opencode-dux 1.0.0
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/LICENSE +21 -0
- package/README.md +452 -0
- package/dist/agents/descriptions.d.ts +6 -0
- package/dist/agents/designer.d.ts +2 -0
- package/dist/agents/explorer.d.ts +2 -0
- package/dist/agents/fixer.d.ts +2 -0
- package/dist/agents/index.d.ts +22 -0
- package/dist/agents/interpreter.d.ts +2 -0
- package/dist/agents/librarian.d.ts +2 -0
- package/dist/agents/oracle.d.ts +2 -0
- package/dist/agents/orchestrator.d.ts +27 -0
- package/dist/agents/overrides.d.ts +18 -0
- package/dist/agents/prompt-blocks.d.ts +97 -0
- package/dist/agents/steward.d.ts +3 -0
- package/dist/cli/config-io.d.ts +24 -0
- package/dist/cli/config-manager.d.ts +4 -0
- package/dist/cli/index.d.ts +2 -0
- package/dist/cli/index.js +1006 -0
- package/dist/cli/install.d.ts +2 -0
- package/dist/cli/mcps.d.ts +13 -0
- package/dist/cli/model-key-normalization.d.ts +1 -0
- package/dist/cli/paths.d.ts +35 -0
- package/dist/cli/providers.d.ts +137 -0
- package/dist/cli/skills.d.ts +22 -0
- package/dist/cli/system.d.ts +5 -0
- package/dist/cli/types.d.ts +38 -0
- package/dist/config/constants.d.ts +12 -0
- package/dist/config/index.d.ts +4 -0
- package/dist/config/loader.d.ts +40 -0
- package/dist/config/runtime-preset.d.ts +12 -0
- package/dist/config/schema.d.ts +281 -0
- package/dist/config/utils.d.ts +10 -0
- package/dist/discovery/local/types.d.ts +79 -0
- package/dist/discovery/local.d.ts +73 -0
- package/dist/discovery/mcp-servers.d.ts +88 -0
- package/dist/discovery/skills.d.ts +94 -0
- package/dist/hooks/apply-patch/codec.d.ts +7 -0
- package/dist/hooks/apply-patch/errors.d.ts +25 -0
- package/dist/hooks/apply-patch/execution-context.d.ts +27 -0
- package/dist/hooks/apply-patch/index.d.ts +15 -0
- package/dist/hooks/apply-patch/matching.d.ts +26 -0
- package/dist/hooks/apply-patch/operations.d.ts +3 -0
- package/dist/hooks/apply-patch/patch.d.ts +2 -0
- package/dist/hooks/apply-patch/prepared-changes.d.ts +17 -0
- package/dist/hooks/apply-patch/resolution.d.ts +19 -0
- package/dist/hooks/apply-patch/rewrite.d.ts +7 -0
- package/dist/hooks/apply-patch/test-helpers.d.ts +6 -0
- package/dist/hooks/apply-patch/types.d.ts +80 -0
- package/dist/hooks/auto-update-checker/cache.d.ts +11 -0
- package/dist/hooks/auto-update-checker/checker.d.ts +32 -0
- package/dist/hooks/auto-update-checker/constants.d.ts +11 -0
- package/dist/hooks/auto-update-checker/index.d.ts +18 -0
- package/dist/hooks/auto-update-checker/types.d.ts +22 -0
- package/dist/hooks/chat-headers.d.ts +16 -0
- package/dist/hooks/context-pressure-reminder/index.d.ts +33 -0
- package/dist/hooks/delegate-task-retry/guidance.d.ts +2 -0
- package/dist/hooks/delegate-task-retry/hook.d.ts +8 -0
- package/dist/hooks/delegate-task-retry/index.d.ts +4 -0
- package/dist/hooks/delegate-task-retry/patterns.d.ts +11 -0
- package/dist/hooks/filter-available-skills/index.d.ts +32 -0
- package/dist/hooks/foreground-fallback/index.d.ts +72 -0
- package/dist/hooks/image-hook.d.ts +5 -0
- package/dist/hooks/index.d.ts +14 -0
- package/dist/hooks/json-error-recovery/hook.d.ts +18 -0
- package/dist/hooks/json-error-recovery/index.d.ts +1 -0
- package/dist/hooks/phase-reminder/index.d.ts +26 -0
- package/dist/hooks/post-file-tool-nudge/index.d.ts +19 -0
- package/dist/hooks/task-session-manager/index.d.ts +52 -0
- package/dist/hooks/todo-continuation/index.d.ts +53 -0
- package/dist/hooks/todo-continuation/todo-hygiene.d.ts +35 -0
- package/dist/index.d.ts +5 -0
- package/dist/index.js +31782 -0
- package/dist/mcp/context7.d.ts +6 -0
- package/dist/mcp/grep-app.d.ts +6 -0
- package/dist/mcp/index.d.ts +13 -0
- package/dist/mcp/types.d.ts +12 -0
- package/dist/mcp/websearch.d.ts +9 -0
- package/dist/skills/registry.d.ts +29 -0
- package/dist/subscriptions/accounts-store.d.ts +57 -0
- package/dist/subscriptions/index.d.ts +13 -0
- package/dist/subscriptions/neuralwatt-scraper.d.ts +14 -0
- package/dist/subscriptions/opencode-go-scraper.d.ts +27 -0
- package/dist/subscriptions/types.d.ts +115 -0
- package/dist/subscriptions/usage-service.d.ts +74 -0
- package/dist/tools/ast-grep/cli.d.ts +15 -0
- package/dist/tools/ast-grep/constants.d.ts +25 -0
- package/dist/tools/ast-grep/downloader.d.ts +5 -0
- package/dist/tools/ast-grep/index.d.ts +10 -0
- package/dist/tools/ast-grep/tools.d.ts +3 -0
- package/dist/tools/ast-grep/types.d.ts +30 -0
- package/dist/tools/ast-grep/utils.d.ts +4 -0
- package/dist/tools/delegate.d.ts +14 -0
- package/dist/tools/index.d.ts +5 -0
- package/dist/tools/preset-manager.d.ts +27 -0
- package/dist/tools/smartfetch/binary.d.ts +3 -0
- package/dist/tools/smartfetch/cache.d.ts +6 -0
- package/dist/tools/smartfetch/constants.d.ts +12 -0
- package/dist/tools/smartfetch/index.d.ts +3 -0
- package/dist/tools/smartfetch/network.d.ts +38 -0
- package/dist/tools/smartfetch/secondary-model.d.ts +28 -0
- package/dist/tools/smartfetch/tool.d.ts +3 -0
- package/dist/tools/smartfetch/types.d.ts +122 -0
- package/dist/tools/smartfetch/utils.d.ts +18 -0
- package/dist/tui-state.d.ts +168 -0
- package/dist/tui.d.ts +37 -0
- package/dist/tui.js +1896 -0
- package/dist/utils/agent-variant.d.ts +63 -0
- package/dist/utils/compat.d.ts +30 -0
- package/dist/utils/env.d.ts +1 -0
- package/dist/utils/index.d.ts +9 -0
- package/dist/utils/internal-initiator.d.ts +6 -0
- package/dist/utils/logger.d.ts +8 -0
- package/dist/utils/polling.d.ts +21 -0
- package/dist/utils/session-manager.d.ts +55 -0
- package/dist/utils/session.d.ts +90 -0
- package/dist/utils/subagent-depth.d.ts +35 -0
- package/dist/utils/system-collapse.d.ts +6 -0
- package/dist/utils/task.d.ts +4 -0
- package/dist/utils/zip-extractor.d.ts +1 -0
- package/index.ts +1 -0
