@zvec/zvec-grep 0.1.5 → 0.1.6
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/README.md +88 -249
- package/README_CN.md +82 -246
- package/dist/authorization/index.d.ts +7 -0
- package/dist/authorization/index.js +7 -0
- package/dist/authorization/index.js.map +1 -0
- package/dist/authorization/manager.d.ts +8 -0
- package/dist/authorization/manager.js +19 -0
- package/dist/authorization/manager.js.map +1 -0
- package/dist/authorization/operation.d.ts +7 -0
- package/dist/authorization/operation.js +50 -0
- package/dist/authorization/operation.js.map +1 -0
- package/dist/authorization/planner.d.ts +21 -0
- package/dist/authorization/planner.js +97 -0
- package/dist/authorization/planner.js.map +1 -0
- package/dist/authorization/prompt.d.ts +11 -0
- package/dist/authorization/prompt.js +62 -0
- package/dist/authorization/prompt.js.map +1 -0
- package/dist/authorization/store.d.ts +20 -0
- package/dist/authorization/store.js +251 -0
- package/dist/authorization/store.js.map +1 -0
- package/dist/authorization/target.d.ts +15 -0
- package/dist/authorization/target.js +49 -0
- package/dist/authorization/target.js.map +1 -0
- package/dist/authorization/types.d.ts +61 -0
- package/dist/authorization/types.js +2 -0
- package/dist/authorization/types.js.map +1 -0
- package/dist/cli/args.d.ts +1 -1
- package/dist/cli/args.js +304 -54
- package/dist/cli/args.js.map +1 -1
- package/dist/cli/auth.d.ts +21 -0
- package/dist/cli/auth.js +212 -0
- package/dist/cli/auth.js.map +1 -0
- package/dist/cli/commands.js +433 -618
- package/dist/cli/commands.js.map +1 -1
- package/dist/cli/errors.js +2 -2
- package/dist/cli/errors.js.map +1 -1
- package/dist/cli/format/context.d.ts +1 -0
- package/dist/cli/format/context.js +241 -45
- package/dist/cli/format/context.js.map +1 -1
- package/dist/cli/format/debug.js +2 -2
- package/dist/cli/format/debug.js.map +1 -1
- package/dist/cli/format/progress.d.ts +4 -0
- package/dist/cli/format/progress.js +219 -46
- package/dist/cli/format/progress.js.map +1 -1
- package/dist/cli/format/status.d.ts +77 -6
- package/dist/cli/format/status.js +302 -106
- package/dist/cli/format/status.js.map +1 -1
- package/dist/cli/help.js +258 -59
- package/dist/cli/help.js.map +1 -1
- package/dist/cli/index.js +0 -0
- package/dist/cli/install.d.ts +4 -0
- package/dist/cli/install.js +845 -0
- package/dist/cli/install.js.map +1 -0
- package/dist/cli/managed-rg.d.ts +10 -0
- package/dist/cli/managed-rg.js +329 -0
- package/dist/cli/managed-rg.js.map +1 -0
- package/dist/cli/types.d.ts +20 -6
- package/dist/cli/types.js.map +1 -1
- package/dist/client/daemon-client.d.ts +19 -0
- package/dist/client/daemon-client.js +180 -0
- package/dist/client/daemon-client.js.map +1 -0
- package/dist/client/mode-router.d.ts +9 -0
- package/dist/client/mode-router.js +28 -0
- package/dist/client/mode-router.js.map +1 -0
- package/dist/client/search-policy.d.ts +11 -0
- package/dist/client/search-policy.js +15 -0
- package/dist/client/search-policy.js.map +1 -0
- package/dist/daemon/backend.d.ts +76 -0
- package/dist/daemon/backend.js +1049 -0
- package/dist/daemon/backend.js.map +1 -0
- package/dist/daemon/change-set.d.ts +25 -0
- package/dist/daemon/change-set.js +101 -0
- package/dist/daemon/change-set.js.map +1 -0
- package/dist/daemon/config.d.ts +24 -0
- package/dist/daemon/config.js +88 -0
- package/dist/daemon/config.js.map +1 -0
- package/dist/daemon/errors.d.ts +5 -0
- package/dist/daemon/errors.js +11 -0
- package/dist/daemon/errors.js.map +1 -0
- package/dist/daemon/http-server.d.ts +30 -0
- package/dist/daemon/http-server.js +287 -0
- package/dist/daemon/http-server.js.map +1 -0
- package/dist/daemon/index-coordinator.d.ts +23 -0
- package/dist/daemon/index-coordinator.js +68 -0
- package/dist/daemon/index-coordinator.js.map +1 -0
- package/dist/daemon/job-scheduler.d.ts +70 -0
- package/dist/daemon/job-scheduler.js +366 -0
- package/dist/daemon/job-scheduler.js.map +1 -0
- package/dist/daemon/logger.d.ts +9 -0
- package/dist/daemon/logger.js +53 -0
- package/dist/daemon/logger.js.map +1 -0
- package/dist/daemon/model-pool.d.ts +45 -0
- package/dist/daemon/model-pool.js +187 -0
- package/dist/daemon/model-pool.js.map +1 -0
- package/dist/daemon/root-lease.d.ts +14 -0
- package/dist/daemon/root-lease.js +188 -0
- package/dist/daemon/root-lease.js.map +1 -0
- package/dist/daemon/root-runtime.d.ts +78 -0
- package/dist/daemon/root-runtime.js +333 -0
- package/dist/daemon/root-runtime.js.map +1 -0
- package/dist/daemon/runtime-manager.d.ts +45 -0
- package/dist/daemon/runtime-manager.js +243 -0
- package/dist/daemon/runtime-manager.js.map +1 -0
- package/dist/daemon/runtime.d.ts +12 -0
- package/dist/daemon/runtime.js +102 -0
- package/dist/daemon/runtime.js.map +1 -0
- package/dist/daemon/server-controller.d.ts +39 -0
- package/dist/daemon/server-controller.js +203 -0
- package/dist/daemon/server-controller.js.map +1 -0
- package/dist/daemon/watch-manager.d.ts +50 -0
- package/dist/daemon/watch-manager.js +293 -0
- package/dist/daemon/watch-manager.js.map +1 -0
- package/dist/daemon/workspace-read-session-cache.d.ts +31 -0
- package/dist/daemon/workspace-read-session-cache.js +115 -0
- package/dist/daemon/workspace-read-session-cache.js.map +1 -0
- package/dist/engine/config.d.ts +30 -8
- package/dist/engine/config.js +171 -45
- package/dist/engine/config.js.map +1 -1
- package/dist/engine/{errors/index.d.ts → errors.d.ts} +6 -1
- package/dist/engine/{errors/details.js → errors.js} +17 -4
- package/dist/engine/errors.js.map +1 -0
- package/dist/engine/extraction/code/adapter.d.ts +1 -1
- package/dist/engine/extraction/code/extractor.d.ts +5 -11
- package/dist/engine/extraction/code/extractor.js +105 -44
- package/dist/engine/extraction/code/extractor.js.map +1 -1
- package/dist/engine/extraction/code/families/c-family.js +1 -1
- package/dist/engine/extraction/code/families/c-family.js.map +1 -1
- package/dist/engine/extraction/code/families/js-ts.d.ts +1 -1
- package/dist/engine/extraction/code/families/metadata.d.ts +1 -1
- package/dist/engine/extraction/code/tree-sitter/grammar.js.map +1 -0
- package/dist/engine/extraction/code/tree-sitter/nodes.js.map +1 -0
- package/dist/engine/extraction/code/tree-sitter/parser.js.map +1 -0
- package/dist/engine/extraction/image/extractor.d.ts +5 -0
- package/dist/engine/extraction/image/extractor.js +35 -0
