tree b4380666259c55230a0ba105c3e4fcb2589aaea1
parent 41637c307ff042e668182e4d89aaa1f0ed608655
author Marc Kupietz <kupietz@ids-mannheim.de> 1781879048 +0200
committer Marc Kupietz <kupietz@ids-mannheim.de> 1781890647 +0200

Pass I5 xenoData metadata blocks to Krill output (issue #54)

I5 <xenoData> blocks carry corpus-provided metadata (e.g. the RPK
denomination) that previously never reached Krill, so it was neither
displayed nor searchable. Mirror KorAP-XML-Krill's KorAP::XML::Meta::I5
handling (commit 24ad3c0): parse each <meta> into a typed Krill field,
mapping the meta type to the field type (string/keyword/text/date ->
type:string/keywords/text/date; attachment/uri -> type:attachement) and
deriving the key as <project>.<name>. For doc/corpus headers the name is
level-prefixed (docRcpnt/corpusRcpnt) so the same meta stays distinct per
level; repeated keywords accumulate, repeated scalars keep the last value.

Fields are collected per level and inherited corpus -> doc -> text, then
attached to each text's Krill stand-off fields (Krill-only by construction).

Tests: parseXenoData unit coverage for the type/key mapping and edge cases
(text/date metas, empty/unknown-type skipping, doc/corpus prefixing), plus
a full-pipeline test over a new ked_sample.zip fixture (packaged from
KorAP-XML-Krill t/real/corpus/KED) that mirrors the Perl ked.t assertions.

Resolves #54

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Change-Id: I2d895bbc05cfe203a6e47ad4932c9ea1eb6fd0a2