open/llm-ingestion-okf
1
0
Fork 0
Code Issues Releases 4 Activity

test(fixtures): the STS fixtures and fixture codes are fictitious

Browse source 
Three STS fixtures still carried the section titles and labels of one real
reference document, and three identifiers were copies of its codes with a
letter or a word swapped. They now describe an invented kitchen counter and
cookbook series: the titles, labels and descriptions of sts-identity.xml,
sts-inherit.xml and sts-empty-label.xml, the P350/P351 document codes, the
99-0001 delivery prefix and chapter 7 of the image and accounting corpora.
Generated fixtures are regenerated and the witness inventory's per-document
totals are identical before and after; only names and text move.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Kjell Tore Guttormsen 2026-09-23 14:50:44 +02:00
commit 88cf67f12e
65 changed files with 396 additions and 390 deletions
Download patch file Download diff file Expand all files Collapse all files

6
tests/test_soft_hyphen_door.py
View file

@ -1,6 +1,6 @@
"""The soft-hyphen normalisation door, in front of the persist gate.
Operator decision 2026-09-18. A 701-page reference standard was refused WHOLE
Operator decision 2026-09-18. A long reference standard was refused WHOLE
by `llm-ingestion-guard` 1.4.0 -- `output:zero-width-present`, HIGH, an any-tier
carrier and therefore `fail_secure` at every trust level -- because the
publisher's source carries 71 U+00AD SOFT HYPHEN and 0 real zero-width
@ -16,7 +16,7 @@ before -- which is the known-negative every test here carries.
The three alternatives and why they are not here: weakening the guard is the
security repo's call and weakens every consumer's screen; delivering a cleaned
corpus fixes one document and no other; and doing nothing leaves a 701-page
corpus fixes one document and no other; and doing nothing leaves a long
reference standard unreadable for the whole chain.
U+00A0 NBSP is NOT touched (6 633 of them in that document). The guard has no rule
@ -134,7 +134,7 @@ def _build(inbox: Path, out: Path, accounting: Path) -> tuple[int, str]:
_SOURCE = (
"# Kapit{s}tel 84\n\nAr{s}beider med sjoko{s}lade og surdeigs{s}bakst.\n\n"
"# Kapit{s}tel 7\n\nAr{s}beider med sjoko{s}lade og surdeigs{s}bakst.\n\n"
"# Krav\n\nTilsvar{s}ende krav gjelder.\n"
)