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

feat(propose,consume,tools): the type that declares nothing, and the prefix that is not a word

Browse source 
Three of round 9's four measured holes, each closed with a rule chosen on a
measurement rather than named as a limit.

`rtf` GIVES 0 SEGMENTS -> 6 of 6 AUTHORED TITLES over N = 4. The container has
no heading style, so the author's title is bold text. The grammar is markdown,
not `rtf`: the converter already writes that title as `**...**` in the same
output every office row produces, so no `rtf`-only heading form exists. Three
parameters were swept over 47 readable documents and ONE carried -- refusing a
line that ends in terminal punctuation takes false-positive lines from 9-12 to
1-2. A maximum title length (unlimited/40/60/80/120) and a
must-stand-between-blank-lines clause are both FLAT, so neither is in the rule.
The last false positive is closed by G1, the principle `_gate_outline` already
carries: recovery yields to declaration. False positives are then 0 of the 31
declaring documents by construction, and 0 of 27 on the corpus. Reach: 2 of 39
corpus documents, both `docx`, 0 of 33 `pdf` and 0 of 2 `xlsx`. Behind
`--bold-title`, default OFF pending the hit@8 measurement; the default bundle
is byte-identical without it.

BOTH ALTERNATIVES THE ORDER NAMED WERE MEASURED AND FELLED. A fourth hand-laid
fixture DECLARES heading styles in a stylesheet and the converter discards
them, emitting the same bold line -- so "read the declared headings out of the
markdown" has nothing to read. `rtf` -> `docx` -> markdown yields 0 ATX
headings on that same document, because the loss is in the `rtf` READER before
any writer sees the style. Fixtures are hand-laid in `make_k2_office.py` with
the fasit written first; they live in their own directory because Door B walks
a drop directory recursively and `k2-office/` reads its N off the listing.

THE PREFIX OVER-MATCH: THREE CANDIDATES MEASURED, ALL THREE FAILED ON ONE ROW.
Re-measured on the pinned 453-concept bundle with the control run first:
`under` occurs 79 times by equality and matches 172 by prefix, `undersjoisk` 0
and 172, `bilateral` 0 and 400 of 453, `standhaftig` 0 and 219. The two extra
known-negatives were FOUND, not chosen -- every 4-character prefix ranked by
document frequency, then a real word taken from the widest. A longer floor
(5-8), a coverage share (0.5-0.8) and a long-words-only floor (>= 8) each cost
row 1 its rank on the default bundle and the whole row on Arm B. Decomposed:
row 1's token `prisene` reaches its gold document through
`pris|sammenstilling` on four characters -- 0.57 of one word and 0.22 of the
other -- so the over-match and the wanted match are one mechanism.

THE FOURTH CANDIDATE IS THE ANSWER: the shared prefix must be a WORD the bundle
uses. `pris` is; `bila` and `stan` are not. `bilateral` 400 -> 0 and 512 -> 0,
`standhaftig` 219 -> 56 and 235 -> 33, every hit@8 row keeping rank 1 on BOTH
bundles. `undersjoisk` stops at 162 because `under` IS a word here -- a genuine
Norwegian morpheme, so that residual is a different answer, not a ceiling. ON
by default (`--no-stem-prefix`), pinned with its own known-negative on the
shipped bytes.

THE SHIM: a path importer holds the object `module_from_spec` made, and
`sys.modules[__name__] = _impl` never reaches it. Measured under both counting
methods -- 3 of 76 public names by `vars()`. One line copies the public names
into this file's globals; the dunder filter is load-bearing, because an
unfiltered copy overwrites `__name__` before the next line uses it as the alias
key. It restores attribute ACCESS and not patch-through, which is why the alias
stays. A CHANGELOG note under 0.7.0 and a shim docstring line say so, since
what the consumer asked for was the note.

Suite 1515 -> 1535.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Kjell Tore Guttormsen 2026-09-09 23:05:45 +02:00
commit 191de89f41
16 changed files with 1100 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

