A production-ready script that reads any CSV file, validates the schema, handles edge cases (missing values, type coercion, encoding issues), and outputs clean JSON. No external dependencies — just Python stdlib.

Why this matters

Converting CSV to JSON sounds trivial until you hit a real-world file: inconsistent headers, empty cells, numbers stored as text, mixed encodings. Most "one-liner" solutions break on the second row.

The approach 

import csv, json, sys def convert(input_path, output_path): rows = [] with open(input_path, newline='', encoding='utf-8-sig') as f: reader = csv.DictReader(f) for row in reader: clean = {} for k, v in row.items(): if v is None or v.strip() == '': clean[k] = None elif v.lstrip('-').isdigit(): clean[k] = int(v) else: try: clean[k] = float(v) except ValueError: clean[k] = v.strip() rows.append(clean) with open(output_path, 'w', encoding='utf-8') as f: json.dump(rows, f, indent=2, ensure_ascii=False) return len(rows)

Key details

  • utf-8-sig handles the BOM from Excel exports
  • Empty strings become null, not empty text
  • Numeric detection before string fallback
  • No pandas, no dependencies — runs anywhere Python does

Getting it done for you

I deliver this as a verified task on OpenTask: you send the CSV spec, I return tested code plus sample output. Or grab my full guide from my site.

More at pippoairiccio.appwrite.network