vectra 0.12.3 → 0.14.0
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 +92 -100
- package/lib/BrowserWebFetcher.d.ts +75 -0
- package/lib/BrowserWebFetcher.d.ts.map +1 -0
- package/lib/BrowserWebFetcher.js +290 -0
- package/lib/BrowserWebFetcher.js.map +1 -0
- package/lib/FileFetcher.d.ts.map +1 -1
- package/lib/FileFetcher.js +25 -15
- package/lib/FileFetcher.js.map +1 -1
- package/lib/FileFetcher.spec.d.ts +2 -0
- package/lib/FileFetcher.spec.d.ts.map +1 -0
- package/lib/FileFetcher.spec.js +244 -0
- package/lib/FileFetcher.spec.js.map +1 -0
- package/lib/FolderWatcher.d.ts +91 -0
- package/lib/FolderWatcher.d.ts.map +1 -0
- package/lib/FolderWatcher.js +304 -0
- package/lib/FolderWatcher.js.map +1 -0
- package/lib/FolderWatcher.spec.d.ts +2 -0
- package/lib/FolderWatcher.spec.d.ts.map +1 -0
- package/lib/FolderWatcher.spec.js +308 -0
- package/lib/FolderWatcher.spec.js.map +1 -0
- package/lib/GPT3Tokenizer.spec.d.ts +2 -0
- package/lib/GPT3Tokenizer.spec.d.ts.map +1 -0
- package/lib/GPT3Tokenizer.spec.js +45 -0
- package/lib/GPT3Tokenizer.spec.js.map +1 -0
- package/lib/ItemSelector.d.ts.map +1 -1
- package/lib/ItemSelector.js +19 -8
- package/lib/ItemSelector.js.map +1 -1
- package/lib/ItemSelector.spec.d.ts +2 -0
- package/lib/ItemSelector.spec.d.ts.map +1 -0
- package/lib/ItemSelector.spec.js +204 -0
- package/lib/ItemSelector.spec.js.map +1 -0
- package/lib/LocalDocument.d.ts +1 -1
- package/lib/LocalDocument.d.ts.map +1 -1
- package/lib/LocalDocument.js +5 -45
- package/lib/LocalDocument.js.map +1 -1
- package/lib/LocalDocument.spec.d.ts +2 -0
- package/lib/LocalDocument.spec.d.ts.map +1 -0
- package/lib/LocalDocument.spec.js +214 -0
- package/lib/LocalDocument.spec.js.map +1 -0
- package/lib/LocalDocumentIndex.d.ts +20 -0
- package/lib/LocalDocumentIndex.d.ts.map +1 -1
- package/lib/LocalDocumentIndex.js +16 -52
- package/lib/LocalDocumentIndex.js.map +1 -1
- package/lib/LocalDocumentIndex.spec.d.ts +2 -0
- package/lib/LocalDocumentIndex.spec.d.ts.map +1 -0
- package/lib/LocalDocumentIndex.spec.js +494 -0
- package/lib/LocalDocumentIndex.spec.js.map +1 -0
- package/lib/LocalDocumentResult.d.ts +32 -11
- package/lib/LocalDocumentResult.d.ts.map +1 -1
- package/lib/LocalDocumentResult.js +305 -257
- package/lib/LocalDocumentResult.js.map +1 -1
- package/lib/LocalDocumentResult.spec.d.ts +2 -0
- package/lib/LocalDocumentResult.spec.d.ts.map +1 -0
- package/lib/LocalDocumentResult.spec.js +373 -0
- package/lib/LocalDocumentResult.spec.js.map +1 -0
- package/lib/LocalEmbeddings.d.ts +59 -0
- package/lib/LocalEmbeddings.d.ts.map +1 -0
- package/lib/LocalEmbeddings.js +101 -0
- package/lib/LocalEmbeddings.js.map +1 -0
- package/lib/LocalEmbeddings.spec.d.ts +2 -0
- package/lib/LocalEmbeddings.spec.d.ts.map +1 -0
- package/lib/LocalEmbeddings.spec.js +155 -0
- package/lib/LocalEmbeddings.spec.js.map +1 -0
- package/lib/LocalIndex.d.ts +27 -18
- package/lib/LocalIndex.d.ts.map +1 -1
- package/lib/LocalIndex.js +109 -105
- package/lib/LocalIndex.js.map +1 -1
- package/lib/LocalIndex.spec.js +434 -43
- package/lib/LocalIndex.spec.js.map +1 -1
- package/lib/OpenAIEmbeddings.d.ts +4 -6
- package/lib/OpenAIEmbeddings.d.ts.map +1 -1
- package/lib/OpenAIEmbeddings.js +16 -24
- package/lib/OpenAIEmbeddings.js.map +1 -1
- package/lib/OpenAIEmbeddings.spec.d.ts +2 -0
- package/lib/OpenAIEmbeddings.spec.d.ts.map +1 -0
- package/lib/OpenAIEmbeddings.spec.js +298 -0
- package/lib/OpenAIEmbeddings.spec.js.map +1 -0
- package/lib/TextSplitter.d.ts +2 -0
- package/lib/TextSplitter.d.ts.map +1 -1
- package/lib/TextSplitter.js +154 -111
- package/lib/TextSplitter.js.map +1 -1
- package/lib/TextSplitter.spec.js +289 -61
- package/lib/TextSplitter.spec.js.map +1 -1
- package/lib/TransformersEmbeddings.d.ts +121 -0
- package/lib/TransformersEmbeddings.d.ts.map +1 -0
