FA-80631 / Bidirectional text layout / Open access
Caret visual position: leading edge side · case 01
Carets attached to RTL characters appear on the wrong side.
ROOT CAUSE
The leading-edge formula uses the trailing-edge sides.
VERIFIED REPAIR
Leading edge is left for LTR and right for RTL.
Unsuccessful approach: Always using the right side breaks LTR characters.
Case contract
Input [levels, logical caret offset k, affinity]. Characters are one unit wide and visually ordered by level reversal. Affinity "after" attaches the caret to the leading edge of character k (if k < n), otherwise to the trailing edge of character k-1; offset 0 always uses the leading edge of character 0. The leading edge of an LTR character is its left side, of an RTL (odd) character its right side. Return the visual x.
Why this case matters
Mixed right-to-left and left-to-right text must resolve levels and visual order exactly, or words, numbers and carets land in the wrong place.
1 / The failure
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(x):
def reorder(lv):
vis = list(range(len(lv)))
odd = [l for l in lv if l % 2 == 1]
if not odd:
return vis
for lev in range(max(lv), min(odd) - 1, -1):
i = 0
while i < len(vis):
if lv[vis[i]] >= lev:
j = i
while j < len(vis) and lv[vis[j]] >= lev:
j += 1
vis[i:j] = vis[i:j][::-1]
i = j
else:
i += 1
return vis
levels, k, affinity = x
n = len(levels)
if n == 0:
return 0
vis = reorder(levels)
pos = [0] * n
for v, i in enumerate(vis):
pos[i] = v
if affinity == 'after' and k < n or k == 0:
j = k
leading = True
else:
j = k - 1
leading = False
rtl = levels[j] % 2 == 1
if leading:
return pos[j] if rtl else pos[j] + 1
return pos[j] if rtl else pos[j] + 1
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('regression: leading edge side', [[1, 1, 1, 0], 0, 'after'], 3), ('regression: leading edge side', [[2, 0], 0, 'after'], 0), ('regression: leading edge side', [[2, 2, 2, 2, 2, 3, 3, 2, 1], 0, 'before'], 1), ('line end', [[1, 1, 1], 3, 'after'], 0), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('control layout', [[3, 3, 3, 2, 2, 3, 3, 3, 3, 2], 9, 'before'], 5), ('control layout', [[2, 2, 2, 2, 1, 1, 3], 3, 'before'], 6)], [('regression: leading edge side', [[1, 1, 1, 1, 0, 1, 1, 1], 4, 'after'], 4), ('regression: leading edge side', [[4, 3, 3, 3], 3, 'after'], 1), ('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2], 2, 'after'], 2), ('regression: leading edge side', [[4], 0, 'after'], 0), ('line start before', [[1, 1, 0], 0, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[1, 1, 1, 1, 3, 3, 1, 1], 1, 'before'], 7), ('control layout', [[0, 0, 1, 1, 2, 2, 2, 0, 0], 6, 'before'], 4)], [('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2, 1, 1], 1, 'after'], 3), ('regression: leading edge side', [[2, 2, 2, 2, 1], 4, 'after'], 1), ('regression: leading edge side', [[3, 0, 0, 0, 2, 2, 2], 3, 'after'], 3), ('regression: leading edge side', [[3, 2, 2, 2, 1, 1, 3, 3, 3], 1, 'after'], 6), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('line end', [[1, 1, 1], 3, 'after'], 0), ('control layout', [[1, 1, 1, 1, 3, 3, 3], 4, 'before'], 3), ('control layout', [[0, 0, 0, 1, 3, 3, 3, 