test_csv.py

import pandas as pd
import io

# Test the CSV file parsing
raw_data = open('S1_Venues_Cali(in).csv', 'rb').read()

for encoding in ['utf-8', 'latin-1', 'cp1252', 'iso-8859-1']:
    try:
        df = pd.read_csv(io.StringIO(raw_data.decode(encoding)))
        print(f"\n=== Successfully decoded with: {encoding} ===")
        print(f"Columns: {list(df.columns)}")

        # Clean column names
        cleaned_columns = []
        for col in df.columns:
            cleaned = str(col).strip().replace('\ufeff', '').replace('\u200b', '').replace('\ufffe', '')
            print(f"Original: {repr(col)} -> Cleaned: {repr(cleaned)}")
            cleaned_columns.append(cleaned)
        df.columns = cleaned_columns

        print(f"\nCleaned columns: {list(df.columns)}")

        # Check for instagram
        instagram_col = None
        for col in df.columns:
            print(f"Checking: {repr(col)} -> lower: {repr(col.lower())} -> 'instagram' in lower: {'instagram' in col.lower()}")
            if 'instagram' in col.lower():
                instagram_col = col
                break

        print(f"\nFound Instagram column: {instagram_col}")
        break
    except Exception as e:
        print(f"Failed with {encoding}: {e}")