- package/lib/TransformersEmbeddings.js +176 -0
- package/lib/TransformersEmbeddings.js.map +1 -0
- package/lib/TransformersEmbeddings.spec.d.ts +2 -0
- package/lib/TransformersEmbeddings.spec.d.ts.map +1 -0
- package/lib/TransformersEmbeddings.spec.js +198 -0
- package/lib/TransformersEmbeddings.spec.js.map +1 -0
- package/lib/TransformersTokenizer.d.ts +33 -0
- package/lib/TransformersTokenizer.d.ts.map +1 -0
- package/lib/TransformersTokenizer.js +44 -0
- package/lib/TransformersTokenizer.js.map +1 -0
- package/lib/TransformersTokenizer.spec.d.ts +2 -0
- package/lib/TransformersTokenizer.spec.d.ts.map +1 -0
- package/lib/TransformersTokenizer.spec.js +112 -0
- package/lib/TransformersTokenizer.spec.js.map +1 -0
- package/lib/WebFetcher.d.ts +1 -2
- package/lib/WebFetcher.d.ts.map +1 -1
- package/lib/WebFetcher.js +58 -54
- package/lib/WebFetcher.js.map +1 -1
- package/lib/WebFetcher.spec.d.ts +2 -0
- package/lib/WebFetcher.spec.d.ts.map +1 -0
- package/lib/WebFetcher.spec.js +263 -0
- package/lib/WebFetcher.spec.js.map +1 -0
- package/lib/browser.d.ts +30 -0
- package/lib/browser.d.ts.map +1 -0
- package/lib/browser.js +52 -0
- package/lib/browser.js.map +1 -0
- package/lib/codecs/IndexCodec.d.ts +37 -0
- package/lib/codecs/IndexCodec.d.ts.map +1 -0
- package/lib/codecs/IndexCodec.js +3 -0
- package/lib/codecs/IndexCodec.js.map +1 -0
- package/lib/codecs/JsonCodec.d.ts +19 -0
- package/lib/codecs/JsonCodec.d.ts.map +1 -0
- package/lib/codecs/JsonCodec.js +35 -0
- package/lib/codecs/JsonCodec.js.map +1 -0
- package/lib/codecs/JsonCodec.spec.d.ts +2 -0
- package/lib/codecs/JsonCodec.spec.d.ts.map +1 -0
- package/lib/codecs/JsonCodec.spec.js +66 -0
- package/lib/codecs/JsonCodec.spec.js.map +1 -0
- package/lib/codecs/LocalIndex.protobuf.spec.d.ts +2 -0
- package/lib/codecs/LocalIndex.protobuf.spec.d.ts.map +1 -0
- package/lib/codecs/LocalIndex.protobuf.spec.js +108 -0
- package/lib/codecs/LocalIndex.protobuf.spec.js.map +1 -0
- package/lib/codecs/ProtobufCodec.d.ts +20 -0
- package/lib/codecs/ProtobufCodec.d.ts.map +1 -0
- package/lib/codecs/ProtobufCodec.js +225 -0
- package/lib/codecs/ProtobufCodec.js.map +1 -0
- package/lib/codecs/ProtobufCodec.spec.d.ts +2 -0
- package/lib/codecs/ProtobufCodec.spec.d.ts.map +1 -0
- package/lib/codecs/ProtobufCodec.spec.js +155 -0
- package/lib/codecs/ProtobufCodec.spec.js.map +1 -0
- package/lib/codecs/index.d.ts +5 -0
- package/lib/codecs/index.d.ts.map +1 -0
- package/lib/codecs/index.js +21 -0
- package/lib/codecs/index.js.map +1 -0
- package/lib/codecs/migrateIndex.d.ts +24 -0
- package/lib/codecs/migrateIndex.d.ts.map +1 -0
- package/lib/codecs/migrateIndex.js +119 -0
- package/lib/codecs/migrateIndex.js.map +1 -0
- package/lib/codecs/migrateIndex.spec.d.ts +2 -0
- package/lib/codecs/migrateIndex.spec.d.ts.map +1 -0
- package/lib/codecs/migrateIndex.spec.js +151 -0
- package/lib/codecs/migrateIndex.spec.js.map +1 -0
- package/lib/codecs/schemas/index.proto +34 -0
- package/lib/index.d.ts +9 -1
- package/lib/index.d.ts.map +1 -1
- package/lib/index.js +9 -1
- package/lib/index.js.map +1 -1
- package/lib/internals/Colorize.d.ts.map +1 -1
- package/lib/internals/Colorize.js +20 -15
- package/lib/internals/Colorize.js.map +1 -1
- package/lib/server/IndexManager.d.ts +78 -0
- package/lib/server/IndexManager.d.ts.map +1 -0
- package/lib/server/IndexManager.js +259 -0
- package/lib/server/IndexManager.js.map +1 -0
- package/lib/server/VectraServer.d.ts +40 -0
- package/lib/server/VectraServer.d.ts.map +1 -0
- package/lib/server/VectraServer.js +151 -0
- package/lib/server/VectraServer.js.map +1 -0
- package/lib/server/VectraServer.spec.d.ts +2 -0
- package/lib/server/VectraServer.spec.d.ts.map +1 -0
- package/lib/server/VectraServer.spec.js +322 -0
- package/lib/server/VectraServer.spec.js.map +1 -0
