CSV Deduplicator & Cleaner Cleaner..

The CSV cleaner processes your file in four stages. First, it parses the CSV using a streaming parser that handles quoted fields, escaped characters, and common formatting variations. Column types are auto-detected by analyzing both the header name and a sample of cell values for patterns like @ symbols (email), digit density (phone), and dot-separated segments without spaces (domain).

Second, normalization is applied column by column. Email and domain columns are lowercased, name columns are converted to Title Case (with a toggle to disable), and all fields have whitespace trimmed. Each normalization is counted so you can see exactly how many changes were made.

Third, deduplication runs in two passes. The exact-match pass computes a hash of each complete row and groups identical rows together. The fuzzy-match pass uses a blocking strategy — grouping rows by shared characteristics (email domain, name prefix, domain root) and comparing only within blocks using Levenshtein distance. This avoids the O(n-squared) performance problem of comparing every row to every other row.

Finally, tie-breaking selects which duplicate to keep. The row with the most non-empty cells is preserved. If two duplicates have equal completeness, the first occurrence wins. The exported CSV is sanitized against formula injection by prefixing cells starting with =, +, -, or @ with a single quote.