Sanitize, format, and process apache parquet PII column payloads 100% in-browser with zero cloud upload risk.
Process and sanitize Parquetcolumnshield payloads 100% privately inside browser memory
Drop Local File Here or Click to Browse
Supports CSV, JSON, TXT, XML & Log Streams
parquet-tools cat --json part-00000.snappy.parquet | jq 'del(.customer_email, .ssn, .card_token)' > sample.json⚠️ Why Naive CLI / Regex Fails: Deleting columns with jq del() breaks downstream Spark/DuckDB schema readers expecting fixed struct fields, and parquet-tools meta dumps still leak plaintext customer emails and SSNs inside RowGroup ColumnChunk statistics.min_value and max_value headers.
🛡️ Pre-LLM & Ticket Safety: How to sanitize an Apache Parquet schema/row-group dump or parquet-tools inspect output before pasting into ChatGPT, Claude, or Cursor or sharing with Databricks / Snowflake support
Core capabilities built into ParquetColumnShield
Neutralizes the #1 hidden Apache Parquet privacy trap: plaintext customer emails, SSNs, and account IDs stored inside ColumnChunk min/max statistics and bloom filter headers.
Maintains exact BYTE_ARRAY, INT64, DECIMAL, and TIMESTAMP_MILLIS logical types so Spark, Trino, and DuckDB queries run without schema mismatch errors.
Traverses deeply nested Parquet repetition/definition levels to scrub PII inside user_profile.contacts[] arrays and event_properties map values.
Eliminate GH013 and Gitleaks blocks on dbt/Spark test fixtures and safely debug Parquet schema evolution errors in Claude or ChatGPT.
Zero configuration, zero cloud server dependencies, instant execution.
Paste parquet-tools cat --json records, parquet-tools inspect footer metadata, or PyArrow schema dumps directly into local browser memory.
Type-preserving column scrubbers hash emails, SSNs, IBANs, and bearer tokens across both row records and ColumnChunk min_value/max_value statistics.
Preview 70% of the sanitized Parquet dump immediately or unlock 100% export and the air-gapped Offline PRO HTML workstation for $19 one-time.
See why merchants and prosumers switch from monthly subscriptions to our local buyout.
| Feature / Metric | 🛡️ ParquetColumnShield | ☁️ Cloud SaaS (Dataedo) |
|---|---|---|
| Data Processing Architecture | 🛡️ 100% Local Browser RAM (Zero Network Upload) | ☁️ Cloud SaaS Upload |
| ColumnChunk min/max Statistics PII Handling | 🛡️ Automatic Scrubbing of Row Values AND Footer min/max Bounds | ☁️ Cloud SaaS Upload |
| Spark / DuckDB / Trino Schema Compatibility | 🛡️ Type-Preserving Pseudonymization (Keeps All Columns & Types) | ☁️ Cloud SaaS Upload |
| Pre-LLM & Git GH013 Compliance Readiness | 🛡️ Instant Gitleaks & GH013 Clean Lakehouse Test Fixtures | ☁️ Cloud SaaS Upload |
Eliminate vendor lock-in and cloud upload security risks.
Why pay monthly recurring SaaS fees for simple apache parquet tools? Switch to ParquetColumnShield for a single $19 buyout.
Dataedo uploads private data to cloud servers. We process 100% locally on your machine.
Process large apache parquet files instantly without waiting for cloud server roundtrips.
Deep technical breakdown driving our client-side productivity engine.
Processes ParquetColumnShield structures directly inside isolated browser memory.
Ensures zero ParquetColumnShield bytes touch external API servers or third-party storage.
Parses high-throughput ParquetColumnShield records in real-time with sub-millisecond response times.
Full compliance with international privacy mandates through total data sovereignty.
No monthly bills. No sneaky price hikes. Pay once, use forever.
Live Local Studio Preview
Prosumer & Solo Professional
One-Off Emergency Unlock & Trial
Everything you need to know about privacy, local memory execution, and buyouts.
Every Parquet RowGroup stores ColumnChunk statistics containing the exact min_value and max_value for each column to speed up predicate pushdown. For string columns like email, SSN, or account_number, real customer PII is embedded directly in the file footer.
When diagnosing schema evolution, partition skew, or predicate pushdown bugs in Spark, Trino, or Iceberg, removing columns alters the schema ordinal positions and breaks downstream SQL readers. ParquetColumnShield masks values in place while preserving every column name and type.
Data engineers often commit small parquet-tools JSON dumps or PyArrow schema snapshots as unit test fixtures, which get blocked when custom key_value_metadata contains S3 tokens or row samples contain live tokens. ParquetColumnShield scrubs both.
Yes. It traverses nested JSON representations of Parquet STRUCT, LIST, and MAP columns to mask inner PII fields while keeping repetition and definition structure intact.
Never. All parsing, footer statistic scrubbing, and Canvas column-density visualization execute 100% locally inside your browser memory with zero network egress.