3], 6, 'before'], 5)], [('regression: leading edge side', [[3, 3, 3, 3, 1, 1, 1], 6, 'after'], 1), ('regression: leading edge side', [[1, 1], 1, 'after'], 1), ('regression: leading edge side', [[2, 2, 3, 3, 3, 1, 1, 3], 0, 'after'], 3), ('regression: leading edge side', [[4, 4, 4, 4, 2], 1, 'after'], 1), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 2, 2, 3, 3, 2, 2, 2, 3, 3], 10, 'after'], 8), ('control layout', [[1, 1, 1, 1, 1, 0, 2, 2], 4, 'before'], 1)], [('regression: leading edge side', [[2, 2, 2, 4, 4, 4, 4, 3, 3], 4, 'after'], 6), ('regression: leading edge side', [[1, 1, 1, 1, 1], 2, 'after'], 3), ('regression: leading edge side', [[2, 2, 1, 1], 0, 'after'], 2), ('regression: leading edge side', [[3, 3, 2, 2, 2, 2, 2, 2, 1, 1], 7, 'after'], 9), ('line end', [[1, 1, 1], 3, 'after'], 0), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 3, 4, 1, 1, 1, 1, 2], 5, 'before'], 3), ('control layout', [[1, 1, 1, 3, 3, 3, 3, 1, 1, 1], 9, 'before'], 1)]]
for label, args, expected in fixtures[N - 1]:
check(label, solve(args), expected)
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 |
|---|---|---|---|
| same boundary after | 3 | 4 | Failed |
| regression: leading edge side | 2 | 3 | Failed |
| regression: leading edge side | 1 | 0 | Failed |
| regression: leading edge side | 2 | 1 | Failed |
| line end | 0 | 0 | Passed |
| caret at direction boundary | 2 | 2 | Passed |
| control layout | 5 | 5 | Passed |
| control layout | 6 | 6 | Passed |
SHA-256 / 64ee8ca0dbca149ef312909664a23d78361f05eb194562ce41fa5e75fe44033f
2 / The unsuccessful fix
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(x):
def reorder(lv):
vis = list(range(len(lv)))
odd = [l for l in lv if l % 2 == 1]
if not odd:
return vis
for lev in range(max(lv), min(odd) - 1, -1):
i = 0
while i < len(vis):
if lv[vis[i]] >= lev:
j = i
while j < len(vis) and lv[vis[j]] >= lev:
j += 1
vis[i:j] = vis[i:j][::-1]
i = j
else:
i += 1
return vis
levels, k, affinity = x
n = len(levels)
if n == 0:
return 0
vis = reorder(levels)
pos = [0] * n
for v, i in enumerate(vis):
pos[i] = v
if affinity == 'after' and k < n or k == 0:
j = k
leading = True
else:
j = k - 1
leading = False
rtl = levels[j] % 2 == 1
if leading:
return pos[j] + 1
return pos[j] if rtl else pos[j] + 1
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('regression: leading edge side', [[1, 1, 1, 0], 0, 'after'], 3), ('regression: leading edge side', [[2, 0], 0, 'after'], 0), ('regression: leading edge side', [[2, 2, 2, 2, 2, 3, 3, 2, 1], 0, 'before'], 1), ('line end', [[1, 1, 1], 3, 'after'], 0), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('control layout', [[3, 3, 3, 2, 2, 3, 3, 3, 3, 2], 9, 'before'], 5), ('control layout', [[2, 2, 2, 2, 1, 1, 3], 3, 'before'], 6)], [('regression: leading edge side', [[1, 1, 1, 1, 0, 1, 1, 1], 4, 'after'], 4), ('regression: leading edge side', [[4, 3, 3, 3], 3, 'after'], 1), ('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2], 2, 'after'], 2), ('regression: leading edge side', [[4], 0, 'after'], 0), ('line start before', [[1, 1, 0], 0, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[1, 1, 1, 1, 3, 3, 1, 1], 1, 'before'], 