CSV string double quotes: String length is also considered?

[blackr1234]

I'd like to know why jackson-dataformat-csv:

• Does not quote alphanumeric strings with length <25 (<=24)
• Quotes alphanumeric strings with length >=25

Probably it is documented somewhere. If anyone knows, please let me know.

Thanks.

---

Sample code

public class Main {

    public static void main(String[] args) throws Exception {

        final List<Foo> list = Arrays.asList(
            new Foo("a23456789012345678901234"),
            new Foo("a234567890123456789012345")
        );

        final CsvMapper mapper = new CsvMapper();

        final CsvSchema.Builder schemaBuilder = CsvSchema.builder().setUseHeader(true);
        schemaBuilder.addColumn("name");
        final CsvSchema schema = schemaBuilder.build();

        final ObjectWriter writer = mapper.writerFor(Foo.class).with(schema);
        writer.writeValues(System.out).writeAll(list);

        // name
        // a23456789012345678901234
        // "a234567890123456789012345"
    }
}

class Foo {
    private String name;

    public Foo(String name) {
        setName(name);
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }
}

[cowtowncoder]

Short answer is that CsvEncoder, by default, uses fast heuristics to determine if quoting is or may be needed: 24 happens to be the limit defined as CsvEncoder.MAX_QUOTE_CHECK. I am not sure this is documented in README; PRs are welcome for changes to improve explanations.

There are settings to change this behavior:

• CsvGenerator.Feature.STRICT_CHECK_FOR_QUOTING (default: false) -- if enabled, check will be performed using full String value, no matter the length
• CsvGenerator.Feature.ALWAYS_QUOTE_STRINGS (default: false) -- if enabled, no checks are performance and quoting is always applied.