Colquhoun.AI
← All guides

Research Guides

Count Research CSV Records, Not Text Lines

· Colquhoun.shop editorial

A research export can contain two observations while occupying more than two lines in a text editor. A comment with an embedded line break can explain the difference. Counting physical lines, or splitting every line at each comma, can quietly turn valid evidence into extra records and shifted columns.

This worksheet tests a file exchange you control. It is not a claim about a specific Colquhoun.shop importer, and it does not replace spreadsheet security checks.

1. Agree on the file format before testing

Record the delimiter, character encoding, quote convention and whether a header is present. Do not assume every file labeled CSV uses the same conventions.

RFC 4180 is an informational description of a common CSV format, not an Internet standard. It describes quoted fields containing commas or line breaks, and doubling an internal double quote within a quoted field. It also allows an optional header and a final record without an ending line break. RFC 4180, section 2: CSV format

2. Create a tiny fixture with deliberate awkward values

Use invented records rather than confidential research. Choose three columns: observation ID, page title and review note. In the first record, put a comma in the title and quotation marks around a word in the note. In the second, put a line break inside the note. Write down the intended result: two data records, three fields each, with the original punctuation and line break preserved.

For example, the first title might be Pricing, limits and support, while its note contains the word "included". The second note might contain two sentences separated by an actual line break. Those characters belong to field values; they should not create a fourth column or a third observation.

3. Parse records rather than guessing from lines

Use a CSV-aware importer or parser configured for the agreed format. A plain split at every comma does not understand whether a comma is inside quotes. A line-count command does not establish the number of logical records when fields may contain line breaks.

Compare the parsed output with the fixture's expected IDs and field values. State whether your reported count excludes the header. Include a separate empty-file or header-only case so that “zero observations” is not confused with a failed import.

4. Test the return trip

Export the parsed fixture from the receiving application and read it again with a compatible parser. Compare values, not just how the raw text looks: two valid serializations may quote fields differently. Check that neither note was split, neither title was truncated and no ID was duplicated.

If the application changes line endings within notes or applies another transformation, record that behavior and decide whether it meets the research handoff's requirements. Do not silently call the result lossless. Preserve the original file so the comparison remains possible.

5. Keep an acceptance record

Successful parsing does not establish that the values are factually correct or safe to open as formulas in a spreadsheet. Pair this structural test with the research CSV field-type safety checklist. The immediate goal is narrower: every intended observation remains one record, with its evidence in the right fields.

This guide describes the built-in Web Reader and Keyword Density behavior verified for this publication. Available data depends on the page response; these tools do not measure visits or search rankings.