@zvec/zvec-grep 0.1.4 → 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 +86 -162
- package/README_CN.md +80 -162
- 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 +3 -0
- package/dist/cli/args.js +749 -201
- 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.d.ts +3 -1
- package/dist/cli/commands.js +529 -180
- package/dist/cli/commands.js.map +1 -1
- package/dist/cli/errors.js +7 -6
- package/dist/cli/errors.js.map +1 -1
- package/dist/cli/format/context.d.ts +1 -0
- package/dist/cli/format/context.js +270 -70
- package/dist/cli/format/context.js.map +1 -1
- package/dist/cli/format/debug.js +14 -16
- package/dist/cli/format/debug.js.map +1 -1
- package/dist/cli/format/highlight.js.map +1 -1
- package/dist/cli/format/progress.d.ts +4 -0
- package/dist/cli/format/progress.js +222 -44
- package/dist/cli/format/progress.js.map +1 -1
- package/dist/cli/format/range.js.map +1 -1
- package/dist/cli/format/status.d.ts +77 -6
- package/dist/cli/format/status.js +343 -114
- package/dist/cli/format/status.js.map +1 -1
- package/dist/cli/help.d.ts +1 -1
- package/dist/cli/help.js +368 -84
- package/dist/cli/help.js.map +1 -1
- package/dist/cli/index.js +4 -4
- package/dist/cli/index.js.map +1 -1
- 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 +40 -13
- package/dist/cli/types.js.map +1 -1
- package/dist/cli/version.js +3 -1
- package/dist/cli/version.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 +50 -0
- package/dist/engine/config.js +321 -0
- package/dist/engine/config.js.map +1 -0
- 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/adapter.js.map +1 -1
- package/dist/engine/extraction/code/extractor.d.ts +5 -11
- package/dist/engine/extraction/code/extractor.js +184 -109
- package/dist/engine/extraction/code/extractor.js.map +1 -1
- package/dist/engine/extraction/code/families/c-family.js +13 -7
- 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/js-ts.js +24 -18
- package/dist/engine/extraction/code/families/js-ts.js.map +1 -1
- package/dist/engine/extraction/code/families/metadata.d.ts +1 -1
- package/dist/engine/extraction/code/families/metadata.js +11 -8
- package/dist/engine/extraction/code/families/metadata.js.map +1 -1
- package/dist/engine/extraction/code/families/name-field.js +5 -5
- package/dist/engine/extraction/code/families/name-field.js.map +1 -1
- package/dist/engine/extraction/code/languages/c.js +1 -4
- package/dist/engine/extraction/code/languages/c.js.map +1 -1
- package/dist/engine/extraction/code/languages/cpp.js.map +1 -1
- package/dist/engine/extraction/code/languages/go.js +1 -3
- package/dist/engine/extraction/code/languages/go.js.map +1 -1
- package/dist/engine/extraction/code/languages/java.js.map +1 -1
- package/dist/engine/extraction/code/languages/javascript.js +1 -3
- package/dist/engine/extraction/code/languages/javascript.js.map +1 -1
- package/dist/engine/extraction/code/languages/python.js +5 -6
- package/dist/engine/extraction/code/languages/python.js.map +1 -1
- package/dist/engine/extraction/code/languages/rust.js +1 -5
- package/dist/engine/extraction/code/languages/rust.js.map +1 -1
- package/dist/engine/extraction/code/languages/typescript.js.map +1 -1
- package/dist/engine/extraction/code/tree-sitter/grammar.js.map +1 -0
- package/dist/engine/extraction/{tree-sitter → code/tree-sitter}/nodes.js +2 -1
- 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/ids.js.map +1 -1
- 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 +37 -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} +165 -113
- 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} +168 -102
- 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 -41
- 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 +53 -29
- 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 +399 -147
- package/dist/engine/pipeline/indexing/index.js.map +1 -1
- package/dist/engine/pipeline/indexing/root-paths.js +16 -19
- 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 +318 -66
- 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 +76 -34
- 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/lexical.d.ts +14 -15
- package/dist/engine/service/lexical.js +78 -119
- package/dist/engine/service/lexical.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 +29 -25
- package/dist/engine/service/structure-enrichment.js.map +1 -1
- package/dist/engine/service/types.d.ts +61 -35
- 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 +878 -544
- 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 +110 -49
- package/dist/engine/storage/zvec.js.map +1 -1
- package/dist/engine/types.d.ts +29 -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.d.ts +10 -0
- package/dist/engine/utils/file-selection.js +157 -0
- package/dist/engine/utils/file-selection.js.map +1 -0
- package/dist/engine/utils/glob.d.ts +3 -0
- package/dist/engine/utils/glob.js +116 -18
- package/dist/engine/utils/glob.js.map +1 -1
- package/dist/engine/utils/hash.js.map +1 -1
- package/dist/engine/utils/json.d.ts +6 -2
