0.2.2-rc2 · OFFLINE APPLICATION · NOT FOR PURCHASE
Review two CSV files by an explicit key and deliberate field mappings. Export a complete-scope summary, a bounded sensitive-detail report, or a per-pair exception CSV when its independent budget allows.
Evaluation showcase only. No application download, purchase, paid support or migration certification.
REAL EXPORTS / SYNTHETIC INPUTS
2 records per side. Key sku → sku; fields name → name and price → price. 1 identical / 1 changed / 0 only-source / 0 only-target. Beta → Beta revised changes; price stays 20 → 20.
Full-scope summary · Partial-detail HTML (1 included / 0 omitted) · Exception CSV (1 row)
5,000 records × 50 columns per side; 49 field mappings. Each input is 1,004,080 bytes. All 5,000 changed. The actual CSV is 6,002,582 bytes with all 5,000 exceptions. This one input shape fits; other accepted shapes may not.
Full-scope summary · Partial-detail HTML (50 included / 4,950 omitted) · Full example exception CSV
Leading columns: key, status, source_key_header, target_key_header, source_record_present, target_record_present. Then, for each pair N, four separate columns:
source_header_N, target_header_N, source_value_N, target_value_N
Presence flags distinguish an absent record from an empty value. Pair labels travel with each exception. Read exact schema, example header and text-mitigation rules.
BOUNDED BY DESIGN
Each input: 1 MiB (1,048,576 UTF-8 bytes), 5,000 data records, 50 columns. Each decoded header: at most 256 UTF-8 bytes; aggregate decoded headers: at most 8,192 UTF-8 bytes per file. Each prospective CSV, HTML or live review has an independent 8 MiB (8,388,608-byte) output budget, including escaping and repeated metadata. Inputs within every input limit can still have a blocked output. Blocking is explicit, not silent truncation; CSV, detail HTML, summary and live review are checked independently.
Summary is the default and covers all comparison counts and scope, but retains filenames/headers. Sensitive detail requires opt-in. Detail HTML is limited to 100 exceptions / 5,000 logical cells; at 49 mappings that means only 50 included exceptions. Values are not silently truncated. A selected key/value/label over 4,096 UTF-16 code units blocks detail HTML. Summary recovery requires its own safe plan.
Only Linux ARM Chromium headless has been verified. Normal Linux GUI launch/usability and recipient acceptance remain unverified. Windows, macOS and other browsers are unverified; actual Excel/LibreOffice imports are unverified. No paid platform support or performance promise is made. Full-cap CSV export took about 39 seconds in one QA run; synchronous work can stall the UI. This is not a performance promise.
BEFORE SHARING
The application runs offline from extracted local files. No application server, account, build or installation is required. This page is a static sample showcase, not the app or a purchase page. Hosting this page does not turn it into an online CSV service.
Each input: 1 MiB (1,048,576 UTF-8 bytes), 5,000 data records, 50 columns. Each decoded header: at most 256 UTF-8 bytes; aggregate decoded headers: at most 8,192 UTF-8 bytes per file. Each prospective CSV, HTML or live review has an independent 8 MiB (8,388,608-byte) output budget, including escaping and repeated metadata. Inputs within every input limit can still have a blocked output. Blocking is explicit, not silent truncation; CSV, detail HTML, summary and live review are checked independently.
No. Summary counts and scope cover the complete comparison, without record keys or values. Sensitive detail is opt-in and marked PARTIAL DETAIL: at most 100 exception records / 5,000 logical cells, lower limit wins. At 49 mappings the real example includes 50 records and omits 4,950. No included values are truncated. A selected key/value/label over 4,096 UTF-16 code units blocks detail. Live review separately caps 100 records / 5,000 actual cells.
One row per exception. Six leading columns: key, status, source_key_header, target_key_header, source_record_present, target_record_present. For every pair N: source_header_N, target_header_N, source_value_N, target_value_N. No joined fields. Formula-prefix apostrophe mitigation changes text; not lossless migration import or universal spreadsheet security.
Choose one unique nonblank key on each side. Defaults pair exact-name common non-key headers; review them. Values and keys are exact case-sensitive strings, with no trim or numeric/date normalization. Differing names require explicit mapping confirmation. Duplicate/blank keys block comparison. No merge, edit or migration certification. Changing inputs or mappings invalidates prior results.
Only Linux ARM Chromium headless has been verified. Normal Linux GUI launch/usability and recipient acceptance remain unverified. Windows, macOS and other browsers are unverified; actual Excel/LibreOffice imports are unverified. No paid platform support or performance promise is made. A full-cap short-label CSV download took about 39 seconds in one headless QA run. That is a responsiveness caveat, not a benchmark or speed guarantee; synchronous work can stall the UI.
No. This is 0.2.2-rc2 / NOT FOR PURCHASE. No checkout is active and no application ZIP is public here. Commercial authorization, finalized terms/contact and supported-runtime recipient acceptance are not complete.
Summary omits record keys, values and previews, but retains filenames, column labels, mappings, counts and rules. Those metadata may still be sensitive. Detail HTML and exception CSV contain record data: review before sharing. Reload forgets app state; this is not secure RAM/cache erasure.