7), ('control layout', [[0, 0, 1, 1, 2, 2, 2, 0, 0], 6, 'before'], 4)], [('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2, 1, 1], 1, 'after'], 3), ('regression: leading edge side', [[2, 2, 2, 2, 1], 4, 'after'], 1), ('regression: leading edge side', [[3, 0, 0, 0, 2, 2, 2], 3, 'after'], 3), ('regression: leading edge side', [[3, 2, 2, 2, 1, 1, 3, 3, 3], 1, 'after'], 6), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('line end', [[1, 1, 1], 3, 'after'], 0), ('control layout', [[1, 1, 1, 1, 3, 3, 3], 4, 'before'], 3), ('control layout', [[0, 0, 0, 1, 3, 3, 3, 3], 6, 'before'], 5)], [('regression: leading edge side', [[3, 3, 3, 3, 1, 1, 1], 6, 'after'], 1), ('regression: leading edge side', [[1, 1], 1, 'after'], 1), ('regression: leading edge side', [[2, 2, 3, 3, 3, 1, 1, 3], 0, 'after'], 3), ('regression: leading edge side', [[4, 4, 4, 4, 2], 1, 'after'], 1), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 2, 2, 3, 3, 2, 2, 2, 3, 3], 10, 'after'], 8), ('control layout', [[1, 1, 1, 1, 1, 0, 2, 2], 4, 'before'], 1)], [('regression: leading edge side', [[2, 2, 2, 4, 4, 4, 4, 3, 3], 4, 'after'], 6), ('regression: leading edge side', [[1, 1, 1, 1, 1], 2, 'after'], 3), ('regression: leading edge side', [[2, 2, 1, 1], 0, 'after'], 2), ('regression: leading edge side', [[3, 3, 2, 2, 2, 2, 2, 2, 1, 1], 7, 'after'], 9), ('line end', [[1, 1, 1], 3, 'after'], 0), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 3, 4, 1, 1, 1, 1, 2], 5, 'before'], 3), ('control layout', [[1, 1, 1, 3, 3, 3, 3, 1, 1, 1], 9, 'before'], 1)]]
for label, args, expected in fixtures[N - 1]:
check(label, solve(args), expected)
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 |
|---|---|---|---|
| same boundary after | 4 | 4 | Passed |
| regression: leading edge side | 3 | 3 | Passed |
| regression: leading edge side | 1 | 0 | Failed |
| regression: leading edge side | 2 | 1 | Failed |
| line end | 0 | 0 | Passed |
| caret at direction boundary | 2 | 2 | Passed |
| control layout | 5 | 5 | Passed |
| control layout | 6 | 6 | Passed |
SHA-256 / 717a454435d2c99d2f43f336c70c6c591bbe94ffdefbe4b87370bf773f57ab4c
3 / The verified repair
Exit 0"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(x):
def reorder(lv):
vis = list(range(len(lv)))
odd = [l for l in lv if l % 2 == 1]
if not odd:
return vis
for lev in range(max(lv), min(odd) - 1, -1):
i = 0
while i < len(vis):
if lv[vis[i]] >= lev:
j = i
while j < len(vis) and lv[vis[j]] >= lev:
j += 1
vis[i:j] = vis[i:j][::-1]
i = j
else:
i += 1
return vis
levels, k, affinity = x
n = len(levels)
if n == 0:
return 0
vis = reorder(levels)
pos = [0] * n
for v, i in enumerate(vis):
pos[i] = v
if affinity == 'after' and k < n or k == 0:
j = k
leading = True
else:
j = k - 1
leading = False
rtl = levels[j] % 2 == 1
if leading:
return pos[j] + 1 if rtl else pos[j]
return pos[j] if rtl else pos[j] + 1
