Empty CSV string deserialized as String containing "null" instead of null String

[hanlonjohn23]

Describe the bug
When attempting to deserialize a CSV with empty strings to an object with string fields, the fields are populated with "null" instead of the expected null. Note that this is using the CsvParser.Feature.EMPTY_STRING_AS_NULL feature.

Version information
2.13.3

To Reproduce
My project is using this code to deserialize CSV into a set of POJO objects:

public static <T> Set<T> deserializeSetFromCsvAtPath(Path path, Class<T> clazz) throws IOException {
    Set<T> set = new HashSet<>();

    try (InputStream is = Files.newInputStream(path)) {
        CsvMapper mapper = new CsvMapper();
        mapper.enable(CsvParser.Feature.EMPTY_STRING_AS_NULL);

        CsvSchema headerSchema = CsvSchema.emptySchema().withHeader();

        MappingIterator<T> iterator = mapper
                .readerFor(clazz)
                .with(headerSchema)
                .readValues(is);

        while (iterator.hasNextValue()) {
            set.add(iterator.nextValue());
        }

    } catch (IOException e) {
        throw new IOException("Failed to input stream file at " + path, e);
    }

    return set;
}

We are expecting that given a class like so:

public class Row {
    private final Integer id;
    private final String value;
. . .
}

We would deserialize a CSV like this:

id,value
1,

Into an object containing

id: 1
value: null

Instead, we are experiencing

id: 1
value: "null"

Expected behavior
That given the code above, empty strings in a CSV will be deserialized to a null String object, instead of a String containing "null".

Additional context
Similar to this issue here, only in a different context. Furthermore, I assume that the underlying issue is the same:

Use of parser.getText() unfortunately did include "null" as value. So better choice is parser.getValueAsString() which only coerces scalars, but returns null for JSON null as well as for structured tokens (START_ARRAY, START_OBJECT).

[cowtowncoder]

Will move to the right repository for CSV format issues.

[cowtowncoder]

Thank you for reporting this: it sounds like a possible bug. Thank you for also including reproduction.
I'll see if I can add a unit test and possibly resolve this (or at very least understand what is going on).

[cowtowncoder]

Hmmmh. Unfortunately I am unable to reproduce this on my side, with test like:

    final CsvMapper MAPPER = mapperForCsv();

    static class Row330 {
        public Integer id;
        public String value = "default";
    }

    public void testEmptyStringAsNull330() throws Exception
    {
        CsvSchema headerSchema = CsvSchema.emptySchema().withHeader();
        final String DOC = "id,value\n"
                + "1,\n";

        MappingIterator<Row330> iterator = MAPPER
                .readerFor(Row330.class)
                .with(CsvParser.Feature.EMPTY_STRING_AS_NULL)
                .with(headerSchema)
                .readValues(DOC);        
        Row330 row = iterator.next();

        assertEquals(Integer.valueOf(1), row.id);
        assertNull(row.value);
    }

so I think I'd need some help here.

[hanlonjohn23]

Thanks for your speedy response @cowtowncoder!

So I looked in to this some more this morning, and I managed to track down the source of the issue I was experiencing.
As part of my environment, I was using a custom deserializer like so
Row330.java:

import com.fasterxml.jackson.databind.annotation.JsonDeserialize;

@JsonDeserialize(using = Row330Deserializer.class)
public class Row330 {
    public Integer id;
    public String value = "default";

    public Row330(
            Integer id,
            String value
    ) {
        this.id = id;
        this.value = value;
    }
}

Row330Deserializer.java:

import com.fasterxml.jackson.core.JacksonException;
import com.fasterxml.jackson.core.JsonParser;
import com.fasterxml.jackson.databind.DeserializationContext;
import com.fasterxml.jackson.databind.JsonDeserializer;
import com.fasterxml.jackson.databind.JsonNode;

import java.io.IOException;

public class Row330Deserializer extends JsonDeserializer<Row330> {
    @Override
    public Row330 deserialize(JsonParser p, DeserializationContext ctxt) throws IOException, JacksonException {
        JsonNode node = p.readValueAsTree();
        return new Row330(
                node.get("id").asInt(),
                node.get("value").asText()
        );
    }
}

When using this set up, I believe your test above will fail, as value will be deserialized as "null".
Note that using .textValue() in place of .asText() results in a null String as expected, so I don't know if this is a bug, or simply expected behavior that I couldn't find documented.

[cowtowncoder]

I think it works as expected: asText() would give String "null" for NullNode; this should be specified by Javadocs in JsonNode (haven't checked; PR against Javadocs would be welcome if not).

Whether that makes sense or not would be an issue for jackson-databind, but not something CSV module would have control over.

[hanlonjohn23]

Understandable. Thanks for helping me with my problem @cowtowncoder, even though it turned out to not be a bug!

[cowtowncoder]

No problem @jhanlonhg ! Just glad we were able to figure it out.