- package/dist/engine/extraction/image/extractor.js.map +1 -0
- package/dist/engine/extraction/index.d.ts +3 -44
- package/dist/engine/extraction/index.js +1 -280
- package/dist/engine/extraction/index.js.map +1 -1
- package/dist/engine/extraction/markdown/extractor.d.ts +5 -11
- package/dist/engine/extraction/markdown/extractor.js +35 -16
- package/dist/engine/extraction/markdown/extractor.js.map +1 -1
- package/dist/engine/extraction/runtime.d.ts +4 -0
- package/dist/engine/extraction/runtime.js +26 -0
- package/dist/engine/extraction/runtime.js.map +1 -0
- package/dist/engine/extraction/source.d.ts +18 -0
- package/dist/engine/extraction/source.js +22 -0
- package/dist/engine/extraction/source.js.map +1 -0
- package/dist/engine/extraction/text/extractor.d.ts +7 -0
- package/dist/engine/extraction/text/extractor.js +168 -0
- package/dist/engine/extraction/text/extractor.js.map +1 -0
- package/dist/engine/extraction/types.d.ts +4 -0
- package/dist/engine/{models → extraction}/types.js.map +1 -1
- package/dist/engine/{files/file-type.d.ts → file-type.d.ts} +1 -1
- package/dist/engine/file-type.js.map +1 -0
- package/dist/engine/index-status.d.ts +9 -0
- package/dist/engine/index-status.js +38 -0
- package/dist/engine/index-status.js.map +1 -0
- package/dist/engine/manifest.d.ts +12 -0
- package/dist/engine/manifest.js +120 -0
- package/dist/engine/manifest.js.map +1 -0
- package/dist/engine/models/{providers/llama-cpp/embedding.d.ts → backends/llama-cpp.d.ts} +25 -20
- package/dist/engine/models/{providers/llama-cpp/embedding.js → backends/llama-cpp.js} +127 -106
- package/dist/engine/models/backends/llama-cpp.js.map +1 -0
- package/dist/engine/models/backends/model2vec.d.ts +57 -0
- package/dist/engine/models/backends/model2vec.js +332 -0
- package/dist/engine/models/backends/model2vec.js.map +1 -0
- package/dist/engine/models/backends/qwen.d.ts +44 -0
- package/dist/engine/models/{providers/qwen/embedding.js → backends/qwen.js} +155 -94
- package/dist/engine/models/backends/qwen.js.map +1 -0
- package/dist/engine/models/backends/transformers-js.d.ts +87 -0
- package/dist/engine/models/backends/transformers-js.js +316 -0
- package/dist/engine/models/backends/transformers-js.js.map +1 -0
- package/dist/engine/models/catalog.d.ts +188 -23
- package/dist/engine/models/catalog.js +172 -25
- package/dist/engine/models/catalog.js.map +1 -1
- package/dist/engine/models/download-progress.d.ts +15 -0
- package/dist/engine/models/download-progress.js +48 -0
- package/dist/engine/models/download-progress.js.map +1 -0
- package/dist/engine/models/embeddings.d.ts +58 -18
- package/dist/engine/models/embeddings.js +52 -28
- package/dist/engine/models/embeddings.js.map +1 -1
- package/dist/engine/models/factory.d.ts +2 -5
- package/dist/engine/models/factory.js +41 -56
- package/dist/engine/models/factory.js.map +1 -1
- package/dist/engine/models/index.d.ts +4 -6
- package/dist/engine/models/index.js +4 -4
- package/dist/engine/models/index.js.map +1 -1
- package/dist/engine/models/ranking.d.ts +7 -3
- package/dist/engine/models/resolution.d.ts +8 -0
- package/dist/engine/models/resolution.js +22 -0
- package/dist/engine/models/resolution.js.map +1 -0
- package/dist/engine/pipeline/indexing/index.d.ts +9 -7
- package/dist/engine/pipeline/indexing/index.js +360 -121
- package/dist/engine/pipeline/indexing/index.js.map +1 -1
- package/dist/engine/pipeline/indexing/root-paths.js +6 -6
- package/dist/engine/pipeline/indexing/root-paths.js.map +1 -1
- package/dist/engine/pipeline/indexing/scanner/index.d.ts +6 -1
- package/dist/engine/pipeline/indexing/scanner/index.js +233 -29
- package/dist/engine/pipeline/indexing/scanner/index.js.map +1 -1
- package/dist/engine/pipeline/search/index.d.ts +6 -6
- package/dist/engine/pipeline/search/index.js +8 -10
- package/dist/engine/pipeline/search/index.js.map +1 -1
- package/dist/engine/service/index.d.ts +3 -2
- package/dist/engine/service/index.js +1 -1
- package/dist/engine/service/index.js.map +1 -1
- package/dist/engine/service/root.d.ts +13 -12
- package/dist/engine/service/root.js +35 -24
- package/dist/engine/service/root.js.map +1 -1
- package/dist/engine/service/structure-enrichment.js +12 -14
- package/dist/engine/service/structure-enrichment.js.map +1 -1
- package/dist/engine/service/types.d.ts +32 -37
- package/dist/engine/service/workspace-index.d.ts +24 -0
- package/dist/engine/service/workspace-index.js +151 -0
- package/dist/engine/service/workspace-index.js.map +1 -0
- package/dist/engine/service/zvec-grep.d.ts +12 -1
- package/dist/engine/service/zvec-grep.js +677 -486
- package/dist/engine/service/zvec-grep.js.map +1 -1
- package/dist/engine/storage/index.d.ts +8 -4
- package/dist/engine/storage/index.js +1 -1
- package/dist/engine/storage/index.js.map +1 -1
- package/dist/engine/storage/zvec.d.ts +10 -33
- package/dist/engine/storage/zvec.js +95 -38
- package/dist/engine/storage/zvec.js.map +1 -1
- package/dist/engine/types.d.ts +15 -23
- package/dist/engine/types.js +1 -1
- package/dist/engine/types.js.map +1 -1
- package/dist/engine/utils/daemon-lease.d.ts +17 -0
- package/dist/engine/utils/daemon-lease.js +175 -0
- package/dist/engine/utils/daemon-lease.js.map +1 -0
- package/dist/engine/utils/file-selection.js +38 -1
- package/dist/engine/utils/file-selection.js.map +1 -1
- package/dist/engine/utils/lock.js +1 -1
- package/dist/engine/utils/lock.js.map +1 -1
- package/dist/index-progress.d.ts +10 -0
- package/dist/index-progress.js +159 -0
- package/dist/index-progress.js.map +1 -0
- package/dist/index.d.ts +4 -4
- package/dist/index.js +1 -1
- package/dist/index.js.map +1 -1
- package/dist/mcp/http-transport.d.ts +27 -0
- package/dist/mcp/http-transport.js +182 -0
- package/dist/mcp/http-transport.js.map +1 -0
- package/dist/mcp/input-normalization.d.ts +38 -0
- package/dist/mcp/input-normalization.js +95 -0
- package/dist/mcp/input-normalization.js.map +1 -0
- package/dist/mcp/progress-heartbeat.d.ts +17 -0
- package/dist/mcp/progress-heartbeat.js +46 -0
- package/dist/mcp/progress-heartbeat.js.map +1 -0
- package/dist/mcp/request-metadata.d.ts +2 -0
- package/dist/mcp/request-metadata.js +11 -0
- package/dist/mcp/request-metadata.js.map +1 -0
- package/dist/mcp/request-state.d.ts +40 -0
- package/dist/mcp/request-state.js +210 -0
- package/dist/mcp/request-state.js.map +1 -0