- package/lib/server/handlers/documentHandlers.d.ts +15 -0
- package/lib/server/handlers/documentHandlers.d.ts.map +1 -0
- package/lib/server/handlers/documentHandlers.js +95 -0
- package/lib/server/handlers/documentHandlers.js.map +1 -0
- package/lib/server/handlers/helpers.d.ts +23 -0
- package/lib/server/handlers/helpers.d.ts.map +1 -0
- package/lib/server/handlers/helpers.js +138 -0
- package/lib/server/handlers/helpers.js.map +1 -0
- package/lib/server/handlers/index.d.ts +8 -0
- package/lib/server/handlers/index.d.ts.map +1 -0
- package/lib/server/handlers/index.js +22 -0
- package/lib/server/handlers/index.js.map +1 -0
- package/lib/server/handlers/indexHandlers.d.ts +14 -0
- package/lib/server/handlers/indexHandlers.d.ts.map +1 -0
- package/lib/server/handlers/indexHandlers.js +85 -0
- package/lib/server/handlers/indexHandlers.js.map +1 -0
- package/lib/server/handlers/itemHandlers.d.ts +34 -0
- package/lib/server/handlers/itemHandlers.d.ts.map +1 -0
- package/lib/server/handlers/itemHandlers.js +166 -0
- package/lib/server/handlers/itemHandlers.js.map +1 -0
- package/lib/server/handlers/lifecycleHandlers.d.ts +11 -0
- package/lib/server/handlers/lifecycleHandlers.d.ts.map +1 -0
- package/lib/server/handlers/lifecycleHandlers.js +31 -0
- package/lib/server/handlers/lifecycleHandlers.js.map +1 -0
- package/lib/server/handlers/queryHandlers.d.ts +27 -0
- package/lib/server/handlers/queryHandlers.d.ts.map +1 -0
- package/lib/server/handlers/queryHandlers.js +135 -0
- package/lib/server/handlers/queryHandlers.js.map +1 -0
- package/lib/server/handlers/statsHandlers.d.ts +17 -0
- package/lib/server/handlers/statsHandlers.d.ts.map +1 -0
- package/lib/server/handlers/statsHandlers.js +81 -0
- package/lib/server/handlers/statsHandlers.js.map +1 -0
- package/lib/server/index.d.ts +4 -0
- package/lib/server/index.d.ts.map +1 -0
- package/lib/server/index.js +23 -0
- package/lib/server/index.js.map +1 -0
- package/lib/storage/FileStorage.d.ts +92 -0
- package/lib/storage/FileStorage.d.ts.map +1 -0
- package/lib/storage/FileStorage.js +3 -0
- package/lib/storage/FileStorage.js.map +1 -0
- package/lib/storage/FileStorageUtilities.d.ts +36 -0
- package/lib/storage/FileStorageUtilities.d.ts.map +1 -0
- package/lib/storage/FileStorageUtilities.js +91 -0
- package/lib/storage/FileStorageUtilities.js.map +1 -0
- package/lib/storage/FileStorageUtilities.spec.d.ts +2 -0
- package/lib/storage/FileStorageUtilities.spec.d.ts.map +1 -0
- package/lib/storage/FileStorageUtilities.spec.js +98 -0
- package/lib/storage/FileStorageUtilities.spec.js.map +1 -0
- package/lib/storage/FileType.d.ts +29 -0
- package/lib/storage/FileType.d.ts.map +1 -0
- package/lib/storage/FileType.js +38 -0
- package/lib/storage/FileType.js.map +1 -0
- package/lib/storage/IndexedDBStorage.d.ts +47 -0
- package/lib/storage/IndexedDBStorage.d.ts.map +1 -0
- package/lib/storage/IndexedDBStorage.js +347 -0
- package/lib/storage/IndexedDBStorage.js.map +1 -0
- package/lib/storage/LocalFileStorage.browser.d.ts +19 -0
- package/lib/storage/LocalFileStorage.browser.d.ts.map +1 -0
- package/lib/storage/LocalFileStorage.browser.js +43 -0
- package/lib/storage/LocalFileStorage.browser.js.map +1 -0
- package/lib/storage/LocalFileStorage.d.ts +23 -0
- package/lib/storage/LocalFileStorage.d.ts.map +1 -0
- package/lib/storage/LocalFileStorage.js +152 -0
- package/lib/storage/LocalFileStorage.js.map +1 -0
- package/lib/storage/LocalFileStorage.spec.d.ts +2 -0
- package/lib/storage/LocalFileStorage.spec.d.ts.map +1 -0
- package/lib/storage/LocalFileStorage.spec.js +249 -0
- package/lib/storage/LocalFileStorage.spec.js.map +1 -0
- package/lib/storage/VirtualFileStorage.d.ts +18 -0
- package/lib/storage/VirtualFileStorage.d.ts.map +1 -0
- package/lib/storage/VirtualFileStorage.js +178 -0
