+                        elem_id="upload-img",
+                        show_label=False,
+                        height=512,
+                        width=512
+                    )
+                with gr.Column(scale=2):
+                    with gr.Row():
+                        med_company_name = gr.Dropdown(choices=["NetSol Technologies Ltd.","NetSol Innovation Private Ltd."], interactive=True, multiselect=False)
+                        med_emp_name = gr.Textbox(label="Employee Name")
+                        med_department = gr.Textbox(label="Department")
+                    with gr.Row():
+                        med_designation = gr.Textbox(label="Designation")
+                        med_ext_code = gr.Textbox(label="Extention No.")
+                        med_emp_code = gr.Textbox(label="Employee Code")
+                    medical_upload_btn = gr.Button("Upload and Process")
+                    preview_medical_output = gr.File(label="Download Filled Form")
+
+                    
+                    medical_upload_btn.click(
+                        fn=extract_medical_info,
+                        inputs=[medical_img_input,med_emp_name,med_emp_code,med_department,med_designation,med_company_name,med_ext_code],
+                        outputs=preview_medical_output
+                    )
+
+
+
+
     # CSS:
     gr.HTML("""
         <style>
         #output-box .prose, #output-box .prose pre, #output-box .prose code {
-            font-size: 22px !important;
+            font-size: 30px !important;
         }
         </style>
     """)

form_fill.py

@@ -40,3 +40,54 @@ def fill_child_fee_pdf(
                     annotation.AP = None  # Remove old appearance so new value appears
     PdfWriter().write(output_pdf_path, template_pdf)
     return output_pdf_path
+
+
+def fill_medical_pdf(
+    template_pdf_path,
+    output_pdf_path,
+    company,
+    extension_no,
+    employee_name,
+    employee_code,
+    department,
+    date,
+    total,
+    designation,
+    billing_month,
+    claims  # List of dicts: [{'name': ..., 'relationship': ..., 'category': ..., 'detail': ..., 'amount': ...}]
+):
+    data_dict = {
+        'company': company,
+        'extension_no': extension_no,
+        'employee_name': employee_name,
+        'employee_code': employee_code,
+        'department': department,
+        'designation': designation,
+        'date': date,
+        'billing_month': billing_month,
+        'total': str(total),
+        'current_date': datetime.now().strftime("%d-%b-%Y"),
+    }
+
+    # Map each row of claims to field names
+    for idx, claim in enumerate(claims, start=1):
+        data_dict[f'name_{idx}'] = claim.get('name', '')
+        data_dict[f'relationship_{idx}'] = claim.get('relationship', '')
+        data_dict[f'category_{idx}'] = claim.get('category', '')
+        data_dict[f'detail_{idx}'] = claim.get('detail', '')
+        data_dict[f'amount_{idx}'] = str(claim.get('amount', ''))
+
+    # Fill the PDF
+    template_pdf = PdfReader(template_pdf_path)
+    for page in template_pdf.pages:
+        if not hasattr(page, 'Annots') or not page.Annots:
+            continue
+        for annotation in page.Annots:
+            if annotation.T:
+                key = annotation.T[1:-1]  # Remove parentheses
+                if key in data_dict:
+                    annotation.V = str(data_dict[key])
+                    annotation.AP = None  # Remove old appearance so new value appears
+    PdfWriter().write(output_pdf_path, template_pdf)
+    return output_pdf_path
+

models.py

@@ -27,3 +27,25 @@ class FeeItem(BaseModel):
 class ChildFeeForm(BaseModel):
     items: List[FeeItem]
     total: float  # Calculated after parsing
+
+
+
+
+class MedicalClaimItem(BaseModel):
+    name: str  # Patient name
+    relationship: str  # self, spouse, parent, child
+    category: str  # in-patient, out-patient, maternity(cesarean), maternity(normal)
+    detail: str  # doctor's fee, diagnostic tests, medicines, other hospitalization
+    amount: float
+
+class MedicalReimbursementForm(BaseModel):
+    company: Optional[str] = None
+    extension_no: Optional[str] = None
+    employee_name: str
+    employee_code: str
+    department: Optional[str] = None
+    designation: Optional[str] = None
+    date: Optional[str] = None
+    billing_month: Optional[str] = None
+    claims: List[MedicalClaimItem]
+    total: float

outputs/filled_child_fee_form_e3fba6d8482a476dade270c022cc64e2.pdf

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:fa5b02270d310e55abfdf36281e54e88928ad3b71d19f034b784b1616a2c9eb5
-size 115542

pipeline.py

@@ -5,9 +5,11 @@ import os, uuid
 from PIL import Image
 import base64
 import json
-from models import ReceiptData, ChildFeeForm
-from form_fill import fill_child_fee_pdf
+from models import ReceiptData, ChildFeeForm, MedicalReimbursementForm
+from form_fill import fill_child_fee_pdf, fill_medical_pdf
 from fraud import process_receipt
+from datetime import datetime
+
 
 load_dotenv()
 openai.api_key = os.getenv("OPENAI_API_KEY", "").strip()
@@ -37,6 +39,7 @@ reciept_system_prompt = (
     "- Do not add any explanation or extra text—only the JSON."
 )
 
