@zvec/zvec-grep 0.1.5 → 0.1.7
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 +154 -236
- package/README_CN.md +142 -233
- 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 +338 -55
- package/dist/cli/args.js.map +1 -1
- package/dist/cli/auth.d.ts +21 -0
- package/dist/cli/auth.js +209 -0
- package/dist/cli/auth.js.map +1 -0
- package/dist/cli/commands.js +469 -625
- 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 +4 -0
- package/dist/cli/format/context.js +311 -44
- package/dist/cli/format/context.js.map +1 -1
- package/dist/cli/format/debug.js +12 -6
- 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 +79 -6
- package/dist/cli/format/status.js +341 -111
- package/dist/cli/format/status.js.map +1 -1
- package/dist/cli/help.js +395 -63
- package/dist/cli/help.js.map +1 -1
- package/dist/cli/index.d.ts +1 -1
- package/dist/cli/index.js +1 -1
- package/dist/cli/install.d.ts +4 -0
- package/dist/cli/install.js +963 -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/server-search.d.ts +3 -0
- package/dist/cli/server-search.js +12 -0
- package/dist/cli/server-search.js.map +1 -0
- package/dist/cli/types.d.ts +23 -6
- package/dist/cli/types.js.map +1 -1
- package/dist/client/daemon-client.d.ts +31 -0
- package/dist/client/daemon-client.js +203 -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 +77 -0
- package/dist/daemon/backend.js +1087 -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 +288 -0
- package/dist/daemon/http-server.js.map +1 -0
- package/dist/daemon/index-coordinator.d.ts +24 -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 +46 -0
- package/dist/daemon/runtime-manager.js +246 -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 +254 -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/code-formats.d.ts +3 -0
- package/dist/engine/code-formats.js +14 -0
- package/dist/engine/code-formats.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/adapter.js +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 +116 -48
- 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 +4 -44
- package/dist/engine/extraction/index.js +2 -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 +38 -17
- 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/extraction/vector-content.d.ts +5 -0
- package/dist/engine/extraction/vector-content.js +95 -0
- package/dist/engine/extraction/vector-content.js.map +1 -0
- package/dist/engine/file-size-policy.d.ts +6 -0
- package/dist/engine/file-size-policy.js +20 -0
- package/dist/engine/file-size-policy.js.map +1 -0
- package/dist/engine/file-type.d.ts +15 -0
- package/dist/engine/file-type.js +152 -0
- 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 +13 -0
- package/dist/engine/manifest.js +124 -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-runtime.d.ts +63 -0
- package/dist/engine/models/backends/model2vec-runtime.js +112 -0
- package/dist/engine/models/backends/model2vec-runtime.js.map +1 -0
- package/dist/engine/models/backends/model2vec-tokenizer.d.ts +2 -0
- package/dist/engine/models/backends/model2vec-tokenizer.js +42 -0
- package/dist/engine/models/backends/model2vec-tokenizer.js.map +1 -0
- package/dist/engine/models/backends/model2vec-worker-pool.d.ts +25 -0
- package/dist/engine/models/backends/model2vec-worker-pool.js +264 -0