- package/lib/storage/VirtualFileStorage.js.map +1 -0
- package/lib/storage/VirtualFileStorage.spec.d.ts +2 -0
- package/lib/storage/VirtualFileStorage.spec.d.ts.map +1 -0
- package/lib/storage/VirtualFileStorage.spec.js +302 -0
- package/lib/storage/VirtualFileStorage.spec.js.map +1 -0
- package/lib/storage/index.d.ts +6 -0
- package/lib/storage/index.d.ts.map +1 -0
- package/lib/storage/index.js +22 -0
- package/lib/storage/index.js.map +1 -0
- package/lib/templates/templates/csharp/README.md +48 -0
- package/lib/templates/templates/csharp/VectraClient.cs +234 -0
- package/lib/templates/templates/go/README.md +71 -0
- package/lib/templates/templates/go/vectra_client.go +322 -0
- package/lib/templates/templates/java/README.md +81 -0
- package/lib/templates/templates/java/VectraClient.java +232 -0
- package/lib/templates/templates/python/README.md +37 -0
- package/lib/templates/templates/python/vectra_client.py +279 -0
- package/lib/templates/templates/rust/Cargo.toml +14 -0
- package/lib/templates/templates/rust/README.md +39 -0
- package/lib/templates/templates/rust/build.rs +4 -0
- package/lib/templates/templates/rust/lib.rs +284 -0
- package/lib/templates/templates/typescript/README.md +96 -0
- package/lib/templates/templates/typescript/VectraClient.ts +374 -0
- package/lib/templates/typescript/VectraClient.d.ts +114 -0
- package/lib/templates/typescript/VectraClient.d.ts.map +1 -0
- package/lib/templates/typescript/VectraClient.js +328 -0
- package/lib/templates/typescript/VectraClient.js.map +1 -0
- package/lib/types.d.ts +7 -0
- package/lib/types.d.ts.map +1 -1
- package/lib/utils/index.d.ts +2 -0
- package/lib/utils/index.d.ts.map +1 -0
- package/lib/utils/index.js +18 -0
- package/lib/utils/index.js.map +1 -0
- package/lib/utils/pathUtils.d.ts +40 -0
- package/lib/utils/pathUtils.d.ts.map +1 -0
- package/lib/utils/pathUtils.js +98 -0
- package/lib/utils/pathUtils.js.map +1 -0
- package/lib/vectra-cli.d.ts.map +1 -1
- package/lib/vectra-cli.generate.spec.d.ts +2 -0
- package/lib/vectra-cli.generate.spec.d.ts.map +1 -0
- package/lib/vectra-cli.generate.spec.js +112 -0
- package/lib/vectra-cli.generate.spec.js.map +1 -0
- package/lib/vectra-cli.js +446 -9
- package/lib/vectra-cli.js.map +1 -1
- package/lib/vectra-cli.spec.d.ts +1 -0
- package/lib/vectra-cli.spec.d.ts.map +1 -0
- package/lib/vectra-cli.spec.js +2 -0
- package/lib/vectra-cli.spec.js.map +1 -0
- package/package.json +89 -16
- package/proto/vectra_service.proto +276 -0
- package/src/BrowserWebFetcher.ts +345 -0
- package/src/FileFetcher.spec.ts +234 -0
- package/src/FileFetcher.ts +37 -25
- package/src/FolderWatcher.spec.ts +288 -0
- package/src/FolderWatcher.ts +304 -0
- package/src/GPT3Tokenizer.spec.ts +50 -0
- package/src/ItemSelector.spec.ts +252 -0
- package/src/ItemSelector.ts +163 -150
- package/src/LocalDocument.spec.ts +211 -0
- package/src/LocalDocument.ts +88 -94
- package/src/LocalDocumentIndex.spec.ts +481 -0
- package/src/LocalDocumentIndex.ts +39 -40
- package/src/LocalDocumentResult.spec.ts +373 -0
- package/src/LocalDocumentResult.ts +489 -319
- package/src/LocalEmbeddings.spec.ts +138 -0
- package/src/LocalEmbeddings.ts +120 -0
- package/src/LocalIndex.spec.ts +808 -323
- package/src/LocalIndex.ts +479 -430
- package/src/OpenAIEmbeddings.spec.ts +354 -0
- package/src/OpenAIEmbeddings.ts +26 -27
- package/src/TextSplitter.spec.ts +320 -65
- package/src/TextSplitter.ts +172 -115
- package/src/TransformersEmbeddings.spec.ts +188 -0
- package/src/TransformersEmbeddings.ts +232 -0
- package/src/TransformersTokenizer.spec.ts +143 -0
- package/src/TransformersTokenizer.ts +45 -0
- package/src/WebFetcher.spec.ts +288 -0
- package/src/WebFetcher.ts +184 -186
- package/src/browser.ts +69 -0
- package/src/codecs/IndexCodec.ts +40 -0
- package/src/codecs/JsonCodec.spec.ts +70 -0
