This repository serves several tests for delb against a variety of XML documents, mostly TEI encodings. They are supposed to validate major code change proposals and ensure functionality before releases.
All test corpora preparations and tests are stuffed into a cli application
dit, use its --help output and the Justfile
recipes as references to get acquainted (Just run just for a recipes list).
The get-ready recipe installs the application and fetches the tests that are
included as git submodules.
It doesn't install the delb package / source tree that is supposed to be tested.
- https://www.deutschestextarchiv.de/media/download/dta_kernkorpus_2021-05-13.zip
- to be used in tests, the corpus must be fixed regarding the
lb-bug
- to be used in tests, the corpus must be fixed regarding the
- https://github.com/Brown-University-Library/iip-texts/tree/master/epidoc-filesf
- https://github.com/Brown-University-Library/usep-data/tree/master/xml_inscriptions/transcribed
- https://github.com/simondschweitzer/aed-tei/tree/master/files
- includes hieroglyhs and encoding errors
- https://github.com/orgs/erc-dharma/repositories
- several repos w/ south asian epigraphics, still growing
- https://gams.uni-graz.at/archive/objects/container:mws-gesamt/methods/sdef:Context/get?locale=fr&mode=&context=
- requires a web crawler
All tests can be invoked with just run-tests.
This verifies that an XPath query that targets a TagNode.location_path
attribute yields exactly that TagNode instance.
In order to save time the default --sample-volume is 25, like in percent.
It affects both the number of used random files and tested nodes.
The same document is parsed to a delb and an lxml representation. Both are then compared to have identical contents. This relies on the assumption that lxml / libxml2 are parsing correctly.
This validates that different serializations are parsed back to the identical
document representation.
See delb's test_serialization::test_transparency test for a more elaborated
description.