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('regression: leading edge side', [[1, 1, 1, 0], 0, 'after'], 3), ('regression: leading edge side', [[2, 0], 0, 'after'], 0), ('regression: leading edge side', [[2, 2, 2, 2, 2, 3, 3, 2, 1], 0, 'before'], 1), ('line end', [[1, 1, 1], 3, 'after'], 0), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('control layout', [[3, 3, 3, 2, 2, 3, 3, 3, 3, 2], 9, 'before'], 5), ('control layout', [[2, 2, 2, 2, 1, 1, 3], 3, 'before'], 6)], [('regression: leading edge side', [[1, 1, 1, 1, 0, 1, 1, 1], 4, 'after'], 4), ('regression: leading edge side', [[4, 3, 3, 3], 3, 'after'], 1), ('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2], 2, 'after'], 2), ('regression: leading edge side', [[4], 0, 'after'], 0), ('line start before', [[1, 1, 0], 0, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[1, 1, 1, 1, 3, 3, 1, 1], 1, 'before'], 7), ('control layout', [[0, 0, 1, 1, 2, 2, 2, 0, 0], 6, 'before'], 4)], [('regression: leading edge side', [[2, 2, 2, 2, 2, 2, 2, 1, 1], 1, 'after'], 3), ('regression: leading edge side', [[2, 2, 2, 2, 1], 4, 'after'], 1), ('regression: leading edge side', [[3, 0, 0, 0, 2, 2, 2], 3, 'after'], 3), ('regression: leading edge side', [[3, 2, 2, 2, 1, 1, 3, 3, 3], 1, 'after'], 6), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('line end', [[1, 1, 1], 3, 'after'], 0), ('control layout', [[1, 1, 1, 1, 3, 3, 3], 4, 'before'], 3), ('control layout', [[0, 0, 0, 1, 3, 3, 3, 3], 6, 'before'], 5)], [('regression: leading edge side', [[3, 3, 3, 3, 1, 1, 1], 6, 'after'], 1), ('regression: leading edge side', [[1, 1], 1, 'after'], 1), ('regression: leading edge side', [[2, 2, 3, 3, 3, 1, 1, 3], 0, 'after'], 3), ('regression: leading edge side', [[4, 4, 4, 4, 2], 1, 'after'], 1), ('caret at direction boundary', [[0, 0, 1, 1], 2, 'before'], 2), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 2, 2, 3, 3, 2, 2, 2, 3, 3], 10, 'after'], 8), ('control layout', [[1, 1, 1, 1, 1, 0, 2, 2], 4, 'before'], 1)], [('regression: leading edge side', [[2, 2, 2, 4, 4, 4, 4, 3, 3], 4, 'after'], 6), ('regression: leading edge side', [[1, 1, 1, 1, 1], 2, 'after'], 3), ('regression: leading edge side', [[2, 2, 1, 1], 0, 'after'], 2), ('regression: leading edge side', [[3, 3, 2, 2, 2, 2, 2, 2, 1, 1], 7, 'after'], 9), ('line end', [[1, 1, 1], 3, 'after'], 0), ('same boundary after', [[0, 0, 1, 1], 2, 'after'], 4), ('control layout', [[2, 3, 4, 1, 1, 1, 1, 2], 5, 'before'], 3), ('control layout', [[1, 1, 1, 3, 3, 3, 3, 1, 1, 1], 9, 'before'], 1)]]
for label, args, expected in fixtures[N - 1]:
check(label, solve(args), expected)
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 |
|---|---|---|---|
| same boundary after | 4 | 4 | Passed |
| regression: leading edge side | 3 | 3 | Passed |
| regression: leading edge side | 0 | 0 | Passed |
| regression: leading edge side | 1 | 1 | Passed |
| line end | 0 | 0 | Passed |
| caret at direction boundary | 2 | 2 | Passed |
| control layout | 5 | 5 | Passed |
| control layout | 6 | 6 | Passed |
SHA-256 / 401239653847cdf879c99afac810cce5f65e15871779cc598831741cd04192b1
Verification & scope
A deterministic toy bidi model over stipulated class labels and integer levels; it is inspired by, but does not claim conformance to, any published algorithm. 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 2026-09-29T14:49:55.538675+00:00.
Case digest / 9ad4f9ed13b787cef0f33b3ddf10dc48e7163192deabfaa598c5e75036297818