110
tests/fixtures/make_k2_office.py vendored
View file

@ -26,6 +26,11 @@ from pathlib import Path
HERE = Path(__file__).parent
OUT = HERE / "k2-office"
#: Round 10's rtf variants live in their OWN directory, and that is not tidiness.
#: `test_k2_office_fixtures.py` reads N off `k2-office/` by listing it, because
#: Door B walks a drop directory RECURSIVELY -- anything parked beside the three
#: containers would enter that run and the denominator would stop being three.
RTF_OUT = HERE / "k2-rtf-variants"
TITLE = "Kravspesifikasjon for tunnelbelysning"
INTRO = "Dokumentet samler kravene til belysning i vegtunneler over 500 meter."
@ -307,6 +312,95 @@ def _rtf_row(row: tuple[str, ...]) -> str:
return "\\trowd\\trgaph108" + borders + "\n" + cells + "\\row\n"
# --- the three rtf VARIANTS round 10 added ------------------------------------
#
# WHY THEY EXIST. Round 9 measured the `rtf` row on ONE document and it came
# back at 0 of 0 declared headings, 0 concepts, 1368 of 1368 characters in no
# segment. A repair proposed on N = 1 is a repair fitted to one file, so round
# 10's order required N >= 3 before any rule was written. Three lawful sources
# were offered; this is the first of them, and it is the one this repository's
# committed-fixture policy allows: hand-laid here, in the same file, with the
# fasit written BEFORE the measurement.
#
# WHAT EACH ONE VARIES, and it is one axis each:
#
# -fet-alene the title is bold at BODY size, with no point-size change
# at all. Variant A (`krav-rikt-tekstformat.rtf`) sets its
# title bold AND at 16pt, so without this document a rule
# reading "bold" could be passing on the size instead.
# -fet-i-avsnitt a bold PHRASE inside a running sentence, which must NOT
# become a boundary. The document's own known-negative.
# -stil the container DECLARES heading styles (`\\s1`, `\\s2`) in a
# stylesheet. This is the control for the whole repair, and
# it is a control that FAILED in an informative direction:
# measured, the vendored converter discards the style and
# emits the same bold line, so "read the declared style"
# is not a route that exists for `rtf`. The fixture is kept
# precisely because it pins that.
#
# The bodies are deliberately short. These documents answer where a boundary
# is, not how much text survives -- variant A already carries the coverage and
# fidelity counts for the container.
RTF_VARIANT_TITLE = "Kravspesifikasjon for tunnelbelysning"
RTF_VARIANT_SECTION = "Luminanskrav per trafikklasse"
RTF_VARIANT_BODY_ONE = "Dokumentet samler kravene til belysning i vegtunneler over 500 meter."
RTF_VARIANT_BODY_TWO = "Terskelluminansen er 145 candela og gjelder i hele overgangssonen."
#: The bold RUN in `-fet-i-avsnitt`, set inside a sentence that continues past
#: it. Authored as emphasis, never as a title.
RTF_VARIANT_EMPHASIS = "fravik"
def _rtf_document(paragraphs: list[str]) -> bytes:
"""Wrap hand-written paragraph bodies in the minimal rtf container."""
header = "{\\rtf1\\ansi\\ansicpg1252\\deff0{\\fonttbl{\\f0\\froman Times New Roman;}}"
return (header + "".join(paragraphs) + "}").encode("ascii")
def rtf_bold_only_bytes() -> bytes:
"""Title and section name bold at BODY size -- no point-size change."""
return _rtf_document(
[
"\\pard\\sa180\\b " + _rtf_escape(RTF_VARIANT_TITLE) + "\\b0\\par\n",
"\\pard\\sa180 " + _rtf_escape(RTF_VARIANT_BODY_ONE) + "\\par\n",
"\\pard\\sa180\\b " + _rtf_escape(RTF_VARIANT_SECTION) + "\\b0\\par\n",
"\\pard\\sa180 " + _rtf_escape(RTF_VARIANT_BODY_TWO) + "\\par\n",
]
)
def rtf_bold_inside_paragraph_bytes() -> bytes:
"""One bold title, and one bold phrase mid-sentence that is not a title."""
sentence = (
"Avvik fra kravene krever soknad om "
+ "\\b "
+ _rtf_escape(RTF_VARIANT_EMPHASIS)
+ "\\b0 "
+ " fra byggherren for arbeidet starter."
)
return _rtf_document(
[
"\\pard\\sa180\\b " + _rtf_escape(RTF_VARIANT_TITLE) + "\\b0\\par\n",
"\\pard\\sa180 " + _rtf_escape(RTF_VARIANT_BODY_ONE) + "\\par\n",
"\\pard\\sa180 " + sentence + "\\par\n",
]
)
def rtf_styled_bytes() -> bytes:
"""The container DECLARES heading styles. The converter discards them."""
stylesheet = "{\\stylesheet{\\s0 Normal;}{\\s1\\b\\fs32 heading 1;}{\\s2\\b\\fs28 heading 2;}}"
return _rtf_document(
[
stylesheet,
"\\pard\\s1\\b\\fs32 " + _rtf_escape(RTF_VARIANT_TITLE) + "\\b0\\fs24\\par\n",
"\\pard\\s0\\sa180 " + _rtf_escape(RTF_VARIANT_BODY_ONE) + "\\par\n",
"\\pard\\s2\\b\\fs28 " + _rtf_escape(RTF_VARIANT_SECTION) + "\\b0\\fs24\\par\n",
"\\pard\\s0\\sa180 " + _rtf_escape(RTF_VARIANT_BODY_TWO) + "\\par\n",
]
)
def rtf_bytes() -> bytes:
body = [
"{\\rtf1\\ansi\\ansicpg1252\\deff0",
@ -323,11 +417,15 @@ def rtf_bytes() -> bytes:
if __name__ == "__main__":
OUT.mkdir(parents=True, exist_ok=True)
RTF_OUT.mkdir(parents=True, exist_ok=True)
written = {
"krav-presentasjon.pptx": build_container(pptx_parts()),
"krav-tekstdokument.odt": build_container(odt_parts(), stored_first="mimetype"),
"krav-rikt-tekstformat.rtf": rtf_bytes(),
OUT / "krav-presentasjon.pptx": build_container(pptx_parts()),
OUT / "krav-tekstdokument.odt": build_container(odt_parts(), stored_first="mimetype"),
OUT / "krav-rikt-tekstformat.rtf": rtf_bytes(),
RTF_OUT / "krav-rikt-tekstformat-fet-alene.rtf": rtf_bold_only_bytes(),
RTF_OUT / "krav-rikt-tekstformat-fet-i-avsnitt.rtf": rtf_bold_inside_paragraph_bytes(),
RTF_OUT / "krav-rikt-tekstformat-stil.rtf": rtf_styled_bytes(),
}
for name, payload in sorted(written.items()):
(OUT / name).write_bytes(payload)
print(f"wrote {name} ({len(payload)} bytes)")
for path, payload in sorted(written.items()):
path.write_bytes(payload)
print(f"wrote {path.parent.name}/{path.name} ({len(payload)} bytes)")