@ai-sdk/anthropic 3.0.136 → 3.0.137
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +8 -0
- package/dist/index.js +282 -46
- package/dist/index.js.map +1 -1
- package/dist/index.mjs +282 -46
- package/dist/index.mjs.map +1 -1
- package/dist/internal/index.js +281 -45
- package/dist/internal/index.js.map +1 -1
- package/dist/internal/index.mjs +281 -45
- package/dist/internal/index.mjs.map +1 -1
- package/docs/05-anthropic.mdx +12 -1
- package/package.json +1 -1
- package/src/anthropic-messages-language-model.ts +119 -22
- package/src/has-circular-schema-reference.ts +118 -0
- package/src/sanitize-json-schema.ts +27 -24
- package/src/supports-native-structured-output-schema.ts +138 -0
package/docs/05-anthropic.mdx
CHANGED
|
@@ -194,10 +194,21 @@ default `"auto"` mode uses native structured outputs through
|
|
|
194
194
|
`output_config.format`. These models reject forced tool use, so the
|
|
195
195
|
`"jsonTool"` mode cannot be used with them. If you
|
|
196
196
|
request `"jsonTool"` anyway, the provider falls back to `"outputFormat"` and
|
|
197
|
-
emits a warning.
|
|
197
|
+
emits a warning. The same fallback applies when these models are served through
|
|
198
|
+
providers that otherwise default to the JSON tool, such as the Amazon Bedrock
|
|
199
|
+
Anthropic and Google Vertex Anthropic providers. On those providers, schemas
|
|
200
|
+
that native structured outputs cannot represent keep the JSON tool:
|
|
201
|
+
self-referencing schemas, schemas with more than 10 optional properties, and
|
|
202
|
+
schemas with open objects such as maps (`additionalProperties` set to a schema
|
|
203
|
+
or `true`), which native structured outputs would return as `{}`.
|
|
204
|
+
Likewise, a `required` or named `toolChoice` is downgraded to
|
|
198
205
|
`auto` with a warning; instruct the model to use the tool in your prompt and
|
|
199
206
|
verify that a tool call was made.
|
|
200
207
|
|
|
208
|
+
When the JSON tool cannot be forced (for example with manual thinking) and the
|
|
209
|
+
model answers in text without calling it, the text of its final answer is
|
|
210
|
+
returned. Text before a tool call is still dropped.
|
|
211
|
+
|
|
201
212
|
### Effort
|
|
202
213
|
|
|
203
214
|
Anthropic introduced an `effort` option with `claude-opus-4-5` that affects thinking, text responses, and function calls. Effort defaults to `high` and you can set it to `medium` or `low` to save tokens and to lower time-to-last-token latency (TTLT). `claude-opus-4-7`, `claude-opus-4-8`, `claude-opus-5`, `claude-opus-5-5`, `claude-fable-5`, `claude-fable-5-1`, `claude-sonnet-5`, `claude-sonnet-5-5`, and `claude-haiku-5-5` additionally support `xhigh` for maximum reasoning effort.
|
package/package.json
CHANGED
|
@@ -63,6 +63,7 @@ import {
|
|
|
63
63
|
import { CacheControlValidator } from './get-cache-control';
|
|
64
64
|
import { mapAnthropicStopReason } from './map-anthropic-stop-reason';
|
|
65
65
|
import { sanitizeJsonSchema } from './sanitize-json-schema';
|
|
66
|
+
import { supportsNativeStructuredOutputSchema } from './supports-native-structured-output-schema';
|
|
66
67
|
|
|
67
68
|
function createCitationSource(
|
|
68
69
|
citation: Citation,
|
|
@@ -216,6 +217,47 @@ type AnthropicMessagesConfig = {
|
|
|
216
217
|
supportsStrictTools?: boolean;
|
|
217
218
|
};
|
|
218
219
|
|
|
220
|
+
const TEXT_OR_THINKING_BLOCK_TYPES = new Set<string>([
|
|
221
|
+
'redacted_thinking',
|
|
222
|
+
'text',
|
|
223
|
+
'thinking',
|
|
224
|
+
]);
|
|
225
|
+
|
|
226
|
+
function isContinuationStopReason(
|
|
227
|
+
stopReason: string | null | undefined,
|
|
228
|
+
): boolean {
|
|
229
|
+
return stopReason === 'pause_turn' || stopReason === 'tool_use';
|
|
230
|
+
}
|
|
231
|
+
|
|
232
|
+
/**
|
|
233
|
+
* Text parts to return when the JSON response tool is in use: the text after
|
|
234
|
+
* the last tool call or result, and only if the tool was never called.
