BeanOutputParser improvements

[jonas464646]

Hi

When using the BeanOutputParser to parse dates or numbers, jackson fails fast when the document/input text isn't in the expected format. For example:

com.fasterxml.jackson.core.JsonParseException: Unexpected character ('2' (code 50)): was expecting double-quote to start field name
 at [Source: (String)"{
  "doubleValue": 2,248.02,

java.lang.RuntimeException: com.fasterxml.jackson.databind.exc.InvalidFormatException: Cannot deserialize value of type `java.util.Date` from String "08-Jan-2024": not a valid representation (error: Failed to parse Date value '08-Jan-2024': Cannot parse date "08-Jan-2024": not compatible with any of standard forms ("yyyy-MM-dd'T'HH:mm:ss.SSSX", "yyyy-MM-dd'T'HH:mm:ss.SSS", "EEE, dd MMM yyyy HH:mm:ss zzz", "yyyy-MM-dd"))`

If you have a bean with for example 100 properties, the request fails completely when one property can't be casted.
Could the default be improved to catch the property exception and set only the failing property to null so that the other working properties could be used? To not hide the internal exception a log warning could be added instead.

To improve/customize the parsing, is there already an recommended approach?

• It could be done by optionally adding custom jackson convertors
• or by using tools/openai functions for this