- package/dist/mcp/result-format.d.ts +13 -0
- package/dist/mcp/result-format.js +75 -0
- package/dist/mcp/result-format.js.map +1 -0
- package/dist/mcp/schemas.d.ts +430 -0
- package/dist/mcp/schemas.js +447 -0
- package/dist/mcp/schemas.js.map +1 -0
- package/dist/mcp/tools.d.ts +150 -0
- package/dist/mcp/tools.js +448 -0
- package/dist/mcp/tools.js.map +1 -0
- package/dist/mcp/toolset.d.ts +5 -0
- package/dist/mcp/toolset.js +12 -0
- package/dist/mcp/toolset.js.map +1 -0
- package/dist/observability/trace-context.d.ts +11 -0
- package/dist/observability/trace-context.js +138 -0
- package/dist/observability/trace-context.js.map +1 -0
- package/docs/01-agents.md +111 -0
- package/docs/02-cli.md +229 -0
- package/docs/03-mcp.md +167 -0
- package/docs/04-pipeline.md +218 -0
- package/docs/05-architecture.md +86 -0
- package/docs/06-server.md +210 -0
- package/docs/07-embedding.md +179 -0
- package/docs/08-roadmap.md +85 -0
- package/docs/README.md +53 -0
- package/package.json +17 -8
- package/skills/zvec-grep/SKILL.md +43 -133
- package/skills/zvec-grep/agents/openai.yaml +2 -2
- package/skills/zvec-grep/references/cli-fallback.md +85 -0
- package/dist/cli/mcp.d.ts +0 -2
- package/dist/cli/mcp.js +0 -354
- package/dist/cli/mcp.js.map +0 -1
- package/dist/engine/collection/index.d.ts +0 -55
- package/dist/engine/collection/index.js +0 -710
- package/dist/engine/collection/index.js.map +0 -1
- package/dist/engine/errors/details.d.ts +0 -6
- package/dist/engine/errors/details.js.map +0 -1
- package/dist/engine/errors/index.js +0 -16
- package/dist/engine/errors/index.js.map +0 -1
- package/dist/engine/extraction/tree-sitter/grammar.js.map +0 -1
- package/dist/engine/extraction/tree-sitter/nodes.js.map +0 -1
- package/dist/engine/extraction/tree-sitter/parser.js.map +0 -1
- package/dist/engine/files/file-type.js.map +0 -1
- package/dist/engine/models/providers/index.d.ts +0 -2
- package/dist/engine/models/providers/index.js +0 -3
- package/dist/engine/models/providers/index.js.map +0 -1
- package/dist/engine/models/providers/llama-cpp/embedding.js.map +0 -1
- package/dist/engine/models/providers/llama-cpp/index.d.ts +0 -1
- package/dist/engine/models/providers/llama-cpp/index.js +0 -2
- package/dist/engine/models/providers/llama-cpp/index.js.map +0 -1
- package/dist/engine/models/providers/qwen/embedding.d.ts +0 -42
- package/dist/engine/models/providers/qwen/embedding.js.map +0 -1
- package/dist/engine/models/providers/qwen/index.d.ts +0 -1
- package/dist/engine/models/providers/qwen/index.js +0 -2
- package/dist/engine/models/providers/qwen/index.js.map +0 -1
- package/dist/engine/models/types.d.ts +0 -30
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/grammar.d.ts +0 -0
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/grammar.js +0 -0
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/nodes.d.ts +0 -0
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/nodes.js +0 -0
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/parser.d.ts +0 -0
- /package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/parser.js +0 -0
- /package/dist/engine/{models → extraction}/types.js +0 -0
- /package/dist/engine/{files/file-type.js → file-type.js} +0 -0
|
@@ -0,0 +1,332 @@
|
|
|
1
|
+
import { createWriteStream, existsSync, statSync } from "node:fs";
|
|
2
|
+
import { mkdir, readFile, rename, rm, writeFile } from "node:fs/promises";
|
|
3
|
+
import { basename, dirname, join } from "node:path";
|
|
4
|
+
import { Readable, Transform } from "node:stream";
|
|
5
|
+
import { pipeline } from "node:stream/promises";
|
|
6
|
+
import { AutoTokenizer } from "@huggingface/transformers";
|
|
7
|
+
import { EngineError } from "../../errors.js";
|
|
8
|
+
import { defaultHome } from "../../utils/path.js";
|
|
9
|
+
import { BaseEmbeddingModel, } from "../embeddings.js";
|
|
10
|
+
import { createModelDownloadProgressReporter, } from "../download-progress.js";
|
|
11
|
+
const DEFAULT_MODEL_CACHE_DIR = join(defaultHome(), "models");
|
|
12
|
+
const defaultDependencies = {
|
|
13
|
+
async loadTokenizer(repo, options) {
|
|
14
|
+
return (await AutoTokenizer.from_pretrained(repo, options));
|
|
15
|
+
},
|
|
16
|
+
async loadSafetensors(path, tensorName, dimension) {
|
|
17
|
+
return await readStaticEmbeddingTable(path, tensorName, dimension);
|
|
18
|
+
},
|
|
19
|
+
async download(url, destination, onProgress) {
|
|
20
|
+
const response = await fetch(url, { redirect: "follow" });
|
|
21
|
+
if (!response.ok || !response.body) {
|
|
22
|
+
throw new Error(`HTTP ${response.status} ${response.statusText}`);
|
|
23
|
+
}
|
|
24
|
+
const contentLength = response.headers.get("content-length");
|
|
25
|
+
const parsedTotalBytes = contentLength
|
|
26
|
+
? Number.parseInt(contentLength, 10)
|
|
27
|
+
: undefined;
|
|
28
|
+
const totalBytes = parsedTotalBytes !== undefined &&
|
|
29
|
+
Number.isFinite(parsedTotalBytes) &&
|
|
30
|
+
parsedTotalBytes >= 0
|
|
31
|
+
? parsedTotalBytes
|
|
32
|
+
: undefined;
|
|
33
|
+
let downloadedBytes = 0;
|
|
34
|
+
const progressStream = new Transform({
|
|
35
|
+
transform(chunk, _encoding, callback) {
|
|
36
|
+
downloadedBytes += Buffer.byteLength(chunk);
|
|
37
|
+
onProgress?.({ downloadedBytes, totalBytes });
|
|
38
|
+
callback(null, chunk);
|
|
39
|
+
},
|
|
40
|
+
});
|
|
41
|
+
await pipeline(Readable.fromWeb(response.body), progressStream, createWriteStream(destination));
|
|
42
|
+
},
|
|
43
|
+
};
|
|
44
|
+
export class Model2VecEmbeddingModel extends BaseEmbeddingModel {
|
|
45
|
+
entry;
|
|
46
|
+
info;
|
|
47
|
+
modelCacheDir;
|
|
48
|
+
dependencies;
|
|
49
|
+
tokenizer = null;
|
|
50
|
+
staticTable = null;
|
|
51
|
+
loadPromise = null;
|
|
52
|
+
disposed = false;
|
|
53
|
+
constructor(entry, options, dependencies = {}) {
|
|
54
|
+
super();
|
|
55
|
+
this.entry = entry;
|
|
56
|
+
this.info = {
|
|
57
|
+
reference: entry.reference,
|
|
58
|
+
provider: entry.provider,
|
|
59
|
+
name: entry.model,
|
|
60
|
+
dimension: entry.dimension,
|
|
61
|
+
metric: entry.metric,
|
|
62
|
+
inputKinds: ["text"],
|
|
63
|
+
limits: {
|
|
64
|
+
maxBatchSize: entry.maxBatchSize,
|
|
65
|
+
maxInputTokens: entry.maxInputTokens,
|
|
66
|
+
},
|
|
67
|
+
};
|
|
68
|
+
this.modelCacheDir =
|
|
69
|
+
options.modelCacheDir ??
|
|
70
|
+
process.env.ZVEC_GREP_MODEL_CACHE ??
