REPORT BENCH — WORKING DATA SAMPLES
Self-directed examples, September 2026. All input records are fictional.
Development is AI-assisted. No customer history or business results are claimed.

START HERE
Open cleanup-demo.html or dashboard-demo.html in a modern desktop browser.
Both HTML files contain their own code and sample data. No installation,
hosting account or network connection is needed to run a downloaded tool.
Nothing is uploaded or saved between sessions. Keep the original files.

CSV CLEANUP
The default sample has 16 records. Expected result: 15 retained, 1 exact
duplicate removed, 4 surrounding-space trims, 5 date normalizations and
7 kept rows flagged for review. The cleaned sample matches cleanup-expected.csv.
Use the download buttons for the original, cleaned, flagged and full-audit files.
The audit retains source-record number, physical CSV lines, original values,
output values, each change and each issue.

Only the fixed nine-column schema in cleanup-input.csv is accepted.
Limits: 1,000 data rows; 2 MiB (2,097,152 bytes), UTF-8.
Dates accept valid YYYY-MM-DD and normalize valid YYYY/MM/DD.
Ambiguous date order, numeric notation, required blanks and formula-like text
are flagged and preserved. Identical original parsed rows are deduplicated;
rows that become equal after trimming are not deduplicated automatically.
Same-email jobs are not treated as duplicates.

IMPORTANT: flagged data remains in the cleaned CSV. Review every flagged row.
Formula-like sample cells are preserved, not made safe for spreadsheet execution.
Inspect the JSON audit or import columns as text rather than opening such a CSV
with spreadsheet automatic formula evaluation. This tool does not validate
every possible business rule or certify the source data.

DASHBOARD
The five-column dashboard-input.csv is a separate, already structured fixture.
It is not an automatic downstream conversion of the cleanup example.
Required headers: job_id,job_date,service,status,hours
Limits: 500 data rows; 2 MiB (2,097,152 bytes); 20 service categories.
Unique, trimmed job IDs; real ISO dates within 1900–2099; trimmed service labels;
status completed, pending or cancelled; hours from 0 to 9999.99 with no more
than two decimal places. Quoted commas and UTF-8 are supported.
An invalid record refuses the whole file and clears the old view. No cleaning,
joining, historical storage or financial interpretation happens here.

Independent expected results:
All: 12 jobs; 17.00 hours; 7 completed; 58.3% completed share.
Completed: 7 jobs; 14.25 hours; 100.0% completed share.
Pending: 3 jobs; 2.25 hours; 0.0% completed share.
Cancelled: 2 jobs; 0.50 hours; 0.0% completed share.
All-status chart: Inspection 4, Maintenance 4, Repair 4.
Completed chart: Inspection 2, Maintenance 3, Repair 2.
Pending chart: Inspection 1, Maintenance 1, Repair 1.
Cancelled chart: Inspection 1, Maintenance 0, Repair 1.
Header-only input is valid and has zero jobs, zero hours and no percentage.
The filter applies to every calculation. Recorded hours include all records
in view, including pending/cancelled work; they are recorded inputs.

FILES
cleanup-demo.html — self-contained cleaner.
dashboard-demo.html — self-contained dashboard.
cleanup-input.csv — fictional source with deliberate issues.
cleanup-expected.csv — independently specified cleaned values, review required.
cleanup-audit.json — generated row-by-row audit for the fictional source.
dashboard-input.csv — separate fictional structured source.
README.txt — these instructions.

SCOPE
These are examples of one reusable cleaner and a Basic dashboard with three
metrics, one chart and one filter. They do not establish compatibility with
other formats or delivery of larger customer-specific packages.
Customer work requires agreed rules, example results, capacity and checks.
For Fiverr work, keep messages, files and payment within Fiverr.

VALIDATION
Cleanup module: eight test groups passed. Dashboard module: five test groups
cover fixed totals, filters, empty views, malformed fields and input limits.
These automated checks do not establish browser/device compatibility or an
accessibility certification. No browser interaction QA is claimed for this pack.

Updated demonstration: cleanup includes a field-by-field source-record inspector. Dashboard includes the matching filtered rows and a JSON view download. The All demos link opens the public website; processing itself remains offline.