- package/dist/engine/utils/json.js +50 -12
- package/dist/engine/utils/json.js.map +1 -1
- package/dist/engine/utils/lock.js +7 -7
- package/dist/engine/utils/lock.js.map +1 -1
- package/dist/engine/utils/path.js +2 -2
- package/dist/engine/utils/path.js.map +1 -1
- package/dist/engine/utils/timing.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 +6 -6
- package/dist/index.js +3 -3
- 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 +43 -7
- package/skills/zvec-grep/SKILL.md +43 -123
- package/skills/zvec-grep/agents/openai.yaml +2 -2
- package/skills/zvec-grep/references/cli-fallback.md +85 -0
- 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}/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 +1,168 @@
|
|
|
1
|
+
import { EngineError } from "../../errors.js";
|
|
2
|
+
import { makeEntityId } from "../ids.js";
|
|
3
|
+
import { validateSourceFile } from "../source.js";
|
|
4
|
+
const DEFAULT_TEXT_CHUNK_CHARS = 3600;
|
|
5
|
+
const DEFAULT_TEXT_CHUNK_OVERLAP_CHARS = 540;
|
|
6
|
+
export class TextExtractor {
|
|
7
|
+
async extract(source, options = {}) {
|
|
8
|
+
if (source.kind !== "text") {
|
|
9
|
+
throw new EngineError("Text extractor received a non-text source", {
|
|
10
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_UNSUPPORTED_SOURCE",
|
|
11
|
+
context: `fileId=${source.file.id} sourceKind=${source.kind}`,
|
|
12
|
+
});
|
|
13
|
+
}
|
|
14
|
+
validateSourceFile(source);
|
|
15
|
+
const { maxChunkChars, chunkOverlapChars } = resolveTextChunkOptions(options);
|
|
16
|
+
return extractPlainTextFragments(source, maxChunkChars, chunkOverlapChars);
|
|
17
|
+
}
|
|
18
|
+
}
|
|
19
|
+
function resolveTextChunkOptions(options) {
|
|
20
|
+
const maxChunkChars = options.maxChunkChars ?? DEFAULT_TEXT_CHUNK_CHARS;
|
|
21
|
+
const chunkOverlapChars = options.chunkOverlapChars ?? DEFAULT_TEXT_CHUNK_OVERLAP_CHARS;
|
|
22
|
+
if (!Number.isInteger(maxChunkChars) || maxChunkChars <= 0) {
|
|
23
|
+
throw new EngineError("Text extractor requires a positive integer chunk size", {
|
|
24
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_SIZE",
|
|
25
|
+
context: `maxChunkChars=${maxChunkChars}`,
|
|
26
|
+
});
|
|
27
|
+
}
|
|
28
|
+
if (!Number.isInteger(chunkOverlapChars) ||
|
|
29
|
+
chunkOverlapChars < 0 ||
|
|
30
|
+
chunkOverlapChars >= maxChunkChars) {
|
|
31
|
+
throw new EngineError("Text extractor requires overlap to be smaller than chunk size", {
|
|
32
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_OVERLAP",
|
|
33
|
+
context: `maxChunkChars=${maxChunkChars} chunkOverlapChars=${chunkOverlapChars}`,
|
|
34
|
+
});
|
|
35
|
+
}
|
|
36
|
+
return { maxChunkChars, chunkOverlapChars };
|
|
37
|
+
}
|
|
38
|
+
export function extractPlainTextFragments(source, maxChunkChars, chunkOverlapChars) {
|
|
39
|
+
return chunkText(source.text, maxChunkChars, chunkOverlapChars).map((chunk, index) => ({
|
|
40
|
+
id: makeEntityId(source.file.id, index),
|
|
41
|
+
fileId: source.file.id,
|
|
42
|
+
range: chunk.range,
|
|
43
|
+
content: {
|
|
44
|
+
kind: "text",
|
|
45
|
+
text: chunk.text,
|
|
46
|
+
},
|
|
47
|
+
}));
|
|
48
|
+
}
|
|
49
|
+
function chunkText(text, maxChars, overlapChars) {
|
|
50
|
+
if (text.trim().length === 0) {
|
|
51
|
+
return [];
|
|
52
|
+
}
|
|
53
|
+
const lines = text.split("\n");
|
|
54
|
+
const lineOffsets = computeLineOffsets(lines);
|
|
55
|
+
const chunks = [];
|
|
56
|
+
let startIndex = 0;
|
|
57
|
+
while (startIndex < lines.length) {
|
|
58
|
+
if (lines[startIndex].length + 1 > maxChars) {
|
|
59
|
+
const line = lines[startIndex];
|
|
60
|
+
const lineNumber = startIndex + 1;
|
|
61
|
+
let offset = 0;
|
|
62
|
+
while (offset < line.length) {
|
|
63
|
+
const remaining = line.length - offset;
|
|
64
|
+
const sliceLength = remaining <= maxChars
|
|
65
|
+
? remaining
|
|
66
|
+
: findLineCut(line.slice(offset), maxChars);
|
|
67
|
+
const slice = line.slice(offset, offset + sliceLength);
|
|
68
|
+
if (slice.trim().length > 0) {
|
|
69
|
+
chunks.push({
|
|
70
|
+
text: slice,
|
|
71
|
+
range: {
|
|
72
|
+
kind: "text",
|
|
73
|
+
startLine: lineNumber,
|
|
74
|
+
endLine: lineNumber,
|
|
75
|
+
startOffset: lineOffsets[startIndex] + offset,
|
|
76
|
+
endOffset: lineOffsets[startIndex] + offset + slice.length,
|
|
77
|
+
},
|
|
78
|
+
});
|
|
79
|
+
}
|
|
80
|
+
offset += sliceLength;
|
|
81
|
+
}
|
|
82
|
+
startIndex++;
|
|
83
|
+
continue;
|
|
84
|
+
}
|
|
85
|
+
let usedChars = 0;
|
|
86
|
+
let endIndex = startIndex;
|
|
87
|
+
while (endIndex < lines.length) {
|
|
88
|
+
const lineLength = lines[endIndex].length + 1;
|
|
89
|
+
if (usedChars + lineLength > maxChars && endIndex > startIndex) {
|
|
90
|
+
break;
|
|
91
|
+
}
|
|
92
|
+
usedChars += lineLength;
|
|
93
|
+
endIndex++;
|
|
94
|
+
}
|
|
95
|
+
const chunkLines = lines.slice(startIndex, endIndex);
|
|
96
|
+
const chunk = chunkLines.join("\n");
|
|
97
|
+
if (chunk.trim().length > 0) {
|
|
98
|
+
const endLineIndex = endIndex - 1;