|
|
235
|
+
*/
|
|
236
|
+
function getJsonResponseToolFallbackText<
|
|
237
|
+
PART extends { type: string; name?: unknown },
|
|
238
|
+
>(
|
|
239
|
+
content: ReadonlyArray<PART>,
|
|
240
|
+
stopReason: string | null | undefined,
|
|
241
|
+
): Set<PART> {
|
|
242
|
+
const text = new Set<PART>();
|
|
243
|
+
if (isContinuationStopReason(stopReason)) {
|
|
244
|
+
return text;
|
|
245
|
+
}
|
|
246
|
+
|
|
247
|
+
for (const part of content) {
|
|
248
|
+
if (part.type === 'tool_use' && part.name === 'json') {
|
|
249
|
+
return new Set();
|
|
250
|
+
}
|
|
251
|
+
if (part.type === 'text') {
|
|
252
|
+
text.add(part);
|
|
253
|
+
} else if (!TEXT_OR_THINKING_BLOCK_TYPES.has(part.type)) {
|
|
254
|
+
text.clear();
|
|
255
|
+
}
|
|
256
|
+
}
|
|
257
|
+
|
|
258
|
+
return text;
|
|
259
|
+
}
|
|
260
|
+
|
|
219
261
|
export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
220
262
|
readonly specificationVersion = 'v3';
|
|
221
263
|
|
|
@@ -419,13 +461,15 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
419
461
|
|
|
420
462
|
// The JSON response tool relies on forced tool use, which some models
|
|
421
463
|
// reject (e.g. Opus 5.5, Fable 5.1). Fall back to native structured
|
|
422
|
-
// outputs when the
|
|
464
|
+
// outputs even when the provider disables them by default (Bedrock,
|
|
465
|
+
// Vertex), unless they would reject the schema or lose data from it.
|
|
423
466
|
if (
|
|
424
467
|
!useStructuredOutput &&
|
|
425
468
|
rejectsForcedToolUse &&
|
|
426
|
-
|
|
469
|
+
modelSupportsStructuredOutput &&
|
|
427
470
|
responseFormat?.type === 'json' &&
|
|
428
|
-
responseFormat.schema != null
|
|
471
|
+
responseFormat.schema != null &&
|
|
472
|
+
supportsNativeStructuredOutputSchema(responseFormat.schema)
|
|
429
473
|
) {
|
|
430
474
|
warnings.push({
|
|
431
475
|
type: 'unsupported',
|
|
@@ -1146,11 +1190,20 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
1146
1190
|
const serverToolCalls: Record<string, string> = {}; // tool_use_id -> provider tool name
|
|
1147
1191
|
let isJsonResponseFromTool = false;
|
|
1148
1192
|
|
|
1193
|
+
// The JSON response tool call becomes the text, so other text is dropped,
|
|
1194
|
+
// unless the model never called the tool and answered in text instead.
|
|
1195
|
+
const jsonResponseToolFallbackText = usesJsonResponseTool
|
|
1196
|
+
? getJsonResponseToolFallbackText(response.content, response.stop_reason)
|
|
1197
|
+
: undefined;
|
|
1198
|
+
|
|
1149
1199
|
// map response content to content array
|
|
1150
1200
|
for (const part of response.content) {
|
|
1151
1201
|
switch (part.type) {
|
|
1152
1202
|
case 'text': {
|
|
1153
|
-
if (
|
|
1203
|
+
if (
|
|
1204
|
+
jsonResponseToolFallbackText == null ||
|
|
1205
|
+
jsonResponseToolFallbackText.has(part)
|
|
1206
|
+
) {
|
|
1154
1207
|
const webSearchCitations = part.citations?.filter(
|
|
1155
1208
|
citation => citation.type === 'web_search_result_location',
|
|
1156
1209
|
);
|
|
@@ -1793,6 +1846,21 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
1793
1846
|
let safeguardResults: AnthropicMessageMetadata['safeguardResults'];
|
|
1794
1847
|
let container: AnthropicMessageMetadata['container'] | null = null;
|
|
1795
1848
|
let isJsonResponseFromTool = false;
|
|
1849
|
+
// Text held back while the JSON response tool is in use: it is only the
|
|
1850
|
+
// answer if no tool block follows it and the model never calls the tool.
