File size: 9,020 Bytes
a9e27c0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
"""Regressions for issue #1730 at the shared Claude conversion boundary."""

import re
import unicodedata
from copy import deepcopy

import pytest

from free_claude_code.core.anthropic.conversion import build_base_request_body
from free_claude_code.core.anthropic.models import MessagesRequest
from free_claude_code.core.openai_responses.provider_input import (
    build_responses_provider_request,
)
from free_claude_code.core.reasoning import ReasoningPolicy
from free_claude_code.core.tool_schema_patterns import translate_tool_schema_patterns

ARTIFACT_PATTERN = r'^(?!__.*__$)[^\p{Cc}\p{Cf}\p{Zl}\p{Zp}"\\./[\]]{1,200}$'


@pytest.mark.parametrize("wire_format", ["chat", "responses"])
@pytest.mark.parametrize(
    "pattern", [r"^[^\p{Cc}\p{Cf}\p{Zl}\p{Zp}]{1,200}$", ARTIFACT_PATTERN]
)
def test_claude_converters_preserve_artifact_rules_in_portable_pattern(
    wire_format, pattern
):
    schema = {
        "type": "object",
        "properties": {"name": {"type": "string", "pattern": pattern}},
        "required": ["name"],
    }
    request = MessagesRequest.model_validate(
        {
            "model": "test-model",
            "messages": [{"role": "user", "content": "Call Artifact with demo"}],
            "tools": [
                {"name": "Artifact", "description": "Save", "input_schema": schema}
            ],
        }
    )
    original = deepcopy(request.model_dump())
    if wire_format == "chat":
        body = build_base_request_body(request)
        parameters = body["tools"][0]["function"]["parameters"]
    else:
        body = build_responses_provider_request(
            request, reasoning=ReasoningPolicy.provider_default()
        )
        parameters = body["tools"][0]["parameters"]
    translated = parameters["properties"]["name"]["pattern"]
    assert r"\p{" not in translated
    regex = re.compile(translated)
    assert regex.fullmatch("demo")
    assert not regex.fullmatch("bad\u2028name")
    assert not regex.fullmatch("bad\U000e0001name")
    assert not regex.fullmatch("a" * 201)
    assert not regex.fullmatch("")
    assert parameters["required"] == ["name"]
    assert request.model_dump() == original


def translated_pattern(pattern):
    return translate_tool_schema_patterns({"pattern": pattern})["pattern"]


@pytest.mark.parametrize("wire_format", ["chat", "responses"])
@pytest.mark.parametrize(
    "pattern",
    [r"^[^\p{Cc}]{4294967296}$", "(" * 500 + r"[^\p{Cc}]" + ")" * 500],
    ids=["repeat-overflow", "group-recursion"],
)
def test_compiler_limits_preserve_original_schema_in_both_converters(
    wire_format, pattern
):
    schema = {"type": "string", "pattern": pattern}
    request = MessagesRequest.model_validate(
        {
            "model": "test-model",
            "messages": [{"role": "user", "content": "Test schema passthrough"}],
            "tools": [{"name": "Artifact", "input_schema": schema}],
        }
    )
    original = deepcopy(request.model_dump())
    if wire_format == "chat":
        parameters = build_base_request_body(request)["tools"][0]["function"][
            "parameters"
        ]
    else:
        body = build_responses_provider_request(
            request, reasoning=ReasoningPolicy.provider_default()
        )
        parameters = body["tools"][0]["parameters"]
    assert parameters == schema
    assert request.model_dump() == original


@pytest.mark.parametrize("category", ["Cc", "Cf", "Zl", "Zp"])
def test_category_expansion_preserves_every_unicode_scalar(category):
    regex = re.compile(translated_pattern(r"[^\p{" + category + "}]+"))
    for point in range(0x110000):
        if 0xD800 <= point <= 0xDFFF:
            continue
        char = chr(point)
        assert bool(regex.fullmatch(char)) == (
            unicodedata.category(char) != category
        ), f"U+{point:04X}, category {category}, Unicode {unicodedata.unidata_version}"


@pytest.mark.parametrize(
    "name,accepted",
    [
        ("demo", True),
        ("é名字😀", True),
        ("__name", True),
        ("name__", True),
        ("-name-", True),
        ("a", True),
        ("a" * 200, True),
        ("", False),
        ("a" * 201, False),
        ("__name__", False),
        *[("x" + char + "y", False) for char in r'"\/.[]'],
        *[
            ("x" + char + "y", False)
            for char in [
                "\x00",
                "\x1f",
                "\x7f",
                "\x9f",
                "\u200d",
                "\u2028",
                "\u2029",
                "\U000110bd",
                "\U000e0001",
                "\U000e0020",
                "\U000e007f",
            ]
        ],
    ],
)
def test_artifact_restrictions_survive_translation(name, accepted):
    assert bool(re.fullmatch(translated_pattern(ARTIFACT_PATTERN), name)) == accepted