|
|
99
|
+
chunks.push({
|
|
100
|
+
text: chunk,
|
|
101
|
+
range: {
|
|
102
|
+
kind: "text",
|
|
103
|
+
startLine: startIndex + 1,
|
|
104
|
+
endLine: endIndex,
|
|
105
|
+
startOffset: lineOffsets[startIndex],
|
|
106
|
+
endOffset: lineOffsets[endLineIndex] + lines[endLineIndex].length,
|
|
107
|
+
},
|
|
108
|
+
});
|
|
109
|
+
}
|
|
110
|
+
if (endIndex >= lines.length) {
|
|
111
|
+
break;
|
|
112
|
+
}
|
|
113
|
+
startIndex = computeNextStartLine(lines, startIndex, endIndex, overlapChars);
|
|
114
|
+
}
|
|
115
|
+
return chunks;
|
|
116
|
+
}
|
|
117
|
+
function computeLineOffsets(lines) {
|
|
118
|
+
const offsets = [];
|
|
119
|
+
let offset = 0;
|
|
120
|
+
for (const line of lines) {
|
|
121
|
+
offsets.push(offset);
|
|
122
|
+
offset += line.length + 1;
|
|
123
|
+
}
|
|
124
|
+
return offsets;
|
|
125
|
+
}
|
|
126
|
+
function computeNextStartLine(lines, startIndex, endIndex, overlapChars) {
|
|
127
|
+
if (overlapChars === 0) {
|
|
128
|
+
return endIndex;
|
|
129
|
+
}
|
|
130
|
+
let overlapLines = 0;
|
|
131
|
+
let overlapCount = 0;
|
|
132
|
+
for (let index = endIndex - 1; index >= startIndex && overlapCount < overlapChars; index--) {
|
|
133
|
+
overlapCount += lines[index].length + 1;
|
|
134
|
+
overlapLines++;
|
|
135
|
+
}
|
|
136
|
+
const nextStart = endIndex - overlapLines;
|
|
137
|
+
return nextStart > startIndex ? nextStart : endIndex;
|
|
138
|
+
}
|
|
139
|
+
function findLineCut(line, maxChars) {
|
|
140
|
+
if (line.length <= maxChars) {
|
|
141
|
+
return line.length;
|
|
142
|
+
}
|
|
143
|
+
const minPosition = Math.floor(maxChars * 0.7);
|
|
144
|
+
let bestPosition = -1;
|
|
145
|
+
let bestScore = 0;
|
|
146
|
+
for (let index = minPosition; index < maxChars; index++) {
|
|
147
|
+
const character = line[index];
|
|
148
|
+
let score = 0;
|
|
149
|
+
if (character === "." || character === "!" || character === "?") {
|
|
150
|
+
score = 4;
|
|
151
|
+
}
|
|
152
|
+
else if (character === "," || character === ";" || character === ":") {
|
|
153
|
+
score = 3;
|
|
154
|
+
}
|
|
155
|
+
else if (character === " " || character === "\t") {
|
|
156
|
+
score = 2;
|
|
157
|
+
}
|
|
158
|
+
else if (character === "-" || character === "/" || character === "\\") {
|
|
159
|
+
score = 1;
|
|
160
|
+
}
|
|
161
|
+
if (score > 0 && score >= bestScore) {
|
|
162
|
+
bestScore = score;
|
|
163
|
+
bestPosition = index + 1;
|
|
164
|
+
}
|
|
165
|
+
}
|
|
166
|
+
return bestPosition > 0 ? bestPosition : maxChars;
|
|
167
|
+
}
|
|
168
|
+
//# sourceMappingURL=extractor.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"extractor.js","sourceRoot":"","sources":["../../../../src/engine/extraction/text/extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAE9C,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AACzC,OAAO,EAAE,kBAAkB,EAAgC,MAAM,cAAc,CAAC;AAGhF,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAEtC,MAAM,gCAAgC,GAAG,GAAG,CAAC;AAE7C,MAAM,OAAO,aAAa;IACxB,KAAK,CAAC,OAAO,CACX,MAAc,EACd,UAAwB,EAAE;QAE1B,IAAI,MAAM,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;YAC3B,MAAM,IAAI,WAAW,CAAC,2CAA2C,EAAE;gBACjE,IAAI,EAAE,qDAAqD;gBAC3D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;aAC9D,CAAC,CAAC;QACL,CAAC;QAED,kBAAkB,CAAC,MAAM,CAAC,CAAC;QAC3B,MAAM,EAAE,aAAa,EAAE,iBAAiB,EAAE,GACxC,uBAAuB,CAAC,OAAO,CAAC,CAAC;QAEnC,OAAO,yBAAyB,CAAC,MAAM,EAAE,aAAa,EAAE,iBAAiB,CAAC,CAAC;IAC7E,CAAC;CACF;AAED,SAAS,uBAAuB,CAC9B,OAAqB;IAErB,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,wBAAwB,CAAC;IACxE,MAAM,iBAAiB,GACrB,OAAO,CAAC,iBAAiB,IAAI,gCAAgC,CAAC;IAEhE,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,aAAa,CAAC,IAAI,aAAa,IAAI,CAAC,EAAE,CAAC;QAC3D,MAAM,IAAI,WAAW,CACnB,uDAAuD,EACvD;YACE,IAAI,EAAE,qDAAqD;YAC3D,OAAO,EAAE,iBAAiB,aAAa,EAAE;SAC1C,CACF,CAAC;IACJ,CAAC;IAED,IACE,CAAC,MAAM,CAAC,SAAS,CAAC,iBAAiB,CAAC;QACpC,iBAAiB,GAAG,CAAC;QACrB,iBAAiB,IAAI,aAAa,EAClC,CAAC;QACD,MAAM,IAAI,WAAW,CACnB,+DAA+D,EAC/D;YACE,IAAI,EAAE,wDAAwD;YAC9D,OAAO,EAAE,iBAAiB,aAAa,sBAAsB,iBAAiB,EAAE;SACjF,CACF,CAAC;IACJ,CAAC;IAED,OAAO,EAAE,aAAa,EAAE,iBAAiB,EAAE,CAAC;AAC9C,CAAC;AAED,MAAM,UAAU,yBAAyB,CACvC,MAAkB,EAClB,aAAqB,EACrB,iBAAyB;IAEzB,OAAO,SAAS,CAAC,MAAM,CAAC,IAAI,EAAE,aAAa,EAAE,iBAAiB,CAAC,CAAC,GAAG,CACjE,CAAC,KAAK,EAAE,KAAK,EAAkB,EAAE,CAAC,CAAC;QACjC,EAAE,EAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC;QACvC,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;QACtB,KAAK,EAAE,KAAK,CAAC,KAAK;QAClB,OAAO,EAAE;YACP,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,KAAK,CAAC,IAAI;SACjB;KACF,CAAC,CACH,CAAC;AACJ,CAAC;AAOD,SAAS,SAAS,CAChB,IAAY,EACZ,QAAgB,EAChB,YAAoB;IAEpB,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC7B,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,WAAW,GAAG,kBAAkB,CAAC,KAAK,CAAC,CAAC;IAC9C,MAAM,MAAM,GAAgB,EAAE,CAAC;IAE/B,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,OAAO,UAAU,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QACjC,IAAI,KAAK,CAAC,UAAU,CAAC,CAAC,MAAM,GAAG,CAAC,GAAG,QAAQ,EAAE,CAAC;YAC5C,