- package/dist/engine/models/backends/model2vec-worker-pool.js.map +1 -0
- package/dist/engine/models/backends/model2vec-worker.d.ts +1 -0
- package/dist/engine/models/backends/model2vec-worker.js +41 -0
- package/dist/engine/models/backends/model2vec-worker.js.map +1 -0
- package/dist/engine/models/backends/model2vec.d.ts +40 -0
- package/dist/engine/models/backends/model2vec.js +311 -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 +207 -23
- package/dist/engine/models/catalog.js +191 -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 +59 -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 +409 -165
- package/dist/engine/pipeline/indexing/index.js.map +1 -1
- package/dist/engine/pipeline/indexing/input-budget.d.ts +2 -0
- package/dist/engine/pipeline/indexing/input-budget.js +51 -0
- package/dist/engine/pipeline/indexing/input-budget.js.map +1 -0
- 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 +9 -2
- package/dist/engine/pipeline/indexing/scanner/index.js +291 -35
- 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 +10 -12
- package/dist/engine/service/root.js +29 -25
- package/dist/engine/service/root.js.map +1 -1
- package/dist/engine/service/structure-enrichment.d.ts +1 -1
- package/dist/engine/service/structure-enrichment.js +20 -22
- package/dist/engine/service/structure-enrichment.js.map +1 -1
- package/dist/engine/service/types.d.ts +40 -38
- package/dist/engine/service/workspace-index.d.ts +26 -0
- package/dist/engine/service/workspace-index.js +157 -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 +698 -489
- package/dist/engine/service/zvec-grep.js.map +1 -1
- package/dist/engine/storage/index.d.ts +30 -20
- package/dist/engine/storage/index.js +2 -1
- package/dist/engine/storage/index.js.map +1 -1
- package/dist/engine/storage/layout.d.ts +10 -0
- package/dist/engine/storage/layout.js +29 -0
- package/dist/engine/storage/layout.js.map +1 -0
- package/dist/engine/storage/zvec.d.ts +4 -67
- package/dist/engine/storage/zvec.js +115 -85
- 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.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 +104 -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 +672 -0
- package/dist/mcp/schemas.js +506 -0
- package/dist/mcp/schemas.js.map +1 -0
- package/dist/mcp/stdio-bridge.d.ts +9 -0
- package/dist/mcp/stdio-bridge.js +131 -0
- package/dist/mcp/stdio-bridge.js.map +1 -0
- package/dist/mcp/tools.d.ts +153 -0
- package/dist/mcp/tools.js +482 -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/dist/prompts/zvec-grep-guidance.d.ts +2 -0
- package/dist/prompts/zvec-grep-guidance.js +9 -0
- package/dist/prompts/zvec-grep-guidance.js.map +1 -0
- package/docs/01-agents.md +132 -0
- package/docs/02-cli.md +239 -0
- package/docs/03-mcp.md +185 -0
- package/docs/04-pipeline.md +226 -0
- package/docs/05-architecture.md +86 -0
- package/docs/06-server.md +215 -0
- package/docs/07-embedding.md +182 -0
- package/docs/08-roadmap.md +85 -0
- package/docs/README.md +53 -0
- package/package.json +20 -11
- 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.d.ts +0 -5
- package/dist/engine/files/file-type.js +0 -119
- 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/skills/zvec-grep/SKILL.md +0 -151
