Package org.apache.tika.metadata.schema

Class MigrationTableGenerator

java.lang.Object

org.apache.tika.metadata.schema.MigrationTableGenerator

public final class MigrationTableGenerator
extends Object

TIKA-4797 / TIKA-4799. Produces the 3.x->4.x metadata-key migration table (metadata-migration-3x-4x.json) that LegacyKeyMigrationFilter loads, by joining the committed 3.x and 4.x {class, field, key} field tables on field identity (Class#field):

• same field, changed key -> a RENAME (auto);
• fields the join can't bridge (moves/drops, or fields renamed on both sides) -> taken from the adjudicated migration-overlay.tsv (v4 == DROPPED = 4.x no longer emits it; a third-column EGRESS_ONLY marker = the row applies only V4_TO_V3, for 3.x spellings that are ambiguous or unsafe to rewrite on ingest).

The auto-join is core scope only (org.apache.tika.metadata.*) until the 3.x table is widened past its v1 (the 3.x snapshot only has core-declared fields, so widening the join side would never find extra matches). Overlay-target validation is not core-restricted (TIKA-4816 rename batch): an adjudicated row may rename a bare/legacy key to a Property declared in a parser module (e.g. geotopic:name in GeoParser, grobid:tei:title in TEIDOMParser).

Overlay targets validate against metadata-keys.json (the closed-key registry), not the field table: fieldTable()'s reflection-based field discovery eagerly resolves every declared field's type, so a class with any field typed on a not-transitively-visible dependency (e.g. GeoParser's NameFinderME, which pulls in opennlp-tools only via the provided-scope ctakes-core) throws NoClassDefFoundError and drops the WHOLE class. The closed-key registry has no such gap: it fully loads each class instead of just inspecting field types.

Dependency-free by design: tika-core carries the filter and must stay Jackson-free, so both this generator and the filter emit/parse the flat JSON by hand. MetadataMigrationTableTest regenerates in-memory and asserts the committed tika-core copy has not drifted.

   java ... MigrationTableGenerator <3x-fields.json> <4x-fields.json> <4x-keys.json> <overlay.tsv> <out.json>
 

Method Summary

Modifier and Type	Method	Description
static String	generate(String threeFieldsJson, String fourFieldsJson, String fourKeysJson, String overlayTsv)	Build the sorted v3 -> v4 migration JSON from the four input contents.
static void	main(String[] args)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

generate

public static String generate(String threeFieldsJson,
 String fourFieldsJson,
 String fourKeysJson,
 String overlayTsv)

Build the sorted v3 -> v4 migration JSON from the four input contents.

main

public static void main(String[] args)
                 throws Exception

Throws:

Exception