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC;YAC/B,MAAM,UAAU,GAAG,UAAU,GAAG,CAAC,CAAC;YAClC,IAAI,MAAM,GAAG,CAAC,CAAC;YAEf,OAAO,MAAM,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;gBAC5B,MAAM,SAAS,GAAG,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;gBACvC,MAAM,WAAW,GACf,SAAS,IAAI,QAAQ;oBACnB,CAAC,CAAC,SAAS;oBACX,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,QAAQ,CAAC,CAAC;gBAChD,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,WAAW,CAAC,CAAC;gBAEvD,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAC5B,MAAM,CAAC,IAAI,CAAC;wBACV,IAAI,EAAE,KAAK;wBACX,KAAK,EAAE;4BACL,IAAI,EAAE,MAAM;4BACZ,SAAS,EAAE,UAAU;4BACrB,OAAO,EAAE,UAAU;4BACnB,WAAW,EAAE,WAAW,CAAC,UAAU,CAAC,GAAG,MAAM;4BAC7C,SAAS,EAAE,WAAW,CAAC,UAAU,CAAC,GAAG,MAAM,GAAG,KAAK,CAAC,MAAM;yBAC3D;qBACF,CAAC,CAAC;gBACL,CAAC;gBAED,MAAM,IAAI,WAAW,CAAC;YACxB,CAAC;YAED,UAAU,EAAE,CAAC;YACb,SAAS;QACX,CAAC;QAED,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,QAAQ,GAAG,UAAU,CAAC;QAE1B,OAAO,QAAQ,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC/B,MAAM,UAAU,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YAE9C,IAAI,SAAS,GAAG,UAAU,GAAG,QAAQ,IAAI,QAAQ,GAAG,UAAU,EAAE,CAAC;gBAC/D,MAAM;YACR,CAAC;YAED,SAAS,IAAI,UAAU,CAAC;YACxB,QAAQ,EAAE,CAAC;QACb,CAAC;QAED,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;QACrD,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEpC,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC5B,MAAM,YAAY,GAAG,QAAQ,GAAG,CAAC,CAAC;YAElC,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,KAAK;gBACX,KAAK,EAAE;oBACL,IAAI,EAAE,MAAM;oBACZ,SAAS,EAAE,UAAU,GAAG,CAAC;oBACzB,OAAO,EAAE,QAAQ;oBACjB,WAAW,EAAE,WAAW,CAAC,UAAU,CAAC;oBACpC,SAAS,EAAE,WAAW,CAAC,YAAY,CAAC,GAAG,KAAK,CAAC,YAAY,CAAC,CAAC,MAAM;iBAClE;aACF,CAAC,CAAC;QACL,CAAC;QAED,IAAI,QAAQ,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;YAC7B,MAAM;QACR,CAAC;QAED,UAAU,GAAG,oBAAoB,CAC/B,KAAK,EACL,UAAU,EACV,QAAQ,EACR,YAAY,CACb,CAAC;IACJ,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAwB;IAClD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACrB,MAAM,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;IAC5B,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,oBAAoB,CAC3B,KAAwB,EACxB,UAAkB,EAClB,QAAgB,EAChB,YAAoB;IAEpB,IAAI,YAAY,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,IAAI,YAAY,GAAG,CAAC,CAAC;IACrB,IAAI,YAAY,GAAG,CAAC,CAAC;IAErB,KACE,IAAI,KAAK,GAAG,QAAQ,GAAG,CAAC,EACxB,KAAK,IAAI,UAAU,IAAI,YAAY,GAAG,YAAY,EAClD,KAAK,EAAE,EACP,CAAC;QACD,YAAY,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;QACxC,YAAY,EAAE,CAAC;IACjB,CAAC;IAED,MAAM,SAAS,GAAG,QAAQ,GAAG,YAAY,CAAC;IAE1C,OAAO,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC;AACvD,CAAC;AAED,SAAS,WAAW,CAAC,IAAY,EAAE,QAAgB;IACjD,IAAI,IAAI,CAAC,MAAM,IAAI,QAAQ,EAAE,CAAC;QAC5B,OAAO,IAAI,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,GAAG,GAAG,CAAC,CAAC;IAC/C,IAAI,YAAY,GAAG,CAAC,CAAC,CAAC;IACtB,IAAI,SAAS,GAAG,CAAC,CAAC;IAElB,KAAK,IAAI,KAAK,GAAG,WAAW,EAAE,KAAK,GAAG,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;QACxD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC;QAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YAChE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACvE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACnD,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACxE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;QAED,IAAI,KAAK,GAAG,CAAC,IAAI,KAAK,IAAI,SAAS,EAAE,CAAC;YACpC,SAAS,GAAG,KAAK,CAAC;YAClB,YAAY,GAAG,KAAK,GAAG,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,OAAO,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC;AACpD,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/engine/
|
|
1
|
+
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/engine/extraction/types.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"file-type.js","sourceRoot":"","sources":["../../src/engine/file-type.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAG9C,MAAM,YAAY,GAA2B;IAC3C,IAAI,EAAE,GAAG;IACT,KAAK,EAAE,KAAK;IACZ,MAAM,EAAE,KAAK;IACb,MAAM,EAAE,KAAK;IACb,IAAI,EAAE,KAAK;IACX,MAAM,EAAE,KAAK;IACb,KAAK,EAAE,IAAI;IACX,OAAO,EAAE,MAAM;IACf,KAAK,EAAE,YAAY;IACnB,MAAM,EAAE,KAAK;IACb,MAAM,EAAE,YAAY;IACpB,MAAM,EAAE,YAAY;IACpB,KAAK,EAAE,YAAY;IACnB,MAAM,EAAE,KAAK;IACb,KAAK,EAAE,QAAQ;IACf,KAAK,EAAE,MAAM;IACb,KAAK,EAAE,MAAM;IACb,MAAM,EAAE,KAAK;IACb,QAAQ,EAAE,OAAO;IACjB,KAAK,EAAE,QAAQ;IACf,MAAM,EAAE,QAAQ;IAChB,KAAK,EAAE,QAAQ;IACf,QAAQ,EAAE,OAAO;IACjB,KAAK,EAAE,MAAM;IACb,OAAO,EAAE,MAAM;IACf,MAAM,EAAE,MAAM;IACd,MAAM,EAAE,KAAK;IACb,MAAM,EAAE,KAAK;IACb,OAAO,EAAE,MAAM;IACf,OAAO,EAAE,MAAM;IACf,MAAM,EAAE,KAAK;IACb,SAAS,EAAE,QAAQ;CACpB,CAAC;AAEF,MAAM,YAAY,GAA2B;IAC3C,MAAM,EAAE,KAAK;IACb,OAAO,EAAE,MAAM;IACf,QAAQ,EAAE,MAAM;IAChB,OAAO,EAAE,MAAM;IACf,OAAO,EAAE,MAAM;IACf,MAAM,EAAE,MAAM;CACf,CAAC;AAEF,MAAM,YAAY,GAA2B;IAC3C,KAAK,EAAE,UAAU;IACjB,MAAM,EAAE,UAAU;IAClB,MAAM,EAAE,KAAK;IACb,MAAM,EAAE,MAAM;IACd,OAAO,EAAE,MAAM;IACf,MAAM,EAAE,MAAM;IACd,MAAM,EAAE,KAAK;CACd,CAAC;AAEF,MAAM,aAAa,GAAgC;IACjD,MAAM,EAAE,KAAK;IACb,OAAO,EAAE,MAAM;IACf,MAAM,EAAE,MAAM;IACd,MAAM,EAAE,KAAK;IACb,OAAO,EAAE,MAAM;CAChB,CAAC;AAEF,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC;IAChC,MAAM;IACN,MAAM;IACN,KAAK;IACL,MAAM;IACN,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;IACN,MAAM;IACN,QAAQ;IACR,KAAK;IACL,IAAI;IACJ,IAAI;IACJ,MAAM;IACN,OAAO;IACP,QAAQ;IACR,MAAM;IACN,MAAM;IACN,MAAM;IACN,OAAO;IACP,MAAM;IACN,OAAO;IACP,MAAM;IACN,OAAO;IACP,MAAM;IACN,MAAM;IACN,MAAM;IACN,MAAM;IACN,MAAM;IACN,KAAK;IACL,SAAS;CACV,CAAC,CAAC;AAEH,MAAM,UAAU,cAAc,CAC5B,IAAY;IAEZ,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC;IAE5B,IAAI,IAAI,KAAK,YAAY,EAAE,CAAC;QAC1B,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC;IAChD,CAAC;IAED,IAAI,IAAI,KAAK,UAAU,EAAE,CAAC;QACxB,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC;IAC9C,CAAC;IAED,MAAM,SAAS,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;IAE9C,IAAI,SAAS,IAAI,YAAY,EAAE,CAAC;QAC9B,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,CAAC,SAAS,CAAC,EAAE,CAAC;IAC3D,CAAC;IAED,IAAI,SAAS,IAAI,YAAY,EAAE,CAAC;QAC9B,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,CAAC,SAAS,CAAC,EAAE,CAAC;IAC3D,CAAC;IAED,IAAI,SAAS,IAAI,YAAY,EAAE,CAAC;QAC9B,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,CAAC,SAAS,CAAC,EAAE,CAAC;IAC3D,CAAC;IAED,IAAI,SAAS,IAAI,aAAa,EAAE,CAAC;QAC/B,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,aAAa,CAAC,SAAS,CAAC,EAAE,CAAC;IAC7D,CAAC;IAED,IAAI,iBAAiB,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;QACrC,OAAO,IAAI,CAAC;IACd,CAAC;IAED,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;AAC3E,CAAC"}
|
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
import type { WorkspaceIndexStatus, IndexProgress } from "./types.js";
|
|
2
|
+
export type IndexCompletion = {
|
|
3
|
+
completed: number;
|
|
4
|
+
total: number;
|
|
5
|
+
};
|
|
6
|
+
export declare function indexStatusNeedsRefresh(status: WorkspaceIndexStatus | null | undefined): boolean;
|
|
7
|
+
export declare function indexCompletionFromStatus(status: WorkspaceIndexStatus | null | undefined): IndexCompletion | undefined;
|
|
8
|
+
export declare function mergeIndexCompletion(completion: IndexCompletion | undefined, progress: IndexProgress | undefined): IndexCompletion | undefined;
|
|
9
|
+
export declare function indexCompletionForJob(completion: IndexCompletion | undefined, state: "queued" | "running" | "succeeded" | "failed" | "cancelled" | undefined, progress: IndexProgress | undefined): IndexCompletion | undefined;
|
|
@@ -0,0 +1,38 @@
|
|
|
1
|
+
export function indexStatusNeedsRefresh(status) {
|
|
2
|
+
return Boolean(status &&
|
|
3
|
+
(status.filesAdded > 0 ||
|
|
4
|
+
status.filesModified > 0 ||
|
|
5
|
+
status.filesDeleted > 0 ||
|
|
6
|
+
status.filesPending > 0 ||
|
|
7
|
+
status.filesFailed > 0));
|
|
8
|
+
}
|
|
9
|
+
export function indexCompletionFromStatus(status) {
|
|
10
|
+
if (!status)
|
|
11
|
+
return undefined;
|
|
12
|
+
return {
|
|
13
|
+
completed: status.filesUnchanged,
|
|
14
|
+
total: status.filesScanned,
|
|
15
|
+
};
|
|
16
|
+
}
|
|
17
|
+
export function mergeIndexCompletion(completion, progress) {
|
|
18
|
+
const processed = progress?.filesIndexed;
|
|
19
|
+
const total = progress?.filesTotal;
|
|
20
|
+
const succeeded = processed === undefined
|
|
21
|
+
? undefined
|
|
22
|
+
: Math.max(0, processed - (progress?.filesFailed ?? 0));
|
|
23
|
+
if (!completion) {
|
|
24
|
+
return succeeded === undefined || total === undefined
|
|
25
|
+
? undefined
|
|
26
|
+
: { completed: succeeded, total };
|
|
27
|
+
}
|
|
28
|
+
return {
|
|
29
|
+
completed: Math.min(completion.total, completion.completed + (succeeded ?? 0)),
|
|
30
|
+
total: completion.total,
|
|
31
|
+
};
|
|
32
|
+
}
|
|
33
|
+
export function indexCompletionForJob(completion, state, progress) {
|
|
34
|
+
return state === "running"
|
|
35
|
+
? mergeIndexCompletion(completion, progress)
|
|
36
|
+
: completion;
|
|
37
|
+
}
|
|
38
|
+
//# sourceMappingURL=index-status.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index-status.js","sourceRoot":"","sources":["../../src/engine/index-status.ts"],"names":[],"mappings":"AAOA,MAAM,UAAU,uBAAuB,CACrC,MAA+C;IAE/C,OAAO,OAAO,CACZ,MAAM;QACN,CAAC,MAAM,CAAC,UAAU,GAAG,CAAC;YACpB,MAAM,CAAC,aAAa,GAAG,CAAC;YACxB,MAAM,CAAC,YAAY,GAAG,CAAC;YACvB,MAAM,CAAC,YAAY,GAAG,CAAC;YACvB,MAAM,CAAC,WAAW,GAAG,CAAC,CAAC,CAC1B,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,yBAAyB,CACvC,MAA+C;IAE/C,IAAI,CAAC,MAAM;QAAE,OAAO,SAAS,CAAC;IAC9B,OAAO;QACL,SAAS,EAAE,MAAM,CAAC,cAAc;QAChC,KAAK,EAAE,MAAM,CAAC,YAAY;KAC3B,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,oBAAoB,CAClC,UAAuC,EACvC,QAAmC;IAEnC,MAAM,SAAS,GAAG,QAAQ,EAAE,YAAY,CAAC;IACzC,MAAM,KAAK,GAAG,QAAQ,EAAE,UAAU,CAAC;IACnC,MAAM,SAAS,GACb,SAAS,KAAK,SAAS;QACrB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,GAAG,CAAC,QAAQ,EAAE,WAAW,IAAI,CAAC,CAAC,CAAC,CAAC;IAC5D,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,OAAO,SAAS,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS;YACnD,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,EAAE,SAAS,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACtC,CAAC;IACD,OAAO;QACL,SAAS,EAAE,IAAI,CAAC,GAAG,CACjB,UAAU,CAAC,KAAK,EAChB,UAAU,CAAC,SAAS,GAAG,CAAC,SAAS,IAAI,CAAC,CAAC,CACxC;QACD,KAAK,EAAE,UAAU,CAAC,KAAK;KACxB,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,qBAAqB,CACnC,UAAuC,EACvC,KACyE,EACzE,QAAmC;IAEnC,OAAO,KAAK,KAAK,SAAS;QACxB,CAAC,CAAC,oBAAoB,CAAC,UAAU,EAAE,QAAQ,CAAC;QAC5C,CAAC,CAAC,UAAU,CAAC;AACjB,CAAC"}