|
|
71
|
+
DEFAULT_MODEL_CACHE_DIR;
|
|
72
|
+
this.dependencies = { ...defaultDependencies, ...dependencies };
|
|
73
|
+
}
|
|
74
|
+
async doEmbed(contents, options) {
|
|
75
|
+
return await this.embedBatch(contents, options);
|
|
76
|
+
}
|
|
77
|
+
async embedBatch(contents, options) {
|
|
78
|
+
this.ensureNotDisposed();
|
|
79
|
+
await this.ensureLoaded(options.onProgress);
|
|
80
|
+
try {
|
|
81
|
+
const texts = contents.map((content) => formatText(content.text, options.purpose, this.entry));
|
|
82
|
+
return await this.embedTexts(texts);
|
|
83
|
+
}
|
|
84
|
+
catch (cause) {
|
|
85
|
+
throw new EngineError("Model2Vec embedding failed", {
|
|
86
|
+
code: "ZVEC_GREP.ENGINE.MODELS.MODEL2VEC_EMBED_FAILED",
|
|
87
|
+
context: `model=${this.entry.reference} repo=${this.entry.repo}`,
|
|
88
|
+
cause,
|
|
89
|
+
});
|
|
90
|
+
}
|
|
91
|
+
}
|
|
92
|
+
async dispose() {
|
|
93
|
+
if (this.disposed) {
|
|
94
|
+
return;
|
|
95
|
+
}
|
|
96
|
+
this.disposed = true;
|
|
97
|
+
this.staticTable = null;
|
|
98
|
+
this.tokenizer = null;
|
|
99
|
+
this.loadPromise = null;
|
|
100
|
+
}
|
|
101
|
+
async ensureLoaded(onProgress) {
|
|
102
|
+
if (this.tokenizer && this.staticTable) {
|
|
103
|
+
return;
|
|
104
|
+
}
|
|
105
|
+
if (this.loadPromise) {
|
|
106
|
+
return await this.loadPromise;
|
|
107
|
+
}
|
|
108
|
+
this.loadPromise = this.loadModel(onProgress);
|
|
109
|
+
try {
|
|
110
|
+
await this.loadPromise;
|
|
111
|
+
}
|
|
112
|
+
finally {
|
|
113
|
+
this.loadPromise = null;
|
|
114
|
+
}
|
|
115
|
+
}
|
|
116
|
+
async loadModel(onProgress) {
|
|
117
|
+
const downloadProgress = createModelDownloadProgressReporter(this.entry.reference, onProgress, [basename(this.entry.modelFile), basename(this.entry.tokenizerFile)]);
|
|
118
|
+
downloadProgress.start();
|
|
119
|
+
const [modelPath, tokenizerSource] = await Promise.all([
|
|
120
|
+
this.resolveModelPath(downloadProgress),
|
|
121
|
+
this.resolveTokenizerSource(downloadProgress),
|
|
122
|
+
]);
|
|
123
|
+
const tokenizerPromise = this.dependencies.loadTokenizer(tokenizerSource, {
|
|
124
|
+
cache_dir: this.modelCacheDir,
|
|
125
|
+
revision: this.entry.revision,
|
|
126
|
+
...(tokenizerSource !== this.entry.repo
|
|
127
|
+
? { local_files_only: true }
|
|
128
|
+
: {}),
|
|
129
|
+
});
|
|
130
|
+
const [tokenizer, staticTable] = await Promise.all([
|
|
131
|
+
tokenizerPromise,
|
|
132
|
+
this.dependencies.loadSafetensors(modelPath, this.entry.embeddingTensor, this.info.dimension),
|
|
133
|
+
]);
|
|
134
|
+
this.ensureNotDisposed();
|
|
135
|
+
this.tokenizer = tokenizer;
|
|
136
|
+
this.staticTable = staticTable;
|
|
137
|
+
downloadProgress.finish();
|
|
138
|
+
}
|
|
139
|
+
async resolveModelPath(downloadProgress) {
|
|
140
|
+
const modelPath = join(this.modelCacheDir, "model2vec", this.entry.repo.replaceAll("/", "--"), this.entry.revision, basename(this.entry.modelFile));
|
|
141
|
+
return await this.resolveCachedFile(this.entry.modelFile, modelPath, downloadProgress);
|
|
142
|
+
}
|
|
143
|
+
async resolveTokenizerSource(downloadProgress) {
|
|
144
|
+
const tokenizerDirectory = join(this.modelCacheDir, "model2vec", this.entry.repo.replaceAll("/", "--"), this.entry.revision, "tokenizer");
|
|
145
|
+
await this.resolveCachedFile(this.entry.tokenizerFile, join(tokenizerDirectory, "tokenizer.json"), downloadProgress);
|
|
146
|
+
const configPath = join(tokenizerDirectory, "tokenizer_config.json");
|
|
147
|
+
if (!isUsableModelFile(configPath)) {
|
|
148
|
+
await writeFile(configPath, `${JSON.stringify({ tokenizer_class: "PreTrainedTokenizer" })}\n`);
|
|
149
|
+
}
|
|
150
|
+
return tokenizerDirectory;
|
|
151
|
+
}
|
|
152
|
+
async resolveCachedFile(remoteFile, localPath, downloadProgress) {
|
|
153
|
+
if (isUsableModelFile(localPath)) {
|
|
154
|
+
downloadProgress.skip(basename(remoteFile));
|
|
155
|
+
return localPath;
|
|
156
|
+
}
|
|
157
|
+
await mkdir(dirname(localPath), { recursive: true });
|
|
158
|
+
const partialPath = `${localPath}.part-${process.pid}-${Date.now()}`;
|
|
159
|
+
const url = `https://huggingface.co/${this.entry.repo}/resolve/${this.entry.revision}/${remoteFile}`;
|
|
160
|
+
try {
|
|
161
|
+
await this.dependencies.download(url, partialPath, (progress) => {
|
|
162
|
+
downloadProgress.report({
|
|
163
|
+
artifact: basename(remoteFile),
|
|
164
|
+
...progress,
|
|
165
|
+
});
|
|
166
|
+
});
|
|
167
|
+
if (!isUsableModelFile(partialPath)) {
|
|
168
|
+
throw new Error("Downloaded model file is empty");
|
|
169
|
+
}
|
|
170
|
+
await rename(partialPath, localPath);
|
|
171
|
+
return localPath;
|
|
172
|
+
}
|
|
173
|
+
catch (cause) {
|
|
174
|
+
await rm(partialPath, { force: true });
|
|
175
|
+
throw new EngineError("Unable to download Model2Vec model artifact", {
|
|
176
|
+
code: "ZVEC_GREP.ENGINE.MODELS.MODEL2VEC_DOWNLOAD_FAILED",
|
|
177
|
+
context: `model=${this.entry.reference} url=${url}`,
|
|
178
|
+
cause,
|
|
179
|
+
});
|
|
180
|
+
}
|
|
181
|
+
}
|
|
182
|
+
async embedTexts(texts) {
|
|
183
|
+
const tokenizer = this.tokenizer;
|
|
184
|
+
const staticTable = this.staticTable;
|
|
185
|
+
if (!tokenizer || !staticTable) {
|
|
186
|
+
throw new Error("Model2Vec model is not loaded");
|
|
187
|
+
}
|
|
188
|
+
const truncatedInputIndexes = [];
|
|
189
|
+
const tokenLists = await Promise.all(texts.map(async (text, index) => {
|
|
190
|
+
const encoded = await tokenizer(text, {
|
|
191
|
+
add_special_tokens: false,
|
|
192
|
+
truncation: true,
|
|
193
|
+
max_length: this.entry.maxInputTokens + 1,
|
|
194
|
+
});
|
|
195
|
+
const encodedTokenIds = Array.from(encoded.input_ids.data, Number);
|
|
196
|
+
if (encodedTokenIds.length > this.entry.maxInputTokens) {
|
|
197
|
+
truncatedInputIndexes.push(index);
|
|
198
|
+
}
|
|
199
|
+
const unknownTokenId = tokenizer.unk_token_id;