- package/opencode-dux.schema.json +634 -0
- package/package.json +103 -0
- package/src/agents/descriptions.ts +55 -0
- package/src/agents/designer.test.ts +86 -0
- package/src/agents/designer.ts +154 -0
- package/src/agents/display-name.test.ts +186 -0
- package/src/agents/explorer.test.ts +79 -0
- package/src/agents/explorer.ts +144 -0
- package/src/agents/fixer.test.ts +79 -0
- package/src/agents/fixer.ts +145 -0
- package/src/agents/index.test.ts +472 -0
- package/src/agents/index.ts +248 -0
- package/src/agents/interpreter.ts +136 -0
- package/src/agents/librarian.test.ts +80 -0
- package/src/agents/librarian.ts +145 -0
- package/src/agents/oracle.test.ts +89 -0
- package/src/agents/oracle.ts +184 -0
- package/src/agents/orchestrator.test.ts +116 -0
- package/src/agents/orchestrator.ts +574 -0
- package/src/agents/overrides.ts +95 -0
- package/src/agents/prompt-blocks.test.ts +114 -0
- package/src/agents/prompt-blocks.ts +640 -0
- package/src/agents/steward.ts +146 -0
- package/src/cli/config-io.test.ts +536 -0
- package/src/cli/config-io.ts +473 -0
- package/src/cli/config-manager.test.ts +141 -0
- package/src/cli/config-manager.ts +4 -0
- package/src/cli/index.ts +88 -0
- package/src/cli/install.ts +282 -0
- package/src/cli/mcps.test.ts +62 -0
- package/src/cli/mcps.ts +39 -0
- package/src/cli/model-key-normalization.test.ts +21 -0
- package/src/cli/model-key-normalization.ts +60 -0
- package/src/cli/paths.test.ts +167 -0
- package/src/cli/paths.ts +144 -0
- package/src/cli/providers.test.ts +118 -0
- package/src/cli/providers.ts +141 -0
- package/src/cli/skills.test.ts +111 -0
- package/src/cli/skills.ts +103 -0
- package/src/cli/system.test.ts +91 -0
- package/src/cli/system.ts +180 -0
- package/src/cli/types.ts +43 -0
- package/src/config/constants.ts +58 -0
- package/src/config/index.ts +4 -0
- package/src/config/loader.test.ts +1194 -0
- package/src/config/loader.ts +269 -0
- package/src/config/model-resolution.test.ts +176 -0
- package/src/config/runtime-preset.test.ts +61 -0
- package/src/config/runtime-preset.ts +37 -0
- package/src/config/schema.ts +248 -0
- package/src/config/utils.test.ts +41 -0
- package/src/config/utils.ts +23 -0
- package/src/discovery/local/types.ts +85 -0
- package/src/discovery/local.ts +322 -0
- package/src/discovery/mcp-servers.ts +804 -0
- package/src/discovery/skills.ts +959 -0
- package/src/hooks/apply-patch/codec.test.ts +184 -0
- package/src/hooks/apply-patch/codec.ts +352 -0
- package/src/hooks/apply-patch/errors.ts +117 -0
- package/src/hooks/apply-patch/execution-context.ts +432 -0
- package/src/hooks/apply-patch/hook.test.ts +768 -0
- package/src/hooks/apply-patch/index.ts +126 -0
- package/src/hooks/apply-patch/matching.test.ts +215 -0
- package/src/hooks/apply-patch/matching.ts +586 -0
- package/src/hooks/apply-patch/operations.test.ts +1535 -0
- package/src/hooks/apply-patch/operations.ts +3 -0
- package/src/hooks/apply-patch/patch.ts +9 -0
- package/src/hooks/apply-patch/prepared-changes.ts +400 -0
- package/src/hooks/apply-patch/resolution.test.ts +420 -0
- package/src/hooks/apply-patch/resolution.ts +437 -0
- package/src/hooks/apply-patch/rewrite.ts +496 -0
- package/src/hooks/apply-patch/test-helpers.ts +52 -0
- package/src/hooks/apply-patch/types.ts +111 -0
- package/src/hooks/auto-update-checker/cache.test.ts +179 -0
- package/src/hooks/auto-update-checker/cache.ts +188 -0
- package/src/hooks/auto-update-checker/checker.test.ts +159 -0
- package/src/hooks/auto-update-checker/checker.ts +308 -0
- package/src/hooks/auto-update-checker/constants.ts +33 -0
- package/src/hooks/auto-update-checker/index.test.ts +282 -0
- package/src/hooks/auto-update-checker/index.ts +225 -0
- package/src/hooks/auto-update-checker/types.ts +26 -0
- package/src/hooks/chat-headers.test.ts +236 -0
- package/src/hooks/chat-headers.ts +97 -0
- package/src/hooks/context-pressure-reminder/index.test.ts +179 -0
- package/src/hooks/context-pressure-reminder/index.ts +137 -0
- package/src/hooks/delegate-task-retry/guidance.ts +41 -0
- package/src/hooks/delegate-task-retry/hook.ts +23 -0
- package/src/hooks/delegate-task-retry/index.test.ts +38 -0
- package/src/hooks/delegate-task-retry/index.ts +7 -0
- package/src/hooks/delegate-task-retry/patterns.ts +79 -0
- package/src/hooks/filter-available-skills/index.test.ts +297 -0
- package/src/hooks/filter-available-skills/index.ts +160 -0
- package/src/hooks/foreground-fallback/index.test.ts +624 -0
- package/src/hooks/foreground-fallback/index.ts +374 -0
- package/src/hooks/image-hook.ts +6 -0
- package/src/hooks/index.ts +17 -0
- package/src/hooks/json-error-recovery/hook.ts +73 -0
- package/src/hooks/json-error-recovery/index.test.ts +111 -0
- package/src/hooks/json-error-recovery/index.ts +6 -0
- package/src/hooks/phase-reminder/index.test.ts +74 -0
- package/src/hooks/phase-reminder/index.ts +85 -0
- package/src/hooks/post-file-tool-nudge/index.test.ts +94 -0
- package/src/hooks/post-file-tool-nudge/index.ts +63 -0
- package/src/hooks/task-session-manager/index.test.ts +833 -0
- package/src/hooks/task-session-manager/index.ts +434 -0
- package/src/hooks/todo-continuation/index.test.ts +3026 -0
- package/src/hooks/todo-continuation/index.ts +878 -0
- package/src/hooks/todo-continuation/todo-hygiene.test.ts +204 -0
- package/src/hooks/todo-continuation/todo-hygiene.ts +207 -0
- package/src/index.ts +1672 -0
- package/src/mcp/context7.ts +14 -0
- package/src/mcp/grep-app.ts +11 -0
- package/src/mcp/index.test.ts +96 -0
- package/src/mcp/index.ts +66 -0
- package/src/mcp/types.ts +16 -0
- package/src/mcp/websearch.ts +47 -0
- package/src/skills/codemap/README.md +60 -0
- package/src/skills/codemap/SKILL.md +174 -0
- package/src/skills/codemap/scripts/codemap.mjs +483 -0