- package/skills/zvec-grep/agents/openai.yaml +0 -4
- /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
|
@@ -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 +1,5 @@
|
|
|
1
|
+
import type { Content, EntityFragment, EntityMetadata } from "../types.js";
|
|
2
|
+
import type { ChunkOptions } from "./types.js";
|
|
3
|
+
export declare function vectorContentForFragment(fragment: EntityFragment, maxChars?: number): Content;
|
|
4
|
+
export declare function chunkOptionsForMetadata(options: Required<ChunkOptions>, metadata: EntityMetadata | undefined): Required<ChunkOptions>;
|
|
5
|
+
export declare function fitTextToChars(value: string, maxChars: number): string;
|
|
@@ -0,0 +1,95 @@
|
|
|
1
|
+
import { EngineError } from "../errors.js";
|
|
2
|
+
const MAX_METADATA_BUDGET_RATIO = 0.25;
|
|
3
|
+
export function vectorContentForFragment(fragment, maxChars) {
|
|
4
|
+
if (fragment.content.kind !== "text") {
|
|
5
|
+
return fragment.content;
|
|
6
|
+
}
|
|
7
|
+
const metadata = vectorMetadataText(fragment.metadata, metadataBudget(maxChars));
|
|
8
|
+
if (metadata.length === 0) {
|
|
9
|
+
return checkedVectorContent(fragment, fragment.content, maxChars);
|
|
10
|
+
}
|
|
11
|
+
return checkedVectorContent(fragment, {
|
|
12
|
+
kind: "text",
|
|
13
|
+
text: `${metadata}\n${fragment.content.text}`,
|
|
14
|
+
}, maxChars);
|
|
15
|
+
}
|
|
16
|
+
function checkedVectorContent(fragment, content, maxChars) {
|
|
17
|
+
if (content.kind === "text" &&
|
|
18
|
+
maxChars !== undefined &&
|
|
19
|
+
content.text.length > maxChars) {
|
|
20
|
+
throw new EngineError("Embedding content exceeds the estimated input budget", {
|
|
21
|
+
code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMBEDDING_CONTENT_TOO_LARGE",
|
|
22
|
+
context: `fragmentId=${fragment.id} contentChars=${content.text.length} maxChars=${maxChars}`,
|
|
23
|
+
});
|
|
24
|
+
}
|
|
25
|
+
return content;
|
|
26
|
+
}
|
|
27
|
+
export function chunkOptionsForMetadata(options, metadata) {
|
|
28
|
+
const metadataText = vectorMetadataText(metadata, metadataBudget(options.maxChunkChars));
|
|
29
|
+
const separatorChars = metadataText.length > 0 ? 1 : 0;
|
|
30
|
+
const maxChunkChars = Math.max(1, options.maxChunkChars - metadataText.length - separatorChars);
|
|
31
|
+
const chunkOverlapChars = Math.min(options.chunkOverlapChars, Math.max(0, maxChunkChars - 1));
|
|
32
|
+
return { maxChunkChars, chunkOverlapChars };
|
|
33
|
+
}
|
|
34
|
+
export function fitTextToChars(value, maxChars) {
|
|
35
|
+
if (value.length <= maxChars) {
|
|
36
|
+
return value;
|
|
37
|
+
}
|
|
38
|
+
if (maxChars <= 3) {
|
|
39
|
+
return ".".repeat(Math.max(0, maxChars));
|
|
40
|
+
}
|
|
41
|
+
const rawEnd = maxChars - 3;
|
|
42
|
+
const end = rawEnd > 0 &&
|
|
43
|
+
rawEnd < value.length &&
|
|
44
|
+
isHighSurrogate(value.charCodeAt(rawEnd - 1)) &&
|
|
45
|
+
isLowSurrogate(value.charCodeAt(rawEnd))
|
|
46
|
+
? rawEnd - 1
|
|
47
|
+
: rawEnd;
|
|
48
|
+
return `${value.slice(0, end).trimEnd()}...`;
|
|
49
|
+
}
|
|
50
|
+
function vectorMetadataText(metadata, maxChars) {
|
|
51
|
+
if (!metadata) {
|
|
52
|
+
return "";
|
|
53
|
+
}
|
|
54
|
+
if (metadata.kind === "code") {
|
|
55
|
+
return compactMetadataLines([
|
|
56
|
+
metadata.symbolName
|
|
57
|
+