|
|
1851
|
+
type HeldBackText = { id: string; text: string; citations: Citation[] };
|
|
1852
|
+
const heldBackText: HeldBackText[] = [];
|
|
1853
|
+
const heldBackTextByIndex = new Map<number, HeldBackText>();
|
|
1854
|
+
const createTextEnd = (
|
|
1855
|
+
id: string,
|
|
1856
|
+
citations: Citation[],
|
|
1857
|
+
): LanguageModelV3StreamPart => ({
|
|
1858
|
+
type: 'text-end',
|
|
1859
|
+
id,
|
|
1860
|
+
...(citations.length > 0 && {
|
|
1861
|
+
providerMetadata: { anthropic: { citations } },
|
|
1862
|
+
}),
|
|
1863
|
+
});
|
|
1796
1864
|
let isMessageOpen = false;
|
|
1797
1865
|
let activeMessageId: string | null | undefined;
|
|
1798
1866
|
let hasInvalidMessageSequence = false;
|
|
@@ -1860,11 +1928,23 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
1860
1928
|
|
|
1861
1929
|
blockType = contentBlockType;
|
|
1862
1930
|
|
|
1931
|
+
// text before a tool call or result is preamble, not the answer:
|
|
1932
|
+
if (!TEXT_OR_THINKING_BLOCK_TYPES.has(contentBlockType)) {
|
|
1933
|
+
heldBackText.length = 0;
|
|
1934
|
+
}
|
|
1935
|
+
|
|
1863
1936
|
switch (contentBlockType) {
|
|
1864
1937
|
case 'text': {
|
|
1865
1938
|
// when a json response tool is used, the tool call is returned as text,
|
|
1866
|
-
// so we
|
|
1939
|
+
// so the text content is held back until we know whether it was called:
|
|
1867
1940
|
if (usesJsonResponseTool) {
|
|
1941
|
+
const entry = {
|
|
1942
|
+
id: String(value.index),
|
|
1943
|
+
text: '',
|
|
1944
|
+
citations: [],
|
|
1945
|
+
};
|
|
1946
|
+
heldBackText.push(entry);
|
|
1947
|
+
heldBackTextByIndex.set(value.index, entry);
|
|
1868
1948
|
return;
|
|
1869
1949
|
}
|
|
1870
1950
|
|
|
@@ -2414,17 +2494,12 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
2414
2494
|
|
|
2415
2495
|
switch (contentBlock.type) {
|
|
2416
2496
|
case 'text': {
|
|
2417
|
-
controller.enqueue(
|
|
2418
|
-
|
|
2419
|
-
|
|
2420
|
-
|
|
2421
|
-
|
|
2422
|
-
|
|
2423
|
-
citations: contentBlock.citations,
|
|
2424
|
-
},
|
|
2425
|
-
},
|
|
2426
|
-
}),
|
|
2427
|
-
});
|
|
2497
|
+
controller.enqueue(
|
|
2498
|
+
createTextEnd(
|
|
2499
|
+
String(value.index),
|
|
2500
|
+
contentBlock.citations,
|
|
2501
|
+
),
|
|
2502
|
+
);
|
|
2428
2503
|
break;
|
|
2429
2504
|
}
|
|
2430
2505
|
|
|
@@ -2547,8 +2622,12 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
2547
2622
|
switch (deltaType) {
|
|
2548
2623
|
case 'text_delta': {
|
|
2549
2624
|
// when a json response tool is used, the tool call is returned as text,
|
|
2550
|
-
// so we
|
|
2625
|
+
// so the text content is held back until we know whether it was called:
|
|
2551
2626
|
if (usesJsonResponseTool) {
|
|
2627
|
+
const entry = heldBackTextByIndex.get(value.index);
|
|
2628
|
+
if (entry != null) {
|
|
2629
|
+
entry.text += value.delta.text;
