[TIKA-1680] Add configuration layer to configure, Parsers default configurable properties.

• Type: Improvement
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Component/s: config
• Assignee: Unassigned
• Reporter: Mario Costa
• Created: 15/Jul/15 14:29
• Updated: 04/Aug/15 22:22

Description

In several conditions, we want to change the default tika configuration, for some specific parsers, to change the behaviour of libraries/components wrapped in the parsers.

Currently, the ParseContext class allows, for means of passing this default configuration.

Add means, to load from a configuration database/files default configuration to override, currently hard coded, defaults.

E.g. #TIKA-640, there are no means to change the defaults, by means of configuration files, with out having to change, tika source.

• Sugestion

• Add a ParserConfigurationFactory, to delegate configuration loading, to a database/property based file etc.
• Allow for configuration factory instance to be configured via file, like tika-config.xml for Parsers.

interface ParserConfigurationFactory {
/**

• Create or fille the given ParseContext with the default configuration for the parser class...
  */
  <P extends Parser> ParseContext getDefaultConfig(Class<P> class, ParseContext ctx);
  }

Replace when creating a ParseContext, by

ParseContext ctx = ParserConfigurationFactorySingleton.getSingleton().getDefaultConfig(...) ...

Filling the configuration, given a set of properties, could be done via a class <ParserName>ConfigBuilder that know how to build and set the configuration properties in specific objects (such as MimeConfig), related to 3rd party external libraries ...

Best regards,
mc

Issue Links

• is related to: TIKA-1508 Add uniformity to parser parameter configuration