? `symbol: ${metadata.symbolType} ${metadata.symbolName}`
|
|
58
|
+
: `symbol: ${metadata.symbolType}`,
|
|
59
|
+
metadata.scope ? `scope: ${metadata.scope}` : null,
|
|
60
|
+
metadata.signature ? `signature: ${oneLine(metadata.signature)}` : null,
|
|
61
|
+
metadata.modifiers.length > 0
|
|
62
|
+
? `modifiers: ${metadata.modifiers.join(" ")}`
|
|
63
|
+
: null,
|
|
64
|
+
metadata.doc ? `doc: ${oneLine(metadata.doc)}` : null,
|
|
65
|
+
], maxChars);
|
|
66
|
+
}
|
|
67
|
+
return compactMetadataLines([
|
|
68
|
+
metadata.heading ? `heading: ${metadata.heading}` : null,
|
|
69
|
+
typeof metadata.level === "number"
|
|
70
|
+
? `heading_level: ${metadata.level}`
|
|
71
|
+
: null,
|
|
72
|
+
metadata.scope ? `scope: ${metadata.scope}` : null,
|
|
73
|
+
], maxChars);
|
|
74
|
+
}
|
|
75
|
+
function metadataBudget(maxChars) {
|
|
76
|
+
return maxChars === undefined
|
|
77
|
+
? undefined
|
|
78
|
+
: Math.max(0, Math.floor(maxChars * MAX_METADATA_BUDGET_RATIO));
|
|
79
|
+
}
|
|
80
|
+
function compactMetadataLines(lines, maxChars) {
|
|
81
|
+
const text = lines
|
|
82
|
+
.filter((line) => line !== null && line.trim().length > 0)
|
|
83
|
+
.join("\n");
|
|
84
|
+
return maxChars === undefined ? text : fitTextToChars(text, maxChars);
|
|
85
|
+
}
|
|
86
|
+
function oneLine(value) {
|
|
87
|
+
return value.replace(/\s+/g, " ").trim();
|
|
88
|
+
}
|
|
89
|
+
function isHighSurrogate(value) {
|
|
90
|
+
return value >= 0xd800 && value <= 0xdbff;
|
|
91
|
+
}
|
|
92
|
+
function isLowSurrogate(value) {
|
|
93
|
+
return value >= 0xdc00 && value <= 0xdfff;
|
|
94
|
+
}
|
|
95
|
+
//# sourceMappingURL=vector-content.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"vector-content.js","sourceRoot":"","sources":["../../../src/engine/extraction/vector-content.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAI3C,MAAM,yBAAyB,GAAG,IAAI,CAAC;AAEvC,MAAM,UAAU,wBAAwB,CACtC,QAAwB,EACxB,QAAiB;IAEjB,IAAI,QAAQ,CAAC,OAAO,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;QACrC,OAAO,QAAQ,CAAC,OAAO,CAAC;IAC1B,CAAC;IAED,MAAM,QAAQ,GAAG,kBAAkB,CACjC,QAAQ,CAAC,QAAQ,EACjB,cAAc,CAAC,QAAQ,CAAC,CACzB,CAAC;IACF,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC1B,OAAO,oBAAoB,CAAC,QAAQ,EAAE,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;IACpE,CAAC;IAED,OAAO,oBAAoB,CACzB,QAAQ,EACR;QACE,IAAI,EAAE,MAAM;QACZ,IAAI,EAAE,GAAG,QAAQ,KAAK,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE;KAC9C,EACD,QAAQ,CACT,CAAC;AACJ,CAAC;AAED,SAAS,oBAAoB,CAC3B,QAAwB,EACxB,OAAgB,EAChB,QAA4B;IAE5B,IACE,OAAO,CAAC,IAAI,KAAK,MAAM;QACvB,QAAQ,KAAK,SAAS;QACtB,OAAO,CAAC,IAAI,CAAC,MAAM,GAAG,QAAQ,EAC9B,CAAC;QACD,MAAM,IAAI,WAAW,CACnB,sDAAsD,EACtD;YACE,IAAI,EAAE,yDAAyD;YAC/D,OAAO,EAAE,cAAc,QAAQ,CAAC,EAAE,iBAAiB,OAAO,CAAC,IAAI,CAAC,MAAM,aAAa,QAAQ,EAAE;SAC9F,CACF,CAAC;IACJ,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,UAAU,uBAAuB,CACrC,OAA+B,EAC/B,QAAoC;IAEpC,MAAM,YAAY,GAAG,kBAAkB,CACrC,QAAQ,EACR,cAAc,CAAC,OAAO,CAAC,aAAa,CAAC,CACtC,CAAC;IACF,MAAM,cAAc,GAAG,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACvD,MAAM,aAAa,GAAG,IAAI,CAAC,GAAG,CAC5B,CAAC,EACD,OAAO,CAAC,aAAa,GAAG,YAAY,CAAC,MAAM,GAAG,cAAc,CAC7D,CAAC;IACF,MAAM,iBAAiB,GAAG,IAAI,CAAC,GAAG,CAChC,OAAO,CAAC,iBAAiB,EACzB,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,GAAG,CAAC,CAAC,CAC/B,CAAC;IAEF,OAAO,EAAE,aAAa,EAAE,iBAAiB,EAAE,CAAC;AAC9C,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,KAAa,EAAE,QAAgB;IAC5D,IAAI,KAAK,CAAC,MAAM,IAAI,QAAQ,EAAE,CAAC;QAC7B,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;QAClB,OAAO,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAC;IAC3C,CAAC;IACD,MAAM,MAAM,GAAG,QAAQ,GAAG,CAAC,CAAC;IAC5B,MAAM,GAAG,GACP,MAAM,GAAG,CAAC;QACV,MAAM,GAAG,KAAK,CAAC,MAAM;QACrB,eAAe,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC7C,cAAc,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;QACtC,CAAC,CAAC,MAAM,GAAG,CAAC;QACZ,CAAC,CAAC,MAAM,CAAC;IACb,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,OAAO,EAAE,KAAK,CAAC;AAC/C,CAAC;AAED,SAAS,kBAAkB,CACzB,QAAoC,EACpC,QAAiB;IAEjB,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,IAAI,QAAQ,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;QAC7B,OAAO,oBAAoB,CACzB;YACE,QAAQ,CAAC,UAAU;gBACjB,CAAC,CAAC,WAAW,QAAQ,CAAC,UAAU,IAAI,QAAQ,CAAC,UAAU,EAAE;gBACzD,CAAC,CAAC,WAAW,QAAQ,CAAC,UAAU,EAAE;YACpC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,IAAI;YAClD,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,cAAc,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI;YACvE,QAAQ,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC;gBAC3B,CAAC,CAAC,cAAc,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE;gBAC9C,CAAC,CAAC,IAAI;YACR,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI;SACtD,EACD,QAAQ,CACT,CAAC;IACJ,CAAC;IAED,OAAO,oBAAoB,CACzB;QACE,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,YAAY,QAAQ,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,IAAI;QACxD,OAAO,QAAQ,CAAC,KAAK,KAAK,QAAQ;YAChC,CAAC,CAAC,kBAAkB,QAAQ,CAAC,KAAK,EAAE;YACpC,CAAC,CAAC,IAAI;QACR,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,IAAI;KACnD,EACD,QAAQ,CACT,CAAC;AACJ,CAAC;AAED,SAAS,cAAc,CAAC,QAA4B;IAClD,OAAO,QAAQ,KAAK,SAAS;QAC3B,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,GAAG,yBAAyB,CAAC,CAAC,CAAC;AACpE,CAAC;AAED,SAAS,oBAAoB,CAC3B,KAAiC,EACjC,QAAiB;IAEjB,MAAM,IAAI,GAAG,KAAK;SACf,MAAM,CAAC,CAAC,IAAI,EAAkB,EAAE,CAAC,IAAI,KAAK,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;SACzE,IAAI,CAAC,IAAI,CAAC,CAAC;IACd,OAAO,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;AACxE,CAAC;AAED,SAAS,OAAO,CAAC,KAAa;IAC5B,OAAO,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AAC3C,CAAC;AAED,SAAS,eAAe,CAAC,KAAa;IACpC,OAAO,KAAK,IAAI,MAAM,IAAI,KAAK,IAAI,MAAM,CAAC;AAC5C,CAAC;AAED,SAAS,cAAc,CAAC,KAAa;IACnC,OAAO,KAAK,IAAI,MAAM,IAAI,KAAK,IAAI,MAAM,CAAC;AAC5C,CAAC"}
|
|
@@ -0,0 +1,6 @@
|
|
|
1
|
+
import type { FileKind } from "./types.js";
|
|
2
|
+
export declare const DEFAULT_MAX_CODE_FILE_SIZE_BYTES: number;
|
|
3
|
+
export declare const DEFAULT_MAX_TEXT_FILE_SIZE_BYTES: number;
|
|
4
|
+
export declare const DEFAULT_MAX_DATA_FILE_SIZE_BYTES: number;
|
|
5
|
+
export declare const DEFAULT_MAX_IMAGE_FILE_SIZE_BYTES: number;
|
|
6
|
+
export declare function resolveMaxFileSizeBytes(kind: FileKind, explicitMaxFileSizeBytes?: number): number;
|
|
@@ -0,0 +1,20 @@
|
|
|
1
|
+
export const DEFAULT_MAX_CODE_FILE_SIZE_BYTES = 1 * 1024 * 1024;
|
|
2
|
+
export const DEFAULT_MAX_TEXT_FILE_SIZE_BYTES = 256 * 1024 * 1024;
|
|
3
|
+
export const DEFAULT_MAX_DATA_FILE_SIZE_BYTES = 16 * 1024 * 1024;
|
|
4
|
+