|
|
2630
|
+
}
|
|
2552
2631
|
return; // excluding the text-start will also exclude the text-end
|
|
2553
2632
|
}
|
|
2554
2633
|
|
|
@@ -2659,11 +2738,13 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
2659
2738
|
const citation = value.delta.citation;
|
|
2660
2739
|
const contentBlock = contentBlocks[value.index];
|
|
2661
2740
|
|
|
2662
|
-
if (
|
|
2663
|
-
contentBlock?.type === 'text'
|
|
2664
|
-
|
|
2665
|
-
|
|
2666
|
-
|
|
2741
|
+
if (citation.type === 'web_search_result_location') {
|
|
2742
|
+
if (contentBlock?.type === 'text') {
|
|
2743
|
+
contentBlock.citations.push(citation);
|
|
2744
|
+
}
|
|
2745
|
+
heldBackTextByIndex
|
|
2746
|
+
.get(value.index)
|
|
2747
|
+
?.citations.push(citation);
|
|
2667
2748
|
}
|
|
2668
2749
|
|
|
2669
2750
|
const source = createCitationSource(
|
|
@@ -2709,6 +2790,7 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
2709
2790
|
|
|
2710
2791
|
isMessageOpen = true;
|
|
2711
2792
|
activeMessageId = value.message.id;
|
|
2793
|
+
heldBackTextByIndex.clear();
|
|
2712
2794
|
|
|
2713
2795
|
usage.input_tokens = value.message.usage.input_tokens;
|
|
2714
2796
|
usage.cache_read_input_tokens =
|
|
@@ -2926,6 +3008,21 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV3 {
|
|
|
2926
3008
|
providerMetadata[providerOptionsName] = anthropicMetadata;
|
|
2927
3009
|
}
|
|
2928
3010
|
|
|
3011
|
+
// a message that hands off to a tool or continuation is not the
|
|
3012
|
+
// final answer, so its held-back text is discarded:
|
|
3013
|
+
if (
|
|
3014
|
+
!isJsonResponseFromTool &&
|
|
3015
|
+
!isContinuationStopReason(finishReason.raw)
|
|
3016
|
+
) {
|
|
3017
|
+
for (const { id, text, citations } of heldBackText) {
|
|
3018
|
+
if (text.length === 0) continue;
|
|
3019
|
+
controller.enqueue({ type: 'text-start', id });
|
|
3020
|
+
controller.enqueue({ type: 'text-delta', id, delta: text });
|
|
3021
|
+
controller.enqueue(createTextEnd(id, citations));
|
|
3022
|
+
}
|
|
3023
|
+
}
|
|
3024
|
+
heldBackText.length = 0;
|
|
3025
|
+
|
|
2929
3026
|
controller.enqueue({
|
|
2930
3027
|
type: 'finish',
|
|
2931
3028
|
finishReason,
|
|
@@ -0,0 +1,118 @@
|
|
|
1
|
+
import type { JSONSchema7 } from '@ai-sdk/provider';
|
|
2
|
+
|
|
3
|
+
// Keywords whose values are maps of name -> schema rather than a schema.
|
|
4
|
+
const SCHEMA_MAP_KEYWORDS = new Set([
|
|
5
|
+
'dependentSchemas',
|
|
6
|
+
'patternProperties',
|
|
7
|
+
'properties',
|
|
8
|
+
]);
|
|
9
|
+
|
|
10
|
+
// Keywords holding instance data, which can contain a literal `$ref` key.
|
|
11
|
+
const DATA_KEYWORDS = new Set(['const', 'default', 'enum', 'examples']);
|
|
12
|
+
|
|
13
|
+
/**
|
|
14
|
+
* Whether a JSON schema contains a cycle of local `$ref`s, e.g. a tree node
|
|
15
|
+
* whose children reference the node itself.
