Review column types before you import a data table
Admin ·
Check field evidence before importing records into a typed database or spreadsheet. A column can contain an identifier that looks like a number, a quoted word that looks like a boolean or an integer too large to cast safely.
Check the identifier and amount columns
Choose CSV and Load example. Keep Treat quoted fields as text selected and choose Inspect types. The id field is Mixed. It contains two leading-zero text identifiers, 007 and 042, and one number-like field, 105. Read all three as identifiers if their spelling matters.
The amount field has three number-shaped values. The report flags 9007199254740993 as a precision risk and shows its full original digits. Keep that value as text in a JavaScript number workflow where exact digits matter. The joined field has two valid dates and one blank.
id,active,joined,amount,note
007,true,2024-02-29,9007199254740993,""
042,false,2024-03-01,12.50,"false"
105,null,,0.25,readyRFC 4180, CSV records and quotingRFC 8259, JSON types and number interoperabilityRFC 3339, full-date field syntax
Compare the note field under two quote settings
With the quote setting on, note has three Text values and two quoted fields. Turn off Treat quoted fields as text and inspect the same input again. Note now has one Blank, one Boolean and one Text value. Its suggested type becomes Mixed.
Compare the evidence before choosing an import type. CSV quotes also protect commas inside a field, so they do not settle whether a value is text. The unquoted spellings true, false and null have special meanings in this inspection. A value such as TRUE stays text.
Distinguish missing keys and typed JSON values
Choose JSON, load its example and inspect it. The joined field contains two date-shaped strings. It counts as Text because JSON stores both values in quotes. The extra field contains one object and is missing from one row. The active field has a boolean and an explicit null.
Download column-types.json and verify the counts and examples with the input before creating a schema. Review a mixed field's actual purpose before casting it. The report does not modify the source table.
RFC 8259, JSON types and number interoperabilityRFC 3339, full-date field syntax
Common questions
Can the inspection spot dates that would roll over into another month?
The CSV date check requires a real YYYY-MM-DD calendar date. For example, unquoted 2024-02-29 is a date and 2024-02-30 stays text.
Does the report expand arrays and objects into extra columns?
Each property at the top level of a JSON row becomes a column. Nested objects and arrays appear as evidence types in that column.