export const DEFAULT_MAX_IMAGE_FILE_SIZE_BYTES = 10 * 1024 * 1024;
|
|
5
|
+
export function resolveMaxFileSizeBytes(kind, explicitMaxFileSizeBytes) {
|
|
6
|
+
if (explicitMaxFileSizeBytes !== undefined) {
|
|
7
|
+
return explicitMaxFileSizeBytes;
|
|
8
|
+
}
|
|
9
|
+
switch (kind) {
|
|
10
|
+
case "code":
|
|
11
|
+
return DEFAULT_MAX_CODE_FILE_SIZE_BYTES;
|
|
12
|
+
case "text":
|
|
13
|
+
return DEFAULT_MAX_TEXT_FILE_SIZE_BYTES;
|
|
14
|
+
case "data":
|
|
15
|
+
return DEFAULT_MAX_DATA_FILE_SIZE_BYTES;
|
|
16
|
+
case "image":
|
|
17
|
+
return DEFAULT_MAX_IMAGE_FILE_SIZE_BYTES;
|
|
18
|
+
}
|
|
19
|
+
}
|
|
20
|
+
//# sourceMappingURL=file-size-policy.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"file-size-policy.js","sourceRoot":"","sources":["../../src/engine/file-size-policy.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,gCAAgC,GAAG,CAAC,GAAG,IAAI,GAAG,IAAI,CAAC;AAChE,MAAM,CAAC,MAAM,gCAAgC,GAAG,GAAG,GAAG,IAAI,GAAG,IAAI,CAAC;AAClE,MAAM,CAAC,MAAM,gCAAgC,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC;AACjE,MAAM,CAAC,MAAM,iCAAiC,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC;AAElE,MAAM,UAAU,uBAAuB,CACrC,IAAc,EACd,wBAAiC;IAEjC,IAAI,wBAAwB,KAAK,SAAS,EAAE,CAAC;QAC3C,OAAO,wBAAwB,CAAC;IAClC,CAAC;IAED,QAAQ,IAAI,EAAE,CAAC;QACb,KAAK,MAAM;YACT,OAAO,gCAAgC,CAAC;QAC1C,KAAK,MAAM;YACT,OAAO,gCAAgC,CAAC;QAC1C,KAAK,MAAM;YACT,OAAO,gCAAgC,CAAC;QAC1C,KAAK,OAAO;YACV,OAAO,iCAAiC,CAAC;IAC7C,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
import type { FileFormat, FileKind } from "./types.js";
|
|
2
|
+
export type RecognizedFileType = {
|
|
3
|
+
kind: FileKind;
|
|
4
|
+
format: FileFormat;
|
|
5
|
+
patterns: string[];
|
|
6
|
+
};
|
|
7
|
+
export declare function listRecognizedFileTypes(): RecognizedFileType[];
|
|
8
|
+
export declare function listKnownBinaryExtensionGroups(): {
|
|
9
|
+
label: string;
|
|
10
|
+
extensions: string[];
|
|
11
|
+
}[];
|
|
12
|
+
export declare function detectFileType(path: string): {
|
|
13
|
+
kind: FileKind;
|
|
14
|
+
format: FileFormat;
|
|
15
|
+
} | null;
|
|
@@ -0,0 +1,152 @@
|
|
|
1
|
+
import { basename, extname } from "node:path";
|
|
2
|
+
const CODE_FORMATS = {
|
|
3
|
+
".c": "c",
|
|
4
|
+
".cc": "cpp",
|
|
5
|
+
".cpp": "cpp",
|
|
6
|
+
".cxx": "cpp",
|
|
7
|
+
".h": "cpp",
|
|
8
|
+
".hpp": "cpp",
|
|
9
|
+
".go": "go",
|
|
10
|
+
".java": "java",
|
|
11
|
+
".js": "javascript",
|
|
12
|
+
".jsx": "jsx",
|
|
13
|
+
".mjs": "javascript",
|
|
14
|
+
".cjs": "javascript",
|
|
15
|
+
".ts": "typescript",
|
|
16
|
+
".tsx": "tsx",
|
|
17
|
+
".py": "python",
|
|
18
|
+
".rs": "rust",
|
|
19
|
+
".rb": "ruby",
|
|
20
|
+
".php": "php",
|
|
21
|
+
".swift": "swift",
|
|
22
|
+
".kt": "kotlin",
|
|
23
|
+
".kts": "kotlin",
|
|
24
|
+
".cs": "csharp",
|
|
25
|
+
".scala": "scala",
|
|
26
|
+
".sh": "bash",
|
|
27
|
+
".bash": "bash",
|
|
28
|
+
".zsh": "bash",
|
|
29
|
+
".sql": "sql",
|
|
30
|
+
".css": "css",
|
|
31
|
+
".scss": "scss",
|
|
32
|
+
".less": "less",
|
|
33
|
+
".vue": "vue",
|
|
34
|
+
".svelte": "svelte",
|
|
35
|
+
};
|
|
36
|
+
const DATA_FORMATS = {
|
|
37
|
+
".csv": "csv",
|
|
38
|
+
".json": "json",
|
|
39
|
+
".jsonc": "json",
|
|
40
|
+
".toml": "toml",
|
|
41
|
+
".yaml": "yaml",
|
|
42
|
+
".yml": "yaml",
|
|
43
|
+
};
|
|
44
