redline

app.py

@@ -3,16 +3,15 @@ from langchain_pipeline import pipeline, model_names
 import pdfplumber
 from docx import Document
 from redlines import Redlines
+import tempfile
 
 def pdf_to_word(pdf_path, word_path):
     with pdfplumber.open(pdf_path) as pdf:
-        full_text = ""
+        doc = Document()
         for page in pdf.pages:
-            full_text += page.extract_text() + "\n"
-
-    doc = Document()
-    for para in full_text.split("\n"):
-        doc.add_paragraph(para)
+            for item in page.extract_words():
+                doc.add_paragraph(item['text'])
+            doc.add_page_break()
     doc.save(word_path)
 
 def apply_pipeline(file, model_name, balance_type, apsn_transactions, max_fees_per_day, min_overdrawn_fee, min_transaction_overdraft):
@@ -26,7 +25,7 @@ def apply_pipeline(file, model_name, balance_type, apsn_transactions, max_fees_p
         min_transaction_overdraft
     )
 
-def redline_changes(original_path, revised_path):
+def redline_changes(original_path, revised_path, output_path):
     original_doc = Document(original_path)
     revised_doc = Document(revised_path)
 
@@ -34,11 +33,13 @@ def redline_changes(original_path, revised_path):
     revised_text = "\n".join([para.text for para in revised_doc.paragraphs])
 
     redline = Redlines(original_text, revised_text)
-    diff_html = redline.output_markdown()
+    diff_html = redline.output_html()
 
     diff_doc = Document()
-    diff_doc.add_paragraph(diff_html)
-    diff_doc.save("redlined_document.docx")
+    for line in diff_html.split("\n"):
+        diff_doc.add_paragraph(line)
+
+    diff_doc.save(output_path)
 
 # Streamlit App
 st.title("Canarie AI Prototype")
@@ -61,7 +62,7 @@ uploaded_file = st.file_uploader("Choose a file")
 if uploaded_file is not None:
     with st.spinner('Please wait ...'):
         try:
-            original_word_path = "original_document.docx"
+            original_word_path = tempfile.NamedTemporaryFile(delete=False, suffix=".docx").name
             pdf_to_word(uploaded_file, original_word_path)
 
             diff = apply_pipeline(
@@ -74,14 +75,22 @@ if uploaded_file is not None:
                 min_transaction_overdraft
             )
 
-            revised_word_path = "revised_document.docx"
+            revised_word_path = tempfile.NamedTemporaryFile(delete=False, suffix=".docx").name
             revised_doc = Document()
             for line in diff.split("\n"):
                 revised_doc.add_paragraph(line)
             revised_doc.save(revised_word_path)
 
-            redline_changes(original_word_path, revised_word_path)
+            redlined_output_path = tempfile.NamedTemporaryFile(delete=False, suffix=".docx").name
+            redline_changes(original_word_path, revised_word_path, redlined_output_path)
 
+            with open(redlined_output_path, "rb") as f:
+                st.download_button(
+                    label="Download Redlined Document",
+                    data=f,
+                    file_name="redlined_document.docx",
+                    mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+                )
             st.success("Redlined document created successfully!")
 
         except Exception as e: