[tompollard]

Summary

• Restrict CSV delimiter detection to comma, semicolon, tab, or pipe for all CSV files. csv.Sniffer was interpreting digits in headerless decimal sensor files as delimiters, causing false row-width errors. Files that relied on space or colon as a delimiter will now be parsed as one-column CSV.
• Recognize Empatica E4 Connect sensor files (ACC, BVP, EDA, HR, TEMP) by their timestamp and sample-rate rows. Avoid treating those rows as column headers.
• Recognize E4 IBI.csv by its start-time and IBI first row, and check its rows as time_offset and ibi. An IBI.csv with no detected beats is reported as information.
• Match E4 file names in any letter case. tags.csv finds neighboring sensor or IBI files by listing its folder, so this also works on case-sensitive file systems.
• Find CSV and text files by extension in any letter case in the integrity, quality, and privacy checks. Files such as EDA.CSV or NOTES.TXT were previously skipped.
• Treat an empty tags.csv in a matching E4 session as "no event markers" (informational), not as a broken CSV or an empty-file warning.
• Scan headerless E4 tags as data. Do not label nine- or ten-digit Unix event timestamps as phone numbers, including timestamps shifted to dates before 2000 or after 2100; the exemption applies only to the event-time column, so phone-like values in surplus fields and other values are still reported.
• Check E4 sensor data by position under distinct channel names, skipping timestamp and sample-rate rows.
• Scan surplus fields in malformed CSV rows for private information under an "extra field" column, without aborting privacy checks.
• Report each privacy pattern found in a column, rather than only the first one matched. A column holding both an email address and a phone number now gets two warnings, so privacy warning counts can increase for any CSV dataset.
• Document the behavior and add regression tests for the reported cases and ordinary CSV safeguards.
• Add a small synthetic E4 export as a documented example and test the complete Python and CLI validation workflow against it.
• Resolve CSV layouts once for all validator checks. The E4 detector supplies its layout, while ordinary headered CSV remains the default. A new format can add a detector without changing each check.

This is limited to the legacy E4 Connect layout documented by Empatica: https://www.empatica.com/blog/decoding-wearable-sensor-signals-what-to-expect-from-your-e4-data/. It does not establish that timestamps have been de-identified.

Verification

• pytest -q — 195 passed, 1 skipped
• git diff --check — passed