|
+
const TEXT_FORMATS = {
|
|
45
|
+
".md": "markdown",
|
|
46
|
+
".mdx": "markdown",
|
|
47
|
+
".rst": "rst",
|
|
48
|
+
".txt": "text",
|
|
49
|
+
".html": "html",
|
|
50
|
+
".htm": "html",
|
|
51
|
+
".xml": "xml",
|
|
52
|
+
};
|
|
53
|
+
const IMAGE_FORMATS = {
|
|
54
|
+
".gif": "gif",
|
|
55
|
+
".jpeg": "jpeg",
|
|
56
|
+
".jpg": "jpeg",
|
|
57
|
+
".png": "png",
|
|
58
|
+
".webp": "webp",
|
|
59
|
+
};
|
|
60
|
+
const NAMED_FILE_FORMATS = {
|
|
61
|
+
Dockerfile: { kind: "code", format: "dockerfile" },
|
|
62
|
+
Makefile: { kind: "code", format: "makefile" },
|
|
63
|
+
};
|
|
64
|
+
const FILE_FORMATS_BY_KIND = {
|
|
65
|
+
code: CODE_FORMATS,
|
|
66
|
+
data: DATA_FORMATS,
|
|
67
|
+
text: TEXT_FORMATS,
|
|
68
|
+
image: IMAGE_FORMATS,
|
|
69
|
+
};
|
|
70
|
+
const BINARY_EXTENSION_GROUPS = [
|
|
71
|
+
{
|
|
72
|
+
label: "Archives",
|
|
73
|
+
extensions: [".zip", ".tar", ".gz", ".bz2", ".xz", ".7z", ".rar"],
|
|
74
|
+
},
|
|
75
|
+
{
|
|
76
|
+
label: "Compiled",
|
|
77
|
+
extensions: [
|
|
78
|
+
".exe",
|
|
79
|
+
".dll",
|
|
80
|
+
".dylib",
|
|
81
|
+
".so",
|
|
82
|
+
".a",
|
|
83
|
+
".o",
|
|
84
|
+
".obj",
|
|
85
|
+
".wasm",
|
|
86
|
+
".class",
|
|
87
|
+
".jar",
|
|
88
|
+
],
|
|
89
|
+
},
|
|
90
|
+
{
|
|
91
|
+
label: "Documents",
|
|
92
|
+
extensions: [".pdf", ".doc", ".docx", ".ppt", ".pptx", ".xls", ".xlsx"],
|
|
93
|
+
},
|
|
94
|
+
{
|
|
95
|
+
label: "Media",
|
|
96
|
+
extensions: [".mp3", ".mp4", ".mov", ".avi", ".mkv"],
|
|
97
|
+
},
|
|
98
|
+
{ label: "Databases", extensions: [".db", ".sqlite"] },
|
|
99
|
+
];
|
|
100
|
+
const BINARY_EXTENSIONS = new Set(BINARY_EXTENSION_GROUPS.flatMap((group) => group.extensions));
|
|
101
|
+
export function listRecognizedFileTypes() {
|
|
102
|
+
const recognized = new Map();
|
|
103
|
+
const append = (kind, format, pattern) => {
|
|
104
|
+
const key = `${kind}:${format}`;
|
|
105
|
+
const existing = recognized.get(key);
|
|
106
|
+
if (existing) {
|
|
107
|
+
existing.patterns.push(pattern);
|
|
108
|
+
return;
|
|
109
|
+
}
|
|
110
|
+
recognized.set(key, { kind, format, patterns: [pattern] });
|
|
111
|
+
};
|
|
112
|
+
for (const kind of Object.keys(FILE_FORMATS_BY_KIND)) {
|
|
113
|
+
for (const [extension, format] of Object.entries(FILE_FORMATS_BY_KIND[kind])) {
|
|
114
|
+
append(kind, format, extension);
|
|
115
|
+
}
|
|
116
|
+
}
|
|
117
|
+
for (const [name, type] of Object.entries(NAMED_FILE_FORMATS)) {
|
|
118
|
+
append(type.kind, type.format, name);
|
|
119
|
+
}
|
|
120
|
+
return [...recognized.values()];
|
|
121
|
+
}
|
|
122
|
+
export function listKnownBinaryExtensionGroups() {
|
|
123
|
+
return BINARY_EXTENSION_GROUPS.map((group) => ({
|
|
124
|
+
label: group.label,
|
|
125
|
+
extensions: [...group.extensions],
|
|
126
|
+
}));
|
|
127
|
+
}
|
|
128
|
+
export function detectFileType(path) {
|
|
129
|
+
const name = basename(path);
|
|
130
|
+
const namedType = NAMED_FILE_FORMATS[name];
|
|
131
|
+
if (namedType) {
|
|
132
|
+
return { ...namedType };
|
|
133
|
+
}
|
|
134
|
+
const extension = extname(name).toLowerCase();
|
|
135
|
+
if (extension in CODE_FORMATS) {
|
|
136
|
+
return { kind: "code", format: CODE_FORMATS[extension] };
|
|
137
|
+
}
|
|
138
|
+
if (extension in DATA_FORMATS) {
|
|
