/tools/csv-viewer
CSV Viewer Online for Large Files
Inspect RFC 4180 CSV and TSV as logical records, keep identifiers as text, make bounded edits and export locally.
1. Source
Choose local CSV or TSV
UTF-8 · one file · source remains immutable
2. Search and filters
Find rows without changing the index
3. Virtual table
Rows and columns
4. Row details and edit overlay
Selected row
5. Profiles
Column evidence
6. Export
Preview, redact and export
No redaction category is selected automatically. Unchanged CSV records are copied byte-for-byte; edited, redacted or column-filtered records are reserialised.
Preview the export to confirm its active filters, edits and redaction categories.
The first 20 matching records will appear here after preview.
/architecture
A local viewer built around CSV records
Logical record streaming
One MiB slices preserve escaped quotes, split Unicode, CRLF and embedded newlines while sparse checkpoints support seeking.
Bounded DOM
Only a small window of rows and columns exists in the page. The selected row is loaded separately for full inspection.
Text stays text
Leading zeroes, long integers and date-like identifiers remain exact strings unless JSON inference is explicitly enabled.
Immutable source
Edits and exclusions live in a bounded overlay. Downloads receive deterministic filtered or redacted names.
/viewer and editor
View, edit and export without changing the source
| Task | How it works | Representative input |
|---|---|---|
| Inspect RFC 4180 rows | Quoted commas, escaped quotes and embedded newlines remain one logical record. | |
| Open CSV or TSV | Comma, tab, semicolon and pipe delimiters are detected or selected explicitly. | |
| Edit selected fields | Select a row, edit its fields in the detail panel, then undo or redo from the bounded overlay. | |
| Review likely exposure | Masked secret, token, cookie and likely-PII signals can be filtered before explicit redaction. | |
| Export a derivative | Preview active filters, edits, excluded columns and redaction categories before local CSV, JSON or JSONL export. | Source bytes remain unchanged. |
/evidence
Maintained CSV and TSV test evidence
Data Demon Systems maintains the built-in multiline CSV, TSV and exposure-review examples as versioned test inputs. They cover leading zeroes, quoted commas, escaped quotes, embedded newlines, CRLF records, tab detection and inert token-like values.
The public CSV fixtures and methodology reproduce those cases and document the generated large-file test. CSV record handling follows RFC 4180.
For one complete JSON value per line, use the JSONL & NDJSON Viewer Online.
/limits
Deliberate large-file boundaries
Selected files are capped at 4 GiB and read in 1 MiB slices. Individual record inspection is bounded at 8 MiB, profiling covers the first 2,000 columns, rendered cells remain below 3,000, and a browser without Save File support uses a 100 MB Blob fallback.
/faq
CSV viewer and editor questions
Can it open TSV and semicolon-delimited files?
Yes. Comma, tab, semicolon and pipe detection is automatic, with a manual override that rebuilds the index.
How do I edit a CSV cell?
Select a row in the virtual table. Its fields appear in the selected-row panel, where edits remain in a bounded overlay until you export a new file.
Can I sort or insert rows?
Not in version one. Global sorting, insertion and reordering conflict with bounded operation. Cell and header edits plus row and column exclusions are supported.
Are exposure findings proof?
No. They are conservative, masked signals for operator review. Redaction never activates without an explicit category choice.