Added Gemini + Fitz pipeline for PA form automation

.env

@@ -0,0 +1 @@
+GOOGLE_API_KEY=AIzaSyBr1SJSG3RHKOTP-lymDuixN2-xcwPW0Gw

data/Akshay.json

@@ -0,0 +1,10 @@
+{
+  "patient_first": "Akshay",
+  "patient_last": "Chaudhari",
+  "dob": "02/17/1987",
+  "diagnosis": "Crohn's disease",
+  "provider_name": "Timothy Adam, MD",
+  "npi": "1331124163",
+  "insurance_id": "1:5VA",
+  "medication": "Skyrizi"
+}

src/fill_form.py

@@ -0,0 +1,34 @@
+import fitz  # PyMuPDF
+import json
+import os
+
+# Load extracted JSON data
+with open("data/Akshay.json") as f:
+    data = json.load(f)
+
+# Open the PDF
+pdf_path = "Input Data/Akshay/PA.pdf"
+doc = fitz.open(pdf_path)
+
+# Define field -> value mapping
+field_values = {
+    "T11": data.get("patient_first", ""),
+    "T12": data.get("patient_last", ""),
+    "T13": data.get("dob", ""),
+    "T21F": data.get("diagnosis", ""),
+    "Request by T": data.get("medication", "")
+}
+
+# Fill the form
+for page in doc:
+    for field_name, value in field_values.items():
+        for widget in page.widgets():
+            if widget.field_name == field_name:
+                widget.field_value = value
+                widget.update()
+
+# Save output
+os.makedirs("output", exist_ok=True)
+output_path = "output/Akshay_filled.pdf"
+doc.save(output_path)
+print(f"✅ Filled form")

src/gemini_analysis.py

@@ -0,0 +1,37 @@
+import os
+import pathlib
+import json
+from dotenv import load_dotenv
+import google.generativeai as genai
+
+load_dotenv()
+genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
+model = genai.GenerativeModel("gemini-1.5-flash")
+
+def extract_key_data(pdf_path, prompt):
+    pdf_bytes = pathlib.Path(pdf_path).read_bytes()
+    response = model.generate_content([
+        {"mime_type": "application/pdf", "data": pdf_bytes},
+        prompt
+    ])
+    return response.text
+
+pdf_path = "Input Data/Akshay/referral_package.pdf"
+prompt = (
+    "Extract patient first name, last name, DOB, diagnosis, provider name, NPI, insurance ID, and medication. "
+    "Return result as JSON with keys: patient_first, patient_last, dob, diagnosis, provider_name, npi, insurance_id, medication."
+)
+
+result = extract_key_data(pdf_path, prompt)
+
+# Clean up response
+cleaned_result = result.strip()
+if cleaned_result.startswith("```json"):
+    cleaned_result = cleaned_result.replace("```json", "").strip()
+if cleaned_result.endswith("```"):
+    cleaned_result = cleaned_result[:-3].strip()
+
+# Now parse
+with open("data/Akshay.json", "w") as f:
+    json.dump(json.loads(cleaned_result), f, indent=2)
+