|
|
@@ -0,0 +1,12 @@
|
|
|
1
|
+
import type { EmbeddingRuntimeConfig } from "./config.js";
|
|
2
|
+
import type { WorkspaceIndexInfo } from "./types.js";
|
|
3
|
+
export declare const WORKSPACE_MANIFEST_FILE = "manifest.json";
|
|
4
|
+
export declare const CURRENT_MANIFEST_VERSION = 1;
|
|
5
|
+
export type WorkspaceManifest = WorkspaceIndexInfo & {
|
|
6
|
+
manifestVersion: typeof CURRENT_MANIFEST_VERSION;
|
|
7
|
+
embeddingRuntime: EmbeddingRuntimeConfig;
|
|
8
|
+
};
|
|
9
|
+
export declare function workspaceManifestPath(home: string): string;
|
|
10
|
+
export declare function readWorkspaceManifest(home: string): WorkspaceManifest | null;
|
|
11
|
+
export declare function writeWorkspaceManifest(home: string, manifest: WorkspaceManifest): void;
|
|
12
|
+
export declare function workspaceIndexInfoFromManifest(manifest: WorkspaceManifest): WorkspaceIndexInfo;
|
|
@@ -0,0 +1,120 @@
|
|
|
1
|
+
import { join } from "node:path";
|
|
2
|
+
import { EngineError } from "./errors.js";
|
|
3
|
+
import { readJsonFileSync, writeJsonFileSync } from "./utils/json.js";
|
|
4
|
+
export const WORKSPACE_MANIFEST_FILE = "manifest.json";
|
|
5
|
+
export const CURRENT_MANIFEST_VERSION = 1;
|
|
6
|
+
const WORKSPACE_DIRECTORY_MODE = 0o700;
|
|
7
|
+
const WORKSPACE_MANIFEST_MODE = 0o600;
|
|
8
|
+
export function workspaceManifestPath(home) {
|
|
9
|
+
return join(home, WORKSPACE_MANIFEST_FILE);
|
|
10
|
+
}
|
|
11
|
+
export function readWorkspaceManifest(home) {
|
|
12
|
+
const path = workspaceManifestPath(home);
|
|
13
|
+
const value = readJsonFileSync(path, null);
|
|
14
|
+
if (value === null) {
|
|
15
|
+
return null;
|
|
16
|
+
}
|
|
17
|
+
if (!isWorkspaceManifest(value)) {
|
|
18
|
+
throw new EngineError("Workspace index manifest is invalid", {
|
|
19
|
+
code: "ZVEC_GREP.ENGINE.MANIFEST.INVALID",
|
|
20
|
+
context: `path=${path}`,
|
|
21
|
+
});
|
|
22
|
+
}
|
|
23
|
+
return value;
|
|
24
|
+
}
|
|
25
|
+
export function writeWorkspaceManifest(home, manifest) {
|
|
26
|
+
writeJsonFileSync(workspaceManifestPath(home), manifest, {
|
|
27
|
+
directoryMode: WORKSPACE_DIRECTORY_MODE,
|
|
28
|
+
fileMode: WORKSPACE_MANIFEST_MODE,
|
|
29
|
+
});
|
|
30
|
+
}
|
|
31
|
+
export function workspaceIndexInfoFromManifest(manifest) {
|
|
32
|
+
return {
|
|
33
|
+
id: manifest.id,
|
|
34
|
+
name: manifest.name,
|
|
35
|
+
path: manifest.path,
|
|
36
|
+
rootPaths: manifest.rootPaths,
|
|
37
|
+
indexPolicy: manifest.indexPolicy,
|
|
38
|
+
embedding: manifest.embedding,
|
|
39
|
+
indexVersion: manifest.indexVersion,
|
|
40
|
+
createdTime: manifest.createdTime,
|
|
41
|
+
updatedTime: manifest.updatedTime,
|
|
42
|
+
};
|
|
43
|
+
}
|
|
44
|
+
function isWorkspaceManifest(value) {
|
|
45
|
+
if (!isRecord(value) || value.manifestVersion !== CURRENT_MANIFEST_VERSION) {
|
|
46
|
+
return false;
|
|
47
|
+
}
|
|
48
|
+
return (isNonEmptyString(value.id) &&
|
|
49
|
+
isNonEmptyString(value.name) &&
|
|
50
|
+
isNonEmptyString(value.path) &&
|
|
51
|
+
Array.isArray(value.rootPaths) &&
|
|
52
|
+
value.rootPaths.length > 0 &&
|
|
53
|
+
value.rootPaths.every(isRootPath) &&
|
|
54
|
+
(value.indexPolicy === "enabled" || value.indexPolicy === "disabled") &&
|
|
55
|
+
(value.embedding === null || isEmbeddingSchema(value.embedding)) &&
|
|
56
|
+
(value.indexVersion === null || Number.isInteger(value.indexVersion)) &&
|
|
57
|
+
typeof value.createdTime === "number" &&
|
|
58
|
+
Number.isFinite(value.createdTime) &&
|
|
59
|
+
typeof value.updatedTime === "number" &&
|
|
60
|
+
Number.isFinite(value.updatedTime) &&
|
|
61
|
+
isEmbeddingRuntime(value.embeddingRuntime));
|
|
62
|
+
}
|
|
63
|
+
function isRootPath(value) {
|
|
64
|
+
return (isRecord(value) &&
|
|
65
|
+
isNonEmptyString(value.absolutePath) &&
|
|
66
|
+
typeof value.recursive === "boolean" &&
|
|
67
|
+
isOptionalStringArray(value.include) &&
|
|
68
|
+
isOptionalStringArray(value.exclude) &&
|
|
69
|
+
isOptionalStringArray(value.globs) &&
|
|
70
|
+
isOptionalStringArray(value.insensitiveGlobs) &&
|
|
71
|
+
isOptionalStringArray(value.fileTypes) &&
|
|
72
|
+
isOptionalStringArray(value.excludedFileTypes) &&
|
|
73
|
+
isOptionalBoolean(value.hidden) &&
|
|
74
|
+
isOptionalBoolean(value.noIgnore) &&
|
|
75
|
+
isOptionalStringArray(value.ignoreFiles) &&
|
|
76
|
+
isOptionalNonNegativeInteger(value.maxDepth) &&
|
|
77
|
+
isOptionalNonNegativeInteger(value.maxFileSizeBytes) &&
|
|
78
|
+
isOptionalBoolean(value.follow));
|
|
79
|
+
}
|
|
80
|
+
function isEmbeddingSchema(value) {
