Build/Build & digital tools
Create a data-cleaning script from messy examples
Define cleanup rules and edge cases before generating a reversible script.
From these inputs to this result.
An authored, illustrative sample using fictional or supplied example data. This reference is not a recorded AI run; your tool’s answer will vary.
See the exact sample inputs
Sample records
row_id,name,email 1, Avery Chen ,AVERY@example.test 2,Avery Chen,avery@example.test 3,Morgan Reed, 4, Casey Park,Casey@example.test
Desired format
CSV columns row_id,name,email. Trim surrounding whitespace. Lowercase email. Deduplicate nonblank email case-insensitively, keeping the first row. Never discard a row just because email is blank. Keep row IDs. For this sample, expect 3 output rows and 1 duplicate removed. Generate a Python standard-library script that writes a NEW output file, refuses to overwrite an existing file, reports input/output/duplicate/blank-email counts, and shows how to run it. Never modify the source file.
| row id | name | |
|---|---|---|
| 1 | Avery Chen | avery@example.test |
| 3 | Morgan Reed | — |
| 4 | Casey Park | casey@example.test |
Ready to use your own material? Make this yours →
WORKING CODE YOU CAN TRY
Run the sample yourself.
This small Python script performs the cleanup described above. It writes a new file, keeps blank-email rows, and refuses to overwrite existing files. It was checked against the sample and awkward CSV values.
python clean-records.py sample-records.csv my-cleaned-records.csv
Requires Python 3. Run in the folder containing the downloads. This checks this specific cleanup rule; it does not decide whether two people are the same person.
Write the cleanup rules before the script.
- 01
Provide a small representative sample with messy records and the exact output format you want.
- 02
Resolve ambiguous dates, empty values, duplicates, and identifiers before generating the transformation.
- 03
Run the script on a copy first, then compare row counts, changed fields, and rejected records.
Your starting prompt
Replace the bracketed details or make it yours using the form.
Help me design a reversible data-cleaning script. REPRESENTATIVE SAMPLE RECORDS [Add sample records] DESIRED OUTPUT FORMAT [Add desired format] First propose explicit transformation rules with before-and-after examples. Flag ambiguities around dates, missing values, duplicate definitions, and identifiers with leading zeros. Ask me to resolve material ambiguities before destructive transformations. Generate a script that reads an input copy, writes a separate output, and reports changed, unchanged, and rejected records. Include a dry-run option and meaningful checks for row counts and edge cases. State the runtime requirements and distinguish code you have actually run from code that remains untested.
Give it a second look.
- The original input is preserved and ambiguous records are reported.
- Identifiers keep meaningful leading zeros, and totals or row counts reconcile.
“Add a dry-run mode that prints proposed changes without writing output. Include one test each for a duplicate, an empty field, and an ambiguous date.”
The capability to look for
A coding assistant or artifact environment; the result still needs to run and be checked.
Find a tool for the jobHow these ideas and prompts are made