|
|
200
|
+
return encodedTokenIds
|
|
201
|
+
.slice(0, this.entry.maxInputTokens)
|
|
202
|
+
.filter((id) => unknownTokenId === undefined ||
|
|
203
|
+
unknownTokenId === null ||
|
|
204
|
+
id !== unknownTokenId);
|
|
205
|
+
}));
|
|
206
|
+
return {
|
|
207
|
+
vectors: embedStaticTokenLists(tokenLists, staticTable, this.info.dimension, this.entry.normalize),
|
|
208
|
+
truncated: truncatedInputIndexes.sort((left, right) => left - right),
|
|
209
|
+
};
|
|
210
|
+
}
|
|
211
|
+
ensureNotDisposed() {
|
|
212
|
+
if (this.disposed) {
|
|
213
|
+
throw new EngineError("Model2Vec embedding model is disposed", {
|
|
214
|
+
code: "ZVEC_GREP.ENGINE.MODELS.MODEL2VEC_DISPOSED",
|
|
215
|
+
context: `model=${this.entry.reference}`,
|
|
216
|
+
});
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
}
|
|
220
|
+
function isUsableModelFile(path) {
|
|
221
|
+
return existsSync(path) && statSync(path).size > 0;
|
|
222
|
+
}
|
|
223
|
+
function formatText(text, purpose, entry) {
|
|
224
|
+
const prefix = purpose === "query"
|
|
225
|
+
? "queryPrefix" in entry && typeof entry.queryPrefix === "string"
|
|
226
|
+
? entry.queryPrefix
|
|
227
|
+
: undefined
|
|
228
|
+
: "documentPrefix" in entry && typeof entry.documentPrefix === "string"
|
|
229
|
+
? entry.documentPrefix
|
|
230
|
+
: undefined;
|
|
231
|
+
return prefix ? `${prefix}${text}` : text;
|
|
232
|
+
}
|
|
233
|
+
async function readStaticEmbeddingTable(path, tensorName, expectedDimension) {
|
|
234
|
+
const file = await readFile(path);
|
|
235
|
+
if (file.byteLength < 9) {
|
|
236
|
+
throw new Error("Safetensors file is too small");
|
|
237
|
+
}
|
|
238
|
+
const headerLength = Number(file.readBigUInt64LE(0));
|
|
239
|
+
const dataStart = 8 + headerLength;
|
|
240
|
+
if (!Number.isSafeInteger(headerLength) || dataStart > file.byteLength) {
|
|
241
|
+
throw new Error("Safetensors header length is invalid");
|
|
242
|
+
}
|
|
243
|
+
let header;
|
|
244
|
+
try {
|
|
245
|
+
header = JSON.parse(file.subarray(8, dataStart).toString("utf8"));
|
|
246
|
+
}
|
|
247
|
+
catch (cause) {
|
|
248
|
+
throw new Error("Safetensors header is invalid JSON", { cause });
|
|
249
|
+
}
|
|
250
|
+
const tensor = header[tensorName];
|
|
251
|
+
if (!tensor ||
|
|
252
|
+
tensor.shape?.length !== 2 ||
|
|
253
|
+
tensor.shape[1] !== expectedDimension ||
|
|
254
|
+
tensor.data_offsets?.length !== 2 ||
|
|
255
|
+
(tensor.dtype !== "F16" && tensor.dtype !== "F32")) {
|
|
256
|
+
throw new Error(`Safetensors tensor '${tensorName}' is missing or incompatible`);
|
|
257
|
+
}
|
|
258
|
+
const [relativeStart, relativeEnd] = tensor.data_offsets;
|
|
259
|
+
const rows = tensor.shape[0];
|
|
260
|
+
const valueCount = rows * expectedDimension;
|
|
261
|
+
const bytesPerValue = tensor.dtype === "F16" ? 2 : 4;
|
|
262
|
+
if (relativeStart < 0 ||
|
|
263
|
+
relativeEnd - relativeStart !== valueCount * bytesPerValue ||
|
|
264
|
+
dataStart + relativeEnd > file.byteLength) {
|
|
265
|
+
throw new Error(`Safetensors tensor '${tensorName}' has invalid offsets`);
|
|
266
|
+
}
|
|
267
|
+
const byteOffset = file.byteOffset + dataStart + relativeStart;
|
|
268
|
+
const data = tensor.dtype === "F16"
|
|
269
|
+
? new Uint16Array(file.buffer, byteOffset, valueCount)
|
|
270
|
+
: new Float32Array(file.buffer, byteOffset, valueCount);
|
|
271
|
+
return {
|
|
272
|
+
data,
|
|
273
|
+
dimension: expectedDimension,
|
|
274
|
+
dtype: tensor.dtype,
|
|
275
|
+
rows,
|
|
276
|
+
};
|
|
277
|
+
}
|
|
278
|
+
function embedStaticTokenLists(tokenLists, table, dimension, normalize) {
|
|
279
|
+
const halfValues = table.dtype === "F16" ? halfFloatValues() : null;
|
|
280
|
+
return tokenLists.map((tokenIds) => {
|
|
281
|
+
const vector = Array.from({ length: dimension }, () => 0);
|
|
282
|
+
if (tokenIds.length === 0) {
|
|
283
|
+
return vector;
|
|
284
|
+
}
|
|
285
|
+
for (const tokenId of tokenIds) {
|
|
286
|
+
if (!Number.isInteger(tokenId) || tokenId < 0 || tokenId >= table.rows) {
|
|
287
|
+
throw new Error(`Tokenizer returned out-of-range token id: id=${tokenId} rows=${table.rows}`);
|
|
288
|
+
}
|
|
289
|
+
const start = tokenId * dimension;
|
|
290
|
+
for (let column = 0; column < dimension; column++) {
|
|
291
|
+
const value = table.data[start + column];
|
|
292
|
+
vector[column] += halfValues ? halfValues[value] : value;
|
|
293
|
+
}
|
|
294
|
+
}
|
|
295
|
+
let squaredNorm = 0;
|
|
296
|
+
for (let column = 0; column < dimension; column++) {
|
|
297
|
+
vector[column] /= tokenIds.length;
|
|
298
|
+
squaredNorm += vector[column] * vector[column];
|
|
299
|
+
}
|
|
300
|
+
if (normalize && squaredNorm > 0) {
|
|
301
|
+
const inverseNorm = 1 / Math.sqrt(squaredNorm);
|
|
302
|
+
for (let column = 0; column < dimension; column++) {
|
|
303
|
+
vector[column] *= inverseNorm;
|
|
304
|
+
}
|
|
305
|
+
}
|
|
306
|
+
return vector;
|
|
307
|
+
});
|
|
308
|
+
}
|
|
309
|
+
let cachedHalfFloatValues = null;
|
|
310
|
+
function halfFloatValues() {
|
|
311
|
+
if (cachedHalfFloatValues) {
|
|
312
|
+
return cachedHalfFloatValues;
|
|
313
|
+
}
|
|
314
|
+
const values = new Float32Array(65_536);
|
|
315
|
+
for (let bits = 0; bits < values.length; bits++) {
|
|
316
|
+
const sign = bits & 0x8000 ? -1 : 1;
|
|
317
|
+
const exponent = (bits >> 10) & 0x1f;
|
|
318
|
+
const fraction = bits & 0x03ff;
|
|
319
|
+
if (exponent === 0) {
|
|
320
|
+
values[bits] = sign * 2 ** -14 * (fraction / 1024);
|
|
321
|
+
}
|
|
322
|
+
else if (exponent === 0x1f) {
|
|
323
|
+
values[bits] = fraction === 0 ? sign * Infinity : Number.NaN;
|
|
324
|
+
}
|
|
325
|
+
else {
|
|
326
|
+
values[bits] = sign * 2 ** (exponent - 15) * (1 + fraction / 1024);
|
|
327
|
+
}
|
|
328
|
+
}
|
|
329
|
+
cachedHalfFloatValues = values;
|
|
330
|
+
return values;
|
|
331
|
+
}
|
|