- package/src/skills/codemap/scripts/codemap.test.ts +129 -0
- package/src/skills/registry.ts +218 -0
- package/src/skills/simplify/README.md +19 -0
- package/src/skills/simplify/SKILL.md +138 -0
- package/src/subscriptions/accounts-store.test.ts +236 -0
- package/src/subscriptions/accounts-store.ts +184 -0
- package/src/subscriptions/index.ts +30 -0
- package/src/subscriptions/neuralwatt-scraper.ts +108 -0
- package/src/subscriptions/opencode-go-scraper.ts +301 -0
- package/src/subscriptions/types.ts +145 -0
- package/src/subscriptions/usage-service.test.ts +202 -0
- package/src/subscriptions/usage-service.ts +651 -0
- package/src/tools/ast-grep/cli.ts +257 -0
- package/src/tools/ast-grep/constants.ts +214 -0
- package/src/tools/ast-grep/downloader.ts +131 -0
- package/src/tools/ast-grep/index.ts +24 -0
- package/src/tools/ast-grep/tools.ts +117 -0
- package/src/tools/ast-grep/types.ts +51 -0
- package/src/tools/ast-grep/utils.ts +126 -0
- package/src/tools/delegate-handoff.test.ts +18 -0
- package/src/tools/delegate.ts +508 -0
- package/src/tools/index.ts +8 -0
- package/src/tools/preset-manager.test.ts +795 -0
- package/src/tools/preset-manager.ts +332 -0
- package/src/tools/smartfetch/binary.ts +58 -0
- package/src/tools/smartfetch/cache.test.ts +34 -0
- package/src/tools/smartfetch/cache.ts +112 -0
- package/src/tools/smartfetch/constants.ts +29 -0
- package/src/tools/smartfetch/index.ts +8 -0
- package/src/tools/smartfetch/network.test.ts +178 -0
- package/src/tools/smartfetch/network.ts +614 -0
- package/src/tools/smartfetch/secondary-model.test.ts +85 -0
- package/src/tools/smartfetch/secondary-model.ts +276 -0
- package/src/tools/smartfetch/tool.test.ts +60 -0
- package/src/tools/smartfetch/tool.ts +832 -0
- package/src/tools/smartfetch/types.ts +135 -0
- package/src/tools/smartfetch/utils.test.ts +24 -0
- package/src/tools/smartfetch/utils.ts +456 -0
- package/src/tui-state.test.ts +867 -0
- package/src/tui-state.ts +1255 -0
- package/src/tui.test.ts +336 -0
- package/src/tui.ts +1539 -0
- package/src/utils/agent-variant.test.ts +244 -0
- package/src/utils/agent-variant.ts +187 -0
- package/src/utils/compat.ts +91 -0
- package/src/utils/env.ts +12 -0
- package/src/utils/index.ts +9 -0
- package/src/utils/internal-initiator.ts +28 -0
- package/src/utils/logger.test.ts +220 -0
- package/src/utils/logger.ts +136 -0
- package/src/utils/polling.test.ts +191 -0
- package/src/utils/polling.ts +67 -0
- package/src/utils/session-manager.test.ts +173 -0
- package/src/utils/session-manager.ts +356 -0
- package/src/utils/session.test.ts +110 -0
- package/src/utils/session.ts +389 -0
- package/src/utils/subagent-depth.test.ts +170 -0
- package/src/utils/subagent-depth.ts +75 -0
- package/src/utils/system-collapse.test.ts +86 -0
- package/src/utils/system-collapse.ts +24 -0
- package/src/utils/task.test.ts +24 -0
- package/src/utils/task.ts +20 -0
- package/src/utils/zip-extractor.ts +102 -0
|
@@ -0,0 +1,832 @@
|
|
|
1
|
+
import os from 'node:os';
|
|
2
|
+
import path from 'node:path';
|
|
3
|
+
import {
|
|
4
|
+
type PluginInput,
|
|
5
|
+
type ToolDefinition,
|
|
6
|
+
tool,
|
|
7
|
+
} from '@opencode-ai/plugin';
|
|
8
|
+
import { buildBinaryResultMessage, saveBinary } from './binary';
|
|
9
|
+
import {
|
|
10
|
+
buildCacheKey,
|
|
11
|
+
CACHE,
|
|
12
|
+
cacheFetchResult,
|
|
13
|
+
isInvalidLlmsResult,
|
|
14
|
+
} from './cache';
|
|
15
|
+
import {
|
|
16
|
+
DEFAULT_TIMEOUT_SECONDS,
|
|
17
|
+
MAX_BINARY_DOWNLOAD_BYTES,
|
|
18
|
+
MAX_LLMS_PROBE_TIMEOUT_MS,
|
|
19
|
+
MAX_RESPONSE_BYTES,
|
|
20
|
+
MAX_TIMEOUT_SECONDS,
|
|
21
|
+
WEBFETCH_DESCRIPTION,
|
|
22
|
+
} from './constants';
|
|
23
|
+
import {
|
|
24
|
+
buildAllowedOrigins,
|
|
25
|
+
buildConditionalHeaders,
|
|
26
|
+
buildPermissionPatterns,
|
|
27
|
+
decodeBody,
|
|
28
|
+
extractHeaderMetadata,
|
|
29
|
+
fetchWithUpgradeFallback,
|
|
30
|
+
getBinaryKind,
|
|
31
|
+
isBinaryContentType,
|
|
32
|
+
isDocsLikeUrl,
|
|
33
|
+
isGenericBinaryMime,
|
|
34
|
+
isHtmlLikeContentType,
|
|
35
|
+
looksLikeHtmlText,
|
|
36
|
+
looksLikeTextBody,
|
|
37
|
+
normalizeUrl,
|
|
38
|
+
probeLlmsText,
|
|
39
|
+
readBodyLimited,
|
|
40
|
+
runWithScopedTimeout,
|
|
41
|
+
} from './network';
|
|
42
|
+
import {
|
|
43
|
+
decideSecondaryModelUse,
|
|
44
|
+
readSecondaryModelFromConfig,
|
|
45
|
+
runSecondaryModelWithFallback,
|
|
46
|
+
} from './secondary-model';
|
|
47
|
+
import type { RedirectStep, SmartfetchOptions } from './types';
|
|
48
|
+
import {
|
|
49
|
+
buildLlmsRequiredMessage,
|
|
50
|
+
buildRedirectResultMessage,
|
|
51
|
+
cleanFetchedText,
|
|
52
|
+
detectQualitySignals,
|
|
53
|
+
escapeHtml,
|
|
54
|
+
extractFromHtml,
|
|
55
|
+
extractHeadingsFromMarkdown,
|
|
56
|
+
frontmatter,
|
|
57
|
+
inferCanonicalUrlFromText,
|
|
58
|
+
joinRenderedContent,
|
|
59
|
+
pickContent,
|
|
60
|
+
renderMessageForFormat,
|
|
61
|
+
trimBlankRuns,
|
|
62
|
+
withTruncationMarker,
|
|
63
|
+
wordCount,
|
|
64
|
+
} from './utils';
|
|
65
|
+
|
|
66
|
+
const z = tool.schema;
|
|
67
|
+
|
|
68
|
+
export function createWebfetchTool(
|
|
69
|
+
pluginCtx: PluginInput,
|
|
70
|
+
options: SmartfetchOptions = {},
|
|
71
|
+
): ToolDefinition {
|
|
72
|
+
const binaryDir =
|
|
73
|
+
options.binaryDir || path.join(os.tmpdir(), 'opencode-smartfetch');
|
|
74
|
+
|
|
75
|
+
return tool({
|
|
76
|
+
description: WEBFETCH_DESCRIPTION,
|
|
77
|
+
args: {
|
|
78
|
+
url: z.httpUrl(),
|
|
79
|
+
format: z.enum(['text', 'markdown', 'html']).default('markdown'),
|
|
80
|
+
timeout: z
|
|
81
|
+
.number()
|
|
82
|
+
.positive()
|
|
83
|
+
.max(MAX_TIMEOUT_SECONDS)
|
|
84
|
+
.optional()
|
|
85
|
+
.describe('Timeout in seconds, max 120.'),
|
|
86
|
+
prompt: z
|
|
87
|