- package/src/codecs/JsonCodec.ts +37 -0
- package/src/codecs/LocalIndex.protobuf.spec.ts +115 -0
- package/src/codecs/ProtobufCodec.spec.ts +166 -0
- package/src/codecs/ProtobufCodec.ts +193 -0
- package/src/codecs/index.ts +4 -0
- package/src/codecs/migrateIndex.spec.ts +176 -0
- package/src/codecs/migrateIndex.ts +125 -0
- package/src/codecs/schemas/index.proto +34 -0
- package/src/index.ts +9 -1
- package/src/internals/Colorize.ts +19 -16
- package/src/server/IndexManager.ts +243 -0
- package/src/server/VectraServer.spec.ts +303 -0
- package/src/server/VectraServer.ts +156 -0
- package/src/server/handlers/documentHandlers.ts +59 -0
- package/src/server/handlers/helpers.ts +93 -0
- package/src/server/handlers/index.ts +7 -0
- package/src/server/handlers/indexHandlers.ts +44 -0
- package/src/server/handlers/itemHandlers.ts +140 -0
- package/src/server/handlers/lifecycleHandlers.ts +26 -0
- package/src/server/handlers/queryHandlers.ts +96 -0
- package/src/server/handlers/statsHandlers.ts +38 -0
- package/src/server/index.ts +3 -0
- package/src/storage/FileStorage.ts +105 -0
- package/src/storage/FileStorageUtilities.spec.ts +106 -0
- package/src/storage/FileStorageUtilities.ts +77 -0
- package/src/storage/FileType.ts +61 -0
- package/src/storage/IndexedDBStorage.ts +365 -0
- package/src/storage/LocalFileStorage.browser.ts +52 -0
- package/src/storage/LocalFileStorage.spec.ts +292 -0
- package/src/storage/LocalFileStorage.ts +98 -0
- package/src/storage/VirtualFileStorage.spec.ts +307 -0
- package/src/storage/VirtualFileStorage.ts +169 -0
- package/src/storage/index.ts +5 -0
- package/src/templates/csharp/README.md +48 -0
- package/src/templates/csharp/VectraClient.cs +234 -0
- package/src/templates/go/README.md +71 -0
- package/src/templates/go/vectra_client.go +322 -0
- package/src/templates/java/README.md +81 -0
- package/src/templates/java/VectraClient.java +232 -0
- package/src/templates/python/README.md +37 -0
- package/src/templates/python/vectra_client.py +279 -0
- package/src/templates/rust/Cargo.toml +14 -0
- package/src/templates/rust/README.md +39 -0
- package/src/templates/rust/build.rs +4 -0
- package/src/templates/rust/lib.rs +284 -0
- package/src/templates/typescript/README.md +96 -0
- package/src/templates/typescript/VectraClient.ts +374 -0
- package/src/types.ts +131 -123
- package/src/utils/index.ts +1 -0
- package/src/utils/pathUtils.ts +106 -0
- package/src/vectra-cli.generate.spec.ts +72 -0
- package/src/vectra-cli.spec.ts +0 -0
- package/src/vectra-cli.ts +687 -246
- package/README.draft.md +0 -499
- package/README.draft.outline.md +0 -160
- package/README.research.md +0 -2159
|
@@ -0,0 +1,288 @@
|
|
|
1
|
+
import assert from "node:assert";
|
|
2
|
+
import { describe, it, beforeEach, afterEach } from "mocha";
|
|
3
|
+
import sinon from "sinon";
|
|
4
|
+
import { WebFetcher } from "./WebFetcher";
|
|
5
|
+
|
|
6
|
+
describe("WebFetcher", () => {
|
|
7
|
+
let sandbox: sinon.SinonSandbox;
|
|
8
|
+
let fetchStub: sinon.SinonStub;
|
|
9
|
+
|
|
10
|
+
let lastUrl: string | undefined;
|
|
11
|
+
let lastInit: RequestInit | undefined;
|
|
12
|
+
|
|
13
|
+
function makeFetchResponse(status: number, data: string, contentType: string): Response {
|
|
14
|
+
return {
|
|
15
|
+
status,
|
|
16
|
+
statusText: status >= 400 ? `HTTP ${status}` : 'OK',
|
|
17
|
+
ok: status >= 200 && status < 300,
|
|
18
|
+
headers: new Headers({ 'content-type': contentType }),
|
|
19
|
+
text: async () => data,
|
|
20
|
+
json: async () => JSON.parse(data),
|
|
21
|
+
} as Response;
|
|
22
|
+
}
|
|
23
|
+
|
|
24
|
+
beforeEach(() => {
|
|
25
|
+
sandbox = sinon.createSandbox();
|
|
26
|
+
fetchStub = sandbox.stub(globalThis, 'fetch').callsFake(async (input: any, init?: any) => {
|
|
27
|
+