139
|
+
return { kind: "data", format: DATA_FORMATS[extension] };
|
|
140
|
+
}
|
|
141
|
+
if (extension in TEXT_FORMATS) {
|
|
142
|
+
return { kind: "text", format: TEXT_FORMATS[extension] };
|
|
143
|
+
}
|
|
144
|
+
if (extension in IMAGE_FORMATS) {
|
|
145
|
+
return { kind: "image", format: IMAGE_FORMATS[extension] };
|
|
146
|
+
}
|
|
147
|
+
if (BINARY_EXTENSIONS.has(extension)) {
|
|
148
|
+
return null;
|
|
149
|
+
}
|
|
150
|
+
return { kind: "text", format: extension ? extension.slice(1) : "text" };
|
|
151
|
+
}
|
|
152
|
+
//# sourceMappingURL=file-type.js.map
|
|
@@ -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,kBAAkB,GAGpB;IACF,UAAU,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,EAAE;IAClD,QAAQ,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE;CAC/C,CAAC;AAEF,MAAM,oBAAoB,GAAG;IAC3B,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,YAAY;IAClB,KAAK,EAAE,aAAa;CAC8B,CAAC;AAErD,MAAM,uBAAuB,GAAG;IAC9B;QACE,KAAK,EAAE,UAAU;QACjB,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,CAAC;KAClE;IACD;QACE,KAAK,EAAE,UAAU;QACjB,UAAU,EAAE;YACV,MAAM;YACN,MAAM;YACN,QAAQ;YACR,KAAK;YACL,IAAI;YACJ,IAAI;YACJ,MAAM;YACN,OAAO;YACP,QAAQ;YACR,MAAM;SACP;KACF;IACD;QACE,KAAK,EAAE,WAAW;QAClB,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,CAAC;KACxE;IACD;QACE,KAAK,EAAE,OAAO;QACd,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KACrD;IACD,EAAE,KAAK,EAAE,WAAW,EAAE,UAAU,EAAE,CAAC,KAAK,EAAE,SAAS,CAAC,EAAE;CAC9C,CAAC;AAEX,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAC/B,uBAAuB,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAC7D,CAAC;AAQF,MAAM,UAAU,uBAAuB;IACrC,MAAM,UAAU,GAAG,IAAI,GAAG,EAA8B,CAAC;IACzD,MAAM,MAAM,GAAG,CACb,IAAc,EACd,MAAkB,EAClB,OAAe,EACT,EAAE;QACR,MAAM,GAAG,GAAG,GAAG,IAAI,IAAI,MAAM,EAAE,CAAC;QAChC,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACrC,IAAI,QAAQ,EAAE,CAAC;YACb,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YAChC,OAAO;QACT,CAAC;QACD,UAAU,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;IAC7D,CAAC,CAAC;IAEF,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,IAAI,CAAC,oBAAoB,CAAe,EAAE,CAAC;QACnE,KAAK,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC,OAAO,CAC9C,oBAAoB,CAAC,IAAI,CAAC,CAC3B,EAAE,CAAC;YACF,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QAClC,CAAC;IACH,CAAC;IACD,KAAK,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,kBAAkB,CAAC,EAAE,CAAC;QAC9D,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;IACvC,CAAC;IAED,OAAO,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,CAAC;AAClC,CAAC;AAED,MAAM,UAAU,8BAA8B;IAI5C,OAAO,uBAAuB,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC7C,KAAK,EAAE,KAAK,CAAC,KAAK;QAClB,UAAU,EAAE,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC;KAClC,CAAC,CAAC,CAAC;AACN,CAAC;AAED,MAAM,UAAU,cAAc,CAC5B,IAAY;IAEZ,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC5B,MAAM,SAAS,GAAG,kBAAkB,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,SAAS,EAAE,CAAC;QACd,OAAO,EAAE,GAAG,SAAS,EAAE,CAAC;IAC1B,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 +1,13 @@
|
|
|
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 deleteWorkspaceManifest(home: string): void;
|
|
13
|
+
export declare function workspaceIndexInfoFromManifest(manifest: WorkspaceManifest): WorkspaceIndexInfo;
|