|
|
16
|
+
*
|
|
17
|
+
* Anthropic's native structured outputs reject such schemas, while the JSON
|
|
18
|
+
* response tool accepts them.
|
|
19
|
+
*/
|
|
20
|
+
export function hasCircularSchemaReference(schema: JSONSchema7): boolean {
|
|
21
|
+
const visiting = new Set<string>();
|
|
22
|
+
const visited = new Set<string>();
|
|
23
|
+
|
|
24
|
+
const visit = (pointer: string): boolean => {
|
|
25
|
+
if (visiting.has(pointer)) {
|
|
26
|
+
return true;
|
|
27
|
+
}
|
|
28
|
+
if (visited.has(pointer)) {
|
|
29
|
+
return false;
|
|
30
|
+
}
|
|
31
|
+
|
|
32
|
+
const target = resolveSchemaPointer(schema, pointer);
|
|
33
|
+
if (target === undefined) {
|
|
34
|
+
return false;
|
|
35
|
+
}
|
|
36
|
+
|
|
37
|
+
visiting.add(pointer);
|
|
38
|
+
const isCircular = collectLocalRefs(target).some(visit);
|
|
39
|
+
visiting.delete(pointer);
|
|
40
|
+
visited.add(pointer);
|
|
41
|
+
|
|
42
|
+
return isCircular;
|
|
43
|
+
};
|
|
44
|
+
|
|
45
|
+
return visit('#');
|
|
46
|
+
}
|
|
47
|
+
|
|
48
|
+
function collectLocalRefs(node: unknown, refs: string[] = []): string[] {
|
|
49
|
+
if (Array.isArray(node)) {
|
|
50
|
+
for (const item of node) {
|
|
51
|
+
collectLocalRefs(item, refs);
|
|
52
|
+
}
|
|
53
|
+
return refs;
|
|
54
|
+
}
|
|
55
|
+
|
|
56
|
+
if (!isRecord(node)) {
|
|
57
|
+
return refs;
|
|
58
|
+
}
|
|
59
|
+
|
|
60
|
+
for (const [key, value] of Object.entries(node)) {
|
|
61
|
+
if (key === '$ref') {
|
|
62
|
+
if (typeof value === 'string' && value.startsWith('#')) {
|
|
63
|
+
refs.push(value);
|
|
64
|
+
}
|
|
65
|
+
} else if (SCHEMA_MAP_KEYWORDS.has(key) && isRecord(value)) {
|
|
66
|
+
for (const subschema of Object.values(value)) {
|
|
67
|
+
collectLocalRefs(subschema, refs);
|
|
68
|
+
}
|
|
69
|
+
} else if (
|
|
70
|
+
// definitions are only reached through a `$ref`:
|
|
71
|
+
key !== '$defs' &&
|
|
72
|
+
key !== 'definitions' &&
|
|
73
|
+
!DATA_KEYWORDS.has(key)
|
|
74
|
+
) {
|
|
75
|
+
collectLocalRefs(value, refs);
|
|
76
|
+
}
|
|
77
|
+
}
|
|
78
|
+
|
|
79
|
+
return refs;
|
|
80
|
+
}
|
|
81
|
+
|
|
82
|
+
export function resolveSchemaPointer(
|
|
83
|
+
schema: JSONSchema7,
|
|
84
|
+
pointer: string,
|
|
85
|
+
): unknown {
|
|
86
|
+
if (pointer === '#') {
|
|
87
|
+
return schema;
|
|
88
|
+
}
|
|
89
|
+
if (!pointer.startsWith('#/')) {
|
|
90
|
+
return undefined;
|
|
91
|
+
}
|
|
92
|
+
|
|
93
|
+
let decodedPointer: string;
|
|
94
|
+
try {
|
|
95
|
+
decodedPointer = decodeURIComponent(pointer.slice(2));
|
|
96
|
+
} catch {
|
|
97
|
+
return undefined;
|
|
98
|
+
}
|
|
99
|
+
|
|
100
|
+
let current: unknown = schema;
|
|
101
|
+