lastUrl = typeof input === 'string' ? input : input.url;
|
|
28
|
+
lastInit = init;
|
|
29
|
+
return makeFetchResponse(200, 'ok', 'text/plain');
|
|
30
|
+
});
|
|
31
|
+
lastUrl = undefined;
|
|
32
|
+
lastInit = undefined;
|
|
33
|
+
});
|
|
34
|
+
|
|
35
|
+
afterEach(() => {
|
|
36
|
+
sandbox.restore();
|
|
37
|
+
});
|
|
38
|
+
|
|
39
|
+
it("constructor uses defaults and merges config", () => {
|
|
40
|
+
const fetcherDefault = new WebFetcher();
|
|
41
|
+
assert.strictEqual(fetcherDefault["_config"].htmlToMarkdown, true);
|
|
42
|
+
assert.strictEqual(fetcherDefault["_config"].summarizeHtml, false);
|
|
43
|
+
|
|
44
|
+
const custom = {
|
|
45
|
+
htmlToMarkdown: false,
|
|
46
|
+
summarizeHtml: true,
|
|
47
|
+
headers: { Accept: "custom/type", "User-Agent": "custom-agent" },
|
|
48
|
+
requestConfig: { signal: AbortSignal.timeout(123) },
|
|
49
|
+
};
|
|
50
|
+
const fetcher = new WebFetcher(custom);
|
|
51
|
+
assert.strictEqual(fetcher["_config"].htmlToMarkdown, false);
|
|
52
|
+
assert.strictEqual(fetcher["_config"].summarizeHtml, true);
|
|
53
|
+
assert.deepStrictEqual(fetcher["_config"].headers, {
|
|
54
|
+
Accept: "custom/type",
|
|
55
|
+
"User-Agent": "custom-agent",
|
|
56
|
+
});
|
|
57
|
+
assert.ok(fetcher["_config"].requestConfig);
|
|
58
|
+
});
|
|
59
|
+
|
|
60
|
+
it("throws on HTTP error (>=400)", async () => {
|
|
61
|
+
fetchStub.resolves(makeFetchResponse(404, '<html></html>', 'text/html'));
|
|
62
|
+
const fetcher = new WebFetcher();
|
|
63
|
+
await assert.rejects(
|
|
64
|
+
() => fetcher.fetch("https://example.com/404", async () => true),
|
|
65
|
+
/Site returned an HTTP status of 404/
|
|
66
|
+
);
|
|
67
|
+
});
|
|
68
|
+
|
|
69
|
+
it("throws on invalid content-type", async () => {
|
|
70
|
+
fetchStub.resolves(makeFetchResponse(200, '…', 'image/png'));
|
|
71
|
+
const fetcher = new WebFetcher();
|
|
72
|
+
await assert.rejects(
|
|
73
|
+
() => fetcher.fetch("https://example.com/img", async () => true),
|
|
74
|
+
/Site returned an invalid content type of image\/png/
|
|
75
|
+
);
|
|
76
|
+
});
|
|
77
|
+
|
|
78
|
+
it("handles text/html with htmlToMarkdown=true with tables, alignments, and link rewriting", async () => {
|
|
79
|
+
const html = `
|
|
80
|
+
<html><body>
|
|
81
|
+
<script>evil()</script>
|
|
82
|
+
<a href="/rel">relative</a>
|
|
83
|
+
<a href="//cdn.example.com/lib.js">proto-rel</a>
|
|
84
|
+
<table>
|
|
85
|
+
<thead>
|
|
86
|
+
<tr>
|
|
87
|
+
<th align="left">H1</th>
|
|
88
|
+
<th align="right">H2</th>
|
|
89
|
+
<th align="center">H3</th>
|
|
90
|
+
</tr>
|
|
91
|
+
</thead>
|
|
92
|
+
<tbody>
|
|
93
|
+
<tr>
|
|
94
|
+
<td>Cell 1</td>
|
|
95
|
+
<td>Cell\t2</td>
|
|
96
|
+
<td>Cell\n3</td>
|
|
97
|
+
</tr>
|
|
98
|
+
</tbody>
|
|
99
|
+
</table>
|
|
100
|
+
<table>
|
|
101
|
+
<tbody>
|
|
102
|
+
<tr><th>A</th><th>B</th></tr>
|
|
103
|
+
<tr><td>a</td><td>b</td></tr>
|
|
104
|
+
</tbody>
|
|
105
|
+
</table>
|
|
106
|
+
</body></html>
|
|
107
|
+
`;
|
|
108
|
+
fetchStub.resolves(makeFetchResponse(200, html, 'text/html'));
|
|
109
|
+
|
|
110
|
+
const fetcher = new WebFetcher({ htmlToMarkdown: true });
|
|
111
|
+
const calls: any[] = [];
|
|
112
|
+
const onDocument = async (uri: string, text: string, docType?: string) => {
|
|
113
|
+
calls.push([uri, text, docType]);
|
|
114
|
+
return true;
|
|
115
|
+
};
|
|
116
|
+
|
|
117
|
+
const ok = await fetcher.fetch("https://example.com/page", onDocument);
|
|
118
|
+
assert.strictEqual(ok, true);
|
|
119
|
+
assert.strictEqual(calls.length, 1);
|
|
120
|
+
|
|
121
|
+
const [uri, md, docType] = calls[0];
|
|
122
|
+
assert.strictEqual(uri, "https://example.com/page");
|
|
123
|
+
assert.strictEqual(docType, "md");
|
|
124
|
+
|
|
125
|
+
assert.ok(!String(md).includes("<script>"));
|
|
126
|
+
assert.ok(String(md).includes("https://example.com/rel"));