|
|
81
|
+
return (isRecord(value) &&
|
|
82
|
+
isNonEmptyString(value.provider) &&
|
|
83
|
+
isNonEmptyString(value.model) &&
|
|
84
|
+
Number.isInteger(value.dimension) &&
|
|
85
|
+
typeof value.dimension === "number" &&
|
|
86
|
+
value.dimension > 0 &&
|
|
87
|
+
(value.metric === "cosine" ||
|
|
88
|
+
value.metric === "dot" ||
|
|
89
|
+
value.metric === "euclidean"));
|
|
90
|
+
}
|
|
91
|
+
function isEmbeddingRuntime(value) {
|
|
92
|
+
if (!isRecord(value)) {
|
|
93
|
+
return false;
|
|
94
|
+
}
|
|
95
|
+
return ((value.apiKey === undefined || typeof value.apiKey === "string") &&
|
|
96
|
+
(value.endpoint === undefined || typeof value.endpoint === "string") &&
|
|
97
|
+
(value.device === undefined ||
|
|
98
|
+
value.device === "auto" ||
|
|
99
|
+
value.device === "cpu" ||
|
|
100
|
+
value.device === "metal" ||
|
|
101
|
+
value.device === "vulkan" ||
|
|
102
|
+
value.device === "cuda"));
|
|
103
|
+
}
|
|
104
|
+
function isOptionalStringArray(value) {
|
|
105
|
+
return (value === undefined ||
|
|
106
|
+
(Array.isArray(value) && value.every((item) => typeof item === "string")));
|
|
107
|
+
}
|
|
108
|
+
function isOptionalBoolean(value) {
|
|
109
|
+
return value === undefined || typeof value === "boolean";
|
|
110
|
+
}
|
|
111
|
+
function isOptionalNonNegativeInteger(value) {
|
|
112
|
+
return value === undefined || (Number.isInteger(value) && Number(value) >= 0);
|
|
113
|
+
}
|
|
114
|
+
function isRecord(value) {
|
|
115
|
+
return typeof value === "object" && value !== null && !Array.isArray(value);
|
|
116
|
+
}
|
|
117
|
+
function isNonEmptyString(value) {
|
|
118
|
+
return typeof value === "string" && value.length > 0;
|
|
119
|
+
}
|
|
120
|
+
//# sourceMappingURL=manifest.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"manifest.js","sourceRoot":"","sources":["../../src/engine/manifest.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AAEjC,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAE1C,OAAO,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,MAAM,iBAAiB,CAAC;AAEtE,MAAM,CAAC,MAAM,uBAAuB,GAAG,eAAe,CAAC;AACvD,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,CAAC;AAE1C,MAAM,wBAAwB,GAAG,KAAK,CAAC;AACvC,MAAM,uBAAuB,GAAG,KAAK,CAAC;AAOtC,MAAM,UAAU,qBAAqB,CAAC,IAAY;IAChD,OAAO,IAAI,CAAC,IAAI,EAAE,uBAAuB,CAAC,CAAC;AAC7C,CAAC;AAED,MAAM,UAAU,qBAAqB,CAAC,IAAY;IAChD,MAAM,IAAI,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAC;IACzC,MAAM,KAAK,GAAG,gBAAgB,CAAU,IAAI,EAAE,IAAI,CAAC,CAAC;IACpD,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QACnB,OAAO,IAAI,CAAC;IACd,CAAC;IAED,IAAI,CAAC,mBAAmB,CAAC,KAAK,CAAC,EAAE,CAAC;QAChC,MAAM,IAAI,WAAW,CAAC,qCAAqC,EAAE;YAC3D,IAAI,EAAE,mCAAmC;YACzC,OAAO,EAAE,QAAQ,IAAI,EAAE;SACxB,CAAC,CAAC;IACL,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED,MAAM,UAAU,sBAAsB,CACpC,IAAY,EACZ,QAA2B;IAE3B,iBAAiB,CAAC,qBAAqB,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE;QACvD,aAAa,EAAE,wBAAwB;QACvC,QAAQ,EAAE,uBAAuB;KAClC,CAAC,CAAC;AACL,CAAC;AAED,MAAM,UAAU,8BAA8B,CAC5C,QAA2B;IAE3B,OAAO;QACL,EAAE,EAAE,QAAQ,CAAC,EAAE;QACf,IAAI,EAAE,QAAQ,CAAC,IAAI;QACnB,IAAI,EAAE,QAAQ,CAAC,IAAI;QACnB,SAAS,EAAE,QAAQ,CAAC,SAAS;QAC7B,WAAW,EAAE,QAAQ,CAAC,WAAW;QACjC,SAAS,EAAE,QAAQ,CAAC,SAAS;QAC7B,YAAY,EAAE,QAAQ,CAAC,YAAY;QACnC,WAAW,EAAE,QAAQ,CAAC,WAAW;QACjC,WAAW,EAAE,QAAQ,CAAC,WAAW;KAClC,CAAC;AACJ,CAAC;AAED,SAAS,mBAAmB,CAAC,KAAc;IACzC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,eAAe,KAAK,wBAAwB,EAAE,CAAC;QAC3E,OAAO,KAAK,CAAC;IACf,CAAC;IAED,OAAO,CACL,gBAAgB,CAAC,KAAK,CAAC,EAAE,CAAC;QAC1B,gBAAgB,CAAC,KAAK,CAAC,IAAI,CAAC;QAC5B,gBAAgB,CAAC,KAAK,CAAC,IAAI,CAAC;QAC5B,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC;QAC9B,KAAK,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC;QAC1B,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,UAAU,CAAC;QACjC,CAAC,KAAK,CAAC,WAAW,KAAK,SAAS,IAAI,KAAK,CAAC,WAAW,KAAK,UAAU,CAAC;QACrE,CAAC,KAAK,CAAC,SAAS,KAAK,IAAI,IAAI,iBAAiB,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;QAChE,CAAC,KAAK,CAAC,YAAY,KAAK,IAAI,IAAI,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC;QACrE,OAAO,KAAK,CAAC,WAAW,KAAK,QAAQ;QACrC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,WAAW,CAAC;QAClC,OAAO,KAAK,CAAC,WAAW,KAAK,QAAQ;QACrC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,WAAW,CAAC;QAClC,kBAAkB,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAC3C,CAAC;AACJ,CAAC;AAED,SAAS,UAAU,CAAC,KAAc;IAChC,OAAO,CACL,QAAQ,CAAC,KAAK,CAAC;QACf,gBAAgB,CAAC,KAAK,CAAC,YAAY,CAAC;QACpC,OAAO,KAAK,CAAC,SAAS,KAAK,SAAS;QACpC,qBAAqB,CAAC,KAAK,CAAC,OAAO,CAAC;QACpC,qBAAqB,CAAC,KAAK,CAAC,OAAO,CAAC;QACpC,qBAAqB,CAAC,KAAK,CAAC,KAAK,CAAC;QAClC,qBAAqB,CAAC,KAAK,CAAC,gBAAgB,CAAC;QAC7C,qBAAqB,CAAC,KAAK,CAAC,SAAS,CAAC;QACtC,qBAAqB,CAAC,KAAK,CAAC,iBAAiB,CAAC;QAC9C,iBAAiB,CAAC,KAAK,CAAC,MAAM,CAAC;QAC/B,iBAAiB,CAAC,KAAK,CAAC,QAAQ,CAAC;QACjC,qBAAqB,CAAC,KAAK,CAAC,WAAW,CAAC;QACxC,4BAA4B,CAAC,KAAK,CAAC,QAAQ,CAAC;QAC5C,4BAA4B,CAAC,KAAK,CAAC,gBAAgB,CAAC;QACpD,iBAAiB,CAAC,KAAK,CAAC,MAAM,CAAC,CAChC,CAAC;AACJ,