for (const rawSegment of decodedPointer.split('/')) {
|
|
102
|
+
const segment = rawSegment.replace(/~1/g, '/').replace(/~0/g, '~');
|
|
103
|
+
|
|
104
|
+
if (Array.isArray(current)) {
|
|
105
|
+
current = current[Number(segment)];
|
|
106
|
+
} else if (isRecord(current)) {
|
|
107
|
+
current = current[segment];
|
|
108
|
+
} else {
|
|
109
|
+
return undefined;
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
|
|
113
|
+
return current;
|
|
114
|
+
}
|
|
115
|
+
|
|
116
|
+
export function isRecord(value: unknown): value is Record<string, unknown> {
|
|
117
|
+
return typeof value === 'object' && value !== null && !Array.isArray(value);
|
|
118
|
+
}
|
|
@@ -53,12 +53,11 @@ function sanitizeDefinition(
|
|
|
53
53
|
|
|
54
54
|
function sanitizeSchema(schema: JSONSchema7): JSONSchema7 {
|
|
55
55
|
const result: JSONSchema7 = {};
|
|
56
|
-
const schemaWithDefs = schema as JSONSchema7 & {
|
|
57
|
-
$defs?: Record<string, JSONSchema7Definition>;
|
|
58
|
-
};
|
|
59
56
|
|
|
57
|
+
// Keep the definitions next to a `$ref` so a top-level reference still
|
|
58
|
+
// resolves (e.g. `{ $ref: '#/definitions/Answer', definitions: {...} }`).
|
|
60
59
|
if (schema.$ref != null) {
|
|
61
|
-
return { $ref: schema.$ref };
|
|
60
|
+
return { $ref: schema.$ref, ...sanitizeSchemaDefinitions(schema) };
|
|
62
61
|
}
|
|
63
62
|
|
|
64
63
|
if (schema.$schema != null) {
|
|
@@ -103,26 +102,7 @@ function sanitizeSchema(schema: JSONSchema7): JSONSchema7 {
|
|
|
103
102
|
result.allOf = schema.allOf.map(sanitizeDefinition);
|
|
104
103
|
}
|
|
105
104
|
|
|
106
|
-
|
|
107
|
-
result.definitions = Object.fromEntries(
|
|
108
|
-
Object.entries(schema.definitions).map(([name, definition]) => [
|
|
109
|
-
name,
|
|
110
|
-
sanitizeDefinition(definition),
|
|
111
|
-
]),
|
|
112
|
-
);
|
|
113
|
-
}
|
|
114
|
-
|
|
115
|
-
if (schemaWithDefs.$defs != null) {
|
|
116
|
-
const resultWithDefs = result as JSONSchema7 & {
|
|
117
|
-
$defs?: Record<string, JSONSchema7Definition>;
|
|
118
|
-
};
|
|
119
|
-
resultWithDefs.$defs = Object.fromEntries(
|
|
120
|
-
Object.entries(schemaWithDefs.$defs).map(([name, definition]) => [
|
|
121
|
-
name,
|
|
122
|
-
sanitizeDefinition(definition),
|
|
123
|
-
]),
|
|
124
|
-
);
|
|
125
|
-
}
|
|
105
|
+
Object.assign(result, sanitizeSchemaDefinitions(schema));
|
|
126
106
|
|
|
127
107
|
if (schema.type === 'object' || schema.properties != null) {
|
|
128
108
|
if (schema.properties != null) {
|
|
@@ -165,6 +145,29 @@ function sanitizeSchema(schema: JSONSchema7): JSONSchema7 {
|
|
|
165
145
|
return result;
|
|
166
146
|
}
|
|
167
147
|
|
|
148
|
+