+
.string()
|
|
88
|
+
.optional()
|
|
89
|
+
.describe(
|
|
90
|
+
'Optional extraction task to run on the fetched content using a cheap secondary model.',
|
|
91
|
+
),
|
|
92
|
+
extract_main: z.boolean().default(true),
|
|
93
|
+
prefer_llms_txt: z.enum(['auto', 'always', 'never']).default('auto'),
|
|
94
|
+
include_metadata: z.boolean().default(true),
|
|
95
|
+
save_binary: z
|
|
96
|
+
.boolean()
|
|
97
|
+
.default(false)
|
|
98
|
+
.describe(
|
|
99
|
+
'Save binary payload to disk when it fits within the active download limit.',
|
|
100
|
+
),
|
|
101
|
+
},
|
|
102
|
+
async execute(args, ctx) {
|
|
103
|
+
const secondaryModels = await readSecondaryModelFromConfig(
|
|
104
|
+
ctx.directory || pluginCtx.directory,
|
|
105
|
+
);
|
|
106
|
+
const normalized = normalizeUrl(args.url);
|
|
107
|
+
const url = new URL(normalized.url);
|
|
108
|
+
const cacheKey = buildCacheKey(
|
|
109
|
+
args.url,
|
|
110
|
+
args.extract_main,
|
|
111
|
+
args.prefer_llms_txt,
|
|
112
|
+
args.save_binary,
|
|
113
|
+
);
|
|
114
|
+
const shouldProbeLlmsTxt =
|
|
115
|
+
args.prefer_llms_txt === 'always' ||
|
|
116
|
+
(args.prefer_llms_txt === 'auto' && isDocsLikeUrl(url));
|
|
117
|
+
const permissionPatterns = buildPermissionPatterns(
|
|
118
|
+
normalized,
|
|
119
|
+
shouldProbeLlmsTxt,
|
|
120
|
+
);
|
|
121
|
+
const allowedOrigins = buildAllowedOrigins(permissionPatterns);
|
|
122
|
+
|
|
123
|
+
await ctx.ask({
|
|
124
|
+
permission: 'webfetch',
|
|
125
|
+
patterns: permissionPatterns,
|
|
126
|
+
always: permissionPatterns,
|
|
127
|
+
metadata: {
|
|
128
|
+
url: normalized.url,
|
|
129
|
+
requested_url: args.url,
|
|
130
|
+
fallback_url: normalized.fallbackUrl,
|
|
131
|
+
llms_probe_enabled: shouldProbeLlmsTxt,
|
|
132
|
+
format: args.format,
|
|
133
|
+
prompt: args.prompt,
|
|
134
|
+
},
|
|
135
|
+
});
|
|
136
|
+
|
|
137
|
+
const timeoutMs = Math.min(
|
|
138
|
+
(args.timeout ?? DEFAULT_TIMEOUT_SECONDS) * 1000,
|
|
139
|
+
MAX_TIMEOUT_SECONDS * 1000,
|
|
140
|
+
);
|
|
141
|
+
const controller = new AbortController();
|
|
142
|
+
const timeout = setTimeout(
|
|
143
|
+
() => controller.abort(new Error(`timeout after ${timeoutMs}ms`)),
|
|
144
|
+
timeoutMs,
|
|
145
|
+
);
|
|
146
|
+
const abortHandler = () => controller.abort(ctx.abort.reason);
|
|
147
|
+
ctx.abort.addEventListener('abort', abortHandler, { once: true });
|
|
148
|
+
|
|
149
|
+
try {
|
|
150
|
+
let fetchResult = CACHE.get(cacheKey);
|
|
151
|
+
if (isInvalidLlmsResult(fetchResult)) {
|
|
152
|
+
CACHE.delete(cacheKey);
|
|
153
|
+
fetchResult = undefined;
|
|
154
|
+
}
|
|
155
|
+
const cacheHit = !!fetchResult;
|
|
156
|
+
if (fetchResult) {
|
|
157
|
+
fetchResult = {
|
|
158
|
+
...fetchResult,
|
|
159
|
+
requestedUrl: args.url,
|
|
160
|
+
cacheHit: true,
|
|
161
|
+
};
|
|
162
|
+
}
|
|
163
|
+
if (!fetchResult) {
|
|
164
|
+
let staleFetchResult = CACHE.get(cacheKey, {
|
|
165
|
+
allowStale: true,
|
|
166
|
+
noDeleteOnStaleGet: true,
|
|
167
|
+
});
|
|
168
|
+
if (isInvalidLlmsResult(staleFetchResult)) {
|
|
169
|
+
CACHE.delete(cacheKey);
|
|
170
|
+
staleFetchResult = undefined;
|
|
171
|
+
}
|
|
172
|
+
let llmsProbeError: string | undefined;
|
|
173
|
+
|
|
174
|
+
if (shouldProbeLlmsTxt) {
|
|
175
|
+
const fallbackOrigin = normalized.fallbackUrl
|
|
176
|
+
? new URL(normalized.fallbackUrl).origin
|
|
177
|
+
: undefined;
|
|
178
|
+
const probeTimeoutMs = Math.max(
|
|
179
|
+
1,
|
|
180
|
+
Math.min(MAX_LLMS_PROBE_TIMEOUT_MS, timeoutMs),
|
|
181
|
+
);
|
|
182
|
+
const llms = await runWithScopedTimeout(
|
|
183
|
+
controller.signal,
|
|
184
|
+
probeTimeoutMs,
|
|
185
|
+
(probeSignal) =>
|
|
186
|
+
probeLlmsText(url, timeoutMs, probeSignal, fallbackOrigin),
|
|
187
|
+
);
|
|
188
|
+
if (llms && 'text' in llms) {
|
|
189
|
+
const llmsHeaders = llms.headers || {};
|
|
190
|
+
const text = trimBlankRuns(llms.text);
|
|
191
|
+
fetchResult = {
|
|
192
|
+
requestedUrl: args.url,
|
|
193
|
+
finalUrl: llms.url,
|
|
194
|
+
statusCode: llms.statusCode,
|
|
195
|
+
contentType: llmsHeaders.contentType || 'text/plain',
|
|
196
|
+
charset: llmsHeaders.charset,
|
|
197
|
+
etag: llmsHeaders.etag,
|
|
198
|
+
lastModified: llmsHeaders.lastModified,
|
|
199
|
+
contentLength: llmsHeaders.contentLength,
|
|
200
|
+
filename: llmsHeaders.filename,
|
|
201
|
+
canonicalUrl: inferCanonicalUrlFromText(text, llms.url),
|
|
202
|
+
headings: extractHeadingsFromMarkdown(text),
|
|
203
|
+
title: undefined,
|
|
204
|
+
rawContent: text,
|
|
205
|
+
markdown: text,
|
|
206
|
+
text,
|
|
207
|
+
html: text,
|
|
208
|
+
extractedMain: false,
|
|
209
|
+
usedLlmsTxt: true,
|
|
210
|
+
sourceKind: 'llms_txt',
|
|
211
|
+
upgradedToHttps: !!llms.upgradedToHttps,
|
|
212
|
+
redirectChain: llms.redirectChain || [],
|
|
213
|
+
truncated: !!llms.truncated,
|
|
214
|
+
wordCount: wordCount(text),
|
|
215
|
+
qualitySignals: detectQualitySignals({
|
|
216
|
+
text,
|
|
217
|
+
markdown: text,
|
|
218
|
+
rawContent: text,
|
|
219
|
+
wordCount: wordCount(text),
|
|
220
|
+
sourceKind: 'llms_txt',
|
|
221
|
+
extractedMain: false,
|
|
222
|
+
}),
|
|
223
|
+
llmsProbeTruncated: !!llms.truncated,
|
|
224
|
+
decodedCharset: llms.decodedCharset,
|
|
225
|
+
decodeFallback: llms.decodeFallback,
|
|
226
|
+
decodeWarning: llms.decodeWarning,
|
|
227
|
+
cacheHit: false,
|
|
228
|
+
};
|
|
229
|
+
cacheFetchResult(
|
|
230
|
+
fetchResult,
|
|
231
|
+
args.extract_main,
|
|
232
|
+
args.prefer_llms_txt,
|
|
233
|
+
args.save_binary,
|
|
234
|
+
);
|
|
235
|
+
} else if (llms?.error) {
|
|
236
|
+
llmsProbeError = llms.error;
|
|
237
|
+
}
|
|
238
|
+
|
|
239
|
+
if (!fetchResult && args.prefer_llms_txt === 'always') {
|
|
240
|
+
const metadata = args.include_metadata
|
|
241
|
+
? frontmatter({
|
|
242
|
+
requested_url: args.url,
|
|
243
|
+
used_llms_txt: false,
|
|
244
|
+
llms_probe_error: llmsProbeError,
|
|
245
|
+
prefer_llms_txt: args.prefer_llms_txt,
|
|
246
|
+
})
|
|
247
|
+
: '';
|
|
248
|
+
return joinRenderedContent(
|
|
249
|
+
metadata,
|
|
250
|
+
renderMessageForFormat(
|
|
251
|
+
buildLlmsRequiredMessage(args.url, llmsProbeError),
|
|
252
|
+
args.format,
|
|
253
|
+
),
|
|
254
|
+
args.format,
|
|
255
|
+
);
|
|
256
|
+
}
|
|
257
|
+
}
|
|
258
|
+
|
|
259
|
+
if (!fetchResult) {
|
|
260
|
+
const { result, upgradedToHttps } = await fetchWithUpgradeFallback(
|
|
261
|
+
normalized,
|
|
262
|
+
timeoutMs,
|
|
263
|
+
args.format,
|
|
264
|
+
controller.signal,
|
|
265
|
+
buildConditionalHeaders(staleFetchResult),
|
|
266
|
+
'GET',
|
|
267
|
+
allowedOrigins,
|
|
268
|
+
);
|
|
269
|
+
if ('blockedRedirect' in result) {
|
|
270
|
+
const metadata = args.include_metadata
|
|
271
|
+
? frontmatter({
|
|
272
|
+
requested_url: args.url,
|
|
273
|
+
redirect_url: result.redirectUrl,
|
|
274
|
+
status_code: result.statusCode,
|
|
275
|
+
redirect_chain: result.redirectChain.map(
|
|
276
|
+
(step: RedirectStep) =>
|
|
277
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
278
|
+
),
|
|
279
|
+
upgraded_to_https: upgradedToHttps,
|
|
280
|
+
})
|
|
281
|
+
: '';
|
|
282
|
+
return joinRenderedContent(
|
|
283
|
+
metadata,
|
|
284
|
+
renderMessageForFormat(
|
|
285
|
+
buildRedirectResultMessage(
|
|
286
|
+
args.url,
|
|
287
|
+
result.redirectUrl,
|
|
288
|
+
result.statusCode,
|
|
289
|
+
),
|
|
290
|
+
args.format,
|
|
291
|
+
),
|
|
292
|
+
args.format,
|
|
293
|
+
);
|
|
294
|
+
}
|
|
295
|
+
|
|
296
|
+
const { response, finalUrl, redirectChain } = result;
|
|
297
|
+
if (response.status === 304 && staleFetchResult) {
|
|
298
|
+
fetchResult = {
|
|
299
|
+
...staleFetchResult,
|
|
300
|
+
requestedUrl: args.url,
|
|
301
|
+
finalUrl,
|
|
302
|
+
statusCode: staleFetchResult.statusCode,
|
|
303
|
+
redirectChain,
|
|
304
|
+
llmsProbeError,
|
|
305
|
+
cacheRevalidated: true,
|
|
306
|
+
upstreamStatusCode: 304,
|
|
307
|
+
cacheHit: false,
|
|
308
|
+
};
|
|
309
|
+
cacheFetchResult(
|
|
310
|
+
fetchResult,
|
|
311
|
+
args.extract_main,
|
|
312
|
+
args.prefer_llms_txt,
|
|
313
|
+
args.save_binary,
|
|
314
|
+
);
|
|
315
|
+
} else {
|
|
316
|
+
if (!response.ok) {
|
|
317
|
+
throw new Error(
|
|
318
|
+
`Request failed with status code: ${response.status}`,
|
|
319
|
+
);
|
|
320
|
+
}
|
|
321
|
+
const headerMetadata = extractHeaderMetadata(
|
|
322
|
+
response.headers,
|
|
323
|
+
finalUrl,
|
|
324
|
+
);
|
|
325
|
+
const explicitBinary = isBinaryContentType(
|
|
326
|
+
headerMetadata.contentType || '',
|
|
327
|
+
);
|
|
328
|
+
const genericBinaryMime = isGenericBinaryMime(
|
|
329
|
+
headerMetadata.contentType || '',
|
|
330
|
+
);
|
|
331
|
+
const binaryDownloadLimit = args.save_binary
|
|
332
|
+
? MAX_RESPONSE_BYTES
|
|
333
|
+
: MAX_BINARY_DOWNLOAD_BYTES;
|
|
334
|
+
if (
|
|
335
|
+
explicitBinary &&
|
|
336
|
+
!genericBinaryMime &&
|
|
337
|
+
typeof headerMetadata.contentLength === 'number' &&
|
|
338
|
+
headerMetadata.contentLength > binaryDownloadLimit
|
|
339
|
+
) {
|
|
340
|
+
try {
|
|
341
|
+
await response.body?.cancel();
|
|
342
|
+
} catch {
|
|
343
|
+
// ignore cancel failures
|
|
344
|
+
}
|
|
345
|
+
fetchResult = {
|
|
346
|
+
requestedUrl: args.url,
|
|
347
|
+
finalUrl,
|
|
348
|
+
statusCode: response.status,
|
|
349
|
+
contentType:
|
|
350
|
+
headerMetadata.contentType || 'application/octet-stream',
|
|
351
|
+
charset: headerMetadata.charset,
|
|
352
|
+
etag: headerMetadata.etag,
|
|
353
|
+
lastModified: headerMetadata.lastModified,
|
|
354
|
+
contentLength: headerMetadata.contentLength,
|
|
355
|
+
filename: headerMetadata.filename,
|
|
356
|
+
canonicalUrl: finalUrl,
|
|
357
|
+
redirectChain,
|
|
358
|
+
upgradedToHttps,
|
|
359
|
+
truncated: false,
|
|
360
|
+
binary: true,
|
|
361
|
+
binaryKind: getBinaryKind(
|
|
362
|
+
headerMetadata.contentType || 'application/octet-stream',
|
|
363
|
+
),
|
|
364
|
+
downloadLimitBytes: binaryDownloadLimit,
|
|
365
|
+
metadataOnly: true,
|
|
366
|
+
data: undefined,
|
|
367
|
+
llmsProbeError,
|
|
368
|
+
llmsProbeTruncated: false,
|
|
369
|
+
cacheHit: false,
|
|
370
|
+
};
|
|
371
|
+
cacheFetchResult(
|
|
372
|
+
fetchResult,
|
|
373
|
+
args.extract_main,
|
|
374
|
+
args.prefer_llms_txt,
|
|
375
|
+
args.save_binary,
|
|
376
|
+
);
|
|
377
|
+
} else {
|
|
378
|
+
const readLimit =
|
|
379
|
+
explicitBinary && !genericBinaryMime
|
|
380
|
+
? binaryDownloadLimit
|
|
381
|
+
: MAX_RESPONSE_BYTES;
|
|
382
|
+
const body = await readBodyLimited(response, readLimit);
|
|
383
|
+
const provisionalDecoded =
|
|
384
|
+
!headerMetadata.contentType ||
|
|
385
|
+
genericBinaryMime ||
|
|
386
|
+
/^text\//i.test(headerMetadata.contentType)
|
|
387
|
+
? decodeBody(
|
|
388
|
+
body.data,
|
|
389
|
+
headerMetadata.charset,
|
|
390
|
+
headerMetadata.contentType,
|
|
391
|
+
)
|
|
392
|
+
: undefined;
|
|
393
|
+
const looksHtmlPayload = provisionalDecoded
|
|
394
|
+
? looksLikeHtmlText(provisionalDecoded.text)
|
|
395
|
+
: false;
|
|
396
|
+
const contentType = headerMetadata.contentType
|
|
397
|
+
? genericBinaryMime && looksLikeTextBody(body.data)
|
|
398
|
+
? looksHtmlPayload
|
|
399
|
+
? 'text/html'
|
|
400
|
+
: 'text/plain'
|
|
401
|
+
: /^text\/plain(?:;|$)/i.test(headerMetadata.contentType) &&
|
|
402
|
+
looksHtmlPayload
|
|
403
|
+
? 'text/html'
|
|
404
|
+
: headerMetadata.contentType
|
|
405
|
+
: looksLikeTextBody(body.data)
|
|
406
|
+
? looksHtmlPayload
|
|
407
|
+
? 'text/html'
|
|
408
|
+
: 'text/plain'
|
|
409
|
+
: 'application/octet-stream';
|
|
410
|
+
|
|
411
|
+
if (isBinaryContentType(contentType)) {
|
|
412
|
+
const binaryTooLarge =
|
|
413
|
+
body.truncated ||
|
|
414
|
+
(typeof headerMetadata.contentLength === 'number' &&
|
|
415
|
+
headerMetadata.contentLength > binaryDownloadLimit);
|
|
416
|
+
fetchResult = {
|
|
417
|
+
requestedUrl: args.url,
|
|
418
|
+
finalUrl,
|
|
419
|
+
statusCode: response.status,
|
|
420
|
+
contentType,
|
|
421
|
+
charset: headerMetadata.charset,
|
|
422
|
+
etag: headerMetadata.etag,
|
|
423
|
+
lastModified: headerMetadata.lastModified,
|
|
424
|
+
contentLength: headerMetadata.contentLength,
|
|
425
|
+
filename: headerMetadata.filename,
|
|
426
|
+
canonicalUrl: finalUrl,
|
|
427
|
+
redirectChain,
|
|
428
|
+
upgradedToHttps,
|
|
429
|
+
truncated: body.truncated,
|
|
430
|
+
binary: true,
|
|
431
|
+
binaryKind: getBinaryKind(contentType),
|
|
432
|
+
downloadLimitBytes: binaryDownloadLimit,
|
|
433
|
+
metadataOnly: binaryTooLarge,
|
|
434
|
+
data: binaryTooLarge ? undefined : body.data,
|
|
435
|
+
llmsProbeError,
|
|
436
|
+
llmsProbeTruncated: false,
|
|
437
|
+
cacheHit: false,
|
|
438
|
+
};
|
|
439
|
+
cacheFetchResult(
|
|
440
|
+
fetchResult,
|
|
441
|
+
args.extract_main,
|
|
442
|
+
args.prefer_llms_txt,
|
|
443
|
+
args.save_binary,
|
|
444
|
+
);
|
|
445
|
+
} else {
|
|
446
|
+
const decoded =
|
|
447
|
+
provisionalDecoded ||
|
|
448
|
+
decodeBody(body.data, headerMetadata.charset, contentType);
|
|
449
|
+
const rawText = decoded.text;
|
|
450
|
+
const extracted = isHtmlLikeContentType(contentType)
|
|
451
|
+
? await extractFromHtml(
|
|
452
|
+
rawText,
|
|
453
|
+
finalUrl,
|
|
454
|
+
args.extract_main,
|
|
455
|
+
)
|
|
456
|
+
: {
|
|
457
|
+
title: undefined,
|
|
458
|
+
rawContent: cleanFetchedText(rawText),
|
|
459
|
+
html: cleanFetchedText(rawText),
|
|
460
|
+
text: cleanFetchedText(rawText),
|
|
461
|
+
markdown: cleanFetchedText(rawText),
|
|
462
|
+
extractedMain: false,
|
|
463
|
+
canonicalUrl: undefined,
|
|
464
|
+
headings: [],
|
|
465
|
+
};
|
|
466
|
+
|
|
467
|
+
fetchResult = {
|
|
468
|
+
requestedUrl: args.url,
|
|
469
|
+
finalUrl,
|
|
470
|
+
statusCode: response.status,
|
|
471
|
+
contentType,
|
|
472
|
+
charset: headerMetadata.charset,
|
|
473
|
+
etag: headerMetadata.etag,
|
|
474
|
+
lastModified: headerMetadata.lastModified,
|
|
475
|
+
contentLength: headerMetadata.contentLength,
|
|
476
|
+
filename: headerMetadata.filename,
|
|
477
|
+
canonicalUrl:
|
|
478
|
+
extracted.canonicalUrl ||
|
|
479
|
+
inferCanonicalUrlFromText(extracted.markdown, finalUrl) ||
|
|
480
|
+
finalUrl,
|
|
481
|
+
headings: extracted.headings?.length
|
|
482
|
+
? extracted.headings
|
|
483
|
+
: extractHeadingsFromMarkdown(extracted.markdown),
|
|
484
|
+
title: extracted.title,
|
|
485
|
+
rawContent: extracted.rawContent,
|
|
486
|
+
markdown: extracted.markdown,
|
|
487
|
+
text: extracted.text,
|
|
488
|
+
html: extracted.html,
|
|
489
|
+
extractedMain: extracted.extractedMain,
|
|
490
|
+
usedLlmsTxt: false,
|
|
491
|
+
sourceKind: isHtmlLikeContentType(contentType)
|
|
492
|
+
? 'html'
|
|
493
|
+
: 'text',
|
|
494
|
+
upgradedToHttps,
|
|
495
|
+
redirectChain,
|
|
496
|
+
truncated: body.truncated,
|
|
497
|
+
wordCount: wordCount(extracted.text),
|
|
498
|
+
qualitySignals: detectQualitySignals({
|
|
499
|
+
text: extracted.text,
|
|
500
|
+
markdown: extracted.markdown,
|
|
501
|
+
rawContent: extracted.rawContent,
|
|
502
|
+
wordCount: wordCount(extracted.text),
|
|
503
|
+
sourceKind: isHtmlLikeContentType(contentType)
|
|
504
|
+
? 'html'
|
|
505
|
+
: 'text',
|
|
506
|
+
extractedMain: extracted.extractedMain,
|
|
507
|
+
}),
|
|
508
|
+
llmsProbeError,
|
|
509
|
+
llmsProbeTruncated: false,
|
|
510
|
+
decodedCharset: decoded.decodedCharset,
|
|
511
|
+
decodeFallback: decoded.decodeFallback,
|
|
512
|
+
decodeWarning: decoded.decodeWarning,
|
|
513
|
+
cacheHit: false,
|
|
514
|
+
};
|
|
515
|
+
cacheFetchResult(
|
|
516
|
+
fetchResult,
|
|
517
|
+
args.extract_main,
|
|
518
|
+
args.prefer_llms_txt,
|
|
519
|
+
args.save_binary,
|
|
520
|
+
);
|
|
521
|
+
}
|
|
522
|
+
}
|
|
523
|
+
}
|
|
524
|
+
}
|
|
525
|
+
}
|
|
526
|
+
|
|
527
|
+
ctx.metadata({
|
|
528
|
+
title:
|
|
529
|
+
('binary' in fetchResult
|
|
530
|
+
? fetchResult.filename
|
|
531
|
+
: fetchResult.title) || fetchResult.finalUrl,
|
|
532
|
+
metadata: {
|
|
533
|
+
url: fetchResult.finalUrl,
|
|
534
|
+
contentType: fetchResult.contentType,
|
|
535
|
+
truncated: fetchResult.truncated,
|
|
536
|
+
},
|
|
537
|
+
});
|
|
538
|
+
|
|
539
|
+
if ('binary' in fetchResult) {
|
|
540
|
+
if (fetchResult.metadataOnly || !fetchResult.data) {
|
|
541
|
+
const metadata = args.include_metadata
|
|
542
|
+
? frontmatter({
|
|
543
|
+
requested_url: fetchResult.requestedUrl,
|
|
544
|
+
final_url: fetchResult.finalUrl,
|
|
545
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
546
|
+
status_code: fetchResult.statusCode,
|
|
547
|
+
source_content_type: fetchResult.contentType,
|
|
548
|
+
charset: fetchResult.charset,
|
|
549
|
+
etag: fetchResult.etag,
|
|
550
|
+
last_modified: fetchResult.lastModified,
|
|
551
|
+
content_length: fetchResult.contentLength,
|
|
552
|
+
filename: fetchResult.filename,
|
|
553
|
+
binary_kind: fetchResult.binaryKind,
|
|
554
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
555
|
+
(step: RedirectStep) =>
|
|
556
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
557
|
+
),
|
|
558
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
559
|
+
llms_probe_error: fetchResult.llmsProbeError,
|
|
560
|
+
cache_revalidated: fetchResult.cacheRevalidated,
|
|
561
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
562
|
+
upstream_status_code: fetchResult.upstreamStatusCode,
|
|
563
|
+
truncated: fetchResult.truncated,
|
|
564
|
+
download_limit_bytes:
|
|
565
|
+
fetchResult.downloadLimitBytes ?? MAX_BINARY_DOWNLOAD_BYTES,
|
|
566
|
+
binary_metadata_only: true,
|
|
567
|
+
})
|
|
568
|
+
: '';
|
|
569
|
+
return joinRenderedContent(
|
|
570
|
+
metadata,
|
|
571
|
+
renderMessageForFormat(
|
|
572
|
+
buildBinaryResultMessage(fetchResult),
|
|
573
|
+
args.format,
|
|
574
|
+
),
|
|
575
|
+
args.format,
|
|
576
|
+
);
|
|
577
|
+
}
|
|
578
|
+
if (!args.save_binary) {
|
|
579
|
+
const metadata = args.include_metadata
|
|
580
|
+
? frontmatter({
|
|
581
|
+
requested_url: fetchResult.requestedUrl,
|
|
582
|
+
final_url: fetchResult.finalUrl,
|
|
583
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
584
|
+
status_code: fetchResult.statusCode,
|
|
585
|
+
source_content_type: fetchResult.contentType,
|
|
586
|
+
charset: fetchResult.charset,
|
|
587
|
+
etag: fetchResult.etag,
|
|
588
|
+
last_modified: fetchResult.lastModified,
|
|
589
|
+
content_length: fetchResult.contentLength,
|
|
590
|
+
filename: fetchResult.filename,
|
|
591
|
+
binary_kind: fetchResult.binaryKind,
|
|
592
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
593
|
+
(step: RedirectStep) =>
|
|
594
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
595
|
+
),
|
|
596
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
597
|
+
truncated: fetchResult.truncated,
|
|
598
|
+
save_binary: false,
|
|
599
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
600
|
+
})
|
|
601
|
+
: '';
|
|
602
|
+
return joinRenderedContent(
|
|
603
|
+
metadata,
|
|
604
|
+
renderMessageForFormat(
|
|
605
|
+
`${fetchResult.binaryKind.toUpperCase()} content fetched but not saved. Re-run with save_binary=true to persist it.`,
|
|
606
|
+
args.format,
|
|
607
|
+
),
|
|
608
|
+
args.format,
|
|
609
|
+
);
|
|
610
|
+
}
|
|
611
|
+
const savedPath = await saveBinary(
|
|
612
|
+
binaryDir,
|
|
613
|
+
fetchResult.data,
|
|
614
|
+
fetchResult.contentType,
|
|
615
|
+
fetchResult.filename,
|
|
616
|
+
);
|
|
617
|
+
const metadata = args.include_metadata
|
|
618
|
+
? frontmatter({
|
|
619
|
+
requested_url: fetchResult.requestedUrl,
|
|
620
|
+
final_url: fetchResult.finalUrl,
|
|
621
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
622
|
+
status_code: fetchResult.statusCode,
|
|
623
|
+
source_content_type: fetchResult.contentType,
|
|
624
|
+
charset: fetchResult.charset,
|
|
625
|
+
etag: fetchResult.etag,
|
|
626
|
+
last_modified: fetchResult.lastModified,
|
|
627
|
+
content_length: fetchResult.contentLength,
|
|
628
|
+
filename: fetchResult.filename,
|
|
629
|
+
binary_kind: fetchResult.binaryKind,
|
|
630
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
631
|
+
(step: RedirectStep) =>
|
|
632
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
633
|
+
),
|
|
634
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
635
|
+
llms_probe_error: fetchResult.llmsProbeError,
|
|
636
|
+
cache_revalidated: fetchResult.cacheRevalidated,
|
|
637
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
638
|
+
upstream_status_code: fetchResult.upstreamStatusCode,
|
|
639
|
+
truncated: fetchResult.truncated,
|
|
640
|
+
download_limit_bytes:
|
|
641
|
+
fetchResult.downloadLimitBytes ?? MAX_BINARY_DOWNLOAD_BYTES,
|
|
642
|
+
saved_path: savedPath,
|
|
643
|
+
})
|
|
644
|
+
: '';
|
|
645
|
+
return joinRenderedContent(
|
|
646
|
+
metadata,
|
|
647
|
+
renderMessageForFormat(
|
|
648
|
+
buildBinaryResultMessage(fetchResult, savedPath),
|
|
649
|
+
args.format,
|
|
650
|
+
),
|
|
651
|
+
args.format,
|
|
652
|
+
);
|
|
653
|
+
}
|
|
654
|
+
|
|
655
|
+
const baseContent = pickContent(fetchResult, args.format);
|
|
656
|
+
const secondaryModelDecision = decideSecondaryModelUse(
|
|
657
|
+
fetchResult,
|
|
658
|
+
args.prompt,
|
|
659
|
+
secondaryModels,
|
|
660
|
+
);
|
|
661
|
+
const metadata = args.include_metadata
|
|
662
|
+
? frontmatter({
|
|
663
|
+
requested_url: fetchResult.requestedUrl,
|
|
664
|
+
final_url: fetchResult.finalUrl,
|
|
665
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
666
|
+
status_code: fetchResult.statusCode,
|
|
667
|
+
source_content_type: fetchResult.contentType,
|
|
668
|
+
charset: fetchResult.charset,
|
|
669
|
+
etag: fetchResult.etag,
|
|
670
|
+
last_modified: fetchResult.lastModified,
|
|
671
|
+
content_length: fetchResult.contentLength,
|
|
672
|
+
filename: fetchResult.filename,
|
|
673
|
+
headings: fetchResult.headings,
|
|
674
|
+
title: fetchResult.title,
|
|
675
|
+
source_kind: fetchResult.sourceKind,
|
|
676
|
+
used_llms_txt: fetchResult.usedLlmsTxt,
|
|
677
|
+
extracted_main: fetchResult.extractedMain,
|
|
678
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
679
|
+
(step: RedirectStep) =>
|
|
680
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
681
|
+
),
|
|
682
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
683
|
+
llms_probe_error: fetchResult.llmsProbeError,
|
|
684
|
+
llms_probe_truncated: fetchResult.llmsProbeTruncated,
|
|
685
|
+
cache_revalidated: fetchResult.cacheRevalidated,
|
|
686
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
687
|
+
upstream_status_code: fetchResult.upstreamStatusCode,
|
|
688
|
+
truncated: fetchResult.truncated,
|
|
689
|
+
word_count: fetchResult.wordCount,
|
|
690
|
+
quality_signals: fetchResult.qualitySignals,
|
|
691
|
+
decoded_charset: fetchResult.decodedCharset,
|
|
692
|
+
decode_fallback: fetchResult.decodeFallback,
|
|
693
|
+
decode_warning: fetchResult.decodeWarning,
|
|
694
|
+
secondary_model: undefined,
|
|
695
|
+
secondary_model_skipped_reason:
|
|
696
|
+
!secondaryModelDecision.use && args.prompt
|
|
697
|
+
? secondaryModelDecision.reason
|
|
698
|
+
: undefined,
|
|
699
|
+
})
|
|
700
|
+
: '';
|
|
701
|
+
|
|
702
|
+
if (!secondaryModelDecision.use) {
|
|
703
|
+
return joinRenderedContent(metadata, baseContent, args.format);
|
|
704
|
+
}
|
|
705
|
+
|
|
706
|
+
if (!secondaryModels.length) {
|
|
707
|
+
return joinRenderedContent(metadata, baseContent, args.format);
|
|
708
|
+
}
|
|
709
|
+
let secondaryRun:
|
|
710
|
+
| Awaited<ReturnType<typeof runSecondaryModelWithFallback>>
|
|
711
|
+
| undefined;
|
|
712
|
+
let secondaryModelError: string | undefined;
|
|
713
|
+
try {
|
|
714
|
+
secondaryRun = await runSecondaryModelWithFallback(
|
|
715
|
+
pluginCtx.client,
|
|
716
|
+
ctx.directory || process.cwd(),
|
|
717
|
+
secondaryModels,
|
|
718
|
+
args.prompt || '',
|
|
719
|
+
fetchResult.markdown,
|
|
720
|
+
);
|
|
721
|
+
} catch (error: unknown) {
|
|
722
|
+
secondaryModelError =
|
|
723
|
+
error instanceof Error ? error.message : String(error);
|
|
724
|
+
}
|
|
725
|
+
|
|
726
|
+
if (!secondaryRun) {
|
|
727
|
+
const degradedMetadata = args.include_metadata
|
|
728
|
+
? frontmatter({
|
|
729
|
+
requested_url: fetchResult.requestedUrl,
|
|
730
|
+
final_url: fetchResult.finalUrl,
|
|
731
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
732
|
+
status_code: fetchResult.statusCode,
|
|
733
|
+
source_content_type: fetchResult.contentType,
|
|
734
|
+
charset: fetchResult.charset,
|
|
735
|
+
etag: fetchResult.etag,
|
|
736
|
+
last_modified: fetchResult.lastModified,
|
|
737
|
+
content_length: fetchResult.contentLength,
|
|
738
|
+
filename: fetchResult.filename,
|
|
739
|
+
headings: fetchResult.headings,
|
|
740
|
+
title: fetchResult.title,
|
|
741
|
+
source_kind: fetchResult.sourceKind,
|
|
742
|
+
used_llms_txt: fetchResult.usedLlmsTxt,
|
|
743
|
+
extracted_main: fetchResult.extractedMain,
|
|
744
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
745
|
+
(step: RedirectStep) =>
|
|
746
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
747
|
+
),
|
|
748
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
749
|
+
llms_probe_error: fetchResult.llmsProbeError,
|
|
750
|
+
llms_probe_truncated: fetchResult.llmsProbeTruncated,
|
|
751
|
+
cache_revalidated: fetchResult.cacheRevalidated,
|
|
752
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
753
|
+
upstream_status_code: fetchResult.upstreamStatusCode,
|
|
754
|
+
truncated: fetchResult.truncated,
|
|
755
|
+
word_count: fetchResult.wordCount,
|
|
756
|
+
quality_signals: fetchResult.qualitySignals,
|
|
757
|
+
decoded_charset: fetchResult.decodedCharset,
|
|
758
|
+
decode_fallback: fetchResult.decodeFallback,
|
|
759
|
+
decode_warning: fetchResult.decodeWarning,
|
|
760
|
+
secondary_model: undefined,
|
|
761
|
+
secondary_model_skipped_reason: 'secondary_model_failed',
|
|
762
|
+
secondary_model_error: secondaryModelError,
|
|
763
|
+
})
|
|
764
|
+
: '';
|
|
765
|
+
return joinRenderedContent(
|
|
766
|
+
degradedMetadata,
|
|
767
|
+
baseContent,
|
|
768
|
+
args.format,
|
|
769
|
+
);
|
|
770
|
+
}
|
|
771
|
+
|
|
772
|
+
const metadataWithSecondary = args.include_metadata
|
|
773
|
+
? frontmatter({
|
|
774
|
+
requested_url: fetchResult.requestedUrl,
|
|
775
|
+
final_url: fetchResult.finalUrl,
|
|
776
|
+
canonical_url: fetchResult.canonicalUrl,
|
|
777
|
+
status_code: fetchResult.statusCode,
|
|
778
|
+
source_content_type: fetchResult.contentType,
|
|
779
|
+
charset: fetchResult.charset,
|
|
780
|
+
etag: fetchResult.etag,
|
|
781
|
+
last_modified: fetchResult.lastModified,
|
|
782
|
+
content_length: fetchResult.contentLength,
|
|
783
|
+
filename: fetchResult.filename,
|
|
784
|
+
headings: fetchResult.headings,
|
|
785
|
+
title: fetchResult.title,
|
|
786
|
+
source_kind: fetchResult.sourceKind,
|
|
787
|
+
used_llms_txt: fetchResult.usedLlmsTxt,
|
|
788
|
+
extracted_main: fetchResult.extractedMain,
|
|
789
|
+
redirect_chain: fetchResult.redirectChain.map(
|
|
790
|
+
(step: RedirectStep) =>
|
|
791
|
+
`${step.status} ${step.from} -> ${step.to}`,
|
|
792
|
+
),
|
|
793
|
+
upgraded_to_https: fetchResult.upgradedToHttps,
|
|
794
|
+
llms_probe_error: fetchResult.llmsProbeError,
|
|
795
|
+
llms_probe_truncated: fetchResult.llmsProbeTruncated,
|
|
796
|
+
cache_revalidated: fetchResult.cacheRevalidated,
|
|
797
|
+
cache_hit: fetchResult.cacheHit ?? cacheHit,
|
|
798
|
+
upstream_status_code: fetchResult.upstreamStatusCode,
|
|
799
|
+
truncated: fetchResult.truncated,
|
|
800
|
+
word_count: fetchResult.wordCount,
|
|
801
|
+
quality_signals: fetchResult.qualitySignals,
|
|
802
|
+
decoded_charset: fetchResult.decodedCharset,
|
|
803
|
+
decode_fallback: fetchResult.decodeFallback,
|
|
804
|
+
decode_warning: fetchResult.decodeWarning,
|
|
805
|
+
secondary_model_input_truncated: secondaryRun.inputTruncated,
|
|
806
|
+
secondary_model_input_chars: secondaryRun.inputChars,
|
|
807
|
+
secondary_model_source_chars: secondaryRun.sourceChars,
|
|
808
|
+
secondary_model: `${secondaryRun.model.providerID}/${secondaryRun.model.modelID}`,
|
|
809
|
+
})
|
|
810
|
+
: '';
|
|
811
|
+
const secondaryRaw =
|
|
812
|
+
secondaryRun.text || 'No response from secondary model.';
|
|
813
|
+
const secondaryContent =
|
|
814
|
+
args.format === 'html'
|
|
815
|
+
? withTruncationMarker(
|
|
816
|
+
`<pre>${escapeHtml(secondaryRaw)}</pre>`,
|
|
817
|
+
'html',
|
|
818
|
+
false,
|
|
819
|
+
)
|
|
820
|
+
: withTruncationMarker(secondaryRaw, args.format, false);
|
|
821
|
+
return joinRenderedContent(
|
|
822
|
+
metadataWithSecondary,
|
|
823
|
+
secondaryContent,
|
|
824
|
+
args.format,
|
|
825
|
+
);
|
|
826
|
+
} finally {
|
|
827
|
+
clearTimeout(timeout);
|
|
828
|
+
ctx.abort.removeEventListener('abort', abortHandler);
|
|
829
|
+
}
|
|
830
|
+
},
|
|
831
|
+
});
|
|
832
|
+
}
|