|
|
127
|
+
assert.ok(String(md).includes("https://cdn.example.com/lib.js"));
|
|
128
|
+
assert.ok(String(md).includes("| H1 | H2 | H3 |"));
|
|
129
|
+
assert.ok(String(md).includes("| :-- | --: | :-: |"));
|
|
130
|
+
assert.ok(String(md).includes("| Cell 1 | Cell 2 | Cell 3 |"));
|
|
131
|
+
assert.ok(String(md).includes("| A | B |"));
|
|
132
|
+
});
|
|
133
|
+
|
|
134
|
+
it("handles text/html with htmlToMarkdown=false (passes raw html, docType 'html')", async () => {
|
|
135
|
+
const html = "<html><body><p>content</p></body></html>";
|
|
136
|
+
fetchStub.resolves(makeFetchResponse(200, html, 'text/html'));
|
|
137
|
+
|
|
138
|
+
const fetcher = new WebFetcher({ htmlToMarkdown: false });
|
|
139
|
+
const calls: any[] = [];
|
|
140
|
+
const onDocument = async (uri: string, text: string, docType?: string) => {
|
|
141
|
+
calls.push([uri, text, docType]);
|
|
142
|
+
return true;
|
|
143
|
+
};
|
|
144
|
+
|
|
145
|
+
const ok = await fetcher.fetch("https://example.com/raw", onDocument);
|
|
146
|
+
assert.strictEqual(ok, true);
|
|
147
|
+
assert.strictEqual(calls.length, 1);
|
|
148
|
+
|
|
149
|
+
const [_, text, docType] = calls[0];
|
|
150
|
+
assert.strictEqual(docType, "html");
|
|
151
|
+
assert.strictEqual(text, html);
|
|
152
|
+
});
|
|
153
|
+
|
|
154
|
+
it("handles application/json; charset=… (docType 'json')", async () => {
|
|
155
|
+
const json = '{"a":1}';
|
|
156
|
+
fetchStub.resolves(makeFetchResponse(200, json, 'application/json; charset=utf-8'));
|
|
157
|
+
const fetcher = new WebFetcher();
|
|
158
|
+
const calls: any[] = [];
|
|
159
|
+
const onDocument = async (...args: any[]) => {
|
|
160
|
+
calls.push(args);
|
|
161
|
+
return true;
|
|
162
|
+
};
|
|
163
|
+
const ok = await fetcher.fetch("https://example.com/data", onDocument);
|
|
164
|
+
assert.strictEqual(ok, true);
|
|
165
|
+
assert.strictEqual(calls[0][2], "json");
|
|
166
|
+
assert.strictEqual(calls[0][1], json);
|
|
167
|
+
});
|
|
168
|
+
|
|
169
|
+
it("handles application/xml (docType 'xml')", async () => {
|
|
170
|
+
const xml = "<root/>";
|
|
171
|
+
fetchStub.resolves(makeFetchResponse(200, xml, 'application/xml'));
|
|
172
|
+
const fetcher = new WebFetcher();
|
|
173
|
+
const calls: any[] = [];
|
|
174
|
+
const onDocument = async (...args: any[]) => {
|
|
175
|
+
calls.push(args);
|
|
176
|
+
return true;
|
|
177
|
+
};
|
|
178
|
+
const ok = await fetcher.fetch("https://example.com/xml", onDocument);
|
|
179
|
+
assert.strictEqual(ok, true);
|
|
180
|
+
assert.strictEqual(calls[0][2], "xml");
|
|
181
|
+
assert.strictEqual(calls[0][1], xml);
|
|
182
|
+
});
|
|
183
|
+
|
|
184
|
+
it("handles application/javascript (docType 'javascript')", async () => {
|
|
185
|
+
const js = "console.log('hi');";
|
|
186
|
+
fetchStub.resolves(makeFetchResponse(200, js, 'application/javascript'));
|
|
187
|
+
const fetcher = new WebFetcher();
|
|
188
|
+
const calls: any[] = [];
|
|
189
|
+
const onDocument = async (...args: any[]) => {
|
|
190
|
+
calls.push(args);
|
|
191
|
+
return true;
|
|
192
|
+
};
|
|
193
|
+
const ok = await fetcher.fetch("https://example.com/app.js", onDocument);
|
|
194
|
+
assert.strictEqual(ok, true);
|
|
195
|
+
assert.strictEqual(calls[0][2], "javascript");
|
|
196
|
+
assert.strictEqual(calls[0][1], js);
|
|
197
|
+
});
|
|
198
|
+
|
|
199
|
+
it("handles text/plain (docType undefined)", async () => {
|
|
200
|
+
const text = "plain";
|
|
201
|
+
fetchStub.resolves(makeFetchResponse(200, text, 'text/plain'));
|
|
202
|
+
const fetcher = new WebFetcher();
|
|
203
|
+
const calls: any[] = [];
|
|
204
|
+
const onDocument = async (...args: any[]) => {
|
|
205
|
+
calls.push(args);
|
|
206
|
+
return true;
|
|
207
|
+
};
|
|
208
|
+
const ok = await fetcher.fetch("https://example.com/txt", onDocument);