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc;IACvC,OAAO,CACL,QAAQ,CAAC,KAAK,CAAC;QACf,gBAAgB,CAAC,KAAK,CAAC,QAAQ,CAAC;QAChC,gBAAgB,CAAC,KAAK,CAAC,KAAK,CAAC;QAC7B,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,SAAS,CAAC;QACjC,OAAO,KAAK,CAAC,SAAS,KAAK,QAAQ;QACnC,KAAK,CAAC,SAAS,GAAG,CAAC;QACnB,CAAC,KAAK,CAAC,MAAM,KAAK,QAAQ;YACxB,KAAK,CAAC,MAAM,KAAK,KAAK;YACtB,KAAK,CAAC,MAAM,KAAK,WAAW,CAAC,CAChC,CAAC;AACJ,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAc;IACxC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACrB,OAAO,KAAK,CAAC;IACf,CAAC;IAED,OAAO,CACL,CAAC,KAAK,CAAC,MAAM,KAAK,SAAS,IAAI,OAAO,KAAK,CAAC,MAAM,KAAK,QAAQ,CAAC;QAChE,CAAC,KAAK,CAAC,QAAQ,KAAK,SAAS,IAAI,OAAO,KAAK,CAAC,QAAQ,KAAK,QAAQ,CAAC;QACpE,CAAC,KAAK,CAAC,MAAM,KAAK,SAAS;YACzB,KAAK,CAAC,MAAM,KAAK,MAAM;YACvB,KAAK,CAAC,MAAM,KAAK,KAAK;YACtB,KAAK,CAAC,MAAM,KAAK,OAAO;YACxB,KAAK,CAAC,MAAM,KAAK,QAAQ;YACzB,KAAK,CAAC,MAAM,KAAK,MAAM,CAAC,CAC3B,CAAC;AACJ,CAAC;AAED,SAAS,qBAAqB,CAAC,KAAc;IAC3C,OAAO,CACL,KAAK,KAAK,SAAS;QACnB,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAC1E,CAAC;AACJ,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc;IACvC,OAAO,KAAK,KAAK,SAAS,IAAI,OAAO,KAAK,KAAK,SAAS,CAAC;AAC3D,CAAC;AAED,SAAS,4BAA4B,CAAC,KAAc;IAClD,OAAO,KAAK,KAAK,SAAS,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;AAChF,CAAC;AAED,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;AAC9E,CAAC;AAED,SAAS,gBAAgB,CAAC,KAAc;IACtC,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;AACvD,CAAC"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import type { Content } from "
|
|
2
|
-
import {
|
|
3
|
-
import type { LlamaCppEmbeddingCatalogEntry
|
|
1
|
+
import type { Content } from "../../types.js";
|
|
2
|
+
import { BaseEmbeddingModel, type CreateEmbeddingModelOptions, type EmbeddingModelInfo, type EmbeddingResult, type NormalizedEmbeddingOptions } from "../embeddings.js";
|
|
3
|
+
import type { LlamaCppEmbeddingCatalogEntry } from "../catalog.js";
|
|
4
4
|
type LlamaEmbedding = {
|
|
5
5
|
vector: ArrayLike<number>;
|
|
6
6
|
};
|
|
@@ -30,35 +30,38 @@ type Llama = {
|
|
|
30
30
|
}): Promise<LlamaModel>;
|
|
31
31
|
dispose?(): Promise<void> | void;
|
|
32
32
|
};
|
|
33
|
+
type LlamaGpuSelection = Exclude<NonNullable<CreateEmbeddingModelOptions["device"]>, "cpu"> | false;
|
|
33
34
|
type NodeLlamaCppModule = {
|
|
34
35
|
getLlama(options: Record<string, unknown>): Promise<Llama>;
|
|
35
36
|
resolveModelFile(model: string, options: {
|
|
36
37
|
directory: string;
|
|
37
38
|
cli?: boolean;
|
|
39
|
+
onProgress?: (status: {
|
|
40
|
+
totalSize: number;
|
|
41
|
+
downloadedSize: number;
|
|
42
|
+
}) => void;
|
|
38
43
|
}): Promise<string>;
|
|
39
44
|
LlamaLogLevel?: {
|
|
40
45
|
error?: unknown;
|
|
41
46
|
};
|
|
42
47
|
};
|
|
43
48
|
type NodeLlamaCppLoader = () => Promise<NodeLlamaCppModule>;
|
|
44
|
-
|
|
45
|
-
|
|
49
|
+
type LlamaCppRuntimeState = {
|
|
50
|
+
failedGpuInitModes: Set<LlamaGpuSelection>;
|
|
51
|
+
cpuCompatibleFallbackWarningShown: boolean;
|
|
52
|
+
};
|
|
53
|
+
type LlamaCppDependencies = {
|
|
54
|
+
loadRuntime: NodeLlamaCppLoader;
|
|
55
|
+
runtimeState: LlamaCppRuntimeState;
|
|
56
|
+
};
|
|
57
|
+
export declare class LlamaCppEmbeddingModel extends BaseEmbeddingModel {
|
|
46
58
|
private readonly entry;
|
|
47
|
-
readonly
|
|
48
|
-
readonly provider: "local";
|
|
49
|
-
readonly model: string;
|
|
50
|
-
};
|
|
51
|
-
readonly dimension: number;
|
|
52
|
-
readonly metric: import("../../types.js").VectorMetric;
|
|
53
|
-
readonly supportedContentKinds: readonly ["text"];
|
|
54
|
-
readonly limits: {
|
|
55
|
-
readonly maxBatchSize: number;
|
|
56
|
-
readonly maxInputTokens: number;
|
|
57
|
-
};
|
|
58
|
-
readonly recommendedIndexConcurrency = 1;
|
|
59
|
+
readonly info: EmbeddingModelInfo;
|
|
59
60
|
private readonly modelCacheDir;
|
|
60
61
|
private readonly gpu;
|
|
61
|
-
private readonly
|
|
62
|
+
private readonly parallelism?;
|
|
63
|
+
private readonly dependencies;
|
|
64
|
+
private runtimeImport;
|
|
62
65
|
private llama;
|
|
63
66
|
private model;
|
|
64
67
|
private contexts;
|
|
@@ -67,11 +70,13 @@ export declare class LlamaCppEmbeddingModel extends EmbeddingModel {
|
|
|
67
70
|
private contextsCreatePromise;
|
|
68
71
|
private usingCpuFallback;
|
|
69
72
|
private disposed;
|
|
70
|
-
constructor(entry: LlamaCppEmbeddingCatalogEntry, options:
|
|
71
|
-
protected doEmbed(contents: readonly Content[], options:
|
|
73
|
+
constructor(entry: LlamaCppEmbeddingCatalogEntry, options: CreateEmbeddingModelOptions, dependencies?: Partial<LlamaCppDependencies>);
|
|
74
|
+
protected doEmbed(contents: readonly Content[], options: NormalizedEmbeddingOptions): Promise<EmbeddingResult>;
|
|
75
|
+
private embedBatch;
|
|
72
76
|
dispose(): Promise<void>;
|
|
73
77
|
private embedTexts;
|
|
74
78
|
private ensureLlama;
|
|
79
|
+
private loadRuntime;
|
|
75
80
|
private loadLlamaWithFallback;
|
|
76
81
|
private loadCpuCompatibleLlama;
|
|
77
82
|
private ensureModel;
|