Update app.py

app.py

@@ -3,6 +3,27 @@ import pandas as pd
 import json
 from io import BytesIO
 import requests
+import re
+from openpyxl import Workbook
+
+def sanitize_value(val):
+    """
+    Convert complex types to a string and remove illegal characters 
+    that Excel does not accept.
+    """
+    if isinstance(val, bytes):
+        try:
+            s = val.decode("utf-8", errors="replace")
+        except Exception:
+            s = str(val)
+        # Remove control characters (except newline and tab if desired)
+        return re.sub(r'[\x00-\x08\x0B\x0C\x0E-\x1F]', '', s)
+    elif isinstance(val, str):
+        return re.sub(r'[\x00-\x08\x0B\x0C\x0E-\x1F]', '', val)
+    elif isinstance(val, (dict, list)):
+        return re.sub(r'[\x00-\x08\x0B\x0C\x0E-\x1F]', '', str(val))
+    else:
+        return val
 
 def dataset_converter(input_file, conversion_type, parquet_url):
     # Initialize variables for file data and extension
@@ -45,7 +66,6 @@ def dataset_converter(input_file, conversion_type, parquet_url):
     elif conversion_type == "CSV to JSONL":
         if input_file is None or file_extension != "csv":
             raise ValueError("For CSV to JSONL conversion, please upload a CSV file. 📄")
-        # Read CSV with latin1 encoding
         df = pd.read_csv(BytesIO(file_bytes), encoding='latin1')
         output_file = "metadata.jsonl"
         total_data = []
@@ -58,7 +78,6 @@ def dataset_converter(input_file, conversion_type, parquet_url):
                 data[column] = row[column]
             row_data = {"file_name": file_name_val, "ground_truth": json.dumps(data)}
             total_data.append(row_data)
-        # Write JSONL output (using write mode so previous data is overwritten)
         with open(output_file, 'w', encoding='utf-8') as f:
             for row_data in total_data:
                 f.write(json.dumps(row_data) + '\n')
@@ -67,19 +86,17 @@ def dataset_converter(input_file, conversion_type, parquet_url):
     
     # Conversion: Parquet to JSONL
     elif conversion_type == "Parquet to JSONL":
-        # Use uploaded file if available; otherwise try the provided URL
         if input_file is not None:
             df = pd.read_parquet(BytesIO(file_bytes))
         elif parquet_url:
             response = requests.get(parquet_url)
-            response.raise_for_status()  # Ensure the request was successful
+            response.raise_for_status()
             df = pd.read_parquet(BytesIO(response.content))
             file_name = "from_url.parquet"
         else:
             raise ValueError("For Parquet to JSONL conversion, please upload a file or provide a URL. 🌐")
         
         output_file = "output.jsonl"
-        # Recursive function to decode bytes to UTF-8 strings
         def recursive_sanitize(val):
             if isinstance(val, bytes):
                 return val.decode("utf-8", errors="replace")
@@ -98,6 +115,29 @@ def dataset_converter(input_file, conversion_type, parquet_url):
         converted_format = "JSONL"
         preview_str = df.head(10).to_string(index=False)
     
+    # Conversion: Parquet to XLS
+    elif conversion_type == "Parquet to XLS":
+        if input_file is not None:
+            df = pd.read_parquet(BytesIO(file_bytes))
+        elif parquet_url:
+            response = requests.get(parquet_url)
+            response.raise_for_status()
+            df = pd.read_parquet(BytesIO(response.content))
+            file_name = "from_url.parquet"
+        else:
+            raise ValueError("For Parquet to XLS conversion, please upload a file or provide a URL. 🌐")
+        
+        output_file = "output.xlsx"
+        wb = Workbook(write_only=True)
+        ws = wb.create_sheet()
+        ws.append(list(df.columns))
+        for row in df.itertuples(index=False, name=None):
+            sanitized_row = [sanitize_value(cell) for cell in row]
+            ws.append(sanitized_row)
+        wb.save(output_file)
+        converted_format = "XLS"
+        preview_str = df.head(10).to_string(index=False)
+    
     else:
         raise ValueError("Invalid conversion type selected. ⚠️")
 
@@ -116,7 +156,7 @@ body {
     font-family: 'Helvetica Neue', Arial, sans-serif;
 }
 .gradio-container {
-    max-width: 900px;
+    max-width: 1000px;
     margin: 40px auto;
     padding: 20px;
     background-color: #ffffff;
@@ -146,7 +186,7 @@ h1, h2 {
 with gr.Blocks(css=custom_css, title="Datasets Convertor") as demo:
     gr.Markdown("# Datasets Convertor 🚀")
     gr.Markdown(
-        "Upload a CSV or Parquet file (or provide a Parquet file URL for Parquet to JSONL conversion) "
+        "Upload a CSV or Parquet file (or provide a Parquet file URL for Parquet to JSONL/XLS conversion) "
         "and select the conversion type. The app converts the file to the desired format and displays a preview of the top 10 rows. ✨"
     )
     
@@ -155,11 +195,10 @@ with gr.Blocks(css=custom_css, title="Datasets Convertor") as demo:
             input_file = gr.File(label="Upload CSV or Parquet File 📄")
         with gr.Column(scale=1):
             conversion_type = gr.Radio(
-                choices=["CSV to Parquet", "Parquet to CSV", "CSV to JSONL", "Parquet to JSONL"],
+                choices=["CSV to Parquet", "Parquet to CSV", "CSV to JSONL", "Parquet to JSONL", "Parquet to XLS"],
                 label="Conversion Type 🔄"
             )
     
-    # Optional URL input for Parquet to JSONL conversion
     parquet_url = gr.Textbox(label="Parquet File URL (Optional) 🌐", placeholder="Enter URL if not uploading a file")
     
     convert_button = gr.Button("Convert ⚡", elem_classes=["gradio-button"])