#176 - smart output file renaming

src/filler.py

@@ -1,7 +1,7 @@
 from pdfrw import PdfReader, PdfWriter
 from src.llm import LLM
 from datetime import datetime
-
+import re
 
 class Filler:
     def __init__(self):
@@ -12,19 +12,36 @@ def fill_form(self, pdf_form: str, llm: LLM):
         Fill a PDF form with values from user_input using LLM.
         Fields are filled in the visual order (top-to-bottom, left-to-right).
         """
-        output_pdf = (
-            pdf_form[:-4]
-            + "_"
-            + datetime.now().strftime("%Y%m%d_%H%M%S")
-            + "_filled.pdf"
-        )
-
         # Generate dictionary of answers from your original function
         t2j = llm.main_loop()
         textbox_answers = t2j.get_data()  # This is a dictionary
-
+
+        target_keys = ["Employee's name", "Name", "Full Name", "Patient Name"]
+        extracted_name = None
+
+        for key in target_keys:
+            if key in textbox_answers and textbox_answers[key]:
+                val = textbox_answers[key]
+                # If the LLM returned a list (plural values), grab the first item
+                extracted_name = str(val[0]) if isinstance(val, list) else str(val)
+                break
+
+        # 3. Create the output filename
+        if extracted_name and extracted_name != "-1":
+            # Clean illegal characters and spaces
+            safe_name = re.sub(r'[^a-zA-Z0-9_\-]', '_', extracted_name.replace(' ', '_'))
+            safe_name = re.sub(r'_+', '_', safe_name).strip('_')
+            output_pdf = f"{pdf_form[:-4]}_{safe_name}.pdf"
+        else:
+            # Fallback to the original timestamp method if no name is found
+            output_pdf = (
+                pdf_form[:-4]
+                + "_"
+                + datetime.now().strftime("%Y%m%d_%H%M%S")
+                + "_filled.pdf"
+            )
+
         answers_list = list(textbox_answers.values())
-
         # Read PDF
         pdf = PdfReader(pdf_form)
 

src/llm.py

@@ -46,7 +46,7 @@ def build_prompt(self, current_field):
 
     def main_loop(self):
         # self.type_check_all()
-        for field in self._target_fields.keys():
+        for field in self._target_fields:
             prompt = self.build_prompt(field)
             # print(prompt)
             # ollama_url = "http://localhost:11434/api/generate"

src/main.py

@@ -3,6 +3,7 @@
 from commonforms import prepare_form 
 from pypdf import PdfReader
 from controller import Controller
+from typing import Union
 
 def input_fields(num_fields: int):
     fields = []
@@ -68,7 +69,7 @@ def run_pdf_fill_process(user_input: str, definitions: list, pdf_form_path: Unio
 if __name__ == "__main__":
     file = "./src/inputs/file.pdf"
     user_input = "Hi. The employee's name is John Doe. His job title is managing director. His department supervisor is Jane Doe. His phone number is 123456. His email is jdoe@ucsc.edu. The signature is <Mamañema>, and the date is 01/02/2005"
-    fields = ["Employee's name", "Employee's job title", "Employee's department supervisor", "Employee's phone number", "Employee's email", "Signature", "Date"]
+    descriptive_fields = ["Employee's name", "Employee's job title", "Employee's department supervisor", "Employee's phone number", "Employee's email", "Signature", "Date"]
     prepared_pdf = "temp_outfile.pdf"
     prepare_form(file, prepared_pdf)
 
@@ -80,4 +81,4 @@ def run_pdf_fill_process(user_input: str, definitions: list, pdf_form_path: Unio
         num_fields = 0
 
     controller = Controller()
-    controller.fill_form(user_input, fields, file)
+    controller.fill_form(user_input, descriptive_fields, file)