function sanitizeSchemaDefinitions(schema: JSONSchema7): Pick<
|
|
149
|
+
JSONSchema7,
|
|
150
|
+
'definitions'
|
|
151
|
+
> & {
|
|
152
|
+
$defs?: Record<string, JSONSchema7Definition>;
|
|
153
|
+
} {
|
|
154
|
+
const { $defs } = schema as { $defs?: Record<string, JSONSchema7Definition> };
|
|
155
|
+
const sanitizeAll = (definitions: Record<string, JSONSchema7Definition>) =>
|
|
156
|
+
Object.fromEntries(
|
|
157
|
+
Object.entries(definitions).map(([name, definition]) => [
|
|
158
|
+
name,
|
|
159
|
+
sanitizeDefinition(definition),
|
|
160
|
+
]),
|
|
161
|
+
);
|
|
162
|
+
|
|
163
|
+
return {
|
|
164
|
+
...(schema.definitions != null && {
|
|
165
|
+
definitions: sanitizeAll(schema.definitions),
|
|
166
|
+
}),
|
|
167
|
+
...($defs != null && { $defs: sanitizeAll($defs) }),
|
|
168
|
+
};
|
|
169
|
+
}
|
|
170
|
+
|
|
168
171
|
function getConstraintDescription(schema: JSONSchema7): string | undefined {
|
|
169
172
|
const descriptions = DESCRIPTION_CONSTRAINT_KEYS.flatMap(key => {
|
|
170
173
|
const value = schema[key];
|
|
@@ -0,0 +1,138 @@
|
|
|
1
|
+
import type { JSONSchema7 } from '@ai-sdk/provider';
|
|
2
|
+
import {
|
|
3
|
+
hasCircularSchemaReference,
|
|
4
|
+
isRecord,
|
|
5
|
+
resolveSchemaPointer,
|
|
6
|
+
} from './has-circular-schema-reference';
|
|
7
|
+
|
|
8
|
+
// Anthropic rejects schemas with more than 24 optional properties in total,
|
|
9
|
+
// and grammar compilation fails earlier as optional properties add up: from 13
|
|
10
|
+
// on Bedrock and the Anthropic API, from 11 on Vertex (compilation timeout).
|
|
11
|
+
const MAX_OPTIONAL_PROPERTIES = 10;
|
|
12
|
+
|
|
13
|
+
const SUBSCHEMA_KEYWORDS = [
|
|
14
|
+
'additionalProperties',
|
|
15
|
+
'contains',
|
|
16
|
+
'else',
|
|
17
|
+
'if',
|
|
18
|
+
'items',
|
|
19
|
+
'not',
|
|
20
|
+
'then',
|
|
21
|
+
];
|
|
22
|
+
const SUBSCHEMA_LIST_KEYWORDS = ['allOf', 'anyOf', 'oneOf', 'prefixItems'];
|
|
23
|
+
const SUBSCHEMA_MAP_KEYWORDS = [
|
|
24
|
+
'dependentSchemas',
|
|
25
|
+
'patternProperties',
|
|
26
|
+
'properties',
|
|
27
|
+
];
|
|
28
|
+
|
|
29
|
+
type SchemaStats = { optionalProperties: number; hasOpenObject: boolean };
|
|
30
|
+
|
|
31
|
+
/**
|
|
32
|
+
* Whether Anthropic's native structured outputs (`output_config.format`) can
|
|
33
|
+
* represent a schema without rejecting it or losing data.
|
|
34
|
+
*
|
|
35
|
+
* Rejected: self-referencing schemas, and schemas with too many optional
|
|
36
|
+
* properties. Lossy: open objects (maps, free-form objects), which native
|
|
37
|
+
* structured outputs close with `additionalProperties: false`, so the model
|
|
38
|
+
* can only return `{}` for them.