332
|
+
//# sourceMappingURL=model2vec.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"model2vec.js","sourceRoot":"","sources":["../../../../src/engine/models/backends/model2vec.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,SAAS,CAAC;AAClE,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE,EAAE,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAC1E,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACpD,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AAClD,OAAO,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AAEhD,OAAO,EAAE,aAAa,EAAE,MAAM,2BAA2B,CAAC;AAC1D,OAAO,EAAE,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAE9C,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EACL,kBAAkB,GAMnB,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EACL,mCAAmC,GAEpC,MAAM,yBAAyB,CAAC;AAqDjC,MAAM,uBAAuB,GAAG,IAAI,CAAC,WAAW,EAAE,EAAE,QAAQ,CAAC,CAAC;AAE9D,MAAM,mBAAmB,GAA0B;IACjD,KAAK,CAAC,aAAa,CAAC,IAAI,EAAE,OAAO;QAC/B,OAAO,CAAC,MAAM,aAAa,CAAC,eAAe,CACzC,IAAI,EACJ,OAAO,CACR,CAA6B,CAAC;IACjC,CAAC;IACD,KAAK,CAAC,eAAe,CAAC,IAAI,EAAE,UAAU,EAAE,SAAS;QAC/C,OAAO,MAAM,wBAAwB,CAAC,IAAI,EAAE,UAAU,EAAE,SAAS,CAAC,CAAC;IACrE,CAAC;IACD,KAAK,CAAC,QAAQ,CAAC,GAAG,EAAE,WAAW,EAAE,UAAU;QACzC,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAC,CAAC;QAC1D,IAAI,CAAC,QAAQ,CAAC,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;YACnC,MAAM,IAAI,KAAK,CAAC,QAAQ,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;QACpE,CAAC;QACD,MAAM,aAAa,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAC;QAC7D,MAAM,gBAAgB,GAAG,aAAa;YACpC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,aAAa,EAAE,EAAE,CAAC;YACpC,CAAC,CAAC,SAAS,CAAC;QACd,MAAM,UAAU,GACd,gBAAgB,KAAK,SAAS;YAC9B,MAAM,CAAC,QAAQ,CAAC,gBAAgB,CAAC;YACjC,gBAAgB,IAAI,CAAC;YACnB,CAAC,CAAC,gBAAgB;YAClB,CAAC,CAAC,SAAS,CAAC;QAChB,IAAI,eAAe,GAAG,CAAC,CAAC;QACxB,MAAM,cAAc,GAAG,IAAI,SAAS,CAAC;YACnC,SAAS,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ;gBAClC,eAAe,IAAI,MAAM,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;gBAC5C,UAAU,EAAE,CAAC,EAAE,eAAe,EAAE,UAAU,EAAE,CAAC,CAAC;gBAC9C,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;YACxB,CAAC;SACF,CAAC,CAAC;QACH,MAAM,QAAQ,CACZ,QAAQ,CAAC,OAAO,CAAC,QAAQ,CAAC,IAA0B,CAAC,EACrD,cAAc,EACd,iBAAiB,CAAC,WAAW,CAAC,CAC/B,CAAC;IACJ,CAAC;CACF,CAAC;AAEF,MAAM,OAAO,uBAAwB,SAAQ,kBAAkB;IAW1C;IAVV,IAAI,CAAqB;IAEjB,aAAa,CAAS;IACtB,YAAY,CAAwB;IAC7C,SAAS,GAAyB,IAAI,CAAC;IACvC,WAAW,GAAgC,IAAI,CAAC;IAChD,WAAW,GAAyB,IAAI,CAAC;IACzC,QAAQ,GAAG,KAAK,CAAC;IAEzB,YACmB,KAAqC,EACtD,OAAoC,EACpC,eAA+C,EAAE;QAEjD,KAAK,EAAE,CAAC;QAJS,UAAK,GAAL,KAAK,CAAgC;QAKtD,IAAI,CAAC,IAAI,GAAG;YACV,SAAS,EAAE,KAAK,CAAC,SAAS;YAC1B,QAAQ,EAAE,KAAK,CAAC,QAAQ;YACxB,IAAI,EAAE,KAAK,CAAC,KAAK;YACjB,SAAS,EAAE,KAAK,CAAC,SAAS;YAC1B,MAAM,EAAE,KAAK,CAAC,MAAM;YACpB,UAAU,EAAE,CAAC,MAAM,CAAC;YACpB,MAAM,EAAE;gBACN,YAAY,EAAE,KAAK,CAAC,YAAY;gBAChC,cAAc,EAAE,KAAK,CAAC,cAAc;aACrC;SACF,CAAC;QACF,IAAI,CAAC,aAAa;YAChB,OAAO,CAAC,aAAa;gBACrB,OAAO,CAAC,GAAG,CAAC,qBAAqB;gBACjC,uBAAuB,CAAC;QAC1B,IAAI,CAAC,YAAY,GAAG,EAAE,GAAG,mBAAmB,EAAE,GAAG,YAAY,EAAE,CAAC;IAClE,CAAC;IAES,KAAK,CAAC,OAAO,CACrB,QAA4B,EAC5B,OAAmC;QAEnC,OAAO,MAAM,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAClD,CAAC;IAEO,KAAK,CAAC,UAAU,CACtB,QAA4B,EAC5B,OAAmC;QAEnC,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACzB,MAAM,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC;QAE5C,IAAI,CAAC;YACH,MAAM,KAAK,GAAI,QAAmC,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CACjE,UAAU,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CACtD,CAAC;YACF,OAAO,MAAM,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;QACtC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,IAAI,WAAW,CAAC,4BAA4B,EAAE;gBAClD,IAAI,EAAE,gDAAgD;gBACtD,OAAO,EAAE,SAAS,IAAI,CAAC,KAAK,CAAC,SAAS,SAAS,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE;gBAChE,KAAK;aACN,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAEQ,KAAK,CAAC,OAAO;QACpB,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAClB,OAAO;QACT,CAAC;QACD,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC;QACrB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;QACxB,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;QACtB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;IAC1B,CAAC;IAEO,KAAK,CAAC,YAAY,CACxB,UAAuD;QAEvD,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACvC,OAAO;QACT,CAAC;QACD,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACrB,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC;QAChC,CAAC;QAED,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;QAC9C,IAAI,CAAC;YACH,MAAM,IAAI,CAAC,WAAW,CAAC;QACzB,CAAC;gBAAS,CAAC;YACT,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;QAC1B,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,SAAS,CACrB,UAAuD;QAEvD,MAAM,gBAAgB,GAAG,mCAAmC,CAC1D,IAAI,CAAC,KAAK,CAAC,SAAS,EACpB,UAAU,EACV,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,CACrE,CAAC;QACF,gBAAgB,CAAC,KAAK,EAAE,CAAC;QACzB,MAAM,CAAC,SAAS,EAAE,eAAe,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YACrD,IAAI,CAAC,gBAAgB,CAAC,gBAAgB,CAAC;YACvC,IAAI,CAAC,sBAAsB,CAAC,gBAAgB,CAAC;SAC9C,CAAC,CAAC;QACH,MAAM,gBAAgB,GAAG,IAAI,CAAC,YAAY,CAAC,aAAa,CAAC,eAAe,EAAE;YACxE,SAAS,EAAE,IAAI,CAAC,aAAa;YAC7B,QAAQ,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ;YAC7B,GAAG,CAAC,eAAe,KAAK,IAAI,CAAC,KAAK,CAAC,IAAI;gBACrC,CAAC,CAAC,EAAE,gBAAgB,EAAE,IAAI,EAAE;gBAC5B,CAAC,CAAC,EAAE,CAAC;SACR,CAAC,CAAC;QACH,MAAM,CAAC,SAAS,EAAE,WAAW,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YACjD,gBAAgB;YAChB,IAAI,CAAC,YAAY,CAAC,eAAe,CAC/B,SAAS,EACT,IAAI,CAAC,KAAK,CAAC,eAAe,EAC1B,IAAI,CAAC,IAAI,CAAC,SAAS,CACpB;SACF,CAAC,CAAC;QACH,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACzB,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC;QAC3B,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC;QAC/B,gBAAgB,CAAC,MAAM,EAAE,CAAC;IAC5B,CAAC;IAEO,KAAK,CAAC,gBAAgB,CAC5B,gBAA+C;QAE/C,MAAM,SAAS,GAAG,IAAI,CACpB,IAAI,CAAC,aAAa,EAClB,WAAW,EACX,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,EAAE,IAAI,CAAC,EACrC,IAAI,CAAC,KAAK,CAAC,QAAQ,EACnB,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAC/B,CAAC;QACF,OAAO,MAAM,IAAI,CAAC,iBAAiB,CACjC,IAAI,CAAC,KAAK,CAAC,SAAS,EACpB,SAAS,EACT,gBAAgB,CACjB,CAAC;IACJ,CAAC;IAEO,KAAK,CAAC,sBAAsB,CAClC,gBAA+C;QAE/C,MAAM,kBAAkB,GAAG,IAAI,CAC7B,IAAI,CAAC,aAAa,EAClB,WAAW,EACX,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,EAAE,IAAI,CAAC,EACrC,IAAI,CAAC,KAAK,CAAC,QAAQ,EACnB,WAAW,CACZ,CAAC;QACF,MAAM,IAAI,CAAC,iBAAiB,CAC1B,IAAI,CAAC,KAAK,CAAC,aAAa,EACxB,IAAI,CAAC,kBAAkB,EAAE,gBAAgB,CAAC,EAC1C,gBAAgB,CACjB,CAAC;QACF,MAAM,UAAU,GAAG,IAAI,CAAC,kBAAkB,EAAE,uBAAuB,CAAC,CAAC;QACrE,IAAI,CAAC,iBAAiB,CAAC,UAAU,CAAC,EAAE,CAAC;YACnC,MAAM,SAAS,CACb,UAAU,EACV,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,eAAe,EAAE,qBAAqB,EAAE,CAAC,IAAI,CAClE,CAAC;QACJ,CAAC;QACD,OAAO,kBAAkB,CAAC;IAC5B,CAAC;IAEO,KAAK,CAAC,iBAAiB,CAC7B,UAAkB,EAClB,SAAiB,EACjB,gBAA+C;QAE/C,IAAI,iBAAiB,CAAC,SAAS,CAAC,EAAE,CAAC;YACjC,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAC;YAC5C,OAAO,SAAS,CAAC;QACnB,CAAC;QAED,MAAM,KAAK,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;QACrD,MAAM,WAAW,GAAG,GAAG,SAAS,SAAS,OAAO,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,EAAE,EAAE,CAAC;QACrE,MAAM,GAAG,GAAG,0BAA0B,IAAI,CAAC,KAAK,CAAC,IAAI,YAAY,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,UAAU,EAAE,CAAC;QACrG,IAAI,CAAC;YACH,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,GAAG,EAAE,WAAW,EAAE,CAAC,QAAQ,EAAE,EAAE;gBAC9D,gBAAgB,CAAC,MAAM,CAAC;oBACtB,QAAQ,EAAE,QAAQ,CAAC,UAAU,CAAC;oBAC9B,GAAG,QAAQ;iBACZ,CAAC,CAAC;YACL,CAAC,CAAC,CAAC;YACH,IAAI,CAAC,iBAAiB,CAAC,WAAW,CAAC,EAAE,CAAC;gBACpC,MAAM,IAAI,KAAK,CAAC,gCAAgC,CAAC,CAAC;YACpD,CAAC;YACD,MAAM,MAAM,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACrC,OAAO,SAAS,CAAC;QACnB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,EAAE,CAAC,WAAW,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;YACvC,MAAM,IAAI,WAAW,CAAC,6CAA6C,EAAE;gBACnE,IAAI,EAAE,mDAAmD;gBACzD,OAAO,EAAE,SAAS,IAAI,CAAC,KAAK,CAAC,SAAS,QAAQ,GAAG,EAAE;gBACnD,KAAK;aACN,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,UAAU,CAAC,KAAe;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC;QACjC,MAAM,WAAW,GAAG,IAAI,CAAC,WAAW,CAAC;QACrC,IAAI,CAAC,SAAS,IAAI,CAAC,WAAW,EAAE,CAAC;YAC/B,MAAM,IAAI,KAAK,CAAC,+BAA+B,CAAC,CAAC;QACnD,CAAC;QAED,MAAM,qBAAqB,GAAa,EAAE,CAAC;QAC3C,MAAM,UAAU,GAAG,MAAM,OAAO,CAAC,GAAG,CAClC,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,EAAE;YAC9B,MAAM,OAAO,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE;gBACpC,kBAAkB,EAAE,KAAK;gBACzB,UAAU,EAAE,IAAI;gBAChB,UAAU,EAAE,IAAI,CAAC,KAAK,CAAC,cAAc,GAAG,CAAC;aAC1C,CAAC,CAAC;YACH,MAAM,eAAe,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;YACnE,IAAI,eAAe,CAAC,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,EAAE,CAAC;gBACvD,qBAAqB,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACpC,CAAC;YACD,MAAM,cAAc,GAAG,SAAS,CAAC,YAAY,CAAC;YAC9C,OAAO,eAAe;iBACnB,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,cAAc,CAAC;iBACnC,MAAM,CACL,CAAC,EAAE,EAAE,EAAE,CACL,cAAc,KAAK,SAAS;gBAC5B,cAAc,KAAK,IAAI;gBACvB,EAAE,KAAK,cAAc,CACxB,CAAC;QACN,CAAC,CAAC,CACH,CAAC;QAEF,OAAO;YACL,OAAO,EAAE,qBAAqB,CAC5B,UAAU,EACV,WAAW,EACX,IAAI,CAAC,IAAI,CAAC,SAAS,EACnB,IAAI,CAAC,KAAK,CAAC,SAAS,CACrB;YACD,SAAS,EAAE,qBAAqB,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,GAAG,KAAK,CAAC;SACrE,CAAC;IACJ,CAAC;IAEO,iBAAiB;QACvB,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAClB,MAAM,IAAI,WAAW,CAAC,uCAAuC,EAAE;gBAC7D,IAAI,EAAE,4CAA4C;gBAClD,OAAO,EAAE,SAAS,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE;aACzC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;CACF;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,OAAO,UAAU,CAAC,IAAI,CAAC,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC;AACrD,CAAC;AAED,SAAS,UAAU,CACjB,IAAY,EACZ,OAA8C,EAC9C,KAAqC;IAErC,MAAM,MAAM,GACV,OAAO,KAAK,OAAO;QACjB,CAAC,CAAC,aAAa,IAAI,KAAK,IAAI,OAAO,KAAK,CAAC,WAAW,KAAK,QAAQ;YAC/D,CAAC,CAAC,KAAK,CAAC,WAAW;YACnB,CAAC,CAAC,SAAS;QACb,CAAC,CAAC,gBAAgB,IAAI,KAAK,IAAI,OAAO,KAAK,CAAC,cAAc,KAAK,QAAQ;YACrE,CAAC,CAAC,KAAK,CAAC,cAAc;YACtB,CAAC,CAAC,SAAS,CAAC;IAClB,OAAO,MAAM,CAAC,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;AAC5C,CAAC;AAED,KAAK,UAAU,wBAAwB,CACrC,IAAY,EACZ,UAAkB,EAClB,iBAAyB;IAEzB,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,CAAC;IAClC,IAAI,IAAI,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;QACxB,MAAM,IAAI,KAAK,CAAC,+BAA+B,CAAC,CAAC;IACnD,CAAC;IAED,MAAM,YAAY,GAAG,MAAM,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,CAAC;IACrD,MAAM,SAAS,GAAG,CAAC,GAAG,YAAY,CAAC;IACnC,IAAI,CAAC,MAAM,CAAC,aAAa,CAAC,YAAY,CAAC,IAAI,SAAS,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;QACvE,MAAM,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;IAC1D,CAAC;IAED,IAAI,MAGH,CAAC;IACF,IAAI,CAAC;QACH,MAAM,GAAG,IAAI,CAAC,KAAK,CACjB,IAAI,CAAC,QAAQ,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAC5B,CAAC;IACrB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,IAAI,KAAK,CAAC,oCAAoC,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;IACnE,CAAC;IACD,MAAM,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAClC,IACE,CAAC,MAAM;QACP,MAAM,CAAC,KAAK,EAAE,MAAM,KAAK,CAAC;QAC1B,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,iBAAiB;QACrC,MAAM,CAAC,YAAY,EAAE,MAAM,KAAK,CAAC;QACjC,CAAC,MAAM,CAAC,KAAK,KAAK,KAAK,IAAI,MAAM,CAAC,KAAK,KAAK,KAAK,CAAC,EAClD,CAAC;QACD,MAAM,IAAI,KAAK,CACb,uBAAuB,UAAU,8BAA8B,CAChE,CAAC;IACJ,CAAC;IAED,MAAM,CAAC,aAAa,EAAE,WAAW,CAAC,GAAG,MAAM,CAAC,YAAY,CAAC;IACzD,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IAC7B,MAAM,UAAU,GAAG,IAAI,GAAG,iBAAiB,CAAC;IAC5C,MAAM,aAAa,GAAG,MAAM,CAAC,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACrD,IACE,aAAa,GAAG,CAAC;QACjB,WAAW,GAAG,aAAa,KAAK,UAAU,GAAG,aAAa;QAC1D,SAAS,GAAG,WAAW,GAAG,IAAI,CAAC,UAAU,EACzC,CAAC;QACD,MAAM,IAAI,KAAK,CAAC,uBAAuB,UAAU,uBAAuB,CAAC,CAAC;IAC5E,CAAC;IAED,MAAM,UAAU,GAAG,IAAI,CAAC,UAAU,GAAG,SAAS,GAAG,aAAa,CAAC;IAC/D,MAAM,IAAI,GACR,MAAM,CAAC,KAAK,KAAK,KAAK;QACpB,CAAC,CAAC,IAAI,WAAW,CAAC,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,UAAU,CAAC;QACtD,CAAC,CAAC,IAAI,YAAY,CAAC,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC;IAC5D,OAAO;QACL,IAAI;QACJ,SAAS,EAAE,iBAAiB;QAC5B,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,IAAI;KACL,CAAC;AACJ,CAAC;AAED,SAAS,qBAAqB,CAC5B,UAA+B,EAC/B,KAA2B,EAC3B,SAAiB,EACjB,SAAkB;IAElB,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,eAAe,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;IACpE,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE;QACjC,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,SAAS,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;QAC1D,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC1B,OAAO,MAAM,CAAC;QAChB,CAAC;QAED,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAC/B,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC,IAAI,OAAO,GAAG,CAAC,IAAI,OAAO,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;gBACvE,MAAM,IAAI,KAAK,CACb,gDAAgD,OAAO,SAAS,KAAK,CAAC,IAAI,EAAE,CAC7E,CAAC;YACJ,CAAC;YACD,MAAM,KAAK,GAAG,OAAO,GAAG,SAAS,CAAC;YAClC,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,EAAE,MAAM,EAAE,EAAE,CAAC;gBAClD,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,GAAG,MAAM,CAAC,CAAC;gBACzC,MAAM,CAAC,MAAM,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC;YAC3D,CAAC;QACH,CAAC;QAED,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,EAAE,MAAM,EAAE,EAAE,CAAC;YAClD,MAAM,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,MAAM,CAAC;YAClC,WAAW,IAAI,MAAM,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC;QACjD,CAAC;QACD,IAAI,SAAS,IAAI,WAAW,GAAG,CAAC,EAAE,CAAC;YACjC,MAAM,WAAW,GAAG,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;YAC/C,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,EAAE,MAAM,EAAE,EAAE,CAAC;gBAClD,MAAM,CAAC,MAAM,CAAC,IAAI,WAAW,CAAC;YAChC,CAAC;QACH,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC,CAAC,CAAC;AACL,CAAC;AAED,IAAI,qBAAqB,GAAwB,IAAI,CAAC;AAEtD,SAAS,eAAe;IACtB,IAAI,qBAAqB,EAAE,CAAC;QAC1B,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IACD,MAAM,MAAM,GAAG,IAAI,YAAY,CAAC,MAAM,CAAC,CAAC;IACxC,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC;QAChD,MAAM,IAAI,GAAG,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACpC,MAAM,QAAQ,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC;QACrC,MAAM,QAAQ,GAAG,IAAI,GAAG,MAAM,CAAC;QAC/B,IAAI,QAAQ,KAAK,CAAC,EAAE,CAAC;YACnB,MAAM,CAAC,IAAI,CAAC,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,QAAQ,GAAG,IAAI,CAAC,CAAC;QACrD,CAAC;aAAM,IAAI,QAAQ,KAAK,IAAI,EAAE,CAAC;YAC7B,MAAM,CAAC,IAAI,CAAC,GAAG,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC;QAC/D,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,QAAQ,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,GAAG,QAAQ,GAAG,IAAI,CAAC,CAAC;QACrE,CAAC;IACH,CAAC;IACD,qBAAqB,GAAG,MAAM,CAAC;IAC/B,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,44 @@
|
|
|
1
|
+
import type { Content } from "../../types.js";
|
|
2
|
+
import { BaseEmbeddingModel, type CreateEmbeddingModelOptions, type EmbeddingModelInfo, type EmbeddingResult, type NormalizedEmbeddingOptions } from "../embeddings.js";
|
|
3
|
+
import type { QwenMultimodalEmbeddingCatalogEntry, QwenTextEmbeddingCatalogEntry } from "../catalog.js";
|
|
4
|
+
type QwenDependencies = {
|
|
5
|
+
fetch: typeof globalThis.fetch;
|
|
6
|
+
};
|
|
7
|
+
type QwenTextEmbeddingSpec = {
|
|
8
|
+
displayName: string;
|
|
9
|
+
errorCodePrefix: string;
|
|
10
|
+
};
|
|
11
|
+
type QwenTextEmbeddingV4CatalogEntry = Extract<QwenTextEmbeddingCatalogEntry, {
|
|
12
|
+
model: "text-embedding-v4";
|
|
13
|
+
}>;
|
|
14
|
+
type Qwen37TextEmbeddingCatalogEntry = Extract<QwenTextEmbeddingCatalogEntry, {
|
|
15
|
+
model: "qwen3.7-text-embedding";
|
|
16
|
+
}>;
|
|
17
|
+
declare abstract class QwenTextEmbeddingModel extends BaseEmbeddingModel {
|
|
18
|
+
readonly info: EmbeddingModelInfo;
|
|
19
|
+
private readonly entry;
|
|
20
|
+
private readonly apiKey;
|
|
21
|
+
private readonly endpoint;
|
|
22
|
+
private readonly displayName;
|
|
23
|
+
private readonly errorCodePrefix;
|
|
24
|
+
private readonly dependencies;
|
|
25
|
+
constructor(entry: QwenTextEmbeddingCatalogEntry, spec: QwenTextEmbeddingSpec, options: CreateEmbeddingModelOptions, dependencies: Partial<QwenDependencies>);
|
|
26
|
+
protected doEmbed(contents: readonly Content[], options: NormalizedEmbeddingOptions): Promise<EmbeddingResult>;
|
|
27
|
+
private errorCode;
|
|
28
|
+
}
|
|
29
|
+
export declare class QwenTextEmbeddingV4Model extends QwenTextEmbeddingModel {
|
|
30
|
+
constructor(entry: QwenTextEmbeddingV4CatalogEntry, options: CreateEmbeddingModelOptions, dependencies?: Partial<QwenDependencies>);
|
|
31
|
+
}
|
|
32
|
+
export declare class Qwen37TextEmbeddingModel extends QwenTextEmbeddingModel {
|
|
33
|
+
constructor(entry: Qwen37TextEmbeddingCatalogEntry, options: CreateEmbeddingModelOptions, dependencies?: Partial<QwenDependencies>);
|
|
34
|
+
}
|
|
35
|
+
export declare class Qwen3VlEmbeddingModel extends BaseEmbeddingModel {
|
|
36
|
+
readonly info: EmbeddingModelInfo;
|
|
37
|
+
private readonly entry;
|
|
38
|
+
private readonly apiKey;
|
|
39
|
+
private readonly endpoint;
|
|
40
|
+
private readonly dependencies;
|
|
41
|
+
constructor(entry: QwenMultimodalEmbeddingCatalogEntry, options: CreateEmbeddingModelOptions, dependencies?: Partial<QwenDependencies>);
|
|
42
|
+
protected doEmbed(contents: readonly Content[], options: NormalizedEmbeddingOptions): Promise<EmbeddingResult>;
|
|
43
|
+
}
|
|
44
|
+
export {};
|