FA-99651 / Chunk citation alignment / Open access

Truncated excerpts retain citations beyond visible text · case 01

Truncated excerpts retain citations beyond visible text.

Verified by executionVariant 1 · 6 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

The end is copied from the original citation although the excerpt was shortened.

THE FAILURE

The end is copied from the original citation although the excerpt was shortened.

Unsuccessful approach: Capping the local end directly fails when the chunk starts away from zero.

Case contract

Input is chunk start b, original global [s,e), retained local length L. Return [s,min(e,b+L)) when s < b+L, otherwise no span.

Why this case matters

A deterministic citation bookkeeping model isolates source attribution from rendering and retrieval.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(base, span, kept):
    s,e=span
    return [s,e]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('cut tail',solve(N,[N+2,N+9],5),[N+2,N+5])
check('fully retained',solve(N,[N+1,N+3],5),[N+1,N+3])
check('beyond excerpt',solve(N,[N+6,N+8],5),None)
check('ends at boundary',solve(N,[N+3,N+5],3),None)
check('zero kept',solve(N,[N,N+1],0),None)
check('empty citation',solve(N,[N+2,N+2],5),None)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
cut tail	`[3, 10]`	`[3, 6]`	Failed
fully retained	`[2, 4]`	`[2, 4]`	Passed
beyond excerpt	`[7, 9]`	`None`	Failed
ends at boundary	`[4, 6]`	`None`	Failed
zero kept	`[1, 2]`	`None`	Failed
empty citation	`[3, 3]`	`None`	Failed

SHA-256 / f413ca0189aa19aaa71a0aab59612f12af739749af952ca931ba33e849de8804

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(base, span, kept):
    s,e=span
    return [s,min(e,kept)] if s<base+kept else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('cut tail',solve(N,[N+2,N+9],5),[N+2,N+5])
check('fully retained',solve(N,[N+1,N+3],5),[N+1,N+3])
check('beyond excerpt',solve(N,[N+6,N+8],5),None)
check('ends at boundary',solve(N,[N+3,N+5],3),None)
check('zero kept',solve(N,[N,N+1],0),None)
check('empty citation',solve(N,[N+2,N+2],5),None)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
cut tail	`[3, 5]`	`[3, 6]`	Failed
fully retained	`[2, 4]`	`[2, 4]`	Passed
beyond excerpt	`None`	`None`	Passed
ends at boundary	`None`	`None`	Passed
zero kept	`None`	`None`	Passed
empty citation	`[3, 3]`	`None`	Failed

SHA-256 / f1a809635fce9aaec948b7ae1fe3598aaeb93b18fbc8a23b6f4575d2c3a75348

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

Inputs are small integer offsets and identifiers under the stated toy contract; this does not define a document format or production citation API. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.

Observations recorded using Python 3.12.14 at 2026-09-30T16:48:26.713582+00:00.

Case digest / f74979eeb69f3842f10c5187495ed29313d504b1f9b0515b4ac7f8066406400f