|
|
39
|
+
*/
|
|
40
|
+
export function supportsNativeStructuredOutputSchema(
|
|
41
|
+
schema: JSONSchema7,
|
|
42
|
+
): boolean {
|
|
43
|
+
if (hasCircularSchemaReference(schema)) {
|
|
44
|
+
return false;
|
|
45
|
+
}
|
|
46
|
+
|
|
47
|
+
const { optionalProperties, hasOpenObject } = getSchemaStats(
|
|
48
|
+
schema,
|
|
49
|
+
schema,
|
|
50
|
+
new Map(),
|
|
51
|
+
);
|
|
52
|
+
|
|
53
|
+
return !hasOpenObject && optionalProperties <= MAX_OPTIONAL_PROPERTIES;
|
|
54
|
+
}
|
|
55
|
+
|
|
56
|
+
function getSchemaStats(
|
|
57
|
+
node: unknown,
|
|
58
|
+
root: JSONSchema7,
|
|
59
|
+
refStats: Map<string, SchemaStats>,
|
|
60
|
+
): SchemaStats {
|
|
61
|
+
const stats: SchemaStats = { optionalProperties: 0, hasOpenObject: false };
|
|
62
|
+
if (!isRecord(node)) {
|
|
63
|
+
return stats;
|
|
64
|
+
}
|
|
65
|
+
|
|
66
|
+
const add = (child: unknown) => {
|
|
67
|
+
const childStats = getSchemaStats(child, root, refStats);
|
|
68
|
+
stats.optionalProperties += childStats.optionalProperties;
|
|
69
|
+
stats.hasOpenObject ||= childStats.hasOpenObject;
|
|
70
|
+
};
|
|
71
|
+
|
|
72
|
+
// references are counted at every use; the schema has no cycles here:
|
|
73
|
+
if (typeof node.$ref === 'string' && node.$ref.startsWith('#')) {
|
|
74
|
+
let resolved = refStats.get(node.$ref);
|
|
75
|
+
if (resolved == null) {
|
|
76
|
+
resolved = getSchemaStats(
|
|
77
|
+
resolveSchemaPointer(root, node.$ref),
|
|
78
|
+
root,
|
|
79
|
+
refStats,
|
|
80
|
+
);
|
|
81
|
+
refStats.set(node.$ref, resolved);
|
|
82
|
+
}
|
|
83
|
+
stats.optionalProperties += resolved.optionalProperties;
|
|
84
|
+
stats.hasOpenObject ||= resolved.hasOpenObject;
|
|
85
|
+
}
|
|
86
|
+
|
|
87
|
+
if (isOpenObject(node)) {
|
|
88
|
+
stats.hasOpenObject = true;
|
|
89
|
+
}
|
|
90
|
+
|
|
91
|
+
if (isRecord(node.properties)) {
|
|
92
|
+
const required = new Set(Array.isArray(node.required) ? node.required : []);
|
|
93
|
+
stats.optionalProperties += Object.keys(node.properties).filter(
|
|
94
|
+
name => !required.has(name),
|
|
95
|
+
).length;
|
|
96
|
+
}
|
|
97
|
+
|
|
98
|
+
for (const keyword of SUBSCHEMA_KEYWORDS) {
|
|
99
|
+
const value = node[keyword];
|
|
100
|
+
if (Array.isArray(value)) {
|
|
101
|
+
value.forEach(add);
|
|
102
|
+
} else {
|
|
103
|
+
add(value);
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
for (const keyword of SUBSCHEMA_LIST_KEYWORDS) {
|
|
107
|
+
const value = node[keyword];
|
|
108
|
+
if (Array.isArray(value)) {
|
|
109
|
+
value.forEach(add);
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
for (const keyword of SUBSCHEMA_MAP_KEYWORDS) {
|
|
113
|
+
const value = node[keyword];
|
|
114
|
+
if (isRecord(value)) {
|
|
115
|
+
Object.values(value).forEach(add);
|
|
116
|
+
}
|
|
117
|
+
}
|
|
118
|
+
|
|
119
|
+
return stats;
|
|
120
|
+
}
|
|
121
|
+
|
|
122
|
+
function isOpenObject(node: Record<string, unknown>): boolean {
|
|
123
|
+
const { additionalProperties, patternProperties, properties, type } = node;
|
|
124
|
+
|
|
125
|
+
if (additionalProperties === true || isRecord(additionalProperties)) {
|
|
126
|
+
return true;
|
|
127
|
+
}
|
|
128
|
+
if (
|
|
129
|
+
isRecord(patternProperties) &&
|
|
130
|
+
Object.keys(patternProperties).length > 0
|
|
131
|
+
) {
|
|
132
|
+
return true;
|
|
133
|
+
}
|
|
134
|
+
|
|
135
|
+
const isObjectType =
|
|
136
|
+
type === 'object' || (Array.isArray(type) && type.includes('object'));
|
|
137
|
+
return isObjectType && properties == null && additionalProperties !== false;
|
|
138
|
+
}
|