[TIKA-1611] Allow RecursiveParserWrapper to catch exceptions from embedded documents

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 1.9
• Component/s: core
• Assignee: Tim Allison
• Reporter: Tim Allison
• Created: 21/Apr/15 14:11
• Updated: 21/Apr/15 17:46
• Resolved: 21/Apr/15 17:30

Description

While parsing embedded documents, currently, if a parser hits an EncryptedDocumentException or anything wrapped in a TikaException, the Exception is swallowed by ParsingEmbeddedDocumentExtractor:

            DELEGATING_PARSER.parse(
                                    newStream,
                                    new EmbeddedContentHandler(new BodyContentHandler(handler)),
                                    metadata, context);
        } catch (EncryptedDocumentException ede) {
            // TODO: can we log a warning that we lack the password?
            // For now, just skip the content
        } catch (TikaException e) {
            // TODO: can we log a warning somehow?
            // Could not parse the entry, just skip the content
        } finally {
            tmp.close();
        }

For some applications, it might be better to store the stack trace of the attachment that caused an exception.

The proposal would be to include the stack trace in the metadata object for that particular attachment.

The user will be able to specify whether or not to store stack traces, and the default will be to store stack traces. This will be a small change to the legacy behavior.