convert2docx

app.py

@@ -1,72 +1,33 @@
+
+FLOWISE_URL = "https://siraflowappl.happywave-0e1819cd.uaenorth.azurecontainerapps.io/api/v1/prediction/0b8a2f94-f83f-4303-9007-abe55420e171"
+#5f2aa074-ae2e-4b06-a4a2-6f8dccbe59bb" Contract.Auditor.2
 import os
-import uuid
 import logging
+import uuid
 import requests
+import docx2txt
 import gradio as gr
-
 from langdetect import detect
-from multilingual_pdf2text.pdf2text import PDF2Text
-from multilingual_pdf2text.models.document_model.document import Document
-
-import pdfplumber
-import docx2txt
-import shutil
-import subprocess
+from pdf2docx import Converter  # <-- new library for PDF to DOCX
 
-print("🔍 Checking if Tesseract OCR is installed...")
-
-tess_path = shutil.which("tesseract")
-if tess_path:
-    print(f"✅ Tesseract found at: {tess_path}")
-    try:
-        version = subprocess.check_output(["tesseract", "--version"]).decode()
-        print(f"🧠 Tesseract version:\n{version}")
-    except Exception as e:
-        print(f"⚠️ Tesseract found but failed to run: {e}")
-else:
-    print("❌ Tesseract not found. Arabic PDF extraction will likely fail.")
-
-
-# --- Config ---
-FLOWISE_URL = "https://siraflowappl.happywave-0e1819cd.uaenorth.azurecontainerapps.io/api/v1/prediction/0b8a2f94-f83f-4303-9007-abe55420e171"
-#5f2aa074-ae2e-4b06-a4a2-6f8dccbe59bb" Contract.Auditor.2
-COUNTRIES = ["Jordan", "Germany", "UAE", "Turkey","USA"]
-LANGUAGES = ["Arabic","English"]
+COUNTRIES = ["USA", "Germany", "UAE", "Turkey", "Jordan"]
+LANGUAGES = ["English", "Arabic"]
 
 logging.basicConfig(level=logging.INFO)
 
-# --- File Text Extraction ---
 def extract_text_from_file(file_path: str) -> str:
     ext = os.path.splitext(file_path)[1].lower()
 
     try:
         if ext == ".pdf":
-            with pdfplumber.open(file_path) as pdf:
-                extracted = "\n".join(
-                    page.extract_text() for page in pdf.pages if page.extract_text()
-                )
-
-            sample = extracted[:1000].strip()
-            lang = detect(sample) if sample else "unknown"
-
-            if lang == "ar" or not sample:
-                arabic_doc = Document(document_path=file_path, language="ara")
-                pdf2text = PDF2Text(document=arabic_doc)
-                extracted = pdf2text.extract()
-
-                if isinstance(extracted, list):
-                    if isinstance(extracted[0], dict):
-                        return "\n".join(item.get("text", "") for item in extracted)
-                    elif isinstance(extracted[0], str):
-                        return "\n".join(extracted)
-                    else:
-                        return "[Error: Unexpected Arabic extractor format.]"
-                elif isinstance(extracted, str):
-                    return extracted
-                else:
-                    return "[Error: Unrecognized Arabic extraction output.]"
-
-            return extracted or "[Warning: No readable text found in PDF.]"
+            # --- Step 1: Convert PDF to DOCX ---
+            temp_docx_path = file_path.replace(".pdf", "_converted.docx")
+            cv = Converter(file_path)
+            cv.convert(temp_docx_path, start=0, end=None)
+            cv.close()
+            
+            # --- Step 2: Read the DOCX file ---
+            return docx2txt.process(temp_docx_path)
 
         elif ext == ".docx":
             return docx2txt.process(file_path)
@@ -82,7 +43,6 @@ def extract_text_from_file(file_path: str) -> str:
         logging.exception("Failed to extract text.")
         return f"[Error reading file: {e}]"
 
-# --- Flowise Query ---
 def send_to_flowise(file, country: str, language: str) -> str:
     if not file or not country or not language:
         return "❗ Please upload a contract, select a country, and choose a response language."
@@ -92,11 +52,11 @@ def send_to_flowise(file, country: str, language: str) -> str:
         if not contract_text:
             return "⚠️ No content extracted from the uploaded file."
 
-        prompt = f"""Please audit the following employment contract in light of {country} labor law. 
-        Only refer to the provided law document (do not use external knowledge), but you don't have to mention the exact source of info while answering.
-        Is the contract compliant? Provide reasoning and highlight any issues, and suggest improvements.
-        Read line by line, be precise, and don't miss any details in the contract. 
-        Please respond in {language}.
+        prompt = f"""Please audit the following employment contract in light of {country} labor law.
+Only refer to the provided law document (do not use external knowledge). Do not mention the law NO. in your response.
+Is the contract compliant? Provide reasoning and highlight any issues, and suggest improvements.
+Be precise and don't miss any details in the contract.
+Please respond in {language}.
 <CONTRACT START>
 {contract_text}
 <CONTRACT END>"""
@@ -140,7 +100,6 @@ with gr.Blocks(title="Contract Auditor") as demo:
         inputs=[file_input, country_dropdown, language_dropdown],
         outputs=output
     )
-
 if __name__ == "__main__":
     demo.launch(
         server_name="0.0.0.0",