+
 fee_bill_system_prompt = (
     "You are an expert at extracting data from fee bills. "
     "Read the provided image of a child fee bill and return a JSON object that matches the following Pydantic model:\n"
@@ -58,6 +61,29 @@ fee_bill_system_prompt = (
 )
 
 
+medical_form_system_prompt = (
+    "You are an expert at extracting structured data from tabular forms containing sample data. "
+    "Your task is to read the provided form and return a JSON object that matches the following Pydantic model:\n"
+    "class Item(BaseModel):\n"
+    "    name: str #the patient name\n"
+    "    relationship: # self, spouse, parent, child\n"
+    "    category: # in-patient, out-patient, maternity(cesarean), maternity(normal)\n"
+    "    detail: # doctor's fee, diagnostic tests, medicines, other hospitalization\n"
+    "    bill_month: Optional[str] = None # Bill Month Field, if not directly stated, find the date and infer the month from that, if not found return null\n"
+    "    amount: float\n"
+    "class Form(BaseModel):\n"
+    "    claims: List[Item]\n"
+    "    total: float\n"
+    "- Extract only the above information. If a value is missing, set it to null, \"\", or an empty list as appropriate.\n"
+    "- For the claims field, provide a list of objects with name, relationship, category, detail, and amount.\n"
+    "- The total field must be the sum of all amount values in claims.\n"
+    "- Only return a valid JSON object matching the model above.\n"
+    "- Do not add any explanation or extra text—only the JSON."
+    "- Try your very best to extract this information as it is very important that you do so\n"
+    "- If you are unable to extract information, return an empty json in the format requested above, never give a response other than a json"
+)
+
+
 
 
 def pil_to_bytes(pil_img, quality=70):
@@ -170,3 +196,79 @@ def extract_child_fee_info(img_input, emp_name, emp_code, department):
     except Exception as e:
         print("ERROR:", e)
         return None  # or f"Error: {str(e)}"
+
+
+
+def extract_info_batch(file_list):
+    """
+    Accepts a list of file objects/paths, processes each as a PIL image, and returns results.
+    """
+    results = []
+    for file in file_list:
+        img = Image.open(file)
+        results.append(extract_info(img))
+    return "\n\n".join(results)
+
+
+
+def extract_medical_info(pil_img, emp_name, emp_code, department, designation, company, extension_no,):
+    processed_image = preprocess_image(pil_img)
+    img_bytes = pil_to_bytes(processed_image)
+    img_base64 = base64.b64encode(img_bytes.getvalue()).decode("utf-8")
+    response = openai.chat.completions.create(
+        model="gpt-4o",
+        messages=[
+            {"role": "system", "content": medical_form_system_prompt},
+            {"role": "user",
+             "content": [
+                {"type": "text", "text": "Here is a child fee bill image:"},
+                {"type": "image_url", "image_url": {"url": "data:image/png;base64," + img_base64}}
+             ]}
+        ]
+    )
+    raw_output = response.choices[0].message.content
+    print(raw_output)
+    try:
+        if raw_output.startswith("```"):
+            raw_output = raw_output.strip("` \n")
+            if raw_output.startswith("json"):
+                raw_output = raw_output[4:].strip()
+        data = json.loads(raw_output)
+        print(data)
+        # Validate if needed:
+        # ChildFeeForm(**data)
+
+        claims = data.get("claims", [])
+        bill_month = ""
+        if claims and "bill_month" in claims[0]:
+            bill_month = claims[0]["bill_month"]
+
+        date = datetime.now().strftime("%d-%b-%Y")  # e.g., "10-Jun-2024"
+        total = data.get("total", 0)
+
+        print("bill month:",bill_month)
+
+        print("total:",total)
+        os.makedirs("outputs", exist_ok=True)
+        output_pdf_path = f"outputs/filled_medical_form_{uuid.uuid4().hex}.pdf"
+
+
+        filled_pdf_path = fill_medical_pdf(
+            template_pdf_path="Medical Reim. Form.pdf",
+            output_pdf_path=output_pdf_path,
+            company=company,
+            employee_name=emp_name,
+            employee_code=emp_code,
+            department=department,
+            designation=designation,
+            extension_no=extension_no,
+            billing_month=bill_month,
+            claims=claims,
+            date= date,
+            total=total
+        )
+
+        return filled_pdf_path # Return path to Gradio for download
+    except Exception as e:
+        print("ERROR:", e)
+        return None  # or f"Error: {str(e)}"

outputs/filled_child_fee_form_6dfe1052f5be4bbc8a78711a456a3cc9.pdf → templates/Medical Reim. Form.pdf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:48066dd238b5a6eda1e8cd8647afc31074c61b7b95946e6b4329426f4a818331
-size 115543
+oid sha256:c42ef819cd41595124559fb1d5ae4bd290eb5152d477cc0e61ec89f7d13f2380
+size 103497

templates/medical_form.html

@@ -0,0 +1,286 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Medical Reimbursement Form (Applicable for Outpatient)</title>
+    <style>
+        body {
+            font-family: 'Times New Roman', Times, serif;
+            background: #fff;
+            margin: 0;
+            padding: 40px 0;
+        }
+        .form-container {
+            width: 900px;
+            margin: 0 auto;
+            border: 2px solid #000;
+            background: #fff;
+        }
+        .form-header {
+            display: flex;
+            align-items: flex-start;
+            padding: 15px 30px 0 20px;
+        }
+        .logo-placeholder {
+            width: 90px;
+            height: 90px;
+            background: #f2f2f2;
+            border: 1px solid #ddd;
+            margin-right: 18px;
+        }
+        .header-title {
+            flex: 1;
+            text-align: center;
+        }
+        .header-title h2 {
+            font-size: 24px;
+            margin: 0;
+            font-weight: bold;
+        }
+        .header-title h3 {
+            font-size: 18px;
+            margin: 0;
+            font-weight: normal;
+        }
+        .form-details {
+            display: flex;
+            flex-wrap: wrap;
+            padding: 10px 30px 0 30px;
+            font-size: 17px;
+        }
+        .form-details-col {
+            flex: 1 1 50%;
+        }
+        .form-details-row {
+            display: flex;
+            align-items: center;
+            margin-bottom: 7px;
+        }
+        .form-details-label {
+            width: 140px;
+            display: inline-block;
+        }
+        .form-details-input {
+            border: none;
+            border-bottom: 1.2px solid #000;
+            width: 250px;
+            height: 22px;
+            font-size: 16px;
+            background: none;
+            margin-left: 4px;
+            margin-right: 4px;
+        }
+        .form-details-input:focus {
+            outline: none;
+            background: #eef;
+        }
+        .form-table-container {
+            margin: 25px 20px 0 20px;
+        }
+        table {
+            width: 100%;
+            border-collapse: collapse;
+            margin-bottom: 0;
+        }
+        th, td {
+            border: 1px solid #000;
+            padding: 7px 4px;
+            font-size: 16px;
+            text-align: center;
+        }
+        th {
+            background: #ddd;
+            font-weight: bold;
+        }
+        .table-title-row th {
+            border-bottom: 2px solid #000;
+        }
+        .table-claim-col {
+            width: 120px;
+        }
+        .table-input {
+            width: 95%;
+            font-size: 15px;
+            border: none;
+            border-bottom: 1px solid #555;
+            background: none;
+            text-align: center;
+        }
+        .table-input:focus {
+            outline: none;
+            background: #eef;
+        }
+        .total-row td {
+            border: none;
+            text-align: right;
+            font-weight: bold;
+            font-size: 17px;
+            padding-right: 24px;
+        }
+        .signature-row {
+            height: 60px;
+        }
+        .signature-cell {
+            text-align: right;
+            padding-right: 90px;
+            vertical-align: bottom;
+        }
+        .notes-section {
+            font-size: 14px;
+            margin: 8px 22px 10px 24px;
+        }
+        .notes-section ol,
+        .notes-section ul {
+            margin-top: 0;
+            margin-bottom: 0;
+        }
+        .notes-section li {
+            margin-bottom: 4px;
+        }
+        .bottom-line {
+            border-bottom: 1.5px solid #000;
+            margin-top: 30px;
+            width: 250px;
+        }
+        @media (max-width: 950px) {
+            .form-container {
+                width: 98vw;
+            }
+        }
+        @media (max-width: 650px) {
+            .form-header, .form-details, .form-table-container {
+                padding-left: 5px !important;
+                padding-right: 5px !important;
+            }
+            .form-container {
+                padding: 0 1vw;
+            }
+        }
+    </style>
+</head>
+<body>
+    <div class="form-container">
+        <div class="form-header">
+            <div class="logo-placeholder">
+                <!-- Logo goes here -->
+            </div>
+            <div class="header-title">
+                <h2>Medical Reimbursement Form</h2>
+                <h3>(Applicable for Outpatient )</h3>
+            </div>
+        </div>
+        <form>
+        <div class="form-details">
+            <div class="form-details-col">
+                <div class="form-details-row">
+                    <span class="form-details-label">Company:</span>
