Hi,
I'm working on a preflight refactoring [1]. Currently, the validation class implements the PdfAValidator interface that takes a DataSource in parameter. This class : - checks the syntax with a JavaCC parser - loads a PDDocument Object - runs a set of Helper on each object present in the PDF. - returns a ValidationResult object that contains all errors and the PDDocument. Here is what I would like to do: - Create a PdfAParser that inherits an existing parser (NonSequential or PdfParser) to process some validations during the load process. (ex : Stream length validation) - this specific parser will provide an instance of PdfADocument that inherits from PDDocument and contains the list of validation errors. - Create new helpers that check the PDF in a logical way (page by page) instead of object by object - Create a new implementation of the PdfAValidator that uses this new parser to keep compatibility with old version, but the PdfAParser should be the new entry point. Due to the validation path that the new parser must follow (validate the pdf in a logical way instead of object by object), I will create new classes and I will use as much as possible existing classes to avoid breaking the current implementation. All classes that become useless will be marked as Deprecated. I have two questions : - Do you think that the new validation seems better ? - Does the end of validation process must be done in a method "validate" of the PdfADocument or at the end of the parse method ? In the first case, the "validate" method could be call during the "save" method and check if a new PDF is a PDF/A, but because the validation starts in the Parser (syntax and stream length check) may be a "validate" method isn't the right things to do... If you have some questions do not hesitate. BR, Eric [1]https://issues.apache.org/jira/browse/PDFBOX-1312