@pytest.mark.parametrize(
    "pattern",
    [
        r"^[a-z]+$",
        r"[^\\p{Cc}]",
        r"\\p{Cc}",
        r"[\p{Cc}]",
        r"\p{Cc}",
        r"[^\P{Cc}]",
        r"[^\p{L}]",
        r"[^\p{cc}]",
        r"[^\p{Cc}\p{L}]",
        r"[^\p{Cc}]\P{Cf}",
        r"[^\p{Cc}]\p{Cf}",
        r"[^\p{Cc}\p{General_Category=Format}]",
        r"[^\p{Cc}\p{Cf]",
        r"[^\p{Cc}",
        r"[^\p{Cc}]" + "\\",
        r"([^\p{Cc}]",
        r"[^\p{Cc}-z]",
        r"[^a-\p{Cc}]",
        r"[^\p{Cc}--a]",
        r"[^\p{Cc}&&a]",
        r"[^\p{Cc}~~a]",
        r"[^\p{Cc}||a]",
        r"[^\p{Cc}[a]]",
        r"\[^\p{Cc}]",
    ],
)
def test_unsupported_or_literal_patterns_remain_unchanged(pattern):
    schema = {"type": "string", "pattern": pattern}
    assert translate_tool_schema_patterns(schema) is schema


@pytest.mark.parametrize(
    "pattern,excluded,accepted",
    [
        (r"[^\\\p{Cc}]", "\\\x00", "p{}C"),
        (r"[^\p{Cc}\-]", "-\x00", "az"),
        (r"[^\]\p{Cc}]", "]\x00", "["),
        (r"[^\p{Cc}[\]]", "[]\x00", "az"),
        (r"[^\p{Zl}][^\p{Zp}]", "\u2028x", "ab"),
    ],
)
def test_class_escaping_is_preserved(pattern, excluded, accepted):
    translated = translated_pattern(pattern)
    assert translated != pattern
    regex = re.compile(translated)
    if pattern == r"[^\p{Zl}][^\p{Zp}]":
        assert regex.fullmatch(accepted)
        assert not regex.fullmatch(excluded)
    else:
        assert all(not regex.fullmatch(char) for char in excluded)
        assert all(regex.fullmatch(char) for char in accepted)


@pytest.mark.parametrize(
    "container",
    [
        "$defs",
        "definitions",
        "properties",
        "patternProperties",
        "dependentSchemas",
        "dependencies",
    ],
)
def test_schema_maps_visit_values_and_preserve_names(container):
    child = {"type": "string", "pattern": r"[^\p{Cc}]"}
    schema = {container: {"pattern": child, r"\p{Cc}": True, "required": ["name"]}}
    original = deepcopy(schema)
    converted = translate_tool_schema_patterns(schema)
    assert converted[container]["pattern"]["pattern"] != child["pattern"]
    assert converted[container][r"\p{Cc}"] is True
    assert converted[container]["required"] == ["name"]
    assert schema == original


@pytest.mark.parametrize(
    "keyword",
    [
        "additionalProperties",
        "additionalItems",
        "unevaluatedProperties",
        "unevaluatedItems",
        "items",
        "contains",
        "propertyNames",
        "if",
        "then",
        "else",
        "not",
        "contentSchema",
    ],
)
def test_single_schema_keywords_are_translated(keyword):
    schema = {keyword: {"pattern": r"[^\p{Cf}]"}}
    converted = translate_tool_schema_patterns(schema)
    assert converted[keyword]["pattern"] != schema[keyword]["pattern"]


@pytest.mark.parametrize("keyword", ["allOf", "anyOf", "oneOf", "prefixItems", "items"])
def test_schema_arrays_preserve_boolean_and_unchanged_children(keyword):
    child = {"type": "string", "pattern": r"[a-z]"}
    schema = {keyword: [child, False, {"pattern": r"[^\p{Cf}]"}]}
    converted = translate_tool_schema_patterns(schema)
    assert converted[keyword][0] is child
    assert converted[keyword][1] is False
    assert converted[keyword][2]["pattern"] != r"[^\p{Cf}]"


def test_instance_data_is_untouched_and_translation_is_idempotent():
    literal = {"pattern": r"[^\p{Cc}]"}
    schema = {
        "type": "object",
        "properties": {"name": literal},
        "default": literal,
        "const": literal,
        "enum": [literal],
        "examples": [literal],
        "x-custom": literal,
        "description": r"\p{Cc}",
    }
    before = deepcopy(schema)
    converted = translate_tool_schema_patterns(schema)
    for key in ("default", "const", "enum", "examples", "x-custom", "description"):
        assert converted[key] is schema[key]
    assert schema == before
    assert translate_tool_schema_patterns(converted) is converted


@pytest.mark.parametrize(
    "schema", [True, False, {}, {"pattern": 123}, {"type": "string"}]
)
def test_non_patterns_remain_unchanged(schema):
    assert translate_tool_schema_patterns(schema) is schema