FA-11871 / Search retrieval semantics / Open access

Filtered search returns an underfilled result page · case 01

Filtered search returns an underfilled result page.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

The rank cutoff is applied before eligibility filtering.

VERIFIED REPAIR

Filter all candidates, sort by descending score and identifier, then truncate.

Unsuccessful approach: Oversampling a fixed multiple still misses eligible hits farther down the candidate list.

Case contract

Hits are [id,score,eligible], IDs unique. Return the highest ranked k eligible IDs with lexical score ties; k is nonnegative.

Why this case matters

An offline deterministic retrieval model isolates this search contract from tokenization, storage, and network behavior.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(hits, k):
    return [h[0] for h in sorted(hits,key=lambda h:(-h[1],h[0]))[:k] if h[2]]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('deep eligible hit', solve([[str(i),100-i,False] for i in range(N+3)]+[['ok',0,True]],1),['ok'])
check('eligible score order', solve([['b',N,True],['a',N+1,True]],2),['a','b'])
check('score ties', solve([['b',N,True],['a',N,True]],1),['a'])
check('zero limit', solve([['a',N,True]],0),[])
check('empty candidates', solve([],N),[])
check('all rejected', solve([['a',N,False]],1),[])
check('short page exhausted', solve([['a',N,True]],5),['a'])
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
deep eligible hit	`[]`	`['ok']`	Failed
eligible score order	`['a', 'b']`	`['a', 'b']`	Passed
score ties	`['a']`	`['a']`	Passed
zero limit	`[]`	`[]`	Passed
empty candidates	`[]`	`[]`	Passed
all rejected	`[]`	`[]`	Passed
short page exhausted	`['a']`	`['a']`	Passed

SHA-256 / 4ce2d552f508e500a8f0c1986b6258f9885813fdb27ba33c08267f1c89649003

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(hits, k):
    return [h[0] for h in sorted(hits,key=lambda h:(-h[1],h[0]))[:2*k] if h[2]][:k]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('deep eligible hit', solve([[str(i),100-i,False] for i in range(N+3)]+[['ok',0,True]],1),['ok'])
check('eligible score order', solve([['b',N,True],['a',N+1,True]],2),['a','b'])
check('score ties', solve([['b',N,True],['a',N,True]],1),['a'])
check('zero limit', solve([['a',N,True]],0),[])
check('empty candidates', solve([],N),[])
check('all rejected', solve([['a',N,False]],1),[])
check('short page exhausted', solve([['a',N,True]],5),['a'])
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
deep eligible hit	`[]`	`['ok']`	Failed
eligible score order	`['a', 'b']`	`['a', 'b']`	Passed
score ties	`['a']`	`['a']`	Passed
zero limit	`[]`	`[]`	Passed
empty candidates	`[]`	`[]`	Passed
all rejected	`[]`	`[]`	Passed
short page exhausted	`['a']`	`['a']`	Passed

SHA-256 / 85b3d177f5556cd37960ff38ddfaad8c9b4c6e667caad09ba6587b8c752b14a8

3 / The verified repair

Exit 0

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

N = 1
observations = []
def solve(hits, k):
    return [h[0] for h in sorted((h for h in hits if h[2]),key=lambda h:(-h[1],h[0]))[:k]]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('deep eligible hit', solve([[str(i),100-i,False] for i in range(N+3)]+[['ok',0,True]],1),['ok'])
check('eligible score order', solve([['b',N,True],['a',N+1,True]],2),['a','b'])
check('score ties', solve([['b',N,True],['a',N,True]],1),['a'])
check('zero limit', solve([['a',N,True]],0),[])
check('empty candidates', solve([],N),[])
check('all rejected', solve([['a',N,False]],1),[])
check('short page exhausted', solve([['a',N,True]],5),['a'])
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
deep eligible hit	`['ok']`	`['ok']`	Passed
eligible score order	`['a', 'b']`	`['a', 'b']`	Passed
score ties	`['a']`	`['a']`	Passed
zero limit	`[]`	`[]`	Passed
empty candidates	`[]`	`[]`	Passed
all rejected	`[]`	`[]`	Passed
short page exhausted	`['a']`	`['a']`	Passed

SHA-256 / 05454f9b80779229ef491dfd8a92bab66cefdd406dccbee9b6dd2f2485da9c31

Verification & scope

Inputs are already tokenized or scored; this model makes no claim about production engine performance or linguistic analysis. 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 2026-09-29T14:38:51.749573+00:00.

Case digest / 071dae902cee60c74f1014bdd3841de520328b0c9813c8701208f00692de9ab9