|
|
209
|
+
assert.strictEqual(ok, true);
|
|
210
|
+
assert.strictEqual(calls[0][2], undefined);
|
|
211
|
+
assert.strictEqual(calls[0][1], text);
|
|
212
|
+
});
|
|
213
|
+
|
|
214
|
+
it("sets Host and Alt-Used to request hostname; merges and does not mutate caller headers", async () => {
|
|
215
|
+
let capturedUrl: string | undefined;
|
|
216
|
+
let capturedInit: RequestInit | undefined;
|
|
217
|
+
fetchStub.callsFake(async (url: string, init?: RequestInit) => {
|
|
218
|
+
capturedUrl = url;
|
|
219
|
+
capturedInit = init;
|
|
220
|
+
return makeFetchResponse(200, 'ok', 'text/plain');
|
|
221
|
+
});
|
|
222
|
+
|
|
223
|
+
const userHeaders: Record<string, string> = {
|
|
224
|
+
Accept: "custom/type",
|
|
225
|
+
"User-Agent": "custom-agent",
|
|
226
|
+
};
|
|
227
|
+
const fetcher = new WebFetcher({ headers: userHeaders });
|
|
228
|
+
|
|
229
|
+
const onDocument = async () => true;
|
|
230
|
+
const url = "https://host.example.com/path?q=1";
|
|
231
|
+
const ok = await fetcher.fetch(url, onDocument);
|
|
232
|
+
assert.strictEqual(ok, true);
|
|
233
|
+
|
|
234
|
+
assert.strictEqual(capturedUrl, url);
|
|
235
|
+
assert.ok(capturedInit);
|
|
236
|
+
assert.ok(capturedInit!.headers);
|
|
237
|
+
|
|
238
|
+
const headers = capturedInit!.headers as Record<string, string>;
|
|
239
|
+
assert.strictEqual(headers.Host, "host.example.com");
|
|
240
|
+
assert.strictEqual(headers["Alt-Used"], "host.example.com");
|
|
241
|
+
|
|
242
|
+
assert.strictEqual(headers.Accept, "custom/type");
|
|
243
|
+
assert.strictEqual(headers["User-Agent"], "custom-agent");
|
|
244
|
+
|
|
245
|
+
assert.strictEqual((userHeaders as any).Host, undefined);
|
|
246
|
+
assert.strictEqual((userHeaders as any)["Alt-Used"], undefined);
|
|
247
|
+
});
|
|
248
|
+
|
|
249
|
+
it("merges requestConfig options into fetch call", async () => {
|
|
250
|
+
let capturedInit: RequestInit | undefined;
|
|
251
|
+
fetchStub.callsFake(async (_url: string, init?: RequestInit) => {
|
|
252
|
+
capturedInit = init;
|
|
253
|
+
return makeFetchResponse(200, 'ok', 'text/plain');
|
|
254
|
+
});
|
|
255
|
+
|
|
256
|
+
const fetcher = new WebFetcher({
|
|
257
|
+
requestConfig: { keepalive: true },
|
|
258
|
+
});
|
|
259
|
+
const ok = await fetcher.fetch("https://example.com", async () => true);
|
|
260
|
+
assert.strictEqual(ok, true);
|
|
261
|
+
assert.strictEqual(capturedInit!.keepalive, true);
|
|
262
|
+
});
|
|
263
|
+
|
|
264
|
+
it("htmlToMarkdown trims overly long header text when first space/newline index > 64", () => {
|
|
265
|
+
const fetcher = new WebFetcher();
|
|
266
|
+
const longText = "a".repeat(70) + " rest\nmore";
|
|
267
|
+
const html = `<html><body><p>${longText}</p></body></html>`;
|
|
268
|
+
const md = (fetcher as any)["htmlToMarkdown"](html, "https://example.com");
|
|
269
|
+
assert.ok(!md.startsWith("a".repeat(70)));
|
|
270
|
+
assert.ok(md.includes("rest"));
|
|
271
|
+
});
|
|
272
|
+
|
|
273
|
+
it("htmlToMarkdown leaves short content unchanged", () => {
|
|
274
|
+
const fetcher = new WebFetcher();
|
|
275
|
+
const html = `<html><body><p>short text</p></body></html>`;
|
|
276
|
+
const md = (fetcher as any)["htmlToMarkdown"](html, "https://example.com");
|
|
277
|
+
assert.ok(md.includes("short text"));
|
|
278
|
+
});
|
|
279
|
+
|
|
280
|
+
it("propagates onDocument return value (true/false)", async () => {
|
|
281
|
+
fetchStub.resolves(makeFetchResponse(200, 'ok', 'text/plain'));
|
|
282
|
+
const fetcher = new WebFetcher();
|
|
283
|
+
const yes = await fetcher.fetch("https://example.com/yes", async () => true);
|
|
284
|
+
const no = await fetcher.fetch("https://example.com/no", async () => false);
|
|
285
|
+
assert.strictEqual(yes, true);
|
|
286
|
+
assert.strictEqual(no, false);
|
|
287
|
+
});
|
|
288
|
+
});
|