+                    <input class="form-details-input" type="text" name="company">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Name:</span>
+                    <input class="form-details-input" type="text" name="name">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Department:</span>
+                    <input class="form-details-input" type="text" name="department">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Designation:</span>
+                    <input class="form-details-input" type="text" name="designation">
+                </div>
+            </div>
+            <div class="form-details-col">
+                <div class="form-details-row">
+                    <span class="form-details-label">Extension No:</span>
+                    <input class="form-details-input" type="text" name="extension">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Employee Code:</span>
+                    <input class="form-details-input" type="text" name="employee_code">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Date:</span>
+                    <input class="form-details-input" type="date" name="date" style="width:180px;">
+                </div>
+                <div class="form-details-row">
+                    <span class="form-details-label">Billing Month:</span>
+                    <input class="form-details-input" type="text" name="billing_month" placeholder="mm/yy" style="width:90px;">
+                    <span style="margin-left:10px;">(mm/yy)</span>
+                </div>
+            </div>
+        </div>
+        <div class="form-table-container">
+            <table>
+                <tr class="table-title-row">
+                    <th>Name</th>
+                    <th>Relationship</th>
+                    <th>Category</th>
+                    <th class="table-claim-col">Detail</th>
+                    <th class="table-claim-col">Amount</th>
+                </tr>
+                <tbody>
+                    {{claims_rows}}
+                </tbody>
+            </table>
+            <table style="margin-top: 0;">
+                <tr class="total-row">
+                    <td colspan="4" style="border: none;"></td>
+                    <td style="border-top: none; border-left: 1px solid #000;">Total -
+                        <input style="width: 80px; border: none; border-bottom: 1px solid #000; font-size: 16px; background: none; text-align: center;" type="number" step="0.01" min="0" name="total" readonly id="total_amount">
+                    </td>
+                </tr>
+                <tr class="signature-row">
+                    <td colspan="5" class="signature-cell">
+                        <div style="margin-top:20px;">
+                            ________________________________<br>
+                            Employee's Signature
+                        </div>
+                    </td>
+                </tr>
+            </table>
+        </div>
+        </form>
+        <div class="notes-section">
+            <ol>
+                <li>No claim will be processed in case:
+                    <ul style="list-style-type: lower-alpha; margin-left: 20px;">
+                        <li>Supporting documents are not attached.</li>
+                        <li>Form is incomplete.</li>
+                        <li>Claim is time barred <span style="text-decoration:underline;">i.e.</span> not submitted within thirty (30) days after the expense is incurred.</li>
+                        <li>Any violation of applicable policy and procedure.</li>
+                    </ul>
+                </li>
+                <li>
+                    <span style="text-decoration:underline;">"Category"</span> refers to type of the claim <span style="text-decoration:underline;">i.e.</span> in-patient, out-patient, maternity (cesarean), maternity (normal).
+                </li>
+                <li>
+                    <span style="text-decoration:underline;">"Detail"</span> refers to type of the expenditure <span style="text-decoration:underline;">i.e.</span> doctor's fee, diagnostic tests, medicines, other hospitalization charges.
+                </li>
+                <li>
+                    Expenses incurred during probation are considered to be time barred if not claimed within thirty days after confirmation.
+                </li>
+            </ol>
+        </div>
+    </div>
+    <script>
+        // Auto sum the Amount columns
+        const amountFields = [
+            document.querySelector('input[name="claim_amount1"]'),
+            document.querySelector('input[name="claim_amount2"]'),
+            document.querySelector('input[name="claim_amount3"]'),
+            document.querySelector('input[name="claim_amount4"]')
+        ];
+        const totalField = document.getElementById('total_amount');
+        amountFields.forEach(f => {
+            f.addEventListener('input', () => {
+                let total = 0;
+                amountFields.forEach(ff => {
+                    let val = parseFloat(ff.value);
+                    if (!isNaN(val)) total += val;
+                });
+                totalField.value = total > 0 ? total.toFixed(2) : '';
+